Topics

locked PC Panadapter using G90 I/Q output.

W5DXP
 

The noise in the graphics that I previously posted is apparently a ground loop problem among two PCs and the G90. The following panadapter setup works with minimum noise:

G90 I/Q-->--USB Sound Card Dongle -->--Laptop running SDR#

Bob Bryce
 

Thanks for the information W5DXP.
Did you spot that I believe there may also be a problem with only one PC connected - but when TWO cables are connected to the G90....
If I'm running just the IQ output into the computer - no problem - but when I also connect up the CAT cable from the G90 head socket to the computer USB, then the noise level on the IQ output immediately rises. Unfortunately I'm not sure on the source of the noise - it MIGHT be a ground loop somewhere inside or external to the G90, or it just might be noise radiated from the data signals in the USB cable. The G90 is an extremely sensitive instrument when it comes to radiated noise, so I guess I need to do some more experiments....  :-)

W5DXP
 

Bob, I don't see any noise difference between having the CAT cable connected or not. This graphic was taken with the CAT cable connected and FLRig running. Something I don't understand is why the LSB region is a mirror image of the USB region, i.e. the conditions above 0 are duplicated below 0. The large "signals" give a clue to the I/Q bandwidth.

Bob Bryce
 

Hi W5DXP,
Well we certainly are having a different experience.
1. I see no mirror image
2. I see no 'DC' / Low frequency information, unless I also plug in the CAT cable from the display head to the computer.

Please see attached video that shows what I think to be a correct display of FT8 signals in this case.

I believe that the IQ data should be a 24kHz audio bandwidth and that the computer sampling rate you use should be 48,000 samples per second, or higher.

 

Ok folks .. a question: When you plug into the I/Q output for an external display ... does your G90 display change slightly? I'm plugged into what may well be a mono mic input USB new sound card. Useing SDR# or HDSDR I do get a display of the FFT, Waterfall, etc. that looks reasonable. But I've noticed my G90 FFT & waterfall display looks the same on both the left and right sides. In other words the right half of the G90 display is the mirror image of the left half. If I disconnect the cable from the I/Q output the G90 display reverts to normal.

Bob Bryce
 

No change on mine Gary, so it may be a set-up or cabling issue. I wonder if you are accidentally shorting out one of the I or Q signals????

regards
Bob, MM0MMQ

w9ran
 

On 8/21/2019 6:39 AM, W5DXP wrote:
The noise in the graphics that I previously posted is apparently a ground loop problem among two PCs and the G90. The following panadapter setup works with minimum noise:

G90 I/Q-->--USB Sound Card Dongle -->--Laptop running SDR#
I also have been experimenting with the IQ output for several days, and cannot report much success.   In fact I am convinced the radio has some serious issues in this regard, having done similar soundcard SDR connections for a number of years and never running into similar problems  Here is a summary of my findings - please refer to the image here:

https://i.imgur.com/VNETv6Y.jpg

My setup is identical to that listed above, in my case the soundcard is a StarTek ICUSBAUDIO interface which is recommended by HobbyPCB for their HF-RFIQ and which works perfectly with it and my homebrew SDRs up to 96kHz, two channel IQ output.    I am using a 6" long good quality stereo audio cable (many are very poorly shielded) and a Bienno power battery.

Top screen:  Baseline configuration is shown in the top screen: No antenna, battery power - thus the G90 is completely isolated except for the USB interface to the PC.  IQ display is clean and normal in appearance for no-signal condition.

Middle  screen:   This shows what happens when just an antenna is connected (HyTower vertical which is a ground-mounted vertical).     The noise comb in the middle of the screen (around the 0 Hz position) extends out to at least 9 kHz on either side and the numerous spurs result in a constant audio tone in the background.   Signals are present as can be seen but there's a problem - note there are mirror images on each side, indicating there is little or no difference between the I and Q signals thus SDR# is unable to provide proper sideband suppression.   (Yet I can verify the soundcard faithfully produces both channels with good separation, and I've found no way in Windows to make a stereo signal into a mono one).    I've found that this happens about half the time - for reasons unknown.

Bottom screen:  After starting and stopping SDR and unplugging the audio cable, it now shows proper IQ cancellation and the unwanted images are gone.   You can tell that the SSB signals are now all USB although anything near 0 Hz is buried in the crud. I noticed the degree of IQ separation would slowily diminish  over time to the point where images were again seen and determined that this is caused by the IQ correction algorithm in SDR# being overloaded by the amount of crud that is present.   IQ correction is normally turned on for all zero-IF receivers since it will eliminate the center spike, but must be disabled to prevent SDR# from displaying images.   Turning it off does nothing about the center crud or the images at the +/-24 kHz end of the spectrum.   In the bottom creen  I've also tuned SDR# to around -10kHz where it is acting as a low-IF receiver, but a few weak noise spurs are still seen even past 10 kHz.

I'm pretty certain the problem is caused by ground loops - either an antenna, grounded power supply, or other connection between the G90 and outside world will cause this mess to occur.  I have also seen times when the noise can be seen as a spike or hump in the center of the G90 LCD screen as well, which  suggests poor isolation of the IQ signals inside the radio and a lack of buffering that would isolate the radio from the outside world. That may in fact be the solution - I've tried simple capacitive coupling which does not help, however an active buffer might (although it would have to be powered by the G90 side).    The G90 works so well there's no way this junk is present except when the ground loop exists via some external connection.

I also think the level (amplitude) of the IQ signals is low and this contributes to a poor SNR.    I can also confirm that contrary to what is stated in the "improved manual", the "Aux Out Vol." adjustment that is accessed via 2nd-FUNC does NOT have any effect on the level of the IQ signals.   Apparently they do set the analog audio levels but I haven't confirmed that.   Ideally the level of IQ output would also be adjustable from a G90 menu.

The noise about 0 Hz makes the IQ output unusable as a panadaptor, although if I set SDR# to operate as a low-IF receiver at say, 12 kHz IF,  the results are usable since the noise spurs are not audible that far from zero.   But this is not what I anticipate when a radio provides IQ output, and even the simplest QSD receiver has a cleaner output (generally noise around 0 Hz is a function of clock noise and jitter).  I'm very interested in comparing results with any others who have this same problem.

73, Bob W9RAN

Chris Chatelain
 

I noticed that certain cable combinations through the CE-19 adapter will cause that mirroring. I havn't narrowed down which one, but if I only have the i/q plugged in, it's not duplicating the big spikes across the zero point. I need to stick more audio isolation transformers on things and choke some stuff near the sound cards.

--Chris

John H. Long Jr.
 


 The cable, adapter or plug may be converting stereo to mono, or the input to your computer may be mono.


Sent from my Verizon, Samsung Galaxy smartphone

-------- Original message --------
From: Bob Bryce <bob.bryce@...>
Date: 8/21/19 12:03 PM (GMT-07:00)
To: XieguG90@groups.io
Subject: Re: [XieguG90] PC Panadapter using G90 I/Q output.

No change on mine Gary, so it may be a set-up or cabling issue. I wonder if you are accidentally shorting out one of the I or Q signals????

regards
Bob, MM0MMQ

w9ran
 

On 8/21/2019 12:33 PM, Gary Sargent wrote:
I'm plugged into what may well be a mono mic input USB new sound card
You'll be able to demodulate signals Gary, but without both I and Q channels (which your soundcard treats just as if they were left and right stereo audio), you won't be able to suppress the images, which will appear as mirror-imaged versions of the desired signal on the opposite side of 0 Hz.    It's the magic of DSP that allows USB and LSB to be separated (as well as other types of modulation) - but you've got to have both I and Q (quadrature) signals.

Here's the USB soundcard I'm using, which I know has stereo inputs:

https://www.amazon.com/StarTech-com-Sound-SPDIF-Digital-Stereo/dp/B00F7120TQ

If anyone has other recommendations I'd like to know of them. For the Raspberry Pi, the Fe-Pi can't be beat: https://fe-pi.com/products/fe-pi-audio-z-v2

73, Bob W9RAN

 

Thanks for the comments and guidance. I'm sure my laptop sound card and an inexpensive USB sound card are both a mono on the mic inputs. I've just ordered the StarTech model that you use. I like that it looks to be physically small .. Always trying to minimize the clutter!

W5DXP
 

I've been making a rash assumption and that is - that the USB sound card dongles have stereo microphone inputs - I was wrong. The mirror image problem is caused by a mono microphone input which gives only one line of the I/Q signal. My desktop PC has stereo line inputs but my laptop doesn't and neither do any of my three USB sound card dongles. W9RAN above indicates a USB sound card with stereo inputs.

W5DXP
 

Here's my latest results. Note the extreme noise is still there when I turn the G90 off. It appears that the I/Q output needs to be amplified to rise above the noise. On the graphic, time goes from bottom to top.

hjhsoft@...
 
Edited

I've been following this topic and I've made a short video on how the spectrum should look when everything is propperly configured.
https://youtu.be/p_U8rK2N4ZE

Let me start with a short summary of my findings with regard to questions I've seen in this topic.
If you see a signal mirrored around 0Hz (so the left of 0Hz is identical to the right) this is caused by a missing signal. 
You are only processing an I or Q signal. This can have multiple causes, the most common one is using a mono cable or mono audio input

Secondly you need to connect the I/Q signal to a propper soundcard, most (onboard) soundcards generate noise which is fed back into the I/Q output (G90 signal is deteriorated now also) , they have no or little decoupling in the input and/or create a high load on the I/Q output (impedance mismatich). It seems there is no (strong) driver on the I/Q signal which might result in signal loss when connected to a "bad" sound card.

The output levels of the I/Q signals should not pose a problem to a normal Line-In

You need to swap the I/Q signals otherwise the band is upside down.


I've searched arround for an affordable soundcard which supports 192kHz/24bit sampling with a (very) low noise figure, it turns out there are not that many.
The one I'm using and would sugguest you to buy (if you don't have a 192kHz/24bit soundcard allready) is the Asus Xonar U7 MKII.
Sadly you need to install the 120MB software pack before it propperly works and you need to manually adjust some settings.

The provided CAT cable generates noise as mentioned by others.
I've been doing some testing/debugging on this and it seems that I'm unable to filter this noise externally.
It might be possible that the noise is generated internally, I've ruled out a ground loop by using a USB isolator and audio transformers.
I've also applied filtering by adding ferrites on the signal and usb cable with multple turns which have had very little impact on lowering the noise.

Here is an example of a missing I or Q signal (mono input, broken stereo cable, mono cable), as you can see the left of the screen is mirrored onto the right.

No antenna connected only G90 I/Q output noise and 50Hz hum.

Bob Bryce
 

Hi hjhsoft,
I found this to be a very useful post. Thank you.

The confirmation of the CAT USB noise 'problem' was helpful, even if we don't yet have a solution.

On the spec for a sound card, I had understood that the IQ signals had a 24kHz bandwidth. If that is correct then a sample rate of 192Hz may be overkill. Although my audio interface is specified for 192, I get acceptable results at a sample rate of 48. This will open up the number of acceptable audio interfaces / sound cards available on the market.

73s
Bob
MM0MMQ

 

I received my StarTech ICUSBAUDIO2D USB sound card today and have spent a few hours evaluating it. Some comments:
1. I use an external powered USB hub. I have the G90 CAT cable plugged into it, a mono sound card for ACC port I/O and the StarTech sound card for I /Q..
2. When plugged as above, the noise in HDSDR is visable and very bad ... receiving signals is next to impossible.
3. When I unplug the G90 CAT cable, the noise reduces some, but is still unworkable.
4. When I unplug the mono sound card I use for the ACC port, the noise goes away and the HDSDR display looks very good .. much like hjhsoft video, above. The StarTech is stereo mic input .. my prior mirror image display problem is gone and the display is now proper. The StarTech will sample at 96000 and using that the bandwidth displayed is close to 100 KHZ. Nice!
5. My laptop PC is a bit older with a slower Celeron pocessor. SDR# is tasking the PC pretty good ... when other things are running SDR# may bog down. HDSDR is less of a problem in his regard. I prefer HDSDR for this and other reasons. I've eliminated the powered USB hub and the noise remains. 
6. I had hoped that all of these USB devices woud play together, but as it stands I can only view the expanded spectrum I/Q G90 display if the G90 CAT and ACC USB cables are disconnected. 
7. This is usable and worthwhile, but a far cry from my Yaesu FT450D and SDR Play RSP1 running SDRUno on my other PC where I can display the full 40 meter band and use the mouse to tune the 450D by clicking the SDR Uno spectrum display.
8. Sorry for the long post .. the main point is that I've confirmed, as others have stated, the CAT and other USB sound cards as somehow interferring with using the otherwise fine I / Q output port on the G90. 

 

A bit of elaboraton at what is possible with the G90 I /Q properly connected to a PC but without the G90 CAT connected (due to the vertical lines noise that it causes)

1. I'm running HDSDR on my PC and set for an input and output bandwidth of 96000
2. I tune the G90 manually to 7150 LSB (for example)
3. I change the multi-function control to tune in 100 KHZ steps
4. I set the HDSDR frequency to 7150 LSB (to match the G90). Ths HDSDR display now shows a frequency of 7150 smack dab in the center of the display with about 7100 to the left and 7200 to the right.
5. I can use the PC's mouse to click on any of the signals that are on the HDSDR display. Then listen using HDSDR to demodulate the signal and route it through the PC speakers, not the G90.
6. Rotating the MFC a click up or down retunes the G90 by 100 KHZ and the I/Q HDSDR fft display changes accordingly (but the displayed frequency scale in HDSDR does not change, unless you manually change it).
7. Admittedly, this is mostly useful for just receiving. Say you spotted a SSB signal that you wanted to contact. You would have to manually tune in that frequency on the G90 first.
8. Lastly, I took a look at the I/Q signals with my oscilloscope and saw levels in the + / - 50 to 100 mv range. Makes me wonder if some sort of preamp on the I /Q would help. But then I believe USB sound cards with a mic input are set up for low level signals as the I /Q appears to be.

John H. Long Jr.
 

Are there any adjustments / settings for the I/Q output levels?
 
Did you happen to see how much bandwidth the I/Q output covers?
 
We may have to use something like the "WolphiLink".
I wonder what levels it expects?
 
 
 
John H. Long Jr.
KW7A
1370 N 150 E
Nephi, UT 84648
USA
 
(435) 888-LONG
 
 

--------- Original Message ---------
Subject: Re: [XieguG90] PC Panadapter using G90 I/Q output.
From: 'Gary Sargent' <GarySargent@...>
Date: 8/24/19 6:40 am
To: XieguG90@groups.io

A bit of elaboraton at what is possible with the G90 I /Q properly connected to a PC but without the G90 CAT connected (due to the vertical lines noise that it causes)

1. I'm running HDSDR on my PC and set for an input and output bandwidth of 96000
2. I tune the G90 manually to 7150 LSB (for example)
3. I change the multi-function control to tune in 100 KHZ steps
4. I set the HDSDR frequency to 7150 LSB (to match the G90). Ths HDSDR display now shows a frequency of 7150 smack dab in the center of the display with about 7100 to the left and 7200 to the right.
5. I can use the PC's mouse to click on any of the signals that are on the HDSDR display. Then listen using HDSDR to demodulate the signal and route it through the PC speakers, not the G90.
6. Rotating the MFC a click up or down retunes the G90 by 100 KHZ and the I/Q HDSDR fft display changes accordingly (but the displayed frequency scale in HDSDR does not change, unless you manually change it).
7. Admittedly, this is mostly useful for just receiving. Say you spotted a SSB signal that you wanted to contact. You would have to manually tune in that frequency on the G90 first.
8. Lastly, I took a look at the I/Q signals with my oscilloscope and saw levels in the + / - 50 to 100 mv range. Makes me wonder if some sort of preamp on the I /Q would help. But then I believe USB sound cards with a mic input are set up for low level signals as the I /Q appears to be.

 

No adjustments to the I / Q levels that I know of. My sound cards' max sample rate is 96khz and I suppose that's why I see a band width of the approx. 100khz. Others have suggested a higher sample rate sound card might yield a wider bandwidth. But I'm not too informed on that. I wonder if a low noise preamp is used to raise the I / Q signals to something like a line level and then apply that to a USB sound card that had line level inputs .. maybe that would raise the I/Q level enough above the noise that it may be usable with the CAT at the same time. But, I don't plan to do so.

John H. Long Jr.
 

In reference to the Pre-Amp suggestions: 
     I've ordered some (dirt cheap) TS-472 single channel pre-amp(s).

What I plan on doing is use the pre-amps on the I/Q outputs from the G90 and see what happens.
The Pre-amps have adjustment for 0, 10, 20, 30 40 dB gain, so I can also play with the level of gain that might work.

Wish me (us) luck.
73