qs1r latency...

[N2DTS]

Its getting to be radio season for me again, so I picked up a qs1r to replace my sdr-iq I use as a band scope/receiver.
The qs1r is much better then the sdr-iq, although the sdr-iq seemed to do fine in almost every respect for what I do with it.
Its got at least three programs that run it, spectraview (don't like much), HDSDR (good fidelity), and SDR CONSOLE (my favorite).
SDR CONSOLE has nice features, like most things you want to play with on the desk top, like POWER SDR from flex.

The new qs1r uses SDRMAXV which is a good program, but its got small buttons and many things hidden.
The qs1r has an onboard dac audio output so you do not have to use the sound card.
That and the fact it does a lot of the heavy lifting in the very fast FPGA had me thinking the latency through the receiver would be very low, but in my case, its not at all.
On the flex (5000 and 3000) on the same quad core computer, the latency was low enough to monitor my TX signal no problem.
The qs1r seems to have the same latency as the sdr-iq.....1/2 second or more!
I must be doing something wrong, but I do not know what.

I had planned on using the low latency audio output as a monitor, replacing the mod monitor output, which would be interesting, as you might be able to hear things like strong signals or a double AS YOU TRANSMIT.

As it is now, I do not love SDRMAXV software, and I have the latency problem, so the qs1r is no improvement over the sdr-iq.

[AB2EZ]

Brett

The latency associated with a digital filter will be proportional to: [the number of consecutive samples that need to be stored to perform the digital filtering] x [the time (in seconds) between the samples that are input to the digital filter]. I.e. if the digital filter's output is going to depend upon the values of 1024 input samples, then the 1st output will have to be stored (and delayed) until the 1024th input arrives.

In the case of the qsr1... the A/D converter has a 130MSPS sampling rate, so the time between samples is only about 7.7ns ... but the effective number of consecutive samples that must be accumulated to perform all of the digital filtering is huge.

The latency (in seconds) should be approximately N/(delta f) .... where delta f is the frequency change over which the frequency response of the filters in the receiver go from full transmission to essentially zero transmission, and N is a small whole number. I.e. the sharper the filters, the longer the latency. [This is just a mathematical consequence of the principle of causality: the output of the filter cannot occur before the relevant input arrives]

The bottom line is that if you can adjust the sharpness of the cutoff of the digital filters to be less (e.g. going from full transmission to essentially zero transmission in 200Hz rather than 2Hz)... you should be able to reduce the latency.

Stu 

[W3RSW]

1/2 second, --right, something is definitely wrong.
My Qs1r latency is on the order of milliseconds if the onboard DAC is connected directly to my ears, so to speak.  Are you transferring audio directly from the DAC to AF amp or headphones, the setup of least latency?  I can still miss half or a whole first dit at fast code speeds (full break-in) but that's the nature of the DDC/FPGA beast at this time.

Sounds like you have a loop.  Make sure your computer control panel/ sound manager is not also trying to decode the audio via USB, etc.  make sure the other AF line inputs are turned off, etc.  perhaps the DAC is feeding into the sound card and then to speakers or line out. Double digitizing.

I've decoded both audio and, say CWget or MMSTV right through same sound card as well as the GUI.
If settings are not correct you get echos with progressively longer lag times.  Latency "regen
 ", Heh, heh.  Of course I expect latency and get it using only the USB connection to the computer and leaving the On-board DAC jack open.

Oh, and if you still have trouble get on the qs1r yahoo board and ask the troops or Phil n8vb for suggestions.  Most will want to know your lash up and computer.

[N2DTS]

Yes, no one else reports such high latency on the qs1r.
I have the dac output into an audio amp, and also plugged the headhones directly in to the dac port, plenty of audio drive!
I ALSO get audio out of the sound card, and assume its got the same delay, so there is some issue, but I thought the dac output was sort of direct from the fpga and was unable to go through the sound card in the computer.
If I do not plug anything into the sound card jacks, I get error messages in the sdrmaxv screen, but I still get audio out of the dac port...

I will try some experiments tonight.
Maybe disable the sound card completely.

Has anyone tried using a seperate antenna to monitor their transmit audio directly out of the qs1r?
Since it has such a high overload point, it seems like it should work with little or no attenuation, allowing you to hear other signals on the frequency/band while you transmit...
Sort of like a remote monitor receiver without the delay...

[W3RSW]

Do it some times.  Have hung a 50 ohm resistor with 1 inch or so stub into input BNC jack.
Though not much over 100 watt level. Sound is ahead of the gUI.

Watch out for miss coupling and toasting your QS1R.
Did that by trying to load up a 32v2 into an open line and somehow toasted the ADC on the Q'S 1 R board.

[W3RSW]

As to hearing other signals along with your own ( duplex, sort of), the disparate magnitude is to great without some type of switching.  Safer to mute the qs1r and ground the input when transmitting, but everybody's tried the quick and dirty.  Fun and almost works if the band's hot a some qro boys are on.  Make sure to separate or cross polarize your xmit and rcv ants, of course.

The small stub mentioned is fine monitoring only your transmit.  bTW the 0 to 60 MHz wideband GUI view will show some amazing things out of your transmitter.

Same 32v2 looked good and very clean right up through 40mhz or so when transmitting on 3.7mhz, but boy you outta' see the wide swath of only 15 to 20 db down ( if that) crud around 50 MHz.  Talk about thieving power output.  Almost the worst parasitic I've seen but very wideband.  No one traditionally had a clue. Audio superb with MPF'd 102 & d104.


Well your in for a lot of fun.
But I think you may get more particulars on the yahoo forum.

[N2DTS]

I tried everything to get the latency down with zero results.
I might just move it along, since the delay is just as high as the sdr-iq.
I had no problem running a flex 5000 on the same computer.

I checked the usb drivers, removed the usb in device manager, re installed it, no change.
I removed the sound card and got the same results.

Good idea though, before I move it along, I should look at the output of my various transmitters on the wide band display.

[W3RSW]

Hey, Before you sell it see if you can borrow another and try it.  

"I hate it like that" if a good son is wrongly sent to the orphanage.

Conversely try it on another computer without all your other programs.
Start eliminating everything nonessential until you find a conflict.

Surely your not so enamored with your current setup and computer that you'll never change it? You never know what your future lash up will look like post your existing setup.

Phil will provide a schematic to every QS1R owner as well as logic and timing diagrams, USB interface and FPGA logic trails, etc. if it will help ascertain why the latency.

I seriously doubt that a single QS1R board is at fault if everything else works up to spec.
But if it is so simply ask Phil.  He will have had similar problem or have solved it elsewhere.

If it is the board,  send it back to Phil who may find a simple defective chip cap or similar before you dump it at possibly a loss.  And the QS1r world is not large compared to the "dumped linyeear" world.  We talk.

Ok, now for the fan boy diatribe.
Your going to miss a lot of fun, not just the whole 0 to 60 MHz sweep function you want to try, but how about adding the Qs1E xmit board, useful for everything from a transceiver to a sweeping signal generator through home brew filters and stuff like alignment of other stuff.... You name it.
It is a very clean generator.

The qS1R is primarily an experimenter's board.  A Minicircuits preamp and attenuation at lower freqs. gets you performance equal or exceeding most rigs.

Under sampling second order, 60 to 120 MHz, 120 to 240 MHz., etc. gets you FM or whatever by simply providing bandpass filters.

Now that you have it, don't treat it as another Persus or whatever box. Oh, ok I'll take it off your hands for a few sheckles.

[N2DTS]

I have been emailing Phil about the latency.
I think it might be in my usb ports, its a cheap motherboard.
I will try and get a pci-e usb3 card and try that.

I also find the sdr-console program a lot more friendly to use with ham radio operation, and sdrmaxV not so friendly.

I was on 40 meters today, working Clark (KB1NFF) very nice qso, and was switching between the homebrew RX, the qs1r and the sdr-iq. I found I can run both the sdr-iq and the qs1r at the same time, on the same computer, with no problems (same delay though), that is kind of interesting....
Anyway, Clark was running a very good sounding ranger, and signal levels on my end were not that great, and of the 3 receivers, my homebrew had a better copy.
The sdr-iq did fine, it just has more background noise with sdr-console, its better with the hdsdr program, but I like the ease of use with the sdr-console.

I did check out the 2x 4x150 rig on the wideband display, and the big rig (2x 813's) and they were very clean.

I may try the qs1r on my lap top and see what happens.
I do not think the qs1r is the problem or bad, since I only have a delay problem.
But there are things that really bug me about sdrmax.
Real nice display, but it does not remember things, it tunes oddly, lots of things hidden under buttons.
I also could not resize the display so I could stick the filter, band and memory screens under the main screen.
That is another computer glitch I suppose as I think you are supposed to be able to do that.
 
It was good to be back on 40 meters today.

[N2DTS]

Ran the qs1r on the laptop, same delay....

[W3RSW]

very interesting.  Is it really a full 500ms?  +/-

When you compare QS1R with the sdrIQ do you get a delta AF echo or lag between only the two? That is Both rigged up to compare with each other, leaving your home brew rcvr. out of the mix?

If you want,  Send it over to me and I'll compare it simultaneously with my K3. I'll listen to audio from usb / sound card mode and directly from the DAC mode. If I had a dual trace scope I could time the diff.

Have you tried that, getting an actual instead of estimated time diff.?  Curious and understand your not enamored with some of SDRMax features. There are at least five ways you can change freqs on it and it has almost infinitely re sizable windows, etc. as you surmised.  I find it far less cluttered than other GUI's.  btw Phil changes GUI revisions sometimes based on only one user's suggestions if even fairly good.

He recently changed within a week one user's findings of a glitch. I find that remarkable in the SDR marketplace and bodes well for his future offerings.

[AB2EZ]

Brett

I don't have a QSR-1, but I downloaded and installed the SDRMAX-V software.

I suggest that you check the following:

Click on the "set button"

In the "general tab", make sure that "Ps Size" and "Post Ps Size" are both set to 4096 [Please let me know what they were set to when you checked]

Does this reduce the delay?

Also, see if reducing the sampling rate ("SR" tab) has any effect on the delay (it probably won't... but depending upon how the software works, it might)

Stu

[N2DTS]

Stu,
Nice to hear from you!
It was/is set to 4096 and I tried other settings.
I also tried different sample rates.
I get different sound/performance it seems, but the same delay.

And the delay is very long, and very close to what the sdr-iq has.

I ordered a pci-e usb 3.0 card, to eliminate the on the motherboard usb chips and will see if that does anything.

For fun, I will run the scr-iq and the qs1r at the same time, and listen to both audios to see if there is any difference at all. The delay is long, on the screen, and in the audio.
Its got to be very close to 500 ms.


 

Brett

I don't have a QSR-1, but I downloaded and installed the SDRMAX-V software.

I suggest that you check the following:

Click on the "set button"

In the "general tab", make sure that "Ps Size" and "Post Ps Size" are both set to 4096 [Please let me know what they were set to when you checked]

Does this reduce the delay?

Also, see if reducing the sampling rate ("SR" tab) has any effect on the delay (it probably won't... but depending upon how the software works, it might)

Stu

[AB2EZ]

Brett

Try increasing the "filter step size" on the right side of the main display. Set it to a high value, like 500 Hz.

Stu

[N2DTS]

I ran the qs1r, the sdr-iq, and the homebrew at the same time, also using both the sound card output for both sdr's and also used the dac in headphones.

There was very little (I could not hear any) difference between the qs1r dac output and the soundcard output.
The qs1r was a little bit faster then the sdr-iq, both were way behind the homebrew.

It must be the usb ports, as the flex would run fast enough to where I could not hear enough difference to live audio to notice.

I will see what happens when the new usb card arrives.

Its kind of interesting running two sdr programs at the same time on the same computer.

[AB2EZ]

Brett

I think that an important indication of what is causing the delay is the fact that (other than for the delay) the radios (QSR-1 and SDR-IQ) appear to be working fine. I.e. I assume that (except for the large fixed delay) the audio outputs are smooth and artifact-free.

I would expect that if the problem were associated with the USB port (e.g. congestion with respect to the data moving back and forth) or insufficient computing power, then the behavior of the receiver would be erratic.

Again, based on the symptoms you are reporting... it would appear that the digital filter in the QSR-1 is buffering a large number of samples in order to obtain a very sharp cutoff. For any filter, digital or analog, the delay through the filter is inversely proportional to how sharp the transition is between pass band and stop band.

With a digital filter, you can achieve an astoundingly sharp cutoff... at the expense of long delay (a.k.a. latency) between the input and the output of the filter.

[Achieving such a sharp cutoff with an analog filter is impractical... and therefore most of us have probably never heard delay/latency in the audio output of a receiver that employs analog filters. However, we have observed the effects of delay/latency in analog filters in the context of phase shifts that cause oscillations in circuits that employ feedback]

I was looking at the SDRMAX5 application to see if there was a way to set the sharpness of the digital filter. The setting called "filter step size" (which has choices between 1Hz and 500Hz) seemed to me to be a candidate for setting the sharpness of the digital filter that is internal to the QSR-1

Stu

[N2DTS]

Stu,
The filter step size is the step size for changes to the filter bandwidth I think, like the vfo tuning step size, you can set it for 25Hz, or 100Hz, or any step ssize you want.

I know what you are talking about, I forget what it was called in the PSDR software, but it was sharper filter, more latency.
In the SDRMAX software it might be the ps and post ps size settings.

I will try the new usb card, maybe the motherboard one has a huge buffer or something.
The flex 5000 and 3000 were good, but used a firewire plug in card...

If I do not get any difference, I will just move it along, for what I do with it, its no big improvement over the sdr-iq.

And I find it really odd that the simple homebrew RX gives better copy on weak signals...

[AB2EZ]

Brett

My guess is that if the "filter step size" is set to allow you to adjust the filter width in 1Hz steps, then that would imply that the sharpness of the roll off is also 1Hz. Based on the separate control for setting the filter bandwidth (called "filter"), I doubt that the "filter step size" is used to set the filter bandwidth

Stu

[N2DTS]

Well, I will check that out tonight!

[N2DTS]

I did the test, it changes the amount the filter changes, and has no effect on the delay.
The QUALITY of the filter changes the delay in sdr radios, but not the width.

Brett

My guess is that if the "filter step size" is set to allow you to adjust the filter width in 1Hz steps, then that would imply that the sharpness of the roll off is also 1Hz. Based on the separate control for setting the filter bandwidth (called "filter"), I doubt that the "filter step size" is used to set the filter bandwidth

Stu

[AB2EZ]

Okay. Try this:

Tune the receiver to a carrier, and then tune the receiver so the the carrier is right at the edge of the filter pass band. Determine how wide the filter's pass band to stop band transition is by tuning the receiver.

If the width of the passband to stop band transition is (for example) 10Hz, then the latency must be at least 1/10Hz = 100 milliseconds.

If you want the latency to be 10 milliseconds, then the width of the pass band to stop band transition has to be greater than 1/ (10 milliseconds) = 100Hz

Etc.

Stu

[W3RSW]

Gentlemen,
Everyone else's is working at about 10ms latency without intensive instruction or theoretical parsing. - working well from 10kHz or so to 63MHz.

With correct impedance matching exceeds or equals my k3 to 18mhz and with preamp higher.  This with noise figure limited only by atmospherics or wallet.

Send it to me, bwaaa haaaa haaa.

Really, they're giving away quad cores now for half the price of the SDR.

[N2DTS]

I am running it on a quad core, with 8 gig of memory....
I can even run the sdr-iq and the qs1r at the same time with no change in performance to both radios with low cpu usage....

The only thing I can think of is something in the usb ports, so I ordered a dedicated usb card.

The sound card on the computer adds VERY little latency, so that is not it.
I listen on headphones in the dac port and through the sound card to the speakers, and there is almost no difference there...

The flex stuff on the same computer was real time, but using firewire...

[W3RSW]

Oh, sorry 'bout that.
Just saw that you said you had a cheap motherboard.
Hope all works out well.

[N2DTS]

The weak point on this computer is the video system, not real important on sdr stuff I have heard...