Softrock/PowerSDR Anomalies (AKA Bugs) - Distortion on AM - Anyone else notice?

[steve_qix]

I've been using a Softrock (455kHz) coupled to the 1st 455kHz stage of a 75A3 receiver for about a month.  I'm using the latest version of the PowerSDR software with this setup.  The computer hardware is an old Dell Optiplex 1.8gHz with 500MB physical memory running Windows 2000 SP4.   Due to the relatively slow hardware, no other software is installed or in use on the system.

I've identified a number of anomalies in the software, and wonder if anyone else has noticed these.

Now, not all is lost.  The PowerSDR software is usable, and the user interface is quite nice.  There are many features that are not present in other SDR software I've used.  There is other software I have not tried as of yet.  Just so it doesn't look like a big rant !  However, the problems I've identified do affect proper AM detection.

1) The AGC responds to audio.  This is MOST annoying.  Even on the slowest setting, the audio absolutely causes a greater level of AGC.  I have tried all of the settings available in the PowerSDR software, and to no avail.  The only solution is to use the software in the FXD (fixed) mode (no software AGC).  The AGC operation is also audible at times (clicks).

2) The frequency response when using the AM mode is not flat !!     I thought I was hearing things, but when I ran a sweep into a known, flat transmitter and monitored the IF input to the softrock to verify the modulation was indeed flat coming from the receiver IF, the high frequency response begins to roll off at about 1.2kHz when using the AM mode.  It is way down by the time you get to 8kHz.  The IF bandwidth does not change this (unless made more narrow than the modulating frequency).  This does NOT happen when using SAM (Synchronous AM).  The frequency response in this mode is flat.

3) When using SAM (synchronous AM), distortion of the modulating waveform starts to occur at around 70hZ.  The distortion becomes worse as the frequency gets lower.  This does not happen with the standard AM mode, only in synchronous mode.

4) The IF filtering rings (sometimes badly) at the filter edge frequency.  This could be mitigated by using a less-steep slope.  I believe this should be an option.  The steep slopes are nice for many conditions, however listening to AM, particularly in the presence of static, is not one of them.  The static bursts ring badly, making weak signals hard to copy.  The copy is much better on a standard receiver under these conditions.

5) The equalizer should have a more overlap (or a more linear overlap) between the various level adjustments (minor problem).

Anyway, I know there are other folks here using this software.  I wonder if anyone else has noticed any of these issues.

Regards,

Steve

[K9ACT]

I've been using a Softrock (455kHz) coupled to the 1st 455kHz stage of a 75A3 receiver for about a month.  I'm using the latest version of the PowerSDR software with this setup.  The computer hardware is an old Dell Optiplex 1.8gHz with 500MB physical memory running Windows 2000 SP4.   Due to the relatively slow hardware, no other software is installed or in use on the system.

I've identified a number of anomalies in the software, and wonder if anyone else has noticed these.

I have been running a 455 SR on my S40B for about a year and notice practically none of the issues you mention.

I would not trade the combination for the most treasured boat anchor receiver out there.

>1) The AGC responds to audio.

The AGC is a myth.  Set it at mid range and forget it.... just like the Popeil roticerie.

It only causes problems when too high or too low.

>2) The frequency response when using the AM mode is not flat !!

Could be but you sound great on mine.  And SAM is just more wonderful.

> The static bursts ring badly, making weak signals hard to copy.  The copy is much better on a standard receiver under these conditions.

No such ringing here and I have never had a situation when a "standard receiver" sounded better.

Perhaps you need a lousier receiver to start with?

>5) The equalizer should have a more overlap (or a more linear overlap) between the various level adjustments (minor problem).

Right on.  The EQ is the other guy's problem.  You can't make a silk purse out of a sow's ear.

>Anyway, I know there are other folks here using this software.  I wonder if anyone else has noticed any of these issues.

I have tried all the other software out there and none come close to Power SDR in my opinion.

As a point of interest, I have recently acquired a stand alone USB version of the Soft Rock and have grown to love it also.  I was so hide bound in my dedication to the S40B setup that I refused to even consider the USB version.

The friend (K9WEK) who got me involved in SDR got so frustrated at my stubborness that he built a USB version for me and flew up here with it  to make me try it.

It took a bit of getting used to the idea of shooting an old dog but I have not had the S40 on at all for a few weeks now.

They are two different animals but within the bandwidth limits, they out put and control are indistinguishable.  I kept trying to prove that there is some advantage to having a real receiver in front of it but gave up.

The most powerful receiver on earth is now on a card about two inches square.

Why anyone would want or pay big bucks for one of those "classic" boat anchors is beyond me.  They just don't know any better.

js 

[steve_qix]

I have been running a 455 SR on my S40B for about a year and notice practically none of the issues you mention.

I admit I've been digging rather deeply, and running some exacting tests.  But, these issues are definitely real 

>1) The AGC responds to audio.

The AGC is a myth.  Set it at mid range and forget it.... just like the Popeil roticerie.

Well, I don't know about a myth     I found a discussion on how the AGC is implemented.  According to what I read, using the implementation, it will respond to audio on AM.  Another, alternate implementation is required for AM.  I may actually take a stab at writing it myself.

You have a good AGC system ahead of your softrock setup, so the effects of no AGC within the SDR itself are largely mitigated.  However, with no "front end" (so to speak) or one with minimal AGC (like mine, with only one controlled stage ahead of the softrock), the SDR AGC function is highly important.  They audible dynamic range is easily 20dB when listening to very strong and very weak stations in the same conversation.

>2) The frequency response when using the AM mode is not flat !!

Could be but you sound great on mine.  And SAM is just more wonderful.

There is absolutely no doubt, good AM signals sound good !!  The frequency response problem does not exist in SAM (sync AM) mode.  It must just be a bug.  I can't see any reason for doing such a thing on purpose.  I'm going to look into the code and see if I can figure it out.

> The static bursts ring badly, making weak signals hard to copy.  The copy is much better on a standard receiver under these conditions.

No such ringing here and I have never had a situation when a "standard receiver" sounded better.

Perhaps you need a lousier receiver to start with?

I should record this and post the recording.   It only becomes a problem in the presence of broad band noise such as lightening static or background hiss.  The IF filtering generates artifacts at the filter edges which are quite audible (and are in the audio range if the filter is narrow enough).  Under weak-signal conditions, the artifacts rival the modulation !

>5) The equalizer should have a more overlap (or a more linear overlap) between the various level adjustments (minor problem).

Right on.  The EQ is the other guy's problem.  You can't make a silk purse out of a sow's ear.

I was using the 10 band EQ to "work around" the AM frequency response roll-off      I tend to make recordings for people and play them back over the air so folks hear what they sound like.  Since my transmitter is flat to about 10kHz and with very low distortion, I can play something that is quite faithful to the original signal.  I am asked to do this often.  The SDR setup is perfect for recordings, with a seemingly flat IF, and synch detector available.  However, since the AM  IF is not flat, I compensated with the EQ in the SDR to make it so.  Upon performing a frequency sweep, I noticed bumps in the response curve rather than a smooth curve.  Again, in Synch AM mode, the response is flat, but then there is the low frequency distortion in that mode 

>Anyway, I know there are other folks here using this software.  I wonder if anyone else has noticed any of these issues.

I have tried all the other software out there and none come close to Power SDR in my opinion.

As a point of interest, I have recently acquired a stand alone USB version of the Soft Rock and have grown to love it also.  I was so hide bound in my dedication to the S40B setup that I refused to even consider the USB version.

The friend (K9WEK) who got me involved in SDR got so frustrated at my stubborness that he built a USB version for me and flew up here with it  to make me try it.

It took a bit of getting used to the idea of shooting an old dog but I have not had the S40 on at all for a few weeks now.

They are two different animals but within the bandwidth limits, they out put and control are indistinguishable.  I kept trying to prove that there is some advantage to having a real receiver in front of it but gave up.

The most powerful receiver on earth is now on a card about two inches square.

Why anyone would want or pay big bucks for one of those "classic" boat anchors is beyond me.  They just don't know any better.

js 

The SDR is certainly awesome !  The fidelity at this time is still not as good as (at least my) hi-fi AM receiver.  I should point out that this receiver was painstakingly constructed with high fidelity AM in mind all the way.  The IF bandwidth is adjustable to over 20kHz; the detector is a precision rectifier and the AGC is well filtered and does not respond to audio.  However, with software modifications to correct the above mentioned problems, the SDR could certainly replace any standard receiver out there [performance wise], including the one I'm using.

One thing: the SDR setup is larger in size than most standard receivers.  Even with a rack mounted monitor, with the hardware behind it in the rack, and every possible effort to save space, it just takes up a lot of room    However, as specialized hardware interfaces (that resemble standard receivers) are developed [eliminating the need for a mouse and keyboard] and ITX or smaller CPU hardware, this problem will be a thing of the past.

Very interesting !

Regards,

Steve

[steve_qix]

BTW, I would encourage anyone to try SDR if you have not.  This is really good stuff 

[K9ACT]

As a point of interest, I am using the second IF output from the S40B.  I was not getting enough signal from the first.

Not sure where you are coming up with the large space requirement for SDR unless you are including the computer.

The SDR card is about 2 square inches and this is what I like to compare to a boat anchor.  I haven't gotten to this yet but I am going to install it inside the S40 and use the filament for DC supply.  There will be nothing but a cable coming out to the sound card.

On all the other issues, I am curious to know if you hear these things or just measure them.   If the former, I would suggest you have a problem because I hear nothing but good stuff coming out of mine.

Probably the most powerful argument for SDR is the filtering capability and cliff-like drop off at the edges.

Snug it up to a carrier, tweek the other side and QRM is gone.

It was also very educational to demonstrate that either sideband sounds just as good as both.

For your next design project, I would like to suggest a SSB full carrier exciter.  It's over my paygrade but I think it would be a great new mode to tinker with.

js

[k4kyv]

For your next design project, I would like to suggest a SSB full carrier exciter.  It's over my paygrade but I think it would be a great new mode to tinker with.

Why full carrier?  You just need enough carrier for the BFO to lock onto.  Commercial SSB radiotelephone links have used a "pilot carrier" some 20 dB below p.e.p. for decades in order to perfectly synchronise the receiver. 

This is advice I believe the ESSB guys could use as well, despite all the pissing 'n moaning that would bring from the space-shuttle-audio crowd, who already get their knickers in such a twist over the wider frequency response.   

Something else that I think would be great to tinker with would be double-sideband reduced carrier, similar to what W3PHL used to run.  With synchronous detection such as what's in the Softrock system or the Sherwood detector, you could get full fidelity audio with most of the power in the sidebands.

This would require a slow-release AGC in the receiver to avoid a controlled-carrier-like pumping effect with the modulation, which would result from a fast-release AGC reacting to the sidebands at a syllabic rate.

The transmitter could be the W3PHL "upside-down" high level plate modulated balanced modulator circuit, or else the DSBRC could be generated at the 100-watt level using high fidelity components, and amplified with a leen-yar.

1500 watts of audio modulating a 50-watt carrier would tear a hole in the band.

[W1VD]

Steve

My results were significantly different than yours...

http://www.w1vd.com/R-390ASoftRock.html

http://www.w1vd.com/R-390ASoftRockdetails.html

[steve_qix]

Steve

My results were significantly different than yours...

http://www.w1vd.com/R-390ASoftRock.html

http://www.w1vd.com/R-390ASoftRockdetails.html

Hmmmmmm.. that's interesting.  Your frequency response measurements definitely differ from mine, and the sync detector and the "regular" AM detector definitely had different response curves.  I wonder if this anomaly was introduced in the new version?  I notice I am a number of versions later than the version you used for your tests.  This could very well explain the differences.

I will also rerun my tests and include some hard numbers.  Should prove interesting, for sure !

Regards,

Steve

[K9ACT]

I need to correct a mis-statement resulting from a senior moment.

When Steve was talking about AGC, my brain was working on RF gain.

So, everything I said about AGC was nonsense as I never found it to do much of anything good and for some reason set it at "fast" and just leave it there.

The RF gain I find to be kind of a myth as it just seems a variation on the audio gain.  I set it in the middle and leave it alone.

js

[K9ACT]

Why full carrier?  You just need enough carrier for the BFO to lock onto. 

But then you just have another version of the "dead air QSO" you like to talk about.

A carrier is nice.  Sure it's a waste of energy but so is driving a Rolls.

js



 

[steve_qix]

Ok, I re-checked my measurements.

The frequency response is definitely different in standard AM mode as compared to Sync. AM mode, at least on my setup.

It is quite flat in sync mode, and definitely rolls off in standard mode.  The output is down 6 to 8 dB at 8kHz in standard AM mode, and flat in sync mode.

Jay, would it be possible to quickly check this at your setup?  It may be some odd problem that only shows up here, with my sound card, etc., or it may be a real problem.  A sanity check would be greatly appreciated 

Now, on the AGC problem, there is no question about that one.  The AGC system generates periodic "spikes" in the output.  The rate of these spikes increases as the AGC speed is increased.  In Fast mode, the AGC actually adds distortion to low frequency waveforms, however the spikes are there for all speeds/times except for FXD (fixed).  When demodulating an AM signal modulated to about 80% with a sine wave at a kc or 2, the clicks will become quite apparent, particularly in fast mode.

On another note, if you reduce the buffer size, the slope of the selectivity will be less vertical.  Sort of an interesting side-effect.   The ringing is actually reduced by doing this 

Regards,

Steve

[WA1GFZ]

Your levels into the sound card and sound card quality will have a big effect on performance. Also the front end ahead of the soft rock. Any AGC ahead of the softrock is going to cause pumping. When I ran the QSD I had no agc action and a front end with almost 105 dB of dynamic range. The 75A3 is not the best example of performance.
You need a good receiver as much as I need audio processing
  

[WA1GFZ]

Steve sounds like you have a computer horsepower issue. My 1 GHz. machine had the same problem. What % is you CPU running at. I'm usually 10% to 20% at 192k. At 96 I'm around 7%.
When you get over 50% things happen.

[steve_qix]

Your levels into the sound card and sound card quality will have a big effect on performance. Also the front end ahead of the soft rock. Any AGC ahead of the softrock is going to cause pumping. When I ran the QSD I had no agc action and a front end with almost 105 dB of dynamic range. The 75A3 is not the best example of performance.
You need a good receiver as much as I need audio processing
  

Hi Frank,

Well, I did a lot of verification on this one.  I monitored the modulation level at the INPUT to the softrock, which is setting at around 50mV at carrier.  The modulation level is flat to over 10kHz at the SOFTROCK INPUT, so I'm reasonably certain of the flatness of the input to the softrock.  And remember, the performance in Sync AM mode is fine - completely flat at least up to 8kHz, which is where I made my measurements at the high frequency end.

I should have mentioned that this 75A3's AGC and detector are both FAR from stock.  I'm using a well filtered, proper AGC with this receiver, which is not in any way modulalated by the detected audio.  You are sure right about the stock AGC in the 75A3.  It was horrible - one of the worst I've heard anywhere.   It was the first thing I fixed when I got it, along with the detector.  Now it's very good!

An easy check of the AGC problem I'm seeing is to put the PowerSDR software in the FAST AGC mode, and provide an 80% modulated,  1kHz tone, AM signal to the hardware.  At least here, I see small spikes in the output while monitoring on a 'scope.  It is also audible.  Put the AGC in FXD (fixed), and no such spikes occur.  The spikes show up with all AGC settings here (except FXD), but are most noticable on Fast.

The processor load stays at around 20% with the PowerSDR software running.  I suppose it is concievable that there is something going on in the software relative to CPU loading, but I would expect other problems other than just AGC issues.  The software works perfectly - I never have any hangs, glitches or other problems with it (so far!!!).  It doesn't APPEAR to be overloaded, but I've seen odd timing issues in software, so it is not out of the realm of possibility.  Some tests by other folks would certainly be a help !

Talk later and Regards,

Steve

[KB3DKS]

Steve,
Which Optiplex do you have? The GX240 I have, a P4 @ 2 gig with 512 ram uses PC133 memory. A bit of a slow buss rate.
 The Optiplex GX260 I also have is a P4 1.8 gig and uses PC2100 (266 buss speed) 64 bit.
Even tho the 240 processor is faster the machine runs software much slower for the same load..

There's a thought.

 But both machines can be easily upgraded with faster processors. The 260 will take up to a 3 gig cpu and 2 gig of ram. There is also a Bios update that might help. Check the Dell website.

 I'm watching your results here as the one 260 is a possible candidate for my own SDR use. Well built machines. Stay away from the GX270s as they have a capacitor overheating problem on the motherboard.

  Bill, KB3DKS in 1 Land

[W1VD]

Steve

Just ran the audio frequency response again using version 1.10.4 and the results are the same as reported earlier. Well under 1 dB difference between AM and SAM.

Installed 1.10.8 and the results are the same. Well under 1 dB difference between AM and SAM.

Good luck tracking down your problem. Maybe try a wipe and reinstall of the software to rule out corruption?

[steve_qix]

Steve

Just ran the audio frequency response again using version 1.10.4 and the results are the same as reported earlier. Well under 1 dB difference between AM and SAM.

Installed 1.10.8 and the results are the same. Well under 1 dB difference between AM and SAM.

Good luck tracking down your problem. Maybe try a wipe and reinstall of the software to rule out corruption?

It must be some issue with the sound card, which is not a great one (built into the motherboard).... Although, I can't imagine why the response would be different between sync AM mode and regular AM mode.   Assuming a sound card change corrects the problem, I'll definitely be digging into the code to find out WHY     [not that it's a big deal - just would like to know].

I'll let you know !  Thanks for the recheck.  Did you happen to get to check the AGC stuff?  No biggie... just curious !

Thanks and Regards,

Steve

[KD6VXI]

Steve,
Which Optiplex do you have? The GX240 I have, a P4 @ 2 gig with 512 ram uses PC133 memory. A bit of a slow buss rate.
 The Optiplex GX260 I also have is a P4 1.8 gig and uses PC2100 (266 buss speed) 64 bit.
Even tho the 240 processor is faster the machine runs software much slower for the same load..

There's a thought.

 But both machines can be easily upgraded with faster processors. The 260 will take up to a 3 gig cpu and 2 gig of ram. There is also a Bios update that might help. Check the Dell website.

 I'm watching your results here as the one 260 is a possible candidate for my own SDR use. Well built machines. Stay away from the GX270s as they have a capacitor overheating problem on the motherboard.

  Bill, KB3DKS in 1 Land

The overclockers have known this for quite some time.  The CPU clock speed has little to nothing to do with performance on the MAJORITY of applications (yes, some are more effected by CPU speed, but those are typically the ones that they load the instructions, and then their is SEVERE math intensive calc's going on).

Getting your buss speed from 66 to 83 mhz, for instance, during the Pentium days, would almost net you a double in performance.  A Pentium 166 running at 83X2 would eat a Pentium 200 for lunch, running at 66X3.  It was pretty much on par with a 233, running 66X3.5

Thats the perf. trick.  Different CPU's have different ways to play with their clocking, but if you can slow the processor down and get the buss speed up (sometimes increasing the CPU's voltages up by .1 to .3 volts will GREATLY increase stability), you will create a whole new machine.

You can also just crank the buss speed up, while not playing with the multiplier (to get FASTER chip speed), but this really cranks up the heat generated, and I don't reccomend it.

--Shane

[W1DAN]

Steve:

Wonder if any changing phase shift versus frequency between the left and the right channels in the sound card could be possibly causing your frequency response problem?

Dan

[flintstone mop]

Hi Steve
DEFINITELY try to switch the outputs from the softrock to your soundcard. It made a night-to-day difference for me!!!!

I don't think that horsepower is critical for receiving. I'm using a Toshiba Satellite S1905 P4, 512 Ram, 1.5ghz. About 30% CPU. Using an outboard USB Creative labs soundcard. One of the "supported cards" and the recovered audio is beautiful. SAM seems to squash the high end a little.

I can create distortion by activating the EQ and boosting the "level" 2 notches up from the starting point. Check your EQ tool and see if it's been selected "ON"

Are you overloading the I.F. input of the softrock from your RX

Fred

[WA1GFZ]

Steve also look at the I&Q output frequency response to verify it isn't the QSD. I'm not crazy about Tony's trnasformer design. I did my own thing. Any imbalance in the transformer will mess things up. He uses too much wire on a low perm core. I use a little 2 hole type 77 core with three turns trifilar #30.
Also the softrock input Z needs to be 50 ohms or the integrator after the switch will be messed up.
Run a modulated 455 KHz signal into the QSD and test it without the 75A3.
I measured about 97 dB dynamic range directly into the QSD. With a modulated 455 KHz into the SR you can check for 90 degree phase shift with an XY scope.

[Steve - WB3HUZ]

If you are coming off an IF output, you don't need I & Q.

[steve_qix]

If you are coming off an IF output, you don't need I & Q.

Hi Steve,

I think I need them in my setup :-)  The IF (the 1st IF) of the 75A-3 is b-r-o-a-d (and I made it broader by loading the primary with resistance to flatten the response out).

If I don't have the I & Q (and properly balanced), the images are severe !  But, I would say for a normal IF reponse, both channels would not be needed.

Regards,

Steve

[WA1GFZ]

I found even in a tight 20 KHz IF you need I and Q to take out images
unless you generate I and Q in the processing. A couple of my receivers have a 6.4 KHz baseband outputs that I used for a while. 

[w3jn]

Hi Steve,

I think I need them in my setup :-)  The IF (the 1st IF) of the 75A-3 is b-r-o-a-d (and I made it broader by loading the primary with resistance to flatten the response out).

If I don't have the I & Q (and properly balanced), the images are severe !  But, I would say for a normal IF reponse, both channels would not be needed.

Regards,

Steve

That might be part of your problem here, Steve.  By ramming 3o-40 KHz worth of stuff into the I/Q detectors in the softrock,  I would presume that there's probably a fair amount of phase shift between the upper and lower ends of your IF passband (which will manifest itself in not true I/Q for wideband junk like lightning crashes).  Further, this may be beyond the Nyquist rate of the sound card.  Try it at the 2nd IF and see if that doesn't improve things a bit.  Also, by using the 2nd IF you'll see smoother AGC action due to the fact you're tapping at a place that is gain-controlled by 3 stages rather than 2.