Title: [solved] Receiving strong sigs 12.KHZ
Post by: WD8BXS on 19. July 2018, 14:59:18
This is the first time on a V6 radio that I have heard this.
I am hearing strong signals that are -12khz from the tuned frequency.
It does not matter where I place the xlate 0, -6, or -12, the same thing still happens.
Any ideas??
Chuck |
Title: Re:Receiving strong sigs 12.KHZ
Post by: DF8OE on 19. July 2018, 15:51:37
12KHz regardless of Xlate setting? That sounds to be a real signal at that frequency...
If"mirror" does not change by Xlate setting, it is "real signal" - I think.
vy 73
Andreas |
Title: Re:Receiving strong sigs 12.KHZ
Post by: DB4PLE on 19. July 2018, 17:10:12
Hi Chuck,
1. You tune to frequency X.
2. You (or someone else) generate(s) a signal at X-12khz and you can hear it when tuned to X no matter what XLATE is set to.
3. Can you also hear the signal when tuned to X-12khz?
73
Danilo
|
Title: Re:Receiving strong sigs 12.KHZ
Post by: WD8BXS on 19. July 2018, 21:32:38
Yes, I was listening on 7.180, and I heard a station that I know works 7168.
I tuned to 7168, and there he was.
I tuned back to 7180 and there he still was over top of the station on 7180
73
CHuck |
Title: Re:Receiving strong sigs 12.KHZ
Post by: DB4PLE on 19. July 2018, 21:45:21
Hi Chuck,
- Any chance that the station at 7168 was transmitting a second strong signal 12khz from its main frequency?
- Did you had the opportunity to verify with a second receiver that only the 0.6 mchf heard the signal at 7180? If not, you should really try to see if this also happens if use a signal generator or second transmitter.
73
Danilo
|
Title: Re:Receiving strong sigs 12.KHZ
Post by: WD8BXS on 20. July 2018, 11:55:09
I will try to duplicate this again later today.
I will keep you posted.
Thanks,
73,
Chuck |
Title: Re:Receiving strong sigs 12.KHZ
Post by: WD8BXS on 20. July 2018, 15:00:50
I can tune any station on my Icom 7610, and hear them 12khz higher AND 12khx lower on my MCHF !!!
This is regardless of where the translate is set.
Chuck |
Title: Re:Receiving strong sigs 12.KHZ
Post by: WD8BXS on 20. July 2018, 16:03:19
Here is a video of this issue:
https://www.dropbox.com/s/bjcdefj60satuiw/REC_0003.MOV?dl=0 |
Title: Re:Receiving strong sigs 12.KHZ
Post by: DF8OE on 20. July 2018, 16:06:53
I do not have any idea what this can be caused by. It is unlogical, because there is no math that allows that.
Sorry - no hint from my side. If you would have written "...at 6KHz Xlate..." there *is* a math. But at 12KHz Xlate there is no math - the spacing must be 24KHz.
First you should swap RF or UI PCB to another mcHF which is running fine so that you are able to decide on which board it is. And the I would be very pleased if you can give us a feedback what was the cause when you fixed it.
vy 73
Andreas |
Title: Re:Receiving strong sigs 12.KHZ
Post by: WD8BXS on 20. July 2018, 18:05:54
Swapped out a "U" Board, NO Change.
I will have to build an RF Board.
73,
Chuck |
Title: Re:Receiving strong sigs 12.KHZ
Post by: DJ3FV on 21. July 2018, 06:15:00
Hi Chuck,
is your radio a chinese clone (RS-918) based on the V0.6 mcHF? I am asking because I read that in this radios parts have been replaced by cheeper chinese parts. This could be a root cause for such effects.
73, Uwe |
Title: Re:Receiving strong sigs 12.KHZ
Post by: WD8BXS on 21. July 2018, 13:32:23
Hi,
This is NOT a clone! I built it!
73,
Chuck |
Title: Re:Receiving strong sigs 12.KHZ
Post by: WD8BXS on 21. July 2018, 17:26:06
OK, The verdict is in.
I just finished another RF board and it does NOT have the spurious receive!!!
Now what area do I start looking??
Or do I just randomly poke around??
CHuck |
Title: Re:Receiving strong sigs 12.KHZ
Post by: WD8BXS on 21. July 2018, 20:19:17
I was wrong, the new one does it also!!!!
Chuck |
Title: Re:Receiving strong sigs 12.KHZ
Post by: WD8BXS on 21. July 2018, 20:22:50
There is no s meter reading for the extra signal.
Just on the main signal.
Chuck |
Title: Re:Receiving strong sigs 12.KHZ
Post by: DB4PLE on 21. July 2018, 20:53:58
Hi Chuck,
essentially, what you are saying/seeing is: two UI boards in combination with two different RF boards show the same problem.
1. The problem could be in the schematics (unlikely), but in this case others should be able reproduce this.
2. It could be in the software in general, (unlikely), but in this case others should be able reproduce this.
3. It could be in the software in combination with some settings, in this case others should be able reproduce this if they have the same settings.
So for now number 3 is more likely than the other scenarios (apart from some broken / wrong components on the boards) and more easily reproducable, please download and send the settings here or in a Github issue, please use the backup tool, see here https://github.com/df8oe/UHSDR/wiki/Backup-and-Restore#backup--restore-on-external-computer-since-2919 (https://github.com/df8oe/UHSDR/wiki/Backup-and-Restore#backup--restore-on-external-computer-since-2919).
|
Title: Re:Receiving strong sigs 12.KHZ
Post by: WD8BXS on 21. July 2018, 23:23:16
I just found out that this problem started with 2.9.16 firmware.
I loaded 2.9.15 and all is well.
THERE IS A FIRMWARE ISSUE!
73,
Chuck |
Title: Re:Receiving strong sigs 12.KHZ
Post by: DF8OE on 22. July 2018, 05:43:46
| THERE IS A FIRMWARE ISSUE! |
|
If this would be right it would be mandatory that every mcHF would have this problem - but it is not so!
Maybe you do have a scrambled configuration.
It would be great if you do make a backup of this configuration via USB and PC before you proceed. This would help us to identify such scrambled settings during power on cycle.
Please try newest firmware and run mcHF with a clean (== default) configuration. Feedback is welcome.
vy 73
Andreas |
Title: Re:Receiving strong sigs 12.KHZ
Post by: WD8BXS on 22. July 2018, 19:06:54
OK, I reset the EEprom, I set to defaults, I loaded 2.9.43, and all seems to be working OK now!!!
Both radios are OK right now.
I will do calibration on both and see what happens.
Thanks, so Far!
73,
Chuck |
Title: Re:[solved] Receiving strong sigs 12.KHZ
Post by: WD8BXS on 22. July 2018, 21:52:53
All seems OK now that I have cleared the EEprom and reset to defaults, loaded new firmware 2.9.43, and calibrated.
Very confusing, but there must have been something that corrupted the info. Maybe I should do the EEprom reset before doing anything the next time!!
Thanks for putting up with me.
Very 73,
Chuck WD8BXS :D :D :D :D :D |
Title: Re:[solved] Receiving strong sigs 12.KHZ
Post by: DF8OE on 23. July 2018, 04:43:09
Please can you attach the old (scrambled) setting zipped to your next reply? We have programmed possibility to store configurations exactly because we need informations to hunt such "strange things". Without such a scrambeld configuration there is no chance to step on.
There is a routine in firmware which tests each seting at starting up. But it seems there are some possibilities for scrambled values to pass this routine. I would not call this "a bug" - but you see there is something that can be improved. But only with your help and contribution...
If we do not get this contribution you possibly must again and again reset your configuration and setup every setting new ::)
vy 73
Andreas |
Title: Re:[solved] Receiving strong sigs 12.KHZ
Post by: F4HTX on 24. July 2018, 05:46:04
Hello Friends,
This is excactly what I encountered recently. I first thought it was a xslate problem, but it was exactly as Chuck described (see the Rx IQ auto correction problem topic in UHSDR Firmware). Downgrading the firmware also did the trick, but not the same versions.
The problem solved after a full reset and I unfortunately lost my settings in the process, but I have two more TRX with the problem (one mchf, one OVI/mchf RF).
Before performing reset, I will do some measurements with RF generator and this time salvage the settings file.
As soon as the QRM pro slow down a little...
Best Regards,
François |
Title: Re:[solved] Receiving strong sigs 12.KHZ
Post by: DF8OE on 24. July 2018, 07:34:12
It would be great if we could save a setting where the issue exists. I never have had seen it here.
vy 73
Andreas |
Title: Re:[solved] Receiving strong sigs 12.KHZ
Post by: F4HTX on 28. July 2018, 15:27:19
Hi Andreas,
This is the config file from a v0.6 mchf (cpu 413h, BL4.0.0, FW 2.9.46) which exhibits the problem discussed here.
I will post another config file from an OVI40 soon. Sorry, I forgot, I had reseted this unit last week...
Hope this will help to pinpoint this unexpected behaviour.
73's
François |
Title: Re:[solved] Receiving strong sigs 12.KHZ
Post by: DF8OE on 28. July 2018, 15:45:43
Many thanks for your contribution! The problem is not easy - we have already seen it is not one simple setting but a combination of settings (which all will be in their range - seen from only their range). Maybe we can work it out with configs which do show the issue.
vy 73
Andreas |
Title: Re:[solved] Receiving strong sigs 12.KHZ
Post by: F4HTX on 28. July 2018, 16:03:08
Hi Andreas,
Yes, this is a tricky problem. I will freeze this mchf until we fix it. So don't hesitate to ask me for any testing.
If necessary, I might sent it to Germany too, not complicated.
Best Regards,
François |
Title: Re:[solved] Receiving strong sigs 12.KHZ
Post by: DF8OE on 28. July 2018, 16:08:26
Hi François,
with the config json file we are able to reproduce the problem on every mcHF. So we do not need hardware - but luck and time to investigate ;D. Thanks for your offer - not neccessary in this case.
vy 73
Andreas |
Title: Re:[solved] Receiving strong sigs 12.KHZ
Post by: F4HTX on 02. August 2018, 15:51:42
Hi Andreas,
As promised, please find attached a second Config File from another mchf with the same signal ghosts. I hope this will help to pinpoint the troublemaker.
73's
François |
Title: Re:[solved] Receiving strong sigs 12.KHZ
Post by: DF8OE on 02. August 2018, 16:30:46
Hi François,
many thanks. I have stored it and will have a look by time. Actually I am working hard on RF schematics and PCBs so I will not work on this next day / weeks, but it is important to have as many as possible configs with the bad settings...
vy 73
Andreas |
Title: Re:[solved] Receiving strong sigs 12.KHZ
Post by: WD8BXS on 28. August 2018, 21:21:24
Just did a radio, that had the 12khz problem, reset and problem went away.
Went through the calibrate and now the issue is back.
Here is the config file for you to look at. |
Title: Re:[solved] Receiving strong sigs 12.KHZ
Post by: DF8OE on 29. August 2018, 14:15:28
Can you tell us at which configuration elemnts you were when it popped up?
EDIT:
I can confirm that this settings are "infected". But I cannot reproduce or investigate the setting(s) which are "the bad ones". The differences to a clean default setting are already too large.
vy 73
Andreas |
Title: Re:[solved] Receiving strong sigs 12.KHZ
Post by: WD8BXS on 29. August 2018, 15:25:17
Well, no.
I don't check it after every entry with the config.
I guess I could try some of that.
Chuck |
Title: Re:[solved] Receiving strong sigs 12.KHZ
Post by: DF8OE on 29. August 2018, 15:28:14
This is a *very ugly bug* and the only way I see to identify it quickly is by a "manual" how exactly it can be reproduced.
It seems that it is related to a very special combination of settings which I never use that way - and many others not, too.
vy 73
Andreas |
Title: Re:[solved] Receiving strong sigs 12.KHZ
Post by: WD8BXS on 29. August 2018, 16:40:23
Just reset this one, and the problem is still here.
I can hear the station 12 khz up, although not real load.
Here is the config file.
Chuck
Can you send me a clean reset file to load?? |
Title: Re:[solved] Receiving strong sigs 12.KHZ
Post by: WD8BXS on 29. August 2018, 17:40:47
Just did another reset, and I got it working.
I went through the config, and all was well, UNTIL!!
I saved config to Flash, then restored from flash, THAT is when the 12khz returned!!
de Chuck |
Title: Re:[solved] Receiving strong sigs 12.KHZ
Post by: DJ0RG on 29. August 2018, 17:50:31
| Can you save and compare the compare the config before and after flash memory, to ensure it is unchanged and the flash works correctly? |
Title: Re:[solved] Receiving strong sigs 12.KHZ
Post by: WD8BXS on 29. August 2018, 18:17:11
Title: Re:[solved] Receiving strong sigs 12.KHZ
Post by: WD8BXS on 29. August 2018, 18:47:46
| Now I can't get it to go away, reset several times, no joy!! |
Title: Re:[solved] Receiving strong sigs 12.KHZ
Post by: WD8BXS on 29. August 2018, 18:51:25
Finally!! Now I will calibrate and send the file that is working.
Chuck |
Title: Re:[solved] Receiving strong sigs 12.KHZ
Post by: WD8BXS on 29. August 2018, 19:08:57
OK, Here is a good file with the radio working!
Chuck |
Title: Re:[solved] Receiving strong sigs 12.KHZ
Post by: DJ0RG on 29. August 2018, 19:37:40
OK, I guess I deserved this. My wish was that you compare the two config files before saving to flash memory and after loading from flash memory, but I might have been to vague ;-)
Anyway, these two files have 32 differences. Now I'm no way included in any development here, I just threw in a thought I had. I don't know if 32 differences are normal for this process.
In order to avoid misunderstandings:
1) You mentioned that you had a working config, after resetting to default and doing some configuration.
2) When you then saved to flash and loaded from there again, the component was corrupted.
3) The two files you attached in previous posts are from before saving to flash, and after loading from flash.
Is this right so far?
@Developers: I can give a comprehensive list of the changes between the two files, if that helps you.
One quick example:
Good:
{
"addr": 170,
"value": 384
},
Bad:
{
"addr": 170,
"value": 272
},
The value in address 170 changed
from 384 (1 1000 0000, 0x180)
to 272 (1 0001 0000, 0x110)
Two bits flipped.
I am a programmer, but have not seen any source code yet. Don't even know if that is publically available or not. I therefore cannot say anything about this change, if it is normal or not, and if it affects any of the RF operations. However, if you assign tasks to me, I can try to help out.
|
Title: Re:[solved] Receiving strong sigs 12.KHZ
Post by: WD8BXS on 29. August 2018, 19:52:19
OK, Let's do it this way.
I have the radio working fine right now. I will save to flash and restore, and see if the issue starts, then I will send that file.
Chuck |
Title: Re:[solved] Receiving strong sigs 12.KHZ
Post by: WD8BXS on 29. August 2018, 19:57:56
THIS is bizaar!! I just tested it again before I was going to do the save / restore, and now the issue is there!!!
Here is the current file: |
Title: Re:[solved] Receiving strong sigs 12.KHZ
Post by: DB4PLE on 29. August 2018, 19:58:42
Hi,
I also did a diff, and to limit our guess work, here are the changed config parameters
If you see a "-" in front I think the setting should not be related to the issue
If you see a "?" in front it is probably worth looking at the differences
If you see nothing before EEPROM I am not sure (this is only for the filter map)
Now some one should take one by one parameter value of the ? values from the good one, write them by hand into the "bad" file, and load it into a F4 mcHF until it no longer exhibits the problems.
If you found a value which provokes the issue, make note of it.
Now start again with the bad configuration, remove the problematic value first. If we now have the issue gone, good.
If not, add the ? values from the beginning, until you identify the issue being gone. Now we have found at least two parameters required for issue.
With these two set into the bad configuration, check if this causes the issue to go away. If not, repeat procedure, until have identified all the parameters required to make the issue go away.
Cross check by adding the bad values of the identified parameters to the good configuration. This one now should also show the issue.
Let us know the parameter id numbers...
73
Danilo
-EEPROM_BAND_MODE 1
-EEPROM_FREQ_HIGH 2
-EEPROM_FREQ_LOW 3
-EEPROM_FREQ_STEP 4
-EEPROM_AUDIO_GAIN 8
-EEPROM_TX_POWER_LEVEL 12
?EEPROM_TX_IQ_80M_GAIN_BALANCE 17 // TX gain balance
?EEPROM_TX_IQ_80M_PHASE_BALANCE 18 // TX phase balance
-EEPROM_FREQ_CAL 75 // Frequency calibration
-EEPROM_SPECTRUM_TRACE_COLOUR 85 // Custom setting for spectrum scope trace colour
?EEPROM_TX_IQ_10M_GAIN_BALANCE 139 // TX gain balance
?EEPROM_TX_IQ_10M_PHASE_BALANCE 140 // TX phase balance
?EEPROM_TX_IQ_80M_GAIN_BALANCE_TRANS_OFF 168 // IQ Gain balance for AM reception
?EEPROM_FLAGS1 170 // Miscellaneous status flag, saved in EEPROM -
-EEPROM_BAND0_MODE_A 183 // Band/mode/filter memory per-band - bands indexed
-EEPROM_BAND2_MODE_A 185
-EEPROM_BAND2_FREQ_HIGH_A 203
-EEPROM_BAND2_FREQ_LOW_A 221
-EEPROM_BAND2_FREQ_LOW_B 275
?EEPROM_TX_IQ_10M_GAIN_BALANCE_TRANS_OFF 300 // IQ Gain balance for AM transmission
?EEPROM_TX_IQ_10_PHASE_BALANCE_TRANS_OFF 301 // IQ Gain balance for FM transmission
-EEPROM_BEEP_LOUDNESS 309
329, 334, 333, 338, 358 is in EEPROM_FILTER_PATH_MAP_BASE 329
-EEPROM_TX_FILTER 363 // TX_Filter configuration
?EEPROM_AGC_WDSP_THRESH 376
-EEPROM_Scope_TRACE_HL_BW 390
-EEPROM_Scope_TRACE_HL_BW_BGR 391 |
Title: Re:[solved] Receiving strong sigs 12.KHZ
Post by: WD8BXS on 29. August 2018, 20:09:00
I just reloaded the WD8BXSgood file with uhsdr_tool, and it is working again.
Looking at the ? values, I did not change several of these!
Chuck |
Title: Re:[solved] Receiving strong sigs 12.KHZ
Post by: DJ0RG on 29. August 2018, 20:18:20
Trying a side path meanwhile: Is it normal / by design that values like EEPROM_TX_IQ_80M_PHASE_BALANCE change from saving to flash to loading from flash? I wonder if this is a memory problem, either in memory or on flash. Currently config memory would be my guess (not knowing the system at all ;) )
Edit: The unit might be confused by a certain configuration schema, but we also might have two problems here: corrupt memory and a unit being confused by the result. |
Title: Re:[solved] Receiving strong sigs 12.KHZ
Post by: DB4PLE on 29. August 2018, 20:39:22
Hi,
I did not think straight, the TX lG abelled phase / gain values should not play a role in receiving. So this leaves us
with just
?EEPROM_FLAGS1 170 // Miscellaneous status flag, saved in EEPROM -
?EEPROM_AGC_WDSP_THRESH 376
and possibly the filters:
329, 334, 333, 338, 358 is in EEPROM_FILTER_PATH_MAP_BASE 329
I compared the FLAGS1 settings (FLAGS1 is 16 independent yes/no settings) and the results are:
The only differences are:
#define FLAGS1_SWAP_FWDREV_SENSE 0x10 // if FWD/REV A/D inputs from RF power detectors are to be reversed
#define FLAGS1_WFALL_ENABLED 0x80 // 1 = Waterfall display
And I don't think any of this is causing the issue.
Leaves us with the AGC_THRES or the filters...
73
Danilo
|
Title: Re:[not solved] Receiving strong sigs 12.KHZ
Post by: DF8OE on 30. August 2018, 03:57:26
I am absolutely interested in the result. I already have spent hours in changing settings manually and then do a try and error. At the beginnig of my tests I was faked more than one time because of AGC. As I swapped to very old firmware I thought issue was gone even with infected settings but that is not true! So my conclusion is all WDSP AGC settings are not guilty. I tested with firmware which was so old that we do not have archive binary. I must switch off old AGC (because it eats the spurs by reducing sensitivity).
I stepped on by following scheme:
I took the first half of a working config and the second half of an infected config. Merged together, I tested if the issue is gone or not. If it is gone, bad values must reside in the first half of settings, if not, they must be in the second half. Now I divided the half where the issue must reside in again in two halfs and replaced one with working values. So I found that issue is in the second half of all settings. But this is very, very time intensive and I am working on other places - I stuck up.
After a reset serial EEPROM is not recognized cleanly. I saw this several times. Maybe the scrambling occurs at the transfer process from flash to serial EEPROM.
Regarding "saving of settings": I think we always save the settings which are in serial EEPROM or Flash - not the "active ones" on the radio. When you change settings and do NOT store them by switching off and on the radio or press long MENU you will save the settings from flash / serial EEPROM not the active config from RAM!
Regarding source code:
UHSDR is Open Source GPLv3. You can find source code and all commits from the beginning on GitHub (https://github.com/df8oe/UHSDR)
At some points of investigation I wondered if it is easier to expand the USB-config-read/write capability to work with partially files. E.g. a json which only consists of 30 pairs of addresses and values and I can apply this via python script. So you could test without changing whole settings file. But again: I am very busy at other places and this issue I only get by injecting infected settings. I never was able to reproduce the issue by myself on any mcHF I have tested.
EDIT:
May be there *IS* a bug which changes values which should not be modified (e.g. by a pointer issue). All this is highly hypothetically but for investigating the fastest way(I think) would be the possibility to modify small sets of non-continuous settings by PC and USB.
EDITEDIT:
I have read in Yahoo NG that there is a patched binary of recent firmware existing where the issue is fixed. That is the way Open Source should **NOT** work. Modifiying the source and not contributing to the project. I hope it is a hoax and not reality. Otherwise there do exist some "bad boys" - I only can call them this way.
vy 73
Andreas |
Title: Re:[not solved] Receiving strong sigs 12.KHZ
Post by: SP9BSL on 30. August 2018, 08:23:25
| I have read in Yahoo NG thatthere is a patched binary of recent firmware existing where the issue is fixed. That is the way open source shoulderstand ** NOT ** work. Modifiying the source and not contributing to the project. I hope it is a hoax and not reality. Otherwise there do exist some "bad boys" - I only can call them this way. |
|
Hi Andreas,
I've posted note in Yahoo NG about this... |
Title: Re:[not solved] Receiving strong sigs 12.KHZ
Post by: WD8BXS on 30. August 2018, 12:19:05
So, I restored a known good config file.
all was well.
I changed the frequency calibration, no change.
I changed the display to scope, no change.
I swapped SWR meter, no change.
I turned on RX IQ auto, no change.
I turned the radio off and left it til this morning, when I turned it on this morning, the problem is back.
Here is this file for your consideration.
de Chuck,
Thanks for all of your hard work, I hope this can be figured out, I know it takes time, and we all have other
interests. Take care my friends!! |
Title: Re:[not solved] Receiving strong sigs 12.KHZ
Post by: WD8BXS on 30. August 2018, 13:00:58
OK, Next phase, hih
I restored the good file again, changed the setting again.
Backed up that config, all is still good.
used the rig menu and backed up and restored the settings, and all is still OK
Will keep an eye on and report back.
Chuck |
Title: Re:[not solved] Receiving strong sigs 12.KHZ
Post by: DF8OE on 30. August 2018, 15:24:10
Very interesting is that you can get the issue very easy. I can do what I want: I am unable to produce it by myself. Only with the "help" of infected configs. I am beginning to doubt if it isn't a hardware problem at reading/writing from/ro Flash/EEPROM. Writing/reading issues can produce very cool things.
To check if reading / writing is ok we can implement crc-checksum which must not be very strong (no repair of damaged files - just signalize it is damaged).
vy 73
Andreas |
Title: Re:[not solved] Receiving strong sigs 12.KHZ
Post by: WD8BXS on 30. August 2018, 15:51:17
So, Andreas, you are thinking possible bad solder around EEPROM?
I didn't mention that I use the MCM32F429VIT6 MCU,
I do not think that would be the issue.
Thanks,
Chuck
PS radio is still working OK. |
Title: Re:[not solved] Receiving strong sigs 12.KHZ
Post by: DB4PLE on 30. August 2018, 16:50:10
Hi,
let have a look at the facts and what they mean:
Even if the backup and restore from flash is responsible for changing a good config to its bad sibling (which then causes the 12 khz issue), it is not directly relevant here.
Since also the "bad" config is still a 100% valid configuration (as far as I can see). So there may be multiple ways how to get to a config which causes the trouble, the flash backup/restore possibly one of these.
Until we have figured out, what is the root cause of the 12khz problem, even if the EEPROM to Flash transfer is problematic and we fix it, it would not fix the 12khz issue as such but only close one path to get the problem.
BTW, we already have a routine which checks if the content of EEPROM and Flash storage are identical, we use the when we transfer the Flash for the first time into the EEPROM. We could also call this after Backup and Restore and see if there is an issue. No need to run a CRC, that would be only necessary for data corruption between Backup and Restore (i.e. something changes the content of the flash after backup and before restore but this is unlikely).
73
Danilo
|
Title: Re:[not solved] Receiving strong sigs 12.KHZ
Post by: DF8OE on 31. August 2018, 04:55:04
@DJ0RG
I have played around TX IQ settings and I cannot confirm that anything I have cofigured is scrambled by saving / backing up / restoring. All is working rock solid.
vy 73
Andreas |
Title: Re:[not solved] Receiving strong sigs 12.KHZ
Post by: DB4PLE on 01. September 2018, 11:59:09
Hi Chuck,
could you please do the following and report back:
Use a mcHF which has the issue:
- Turn off all DSP functions and also the Noise blanker.
- Use a filter with a bandwidth of 3.8khz or larger on SSB and test if the issue is still there?
- See if the issue is detectable in all of the modes (SSB, AM, FM) or just in SSB?
- Is there any indication on the waterfall or scope of the 12khz signal (I doubt that, but let's be thorough here).
There must be a place in the signal processing chain where we introduce the problem. I suspect the decimation (since it is 12khz we decimate to in SSB for the filters narrower than 3.8khz). The above tests could help us to exclude certain part of the processing chain.
73
Danilo |
Title: Re:[not solved] Receiving strong sigs 12.KHZ
Post by: WD8BXS on 01. September 2018, 13:15:07
EUREKA!!! I ran the test, and as soon as I go from 3.8 filter to 3.6 the issue is there!
Only on filter 3.8 and above is the extra signal not there.
SSB or CW does not matter.
Chuck |
Title: Re:[not solved] Receiving strong sigs 12.KHZ
Post by: WD8BXS on 01. September 2018, 13:15:55
Also , there is no indication on the scope.
Chuck |
Title: Re:[not solved] Receiving strong sigs 12.KHZ
Post by: DB4PLE on 01. September 2018, 13:45:02
Hi Chuck,
thank you! Then my guess that it is somewhat related to the decimation/filters is right.
However, my ability to go further ends here I am afraid, the signal processing specialist have to take over now.
I will have a look into the code, though, maybe I am lucky. I might ask for more tests.
73
Danilo |
Title: Re:[not solved] Receiving strong sigs 12.KHZ
Post by: DB4PLE on 01. September 2018, 14:13:52
Hi Chuck,
please test again with the firmware attached to this post.
Like before, use a configuration which does show the problem.
Now: Use a 3.4 khz, a 3.6 khz and a 3.8 khz filter in SSB and report which of the filters have the issue. It is not necessary to test more than one filter per bandwidth.
73
Danilo |
Title: Re:[not solved] Receiving strong sigs 12.KHZ
Post by: WD8BXS on 01. September 2018, 15:12:32
Tested with this new firmware,
Same results although at a much lower volume level.
It is visible on the scope now
Chuck |
Title: Re:[not solved] Receiving strong sigs 12.KHZ
Post by: DB4PLE on 01. September 2018, 17:53:52
Hi Chuck,
please answer in detail for the "new" firmware:
3.4 khz Filter: Is the issue noticiable? And if yes, how does it compare to the "official" firmware?
Same for 3.6 and 3.8 khz individually.
Sorry for not asking precise enough in the previous post.
Thank you!
Danilo
|
Title: Re:[not solved] Receiving strong sigs 12.KHZ
Post by: WD8BXS on 01. September 2018, 21:57:35
It sounds that the volume goes to about 1/2 on 3.6 and 3.4, then gets a little bit louder on 3.2
But not as loud as the previous firmware.
Thanks,
Chuck |
Title: Re:[not solved] Receiving strong sigs 12.KHZ
Post by: DB4PLE on 01. September 2018, 22:23:18
Hi Chuck,
now let me tell what I did:
3.4 filter: completely unmodified compared to normal firmware
3.6 filter: used a more powerful decimation filter which is normally used on F7 and H7, but not on F4
3.8 filter: completely unmodified compared to normal firmware
So the expected result would have been: 3.4 no change (i.e. issue is noticable as before), 3.6 a change (or no change if issue is not related to the decimation filter itself) and 3.8 no change (i.e. issue is not noticable).
But anyway, from here as I said, the DSP specialist have to be called in, this is out of my league.
73
Danilo
|
Title: Re:[not solved] Receiving strong sigs 12.KHZ
Post by: WD8BXS on 02. September 2018, 00:14:20
Thank you Danilo, glad to have you here!!
I am sure someone will figure it out!!
DE Chuck, 73 |
Title: Re:[not solved] Receiving strong sigs 12.KHZ
Post by: F4HTX on 02. September 2018, 11:57:37
Hi Chuck, Hi Danilo,
Thanks for Chuck pointing out something, On my side I did tenth of tests to try to isolate something reproductible, but nothing until now...
I can confirm I experienced exactly the same results as Chuck with the filters, and with the "special" firmware, I did not notice significative audible difference.
Best Regards,
François
PS : By the way, the problem is still existant on my D2.9.50 mchf freshly resetted, with no modification of any parameters. This lead more to firmware than configuration issue. |
Title: Re:[not solved] Receiving strong sigs 12.KHZ
Post by: F4HTX on 02. September 2018, 12:57:19
Hi Friends,
Trying to narrow the scope, I did some more digging. Here are the results.
For each iteration, I repeated precisely this same process :
1. Standard usb key firmware update
2. F1, F3 & F5 All Defaults parameters reset
3. no change in anything except :
3.1 I changed the band and frequency to track my -73dBm / 10100kHz test signal
3.2 and, indeed, I changed the bandwidth in order to check for the birdies
I iterated by dichotomy the firmware versions and get back pointing to 2.9.9 / 2.9.10, as in a previous thread where I thought wrongly it was a Rx IQ autocorrection problem ( https://www.amateurfunk-sulingen.de/forum/index.php?board=19;action=display;threadid=1120 )
So, with the process I described earlier :
- 2.9.9, no filter problem
- 2.9.10 it's a mess with 6kHz birdies (try it)
I hope this might help to pinpoint in the code.
73's
François |
Title: Re:[not solved] Receiving strong sigs 12.KHZ
Post by: DF8OE on 02. September 2018, 13:11:13
Hi François,
I have just loaded 2.5.57 (!!) and do have 12KHz issue identical to recent firmwares dependent on filter settings (try it!). So I cannot confirm that there is something changed from 2.9.9 to 2.9.10 which causes the issue.
vy 73
Andreas |
Title: Re:[not solved] Receiving strong sigs 12.KHZ
Post by: F4HTX on 02. September 2018, 13:31:52
Hello Andreas,
I just did the exact same procedure as in my previous post with 2.5.57 : not a single birdy overhere !
Except @+24kHz because the "Rx IQ Autocorrection" is off by default.
This issue is starting to be really fun.
I did not look in details the STM cpu family, if we get different results with same procedure, might it be something related to minor version of the chip ? Sometime manufacturers adjust details without widely advertising...
Best Regards,
François |
Title: Re:[not solved] Receiving strong sigs 12.KHZ
Post by: DF8OE on 02. September 2018, 14:29:47
I have tested with -12KHz Xlate (that is my standard Xlate) and the issue is identical on 2.5.57 and recent firmwares - when I use an infected config. That is reproduceable on all mcHFs I own (three). One has a 407VGT6, the others do have 427/439 MCUs.
This issue is a very, very ugly one. And when I take my normal configuration, it is gone on all devices. I can only get it when I inject an infected config.
And, honestly: before I got such a config I was in doubt if it exists.
vy 73
Andreas |
Title: Re:[not solved] Receiving strong sigs 12.KHZ
Post by: F4HTX on 02. September 2018, 15:05:55
Hello Andreas,
My point was to try to identify if the issue appeared with a specific firmware version using the Default Settings, not an infected configuration (in order to identify a reproductible path exhibiting the 12kHz issue).
So, in my specific case, the issue exists on a mchf with a reseted to defaults configuration only with firmware > 2.9.9. Then the filter settings affects the 12kHz birdies.
I tried to restore an infected configuration to a reseted 2.5.57, but unsuccessfully because :
Detecting if UHSDR is connected:
... could not find a connected UHSDR with extended CAT commands (required)
So it would be necessary to upgrade to get the extended cat, then restore an infected configuration and downgrade to 2.5.57. In this process, I can not perform the Default Parameters reset...
My knowledge of the uhsdr software architecture is minimal but without the reset to defaults parameters, I am not sure in exactly which status mchf will really be.
Let me know if it's relevant to proceed this way.
I am available to do any test which might help the cause :)
Best Regards,
François
|
Title: Re:[not solved] Receiving strong sigs 12.KHZ
Post by: DF8OE on 02. September 2018, 15:11:18
Hi François,
I have injected an infected configuration and after that I have made a backup in flash in "Standard Menu". So I can walk through all versions and can rewoke an infectied config by doing "restore" in "Standard Menu".
I cannot reproduce the issue on recent firmwares when I do a "full reset" (and others have confirmed that a full reset has solved the issue for them, too). It is absolutely confusing and I am unable to see a clean trace.
vy 73
Andreas |
Title: Re:[not solved] Receiving strong sigs 12.KHZ
Post by: F4HTX on 02. September 2018, 15:24:09
Re Andreas,
Sorry, I missed this very interesting point you mentioned :
And when I take my normal configuration, it is gone on all devices. I can only get it when I inject an infected config.
I am not sure what it does mean.
If you perform :
1. Standard usb key 2.9.50 firmware update
2. F1, F3 & F5 All Defaults parameters reset
3. no change in anything except injecting a test signal in your mchf
After this process my mchf shows the filter/birdies issue. Does yours too ?
With this minimal process our two mchf's should be in exactly the same state. If we do not have the same results here this is quite interesting.
73's
François
later PS : Sorry Andreas our responses crossed.
|
Title: Re:[not solved] Receiving strong sigs 12.KHZ
Post by: F4HTX on 02. September 2018, 15:32:01
Hello all,
Because Devil is in the Details, overhere the mchf's having the 12kHz issue do not have the RealTime clock enabled (no hardware mod). Who knows...
François |
Title: Re:[not solved] Receiving strong sigs 12.KHZ
Post by: DF8OE on 02. September 2018, 15:36:14
No realtime clock here, also.
I have taken a look at the code which changed from 2.9.9 to 2.9.10: It was only the I2S-Clock which is used for USB audio transmission. I tried reverting the code on recent firmware (which is very simple) but as I suggested: nothing changes. Issue is present with the modification from 2.9.9 --> 2.9.10 and without...
vy 73
Andreas |
Title: Re:[not solved] Receiving strong sigs 12.KHZ
Post by: F4HTX on 02. September 2018, 16:32:54
Hi Andreas,
Thanks for checking the code, on my side I re-perform the process, and finally found that in 2.9.9 the birdies are there but very very low, even with volume fully cranked up. I only noticed them by moving around the frequency to notice it was there in the noise.
Very sorry for the false clue.
As you said ugly's the word.
73's
François |
Title: Re:[not solved] Receiving strong sigs 12.KHZ
Post by: DF8OE on 02. September 2018, 16:40:46
No problem François - we all are working on the same problem but sometimes go wrong.
Please be so kind and recheck your 2.5.57 observations. Crank up your audio and play with old AGC (switch it off !!!) which "eats" the spurs. It would be fine if you can confirm the spurs are reproduceable there for you, too.
vy 73
Andreas |
Title: Re:[not solved] Receiving strong sigs 12.KHZ
Post by: DF8OE on 02. September 2018, 17:19:51
I have just gone back to the earliest binaries I can access at the moment: 1.4.0 from GitHub. And I can confirm that the issue is already present there. I cannot build binaries from the sources with recent gccs - it builds with hundreds of warnings and binaries are not functionally. I must draw back to an very old compiler. But I can state:
This issue is very, very, very old and was covered by bad working AGC and other bad working stages. But it was already there since a long time - and nobody noticed it.
Happy new year! :P
I cannot believe there are bugfixed binareis what is stated in Yahoo NG. I think they have found a method of covering the issue (like the old AGC has done).
vy 73
Andreas |
Title: Re:[not solved] Receiving strong sigs 12.KHZ
Post by: F4HTX on 02. September 2018, 17:21:04
Hi Andreas,
Good guess : on my mchf, with 2.5.57 it's there, almost unnoticeable, but definitely there.
Furthermore, on my ovi-40 with 2.9.50 and 2.9.9 it's also there. It is very faint and not obvious, but if you track it is definitely a birdy.
From my viewpoint this lead to an interesting conclusion : this issue might have been in our firmwares for a very long time but only surfacing in specific conditions. Most of users won't probably notice any birdies problem in real life use considering the usual HF background noise.
And also a second conclusion : the F1F3F5 reset do not get rid of the birdies problem but only make it to dive back in noise. So the parameters configuration might not be the problem, more a chemical revelator of a low noise illness. Not surprising that finally nothing surfaced in examining the backup files.
About my testing conditions : I used my Adret 7100D @ -73dBm, last time I checked it was correctly calibrated and the generator noise and spurs were very low (-136dB). Just to be sure, I'll check it later this week, in case something failed, and also redo the test with another generator (R&S SMT03). Just in case...
Best Regards,
François
Later PS : Crossed responses again :)
|
Title: Re:[not solved] Receiving strong sigs 12.KHZ
Post by: DD4WH on 03. September 2018, 07:03:34
Hi folks!
Sorry for being silent so long, I was busy with other things/hobbies ;-).
it is hard for me to understand the issue, because it is not present on my machine, but I try to summarize the main facts here and then I would like to pose some more questions which would be very nice if you could try to answer them/ carry out some more tests on your machines, thanks in advance!
SUMMARY of the 12kHz problem:
* a wanted signal is received
* you notice the presence of a copy of that wanted signal (called unwanted signal) which is 12kHz above and 12kHz below the frequency of the wanted signal
* the unwanted signal is audible, but not visible with magnify 1x [do not use other magnify modes while we are investigating this problem, please]
* that unwanted signal is well below the wanted signal [Francois reported: -73dBm wanted signal and -136dBm unwanted signal --> 63dB difference; Chuck reported: "there is no S-meter reading"]
* so I assume the unwanted signal is about 50-60dB lower than the wanted signal, is that correct?
* the unwanted signal(s) is/are much louder when using a filter bandwidth lower than 3k8, but they are still present with filter bandwidths higher than 3k6
* AGC ON/OFF has no influence on the presence of the unwanted signal
* NR ON/OFF has no influence on the presence of the unwanted signal
* NB ON/OFF has no influence on the presence of the unwanted signal
* Translate 0/6/12kHz has no influence on the presence of the unwanted signal
* has been present in the firmware since the very first versions by Chris & Clint (at least from 1.4.0 on)
FIRST THOUGHTS on the cause of the problem (from here on, this is massive speculation from my side):
* to me, it seems everything is pretty OK ;-):
* the filter suppresses unwanted signals by 60dB
* that is how the digital filters have been designed from the beginning of the making of the mcHF software: with 60dB stopband attenuation, so I would not expect better attenuation --> some filter details can be found here:
https://github.com/df8oe/UHSDR/wiki/How-does-your-UHSDR-software-DSP-work
* another question would be: are we happy with "only" 60dB: historically, the mcHF was not designed to be a high end TRX, it evolved more and more into that direction, but the filter design stayed the same and it would be a massive task to recalculate all the filters for higher stopband attenuation and on the F4 it would be very hard to achieve that because of memory and processor load constraints
* on the F7/H7 this is a totally different thing and we have to improve the filters there, that is clear
FURTHER QUESTIONS (please help me here):
* I assume the unwanted signal is about 50-60dB lower than the wanted signal, is that correct?
* could you carry out a test under defined conditions:
- 2k7 lowpass filter in USB
- IQ auto correction ON (because we have to eliminate the possibility that it is just a malaligned IQ phase/amplitude); alternatively you can show me on the video that your IQ phase and amplitude is perfectly manually calibrated at 7100kHz
- set frequency translation to +12kHz
- set display to spectrum display only
- wanted signal 7100kHz
- is the unwanted signal in LSB?
- which signal level does the unwanted signal have? is it approximately 60dB down? Please measure with the S-meter/dBm display with magnify 1x
It would be nice if you could do exactly the same test again with the 3k8 lowpass filter.
Thanks a lot for your efforts and tests so far and thanks in advance for helping exploring the cause of the issue!
73 Frank DD4WH
|
Title: Re:[not solved] Receiving strong sigs 12.KHZ
Post by: DD4WH on 03. September 2018, 07:15:01
could anybody post a video showing this problem or the results of your further tests?
Chuck: unfortunately your link does not work anymore (maybe its my dropbox account which is full ;-))
Thanks in advance!
73 Frank |
Title: Re:[not solved] Receiving strong sigs 12.KHZ
Post by: F4HTX on 03. September 2018, 07:25:15
Hello Frank,
Thanks for trying to summurize, here are couple details to adjust :
1. * you notice the presence of a copy of did wanted signal (called unwanted signal) All which is 12 kHz above and 12kHz below the frequency of the wanted signal
It's 1x +/-12kHz ; 2x +/- 12kHz ; 3x +/-12kHz ; nx... I didn't check where n ends.
2. * That unwanted signal is well below the wanted signal [Francois reported: -73dBm wanted signal and unwanted signal -136dBm -> 63dB difference; Chuck reported: "there is no S-meter reading"]
My explaination was not clear : -73dBm is my test signal and -136dBm is the spurs level on my RF generator, not the mchf birdies
3. * so I assume the unwanted signal is about 50-60dB lower than the wanted signal, is did correct?
Definitely NO. In some of my test the +12 is almost as loud as the original signal. That's why people noticed in the first place and so started the issue.
So, no it's definitely not ok to hear a spur as loud as the main signal.
I'll try to find some spare time to set up the test you are asking for. Not easy to reach the 1Hz resolution in frequency. I have to think about a test setup. Do you need a +/-1Hz precision or more ?
73's
François |
Title: Re:[not solved] Receiving strong sigs 12.KHZ
Post by: DF8OE on 03. September 2018, 07:53:12
I never have had heard spurs louder than 50dB lower than wanted signal. When I use infected config it is ~50dB under wanted signal when I use filters < 3.6KHz and 65dB at wider filters. When I use a default (or all of my) configs it is 65db at filters < 3.6KHz and > 80dB at higher filters. So I am in the same situation as Frank: I do not know where you are talking about - I cannot confirm here :'(. There is nothing what bothers in normal working conditions... All infected configs I have got show suppressions better than 40dB. That is slightly too bad - but not "as loud as wanted signal".
EDIT:
Meanwhile I think these spurs are not removeable at F4 processors. Filters cannot be much steeper than now because of missing horse-power. You can possibly switch of waterfall AND scope and then use a slightly steeper filter. But that is not the goal of an SDR transceiver. The fact that it is present on all firmwares is no issue - it is due to the complete concept and given horse power NORMAL. You can get rid of the too bad suppression by choosing a better configuration. There are some (actually unidentified) settings which made suppression much poorer when cranked up. So we must crank down them...
vy 73
Andreas |
Title: Re:[not solved] Receiving strong sigs 12.KHZ
Post by: DD4WH on 03. September 2018, 07:54:27
Salut Francois and thanks for the quick anwer!
Oh, thats not good to have an unwanted signal at that signal level! Then I have to think some more . . .
Dont worry too much about the test setup: it would be sufficient to inject a tone with your excellent signal generator and then zero beat the wanted and unwanted signal(s) with a one Hz resolution of the VFO. I just want to be sure that the difference of wanted and unwanted signal is exactly 12kHz/24kHz/36kHz etc.
73 Frank
|
Title: Re:[not solved] Receiving strong sigs 12.KHZ
Post by: DD4WH on 03. September 2018, 07:59:14
Hmm, Andreas, now that is confusing . . .
Could it be that we are talking about two different things ? :
1.) the twinpeaks issue that is sometimes occuring with unwanted signals as loud as the wanted signal and
2.) the 12k issue reported here with 50/60dB suppressed unwanted signals?
Chuck, how loud are your unwanted signals? You stated that there is "no S-meter reading", so I assume they are not at all as strong as the wanted signal, is that correct?
Francois, is the issue with unwanted signals that are as loud as the wanted signal reproducible after doing a RESET CODEC from the menu (try it several times ;-))?
73 Frank
|
Title: Re:[not solved] Receiving strong sigs 12.KHZ
Post by: DF8OE on 03. September 2018, 08:01:28
Best is to test in Xlate = off. Then you will not see the damned twin peaks problem which pops up randomnly.
vy 73
Andreas |
Title: Re:[not solved] Receiving strong sigs 12.KHZ
Post by: DD4WH on 03. September 2018, 08:05:09
Andreas, I am not sure whether the twinpeaks issue can be totally omitted when using 0kHz translate, but you will not see it anymore (but probably still have the issue).(not totally sure about this)
To be safe, please use 12kHz translate for all tests, then the twinpeaks issue can be easily seen ;-).
73 Frank |
Title: Re:[not solved] Receiving strong sigs 12.KHZ
Post by: DF8OE on 03. September 2018, 08:10:35
If you use Xlate = off you cannot be confused by seeing / hearing it. It will be exactly at the same frequency as the wanted signal and will not produce spurs beside the wanted signal. Maybe some of the testers have seen the twin peaks syndrom and thought it is/was the 12KHz issue.
12KHz issue is completely independent of Xlate settings.
vy 73
Andreas |
Title: Re:[not solved] Receiving strong sigs 12.KHZ
Post by: DF8OE on 03. September 2018, 08:56:52
My measurement of 12 KHz spur:
Wanted signal: -33dBm @7100KHz
spur: -102dBm @7112KHz
all other spurs at n x 12KHz are less stronger: 7124KHz is -108dBm next spur is not detectable in dBm (but I can hear it).
Measured using 2.9KHz BPF in LSB with Xlate OFF.
If I increase the input level only 6dB the mixer is completely overdriven In that case I do have intermixing products over > 500KHz with amplitudes of > -70dB.
That is my standard working configuration (except Xlate=off - I normally use Xlate=-12KHz).
This is a suppression of nearly 70dB and absolutely no problem in normal working conditions.
vy 73
Andreas |
Title: Re:[not solved] Receiving strong sigs 12.KHZ
Post by: DD4WH on 03. September 2018, 09:02:51
has anybody ever seen this problem in modes AM or SAM?
73 Frank |
Title: Re:[not solved] Receiving strong sigs 12.KHZ
Post by: DD4WH on 03. September 2018, 09:05:19
Hi Andreas,
thanks for your measurements! Is this with a "broken" configuration?
73 Frank |
Title: Re:[not solved] Receiving strong sigs 12.KHZ
Post by: F4HTX on 03. September 2018, 09:19:02
Hello Friends,
That's what I meant by "as loud", I do agree it's a little less loud, but far from normal..
https://wetransfer.com/downloads/523f86dda1883e97c76dffeb5220ab8320180903091123/18cd69c66c98d287cc84b7760c9579c220180903091123/51c124
Rx IQ auto correction is off (doesn't matter on +/_12kHz)
I'll try to do the measurements asked by the end of the day.
73's
François |
Title: Re:[not solved] Receiving strong sigs 12.KHZ
Post by: DF8OE on 03. September 2018, 09:21:40
No. It is with my standard working config (same result as default config).
I forgot value for filter > 3.6KHz:
Spur is audible, but no dBm readable.
Here the result for an infected config:
spur: -80dBm @7112KHz
vy 73
Andreas |
Title: Re:[not solved] Receiving strong sigs 12.KHZ
Post by: DD4WH on 03. September 2018, 09:29:06
Hi Francois,
thanks a lot for the video!
Did you test to reset the codec in the menu and whether that helps to eliminate the loud spurs?
73 Frank |
Title: Re:[not solved] Receiving strong sigs 12.KHZ
Post by: F4HTX on 03. September 2018, 09:35:15
Hi Frank,
Just did it, the codec reset doesn't change anything.
73's
François |
Title: Re:[not solved] Receiving strong sigs 12.KHZ
Post by: DD4WH on 03. September 2018, 09:36:59
Francois, another question (sorry ;-)):
which frequency translation did you have for the test? |
Title: Re:[not solved] Receiving strong sigs 12.KHZ
Post by: DF8OE on 03. September 2018, 09:37:05
Hi François,
I am impressed. If input signal is lower than 9+20dB I never can see or hear any spur - regardless if config is infected or not.
Now I am nearly sure there is a hardware related issue like the twin peaks syndrom. We have had this ~1.5...2 years ago and have found it is a hardware problem INSIDE audio cedec - not removeable. If it is active, phase of I and Q is witched one byte so you see massive spurs like complete debalanced IQ setting. That is why I proposed to test in non Xlate mode: if you use non Xlate your wanted signal is attacked - but no spurs are added. So idf the problem is gone in non Xlate mode it was "twin peaks syndrom". If it is already there: I have no words, no idea and no soultion. I never heard it.
vy 73
Andreas |
Title: Re:[not solved] Receiving strong sigs 12.KHZ
Post by: DD4WH on 03. September 2018, 09:48:43
I guess we are having some problems in defining our standard test conditions ;-).
Francois, could you please do the test again (and make a video again :-)):
* IQ auto correction ON (because we have to eliminate the possibility that it is just a malaligned IQ phase/amplitude); alternatively you can show me on the video that your IQ phase and amplitude is perfectly manually calibrated at 10100kHz
* set frequency translation to +12kHz
* set display to spectrum display only
* could you tune a little slower please, and tune from 10100kHz to 10126kHz and down to 10074kHz
* after that, could you reset the codec several times and do the test again
Sorry to be so demanding, but I would really like to thoroughly understand the problem (and potentially provide a solution to it :-)).
73 Frank
|
Title: Re:[not solved] Receiving strong sigs 12.KHZ
Post by: DF8OE on 03. September 2018, 09:57:38
Here is the result of the German jury ::) :
Video (https://www.amateurfunk-sulingen.de/data/uploads/VID_20180903_114439.mp4)
??? I do not understand anything.
vy 73
Andreas |
Title: Re:[not solved] Receiving strong sigs 12.KHZ
Post by: F4HTX on 03. September 2018, 09:59:24
Hi Friends,
Yes, this is a very unusual problem indeed.
Frank : the video is with -12kHz, but no change for any setting, including off
Andreas : How do you perform the spur measurement ? I am not sure the s-meter is reliable, the problem might be downstream.
I had this level of birdies on the two mchf I crafted (v0.6) and also on an ovi-40 but I cannot reproduce it on the ovi-40 since I did the F1F3F5.
73's
François
|
Title: Re:[not solved] Receiving strong sigs 12.KHZ
Post by: DF8OE on 03. September 2018, 10:01:12
Display in dBm is absolutely reliable. You can use it for measurements. Actually I do have here 7 mcHFs (three of my own and 4 repairs). None of them shows the issue, and on my two OVI40 I never have had seen (heard) them, too.
vy 73
Andreas |
Title: Re:[not solved] Receiving strong sigs 12.KHZ
Post by: DD4WH on 03. September 2018, 10:01:26
Thanks, Andreas!
Could you also extend your tuning range from 10100kHz to 10126kHz and down to 10074kHz?
|
Title: Re:[not solved] Receiving strong sigs 12.KHZ
Post by: DD4WH on 03. September 2018, 10:05:09
Hi Francois,
thanks for the info! Very helpful!
But it could be that the problem you are observing is fully due to a misaligned IQ phase & amplitude correction.
So could you do the video again with automatic IQ correction set to ON?
dBm display/S-meter display was thoroughly tested by Christian:
http://funkamateure-dresden-ov-s06.de/index.php?article_id=462&clang=0
73 Frank |
Title: Re:[not solved] Receiving strong sigs 12.KHZ
Post by: DF8OE on 03. September 2018, 10:22:05
extended range (https://www.amateurfunk-sulingen.de/data/uploads/2018-09-03-121545.webm)
EDIT:
Again: if you want to remove the influence of (probably) bad IQ settings --> do the test with Xlate off. This will remove additional the following influences:
- twin peaks syndrom
- bad hardware in audio IQ chain
- missing or crippled LO signal(s)
If you do the test using Xlate=off there rests only firmware or setting related issues.
vy 73
Andreas
|
Title: Re:[not solved] Receiving strong sigs 12.KHZ
Post by: DD4WH on 03. September 2018, 10:39:16
Hi Andreas,
thanks for the video!
From my point of view, it shows the normal behaviour and I cannot see any malfunction. Unfortunately you do not state, which translate frequency you used . . .
Thats funny, but because you insist on setting translate to OFF, I cannot interpret anything, because you do not write which settings you have . . .
My proposal for all further tests would be:
PLEASE use exactly the test conditions from my post above, so I can interpret all the tests and videos. If only one condition is different and you do not state which one, I cannot interpret them anymore . . .
73 Frank
|
Title: Re:[not solved] Receiving strong sigs 12.KHZ
Post by: DF8OE on 03. September 2018, 10:43:43
I have tested in Xlate=-12KHz (my normal working Xlate). Changing of Xlate has no influence on 12KHz spurs because there is no hardware issue in my mcHFs. But if I would have hardware issue or twin peaks syndrom there would be a huge difference in behaviour! If everything is ok there is no difference.
Auto IQ is set to "ON". If I switch it off my manual adjustments take over. They are exactly done as described in Wiki. So the difference between AUTO IQ ON and OFF is very small.
EDIT:
And YES, I cannot see any malfunction, too. Even with infected config the issue pops up if the input signal is much louder than 9+20dB. Below that NO SPURS AUDIBLE. Under normal working conditions: If I get a signal louder than 9+20dB (and it is *not* my neighbour) all other signals and QRM is higher, too and the resulting spurs are buried in the noise...
vy 73
Andreas |
Title: Re:[not solved] Receiving strong sigs 12.KHZ
Post by: DD4WH on 03. September 2018, 11:03:25
Thanks, Andreas!
Totally agreed!
Lets see what the problem looks like in Chucks and Francois´ radios when they use the same test conditions.
73 Frank
|
Title: Re:[not solved] Receiving strong sigs 12.KHZ
Post by: DF8OE on 03. September 2018, 11:07:12
I think I am "bounced out" now for further testings presenting the issue. I thought I have had it using infected config: but it seems that never was so. The suppression always was > 40dB. What I have heard from the video of François is a complete different behaviour of all I have seen. So I can only state: not reproduceable for me - under no circumstances.
What I have seen is the normal filter suppression slightly impacted by unluckily settings.
vy 73
Andreas |
Title: Re:[not solved] Receiving strong sigs 12.KHZ
Post by: DD4WH on 03. September 2018, 11:11:47
Andreas, I think we have several issues mixed here:
* decimation/Hilbert filter software issue (I found a minor filter problem in the firmware and will test an improvement in the next days)
* maybe an EEPROM setting issue (I sent an Email to Danilo for that)
* maybe IQ settings issues (lets see the outcome of the tests by Chuck and Francois)
Lets see :-)
73 Frank |
Title: Re:[not solved] Receiving strong sigs 12.KHZ
Post by: DF8OE on 03. September 2018, 11:18:03
Hi Frank,
full ack. I can confirm that there is a level jump if I change filter from 3.6K to 3.8K. With input signal -30dBm I can hear it using filter 3.6K (or smaller) but when I switch to 3.8K it is completely gone. It is not audible - as deep as I try to find it. If this can be fixed behaviour is perfect. Now it is "near perfect" because the spurs never can mislead me under normal operating conditions - much too low.
vy 73
Andreas |
Title: Re:[not solved] Receiving strong sigs 12.KHZ
Post by: F4HTX on 03. September 2018, 12:22:49
Hi Friends,
@ Andreas : I did not say the meter measurement is unreliable, I was just asking if the birdies problem, because it’s a locally generated artifact, might elude the s-meter measuring in the signal processing chain. But this might be not relevant because my ignorance about the software architecture, please forgive...
@ Frank : Hereunder a link for the two videos you asked for (sorry 600MB, I should find sometime a way to compress those vids) :
https://wetransfer.com/downloads/f1a3122969e8be3c1a77bebb66c7744e20180903110426/138a4684914b84e908a4ab0f6b26961520180903110426/aec5cd
This time +12kHz and RxIQon, I reseted the codec multiple time on the second video.
I would like to add a couple word about the history : I crafted two mchf v0.6, gifted one to my friend F6DPT (50 years experienced ham), and we were very happy playing with those TRX. After few months I noticed that I listened multiple occurrences of strong signals on a x12kHZ modulo. It was new and loud and I am quite sure I would have noticed it if it existed earlier. A few time later my friend called me because he noticed the same problem. We were doing the firmware update when they appeared and I am totally unsure when the change occurred. But we noticed this almost at the same time, and it appears on the web shortly later. I do agree (because of the firmware downgrade tests we did yesterday) that this is probably a long time existing problem rising up because of the better AGC. We could push the dust under the rug, but, for sure, it will came out again.
The question I would love to get an answer is : which part of the processing is able to generate such artifacts (nx +/-12kHz birdies) ? Does the maths predict this or is that a coding issue ?
There is also a possibility of hardware generated artifacts, ADC and DAC are far from perfect, and their implementation could be very tricky when you are playing with large dynamic range (routing, decoupling, tracks coupling, etc.). So far, we have no control on this aspect of the mchf.
I understand how frustrating and annoying this issue might be for the lead team, but without knowing where it does come from, I think we have a time bomb in the uhsdr.
Once again, we have to go « where no man… » And remember, people in red uniform die first !
73’s
François |
Title: Re:[not solved] Receiving strong sigs 12.KHZ
Post by: DD4WH on 03. September 2018, 12:56:58
Hi Francois,
excellent! Thanks a lot for the videos, that is extremely interesting!
I had to watch the videos three times each to write down what is happening there . . . because on the second video the behaviour is different!
1st video: with automatic IQ correction, 3k6 BPF:
* what you can see is the following:
10.053kHz very faint signal
10.065kHz loud unwanted signal, but not visible
10.077kHz loud unwanted signal, but not visible
10.089kHz NO SIGNAL
10.101kHz wanted signal with -74dBm
10.113kHz NO SIGNAL
10.125kHz NO SIGNAL
10.137kHz NO SIGNAL
10.149kHz NO SIGNAL
--> no unwanted signals above the wanted signal
--> exactly TWO unwanted loud signals below the wanted signal spaced 12kHz in a distance of 24kHz of the wanted signal
--> unwanted signals NOT visible (and no dBm display possible)
2nd video: with automatic IQ correction, 3k6 BPF:
* what you can see is the following:
10.053kHz very faint signal
10.065kHz loud unwanted signal, but not visible
10.077kHz loud unwanted signal, but not visible
10.089kHz loud unwanted signal, but not visible
10.101kHz wanted signal with -74dBm
10.113kHz loud unwanted signal, but not visible
10.125kHz NO SIGNAL
10.137kHz NO SIGNAL
10.149kHz NO SIGNAL
--> YES, unwanted signal above the wanted signal
--> exactly FOUR unwanted loud signals below the wanted signal spaced 12kHz in a distance of 12kHz of the wanted signal, three below and one above
--> unwanted signals NOT visible (and no dBm display possible)
Honestly, I have no idea what that problem is!
It has nothing to do with the things I discussed above in this thread. You can easily see that the problems in your former video posted above in this thread were 100% caused by the IQ correction that was set to OFF. They have disappeared. But I cannot explain the new problems noted above with any software issue.
The most probable solution that comes to my mind is an erroneous codec that does what it wants depending on the reset.
You could do the following:
* set VFO to one of the unwanted signals and listen to the loud tone
* go into the menu and reset codec
* if the tone goes away (and maybe changes its position), the problem has to do with the codec hardware
Where did you buy the codecs that are inside your machine and the machine that you gave to your HAM radio colleague?
Maybe other OMs have an idea how this could be caused by software? I have no clue at all.
73 Frank
|
Title: Re:[not solved] Receiving strong sigs 12.KHZ
Post by: DF8OE on 03. September 2018, 13:05:24
If we
- use same firmware binaries
- use identical settings (default ones or injected via Python)
...and get different behaviour:
I do not see any chance it is a firmware problem.
It would be very exiting to have such a radio here on my desk...
I have listened to the videos and see:
first video:
- you do have a very worse "wrong sideband suppression". Mine uis much, much better - at 30dB higher levels as you mine are "in the noise"
- the spur 12KHz under the signal is missing completely
- the spurs above are missing completely, too
second video differs!!!! (== codec reset changes something).
That looks like an IQ problem - swapped LO signals...
EDIT:
Or other defects in hardware. I have calculated all possible errors in firmware and nothing matches the above results.
I would be interested in the result you get repeating the test (without need of codec reset) with Xlate=off.
EDITEDIT:
I try to look into the future:
Using Xlate=off you do not see any spurs...
vy 73
Andreas
|
Title: Re:[not solved] Receiving strong sigs 12.KHZ
Post by: F4HTX on 03. September 2018, 14:21:02
Hi again Andreas and Frank,
@ Frank : my usual configuration is with -12kHz, so I am not familiar with the +12k test I performed, but I concur on the fact : WTF !!
I did the codec reset, here is the video (Spoiler : you will LOVE this one. This is fun, hold on to your wigs !!) :
https://we.tl/t-XIUe8XkEhX
When the codec reset seems to be efficient, the birdy goes down to where it should stay, in the noise. But, most of the time, it is like the reset is not performed correctly.
It might be hardware problem of the codec. All my components came from large reseller, in this particular case the codec is from Mouser. I might buy from another source and swap the chip. This seems to be a long shot as two mchf with two UI with the same chip problem...
I attached a pic of this area of the UI board I shot on Nov.2017. I will tear apart the mchf and put the UI board under my binocular. Two codecs wrongly soldered… Why not...
@ Andreas : here is the vid you requested with Xlate off
EDITEDIT:
I try to look into the future:
Using Xlate=off you do not see any spurs...
Nope : unfortunately your crystal ball needs upgrade :)
https://we.tl/t-bGBuLDbGj2
73’s
François
|
Title: Re:[not solved] Receiving strong sigs 12.KHZ
Post by: DF8OE on 03. September 2018, 14:47:27
The "twin peaks syndrom" (we called it so) is a result in a shift (1 byte) between I and Q. Under normal usage conditions for the chip (e.g. in mp3 players) it makes no difference if there is a shift or not. So designer of the chip never has recognized that there might be a problem. This shift occurs "by chance". The probility that the issue occurs is dependent on things we cannot identify. Different firmware versions (e.g. only the colour of the fonts was changed) do show higher or lower probility. But there was NO firmware where the issue was always there or never.
So Frank implemented an "autodetection" if the issue is present and if so, it resets the chip as long as it is gone. It seems so that this autodetection does no longer work on all chips. I have not seen this since over a year.
But why you do see 12KHz mirrors when Xlate=off is a miracle. There is nothing that is using 12KHz if Xlate is off. Maybe Frank can interpret this - I cannot.
vy 73
Andreas |
Title: Re:[not solved] Receiving strong sigs 12.KHZ
Post by: WD8BXS on 03. September 2018, 14:52:39
Hello Andreas,
How can it be a hardware issue if you can load a config file in your radio and get it to happen?
If you do not have the issue, send me a copy of your config file and I will test it here.
Chuck |
Title: Re:[not solved] Receiving strong sigs 12.KHZ
Post by: DF8OE on 03. September 2018, 14:59:05
That was an error by me. I never have seen the behaviour François showed in his videos. I have seen a slightly worse suppression at some configs (suppression is only 40...45dB). But the spurs are detectable by dBm display and can be suppressed by resetting the config. That is a problem, too - but it is not the problem which is discussed in the last posts. So I must state I never have had seen the issue at any device using any config.
The problem I have seen is getting lower at higher BW filters (> 3.6K) - the spurs are moving into the noise (included "infected configs"). And they never reached amplitudes like in François' videos.
vy 73
Andreas |
Title: Re:[not solved] Receiving strong sigs 12.KHZ
Post by: DD4WH on 03. September 2018, 14:59:13
Hi Francois, hi Andreas,
thanks a lot for your efforts!
fully agreed ;-). Nice video with a multiresistent codec (although you seem to be a little fast on the encoder, give the automatic IQ correction a little more time to do its job properly before resetting again ;-))
It seems the cause has now been identified as hardware-related. Whether the software can deal with this, is another question. But if the codec is so resistent to reset and the shift seems to be randomly distributed (as can be seen in your two videos having different frequencies with unwanted signals), software correction is not straightforward.
I have seen the same phenomenon with a totally different codec (SGTL5000), so this seems to be common and as Andreas said, would go totally unnoticed in every quality control at the chip factory.
It is really good, Francois, that you built two mcHFs, because that was the final clue that put us on the right track with the codec ;-).
So, yes, hardware check (soldering) or even codec exchange would be the right solution here, I think.
The good thing is, that in the course of the search for a firmware issue I detected a bug in the audio chain, so maybe this will improve suppression of birdies a little more, but that has to be tested. We will keep you informed, if and when a new version is issued.
73 Frank
|
Title: Re:[not solved] Receiving strong sigs 12.KHZ
Post by: DD4WH on 03. September 2018, 15:05:42
Hi Chuck,
it would be very interesting to have a look at your problem and decide whether it has the same cause. Could you make a video of your machine with the following settings ? :
* inject a signal at 10100kHz into your radio
* IQ auto correction ON (because we have to eliminate the possibility that it is just a malaligned IQ phase/amplitude); alternatively you can show me on the video that your IQ phase and amplitude is perfectly manually calibrated at 10100kHz
* set frequency translation to +12kHz
* set display to spectrum display only
* could you tune slowly from 10100kHz to 10126kHz and down to 10074kHz
* after that, could you navigate to one of the unwanted signals and listen to the tone
* then go into the menu and try to restart the codec many times (leave some time after the restart for the automatic IQ correction, one second is enough)
* if it is the same issue, the tone will disappear after a restart (it could be that many restarts are necessary --> see the video by Francois)
Thanks a lot in advance!
73 Frank |
Title: Re:[not solved] Receiving strong sigs 12.KHZ
Post by: F4HTX on 03. September 2018, 15:08:59
Hello Andreas and Frank,
Thanks for the explanations, so, this is what I will do now :
1. Thorough exam of my UI board under the binocular
2. I will swap the codec with one of the ovi-40 kit still not soldered
3. Report to the team…
Many thanks for your dedication folks.
Best Regards,
François
|
Title: Re:[not solved] Receiving strong sigs 12.KHZ
Post by: DD4WH on 03. September 2018, 15:10:59
| Yes, Francois, we are very interested whether the problem goes away with a fresh codec . . . |
Title: Re:[not solved] Receiving strong sigs 12.KHZ
Post by: F4HTX on 03. September 2018, 16:33:23
Hello Frank, Hello Andreas,
I checked with the bino, nothing to report.
I swap my codec, with one new from an ovi-40 kit : not even a small difference ! All the tests are the same.
Too bad, I really thought we found something...
73's
François |
Title: Re:[not solved] Receiving strong sigs 12.KHZ
Post by: DF8OE on 03. September 2018, 16:49:25
Additional sources of devil can be:
- codec voltage supply (both digital AND analog)
- I2S-lines to MCU
- clock input from MCU
- I2C communication (hardware: connections to MCU or software: too fast I2C-speed)
You can try leave away R6 and R7 and/or reduce R20...R22. I never undestood the sense of R6 and R7.
vy 73
Andreas
|
Title: Re:[not solved] Receiving strong sigs 12.KHZ
Post by: F4HTX on 03. September 2018, 17:04:48
Hi Andreas,
Yes, I don't neither understand the pull'up resistors there. This might slow the edge of the signal because of the consumption. I'll monitor that with the scope.
I'll check that later tonight or tomorrow, family time now.
Nice evening,
François |
Title: Re:[not solved] Receiving strong sigs 12.KHZ
Post by: DF8OE on 03. September 2018, 17:25:54
I forgot: power supply of MCU is important, too. If it is 3.1V that might cause problems in external communication, too.
vy 73
Andreas |
Title: Re:[not solved] Receiving strong sigs 12.KHZ
Post by: S53DZ on 03. September 2018, 18:33:59
Andreas et al,
...
You can try leave away R6 and R7 and/or reduce R20...R22. I never undestood the sense of R6 and R7.
...
|
|
but now I am confused. :-\ You have those pull-up and suppressor resistors on the OVI schematics and it works well for me. But for bug-hunting, yes it is worth trying.
I wonder what can be done on mcHF RF PCB rev04 to equalize the LO I and Q path to get better results. Maybe this can help also here. Although in my case all those spurious signals mentioned here are well suppressed.
As an example, (using -12kHz, 2k3 BPF, CW at input):
10.100 = -47 dBm
10.112 = -114 dBm --> -67 dB
10.124 = -101 dBm --> -54 dB
73 Bojan
EDIT
PS: I suppose pull-ups can help holding the defined levels on bus lines when going between Hi-res state. |
Title: Re:[not solved] Receiving strong sigs 12.KHZ
Post by: WD8BXS on 03. September 2018, 18:38:30
Hello Frank,
I did not video, but injecting a signal at 10.100, I hear the same signal at 10.120.
Restarting codec has no effect!
I do not know what you mean by "your IQ phase and amplitude is perfectly manually calibrated at 10100kHz"
I have no idea what you are talking about hihi
I have never seen this talked about.
I have no other tones , just the one at 10.120.
So my issue must be different from Franks.
I have 2 radios that are doing this.
I can't believe that they are defective codecs.
That would be strange. Maybe there are different codecs?? Different specs??
Thanks,
Chuck |
Title: Re:[not solved] Receiving strong sigs 12.KHZ
Post by: DD4WH on 03. September 2018, 19:12:28
hi Chuck,
interesting! An unwanted signal 20kHz up the wanted signal would be something new and well worth putting an eye on ;-).
You could help us solving your problem by making a video with exactly the same settings as stated in my post this afternoon.
Otherwise it would be very hard -if not impossible- to interpret whats going on on your machine.
if you only have one unwanted tone 20kHz up the wanted signal, then I am pretty sure you have a totally different issue than Francois.
73 Frank DD4WH |
Title: Re:[not solved] Receiving strong sigs 12.KHZ
Post by: DD4WH on 03. September 2018, 19:18:12
Hi Bojan,
thanks for your report!
I wonder what can be done on mcHF RF PCB rev04 to equalize the LO I and Q path to get better results. Maybe this can help also here. Although in my case all those spurious signals mentioned here are well suppressed.
As an example, (using -12kHz, 2k3 BPF, CW at input):
10.100 = -47 dBm
10.112 = -114 dBm --> -67 dB
10.124 = -101 dBm --> -54 dB
|
|
Yes, thats the kind of performance that is exactly as coded in the digital filters, which have 60dB of stopband suppression.
The equalization of the I & Q path is done in software, no need to tune the hardware! The automatic IQ correction is capable of 65dB sideband suppression. Did you have the automatic IQ correction switched to ON when you did your measurements?
We also have another minor issue, that is being fixed at the moment, maybe your figures can be optimized by that fix, not sure.
Keep tuned!
73 Frank
|
Title: Re:[not solved] Receiving strong sigs 12.KHZ
Post by: S53DZ on 03. September 2018, 19:45:16
Frank,
yes, auto RX IQ optimization was ON,
FW was "an old one" .47.
And, hm, this was measured on OVI UI with mcHF v04 RF.
73 Bojan |
Title: Re:[not solved] Receiving strong sigs 12.KHZ
Post by: WD8BXS on 03. September 2018, 21:00:54
OH My, I can't believe I typed that hihi
It was 10.100 and 10.112
SORRY!!!
And I must wonder, if there is a firmware out there that says it has fixed it, wouldn't it be a wise decision to at least have a look at it???
Chuck |
Title: Re:[not solved] Receiving strong sigs 12.KHZ
Post by: DB4PLE on 03. September 2018, 21:19:41
Hi Chuck,
these guys unfortunately did not release the source code. So there is nothing to look at for us to learn from. I am very disappointed.
This not how open source should be done (and in breach of the UHSDR open source license) and I don't understand why they do this.
73
Danilo
|
Title: Re:[not solved] Receiving strong sigs 12.KHZ
Post by: WD8BXS on 03. September 2018, 22:13:55
That is a sad situation. I hope things can get better.
It will be a grand day when this issue is solved.
I hope it can be soon.
I am very much considering sending a radio to Andreas for him to check out.
Chuck |
Title: Re:[not solved] Receiving strong sigs 12.KHZ
Post by: DD4WH on 04. September 2018, 08:40:39
Hi!
Could please everyone who still has a problem with unwanted signals at 12kHz spacing try the new daily build 2.9.51 and report back here, please?
Thanks in advance!
For the test please use:
* automatic IQ correction ON
* translate + or -12kHz
73 Frank |
Title: Re:[not solved] Receiving strong sigs 12.KHZ
Post by: F4HTX on 04. September 2018, 09:15:33
Hi Frank,
Congrats : 2.9.51 does the trick !!!
No more birdies on my mchf, I checked +/-6 & 12kHz.
Well, it will have to be confirmed by daily use, but birdies are really gone, not even listenable low in the noise.
Now : what did you do ???
73's,
François
Edit : I carefully listened in +12kHz RXIQon (same as videos)
10.065 birdy -119dBm
10.077 nothing audible
10.089 motor like noise -109dBm
10.100 as usual signal -73dBm
10.113 nothing audible
10.125 nothing (RX IQ on...)
10.137 nothing audible |
Title: Re:[not solved] Receiving strong sigs 12.KHZ
Post by: DD4WH on 04. September 2018, 09:50:37
Hi Francois,
really happy that it solved your problem!
There was a bug in the audio chain, formerly it was like this:
* light low pass filter with only 4 taps @48ksps sample rate
* downsampling-by-4
* heavy Hilbert lowpass filter with +45 and -45 degrees phase shift for I & Q respectively @12ksps sample rate, 199 taps
I changed it to:
* Hilbert lowpass filter with +45 and -45 degrees phase shift for I & Q respectively @48ksps sample rate, 121 taps
* light low pass filter with 4 taps
* downsampling-by-4
The Hilbert lowpass filter with 121 taps prevents the aliasing and suppresses the unwanted signals. Thats DSP standard, so its amazing that it remained unnoticed for so long. Probably my fault :-).
I suspect, the implemented change will also lead to a slightly worse sideband suppression, but as long as nobody notices it, . . . ;-).
In the next days I will finetune a few more things around this bugfix.
Please report anything that could be related to this bugfix.
73 Frank |
Title: Re:[not solved] Receiving strong sigs 12.KHZ
Post by: DD4WH on 04. September 2018, 09:55:54
Francois: thanks for your measurements!
The only remaining birdy 24kHz below the wanted signal is due to IQ imbalance, so has a totally different cause than the aliases caused by the fixed filter problem.
73 Frank |
Title: Re:[not solved] Receiving strong sigs 12.KHZ
Post by: F4HTX on 04. September 2018, 10:10:51
Hi Frank,
Anti-aliasing is always a source of unexpected problems...
Thanks for the explanation, for sure aliasing on a downsampled signal was a little rude ;D
Anyway I wonder why this appears suddenly on some of the mchfs and not others. There still might have some dependency on hardware implementation.
73's
François
@ Andreas : I will carefully check the SDIN & SCLK signals and publish, but a few time later. Pro's awaiting... |
Title: Re:[not solved] Receiving strong sigs 12.KHZ
Post by: S53DZ on 04. September 2018, 10:24:05
| Anyway I wonder why this appears suddenly on some of the mchfs and not others. There still might have some dependency on hardware implementation. |
|
I am asking myself the very same question.
No matter if I had no troubles with aliases here (perhaps because I did not use a wide-bandwidth BPFs > 2k7?) I will do some more measurements to check the behavior after implementing FW 2.9.51 in due time.
73 Bojan |
Title: Re:[not solved] Receiving strong sigs 12.KHZ
Post by: WD8BXS on 04. September 2018, 13:30:02
Well Done, Well Done,Well Done,Well Done,Well Done,Well Done,!!!
That fixed it here as well.
Congratulations to all who helped!!
73, de Chuck WD8BXS |
Diskussions- und Newsboard des DARC-Ortsverbandes I40 | Powered by YaBB SE
© 2001-2003, YaBB SE Dev Team. All Rights Reserved.
|