Title: OVI 40 canal I/Q (RX)
Post by: SP3OSJ on 13. April 2018, 12:30:10
Sound in OVI40 channels is different for channel I for channel Q
Is this good?
Maybe this is the reason why RX has more noise in OVI40 than with the mcHF controller.
A movie with this problem:
https://youtu.be/moNGCHZxbyM (https://youtu.be/moNGCHZxbyM) |
Title: Re:OVI 40 canal I/Q (RX)
Post by: DB4PLE on 13. April 2018, 13:44:56
Hi,
interesting, but you do think this is a problem of your particular hardware or do you think this is a OVI40 design problem (it could also be a software problem, but I checked the code and it is identical in that respect for mcHF and OVI40)?
73
Danilo
|
Title: Re:OVI 40 canal I/Q (RX)
Post by: DF8OE on 13. April 2018, 14:05:49
Hm. All of my investigations are showing noise of OVI40 is much lower than of mcHF and sound of I and Q channel is similar... How have you got your suggestions?
vy 73
Andreas |
Title: Re:OVI 40 canal I/Q (RX)
Post by: SP3OSJ on 13. April 2018, 14:22:02
After connecting the PCB RF the radio is working properly.
A good SSB signal (USB and LSB) is decoded
PCB RF connected to the mcHF controller gives less noise
than the same RF PCB connected to the OVI40 controller.
Why?
LCD is more noise and more noise other processor?
Is this a software error?
Check if the controller has the same noise (I/Q) at the converter A/D (RX) input (speaker).
I will be fine if I have a problem :)
|
Title: Re:OVI 40 canal I/Q (RX)
Post by: DF8OE on 13. April 2018, 17:59:10
I do have mcHF v0.4 and OVI40 with new alpha stage rf pcb connected via 3dB splitter to the same antenna. If there is a signal I cannot read 100% at mcHF I always can read this signal without any problems on OVI40. mcHF does have "helicopter noise" on many band regions - OVI40 does not have. I have not made measurements yet but subjective impression is that OVi40 performs >= 6dB better than mcHF.
I do not have the possibility to connect mcHF rf PCB because I do not have one which is not fitted in a case.
vy 73
Andreas |
Title: Re:OVI 40 canal I/Q (RX)
Post by: SP3OSJ on 13. April 2018, 19:33:51
Adreas, this is all true.
OVI has a better receiver signal.
However, the display shows something else.
Look:https://www.amateurfunk-sulingen.de/forum/index.php?board=18;action=display;threadid=1030 (https://www.amateurfunk-sulingen.de/forum/index.php?board=18;action=display;threadid=1030)
If someone has assembled VIO40 controller and speaker (headphones) please let's check to see if the same noise is touching the pin 4 and 5
OK? |
Title: Re:OVI 40 canal I/Q (RX)
Post by: DB4PLE on 13. April 2018, 20:18:54
Hi,
just did an experiment:
Connected same RF board to dummy load, and then use a mcHF 0.4 UI 3.2 display SPI 320x240 and a OVI40 UI parallel display with 480x320 on it using the same frequency and mode, filter etc. settings.
dbm/hz for mcHF 0.4 -> -144, OVI40 -> -123 . But wait: Now I applied a well known source of noise (an antenna)
dbm/hz for mcHF 0.4 -> -114, OVI40 -> -98
Assuming that both get same signal. Now I calibrate both to show -98 dbm/Hz. And repeat above measurement, now (no surprise here) the mcHF has -128 dbm/Hz and the OVI40 has still -123. That is only 5db difference. While still significant, nowhere near has high as shown.
Unfortunately, my display & heat sink construction makes it difficult to measure the individual channels as asked for.
73
Danilo
|
Title: Re:OVI 40 canal I/Q (RX)
Post by: DF8OE on 14. April 2018, 09:05:11
I have done an experiment, too:
OVI40-TRX (with OVI40-RF alpha stage fitted) and mcHF v0.4.
Both set to 7.120KHz.
Now I applied -73dBm from signal generator first to mcHF and then to OVI40. Because of I have calibrated both they show me S9.
Now I disconnect signal generator and connect 50R dummy load.
OVI40 shows -118dBm, mcHF shows -113dBm.
Now I disconnect dummy load and operate with "open antenna connector".
Noise shown at dBm reading is increasing at both devices ~2...3 dB. OVI40 still gives "white noise" from speaker - but mcHF gives "schrapp-schrapp-schrapp-schrapp" (helicopter noise)...
Last I conect my real antenna (double zepp 2 x 27m matched via CG3000):
OVI40 and mcHF now both show -85dBm.
I cannot see any "higher noise" on OVI40 and compared with real signals OVI40 performs better under all circumstances.
When I go to higher bands effects are moving in parallel on mcHF and OVI40 (noise is getting smaller on both). But the decrease is proportional.
When I inject audio via 30p header both channels I and Q do have same behaviour.
vy 73
Andreas |
Title: Re:OVI 40 canal I/Q (RX)
Post by: DB4PLE on 14. April 2018, 10:03:36
Hi,
btw, I should have mentioned, that I have used a OVI40 H7 to run the test, not the normal OVI49 F7 (which I have in a case, so it is hard to use the same RF board for testing. The STM32H7 runs with 400 Mhz and this indeed might cause some extra noise being generated.
73
Danilo
|
Title: Re:OVI 40 canal I/Q (RX)
Post by: SP3OSJ on 14. April 2018, 12:41:38
Andreas is your mcHF does he have a helicopter? all bands, all freq.
The effect on the movie causes the function:
Menu/Configuration Menu/RX IQ Auto Correction -> ON
|
Title: Re:OVI 40 canal I/Q (RX)
Post by: DD4WH on 14. April 2018, 13:43:56
Hi Artur,
very interesting observations that you made!
I am not sure if you observe the same thing if auto IQ correction is OFF?
The auto IQ phase and amplitude correction algorithm is programmed to be quite slow. It needs nearly one second until it is settled, because of the small alpha in the lowpass filter which smooths the determination of phase and amplitude errors.
The difference in reaction of mcHF and OVI40 UI PCBs could be explained by different intitial phase and amplitude errors introduced by the PCB layout --> just guessing here.
However, that cannot explain why Andreas measures 5dB better performance of OVI40 and Danilo measured 5dB better performance of mcHF . . . but that may be the interaction between UI and RF boards? Andreas uses his own RF board and Danilo uses mcHF RF board. Too many variables here ;-).
All the best 73s
Frank DD4WH
|
Title: Re:OVI 40 canal I/Q (RX)
Post by: DD4WH on 14. April 2018, 14:03:15
oh, just had two other ideas!
1. PROBLEM WITH DIFFERENT FFT SIZES FOR SPECTRUM DISPLAY COULD INFLUENCE MEASUREMENT PERFORMANCE
mcHF UI --> small TFT --> spectrum has resolution of 256 (small FFT)
OVI40 UI --> large TFT --> spectrum has resolution of 512 (large FFT)
Maybe we have a problem with our measurements of the signal and not with the UI board influencing the RF performance?
I have not deeply thought about how a doubled FFT size might influence the dBm measurements, but my suspicion is that the cause of the different figures could be the software measurement procedure itself.
We have to look at the dBm code and also perform test measurements with very low signal strengths.
2. DO NOT BELIEVE IN THE ACCURACY OF DBM MEASUREMENTS IN SSB BANDWIDTH WHICH ARE LOWER THAN -102dBm
Christian, DL9NL has shown nicely and very precisely that the mcHF is capable of measuring signals down to -102dBm (SSB-bandwidth 2.4k) accurately --> forget measuring lower signals in SSB bandwidth accurately! In CW bandwidth (300 Hz), signals can be measured accurately down to -119dBm, but not lower.
http://funkamateure-dresden-ov-s06.de/index.php?article_id=462&clang=0
But in our thread here, signals in SSB bandwidth are compared which are as low as -123dBm (or even lower) --> let us not believe these figures. The mcHF/OVI40 is NOT a measurement receiver capable of measuring accurately such extremely low signals . . .
73 Frank DD4WH
|
Title: Re:OVI 40 canal I/Q (RX)
Post by: SP3OSJ on 14. April 2018, 14:07:20
Franc
Andreas measures mcHF on the "helicopter".
I do not have a helicopter at mcHF
I do not have a helicopter in OVI40
Danilo and Reto (HB9TRT) has good measurements.
My opinion is this.
If Andreas does a "super" RF to VIO40, then this "super" RF PCB will be even better on the mcHF controller - if it's compatible. |
Title: Re:OVI 40 canal I/Q (RX)
Post by: DF8OE on 14. April 2018, 16:42:08
1) It is not compatible because of it uses more I2C and SPI busses.
2) I cannot compare both rf boards but I can state that the bottleneck is the antenna and the noise which comes from it (and its resistance). This is much much higher than any influence of hardware.
vy 73
Andreas |
Title: Re:OVI 40 canal I/Q (RX)
Post by: SP3OSJ on 14. April 2018, 17:52:59
mcHF / OV40 has very poor converters AC/CA WM8731 (TLV320AIC23), CS4272 is much better.
See this movie:
https://www.youtube.com/watch?v=mrf9BsNrIHc&t=318s (https://www.youtube.com/watch?v=mrf9BsNrIHc&t=318s)
Listen, what's the noise. |
Title: Re:OVI 40 canal I/Q (RX)
Post by: DF8OE on 14. April 2018, 18:16:27
Artur I do not think they are very poor. Of course CS4272 does offer much higher sampling rates - but they do not have antialiasing filters.
As I already stated: it does not make any sense to discuss about noise which is covered by thermal noise of 50R antenna. I am nearly sure you will not hear any difference if you use CS4272 instead of WM8731 under real working conditions.
vy 73
Andreas |
Title: Re:OVI 40 canal I/Q (RX)
Post by: DD4WH on 14. April 2018, 18:22:41
The discussion should be backed up by real measurements and not be based on the dBm values on the TFT screen. See my last post.
73 Frank |
Title: Re:OVI 40 canal I/Q (RX)
Post by: DF8OE on 14. April 2018, 18:29:58
I am unable tto do that with my "flying constructions" here. But I do not see any reason to doubt here. OVI40 with its rf PCB has the best receiver compared to different Kenwood, ICOM and Yaesu TRX and mcHF I do have here...
mcHF is known to have a very "sensitive RX" - you can find these statements confirmed by many different OMs. OVI40 in comparison can read signals which are unreadable with mcHF - so what?
Of course I will make measurements when everything is working on PCBs and can be fitted to a case. But I do not think it will get worse if it is on PCBs and in a case!
vy 73
Andreas |
Title: Re:OVI 40 canal I/Q (RX)
Post by: DD4WH on 15. April 2018, 11:35:08
Hi Andreas,
you probably misunderstood me.
I was assuming this thread was about (assumed) differences in IQ signal processing in OVI40 and mcHF (and not about comparing RX properties in mcHF RF board and your prototype RF board).
And those IQ "measurements" made here in this thread are not valid, as I stated in my last post, because they have been made with the internal dBm display.
So I would like to see real measurements on those IQ signals, before I can believe there is something wrong in software or hardware.
BTW, the internal dBm display does this:
for (int c = (int)Lbin; c <= (int)Ubin; c++) // sum up all the values of all the bins in the passband
{
sum_db = sum_db + sd.FFT_Samples[c];
}
This means that measurements with 256-point-FFT --> (dBm display in mcHF UI board) and 512-point FFT (dBm display in OVI40 UI) will always differ by 3dB, because the 512-point-FFT has twice the number of bins to sum up ;-).
So we can take home two messages:
* do not compare dBm measurements between mcHF UI and OVI40 UI
* do not believe in measured signal strengths lower than -102dBm in SSB bandwidth or lower than -112dBm in CW bandwidth
All the best 73s
Frank DD4WH
|
Title: Re:OVI 40 canal I/Q (RX)
Post by: DF8OE on 15. April 2018, 11:40:05
Hi Frank,
I understood your post completely 8) . My answer was directed to Artur who assumes that mcHF has better low signal performance of mcHF compared to OVI40.
I cannot reproduce anny differense in IQ handling between mcHF and OVI40 - and that is not astonishing at all:
both use same audio codec and same firmware - why should there be any differences? I am working (developing) with OVI40 UI since half a year and if there is an issue regarding this I 100% would have noticed that.
vy 73
Andreas |
Title: Re:OVI 40 canal I/Q (RX)
Post by: SP3OSJ on 15. April 2018, 12:08:29
I wanted to write that if there is any disturbance, the receiver does not receive the signal.
If the interference is not present, the receiver reads the signal.
The small display does low noise, the large display (more pixels) makes big noise.
Just enough. |
Title: Re:OVI 40 canal I/Q (RX)
Post by: SP9BSL on 15. April 2018, 12:48:58
Hi,
This means that measurements with 256-point-FFT --> (dBm display in mcHF UI board) and 512-point FFT (dBm display in OVI40 UI) will always differ by 3dB, because the 512-point-FFT has twice the number of bins to sum up ;-).
|
|
You're right, and that's why Danilo with 6th reply wrote dBm/Hz which normalizes all measurements:)
* do not believe in Measured signal strengths lower than -102dBm in SSB bandwidth or lower than -112dBm in CW bandwidth
|
|
fully agree
I have other proposal: we can use PC via CAT/DIQ OUT and measure received noise without LCD plugged in for OVI/mcHF. With PC we can have deeper FFT (more acurate) than the one in UHSDR.
Will do experiment with slowdown of PARALLEL bus for LCD (it has VeryHigh speed mode on GPIO).
|
Title: Re:OVI 40 canal I/Q (RX)
Post by: SP3OSJ on 15. April 2018, 13:35:48
| Is the receiver that has -102dBm/2.7kHz a good receiver on the 50/70/144/432MHz band? |
Title: Re:OVI 40 canal I/Q (RX)
Post by: DF8OE on 15. April 2018, 15:09:02
My rf front end does have a preamp for the higher bands with a noise factor of 0.8dB and a gain of 18dB. I agree to Franks post that there is no reliable measurement above ~ -100dBm. So I compared sensitivity and noise with commercial analogue TRX like my IC211E on 2m. OVI40 does perform nealy similar and I think this is very good.
BTW:
I do not know how you get -102dBm. My OVI40 shows -118dBm - and if I preamp VHF with 18dB this will be added (all "in theory". At least it has same sensitivity as analog IC211E).
vy 73
Andreas |
Title: Re:OVI 40 canal I/Q (RX)
Post by: DD4WH on 15. April 2018, 17:48:25
@Slawek: | You're right, and that's why Danilo with 6th reply wrote dBm/Hz which normalizes all measurements:) |
|
You are right, dBm/Hz-display does normalize and account for filter bandwidth. BUT the figure is still wrong: the damage has already been done, because the sum that is calculated for the dBm AND the dBm/Hz display is dependent on the number of FFT bins. It is a bug in the software, that is what I tried to explain in my last post (see line 2180 to 2192 in spectrum.c). We have to substract 3dB from the reading when we use an FFT of size 512, because the intercept constant in the dBm code has been empirically determined for an FFT size of 256 !
| I have other proposal: we can use PC via CAT/DIQ OUT and measure received noise without LCD plugged in for OVI/mcHF. With PC we can have deeper FFT (more acurate) than the one in UHSDR. |
|
That is a very good idea, not to use the dBm display but an SDR program to deliver measurements of signal strength.
@Artur: please read carefully: I said:Christian, DL9NL has shown nicely and very precisely that the mcHF is capable of measuring signals down to -102dBm (SSB-bandwidth 2.4k) accurately --> forget measuring lower signals in SSB bandwidth accurately! In CW bandwidth (300 Hz), signals can be measured accurately down to -119dBm, but not lower.
http://funkamateure-dresden-ov-s06.de/index.php?article_id=462&clang=0 |
|
I never mentioned a receiver having an MDS of -102dBm
@Andreas: this thread is about a potential IQ problem with the OVI40 UI PCB and not about the prototype RX PCB :-). Nice to hear about the features of that OVI40 RX PCB prototype (looking forward to it ;-)), but we cannot reproduce your results (because we do not know your prototype) and they are unrelated to the potential IQ problem.
So, if somebody thinks we have an IQ problem with the OVI40 UI vs the mcHF UI, please provide reliable data that we can reproduce. At the moment I cannot see a problem that is described accurately enough to even identify and reproduce it.
Have fun with the UHSDR,
73 Frank
|
Title: Re:OVI 40 canal I/Q (RX)
Post by: DB4PLE on 15. April 2018, 18:08:00
Hi Frank,
it is absolutely no problem to account for the FFT size and reduce the 512 FFT values accordingly so that both values are comparable.
This is a no brainer. On the other hand, what I did was to calibrate both so that I measured the same value for the same more or less white noise signal from antenna and that should take into account the dbm difference, shouldn't it?
73
Danilo
|
Title: Re:OVI 40 canal I/Q (RX)
Post by: DD4WH on 15. April 2018, 18:38:17
Hi Danilo,
yes, you are right, I missed that bit of your post.
If you calibrate the 256FFT and the 512FFT to the same signal, their subsequent measurements should be comparable.
I remember we already had a thread on S-Meter reading confusions ;-). Must be somewhere in this forum.
If time allows, I will have a look into other peoples code, how they solved the problem of S-meter readings and if there is potential for improvement (it seems many different measurement types are used even in semiprofessional RXs). First of all, we should add a 3dB subtraction in the FFT512 dBm code.
73 Frank |
Title: Re:OVI 40 canal I/Q (RX)
Post by: DF8OE on 15. April 2018, 18:43:33
@Frank
Yes, so it is. Addition of 3dB will do the job. And all my postings have one aim: To tell that there is no I/Q problem in hardware or firmware and that there is not "more noise" on OVI40 as on mcHF. I will stop this discussion now because there is nothing to investigate related to both of this.
vy 73
Andreas |
Diskussions- und Newsboard des DARC-Ortsverbandes I40 | Powered by YaBB SE
© 2001-2003, YaBB SE Dev Team. All Rights Reserved.
|