Hi all,
Just hoping to confirm the max frequency limit for the FR test? I can’t seem to get more than 22kHz and wanted to confirm that was correct.
Many thanks,
Greg
Hi @swordfish, you can put the QA402 into loopback and run Automated Test->AMP Frequency Response Chirp at each sample frequency and generate the following:
The setting used are as follows:
Let me know if you’d like help recreating the graph.
Oh brilliant, thanks very much!
On a related issue, how do I make the THD vs frequency test go to a higher frequency? Mine seems to stop at 10k - see output and all inputs below. This is in loopback mode with 6" cables.
Many thanks,
Greg
Hi @swordfish, remember when making a THD measurement that you need to measure several harmonics. When you get to any frequency beyond 10 kHz, the harmonics are going to be higher than 20 kHz. So, if your measurement stop frequency (specified in THD or THD+N context menus) are at the default value of 20K, the THD measurement won’t make sense.
I see you have your sample rate at 192K. Can you change your Measurement Stop Frequency to 90k and see if the test proceed beyond 10K as you expect?
Thanks Matt, got it now. I love this thing, getting some very cool results off it.
Sorry only two more questions…
My FFT plots show lots of hash from 50kHz to 100 kHz - see attached.
Is any of this normal? I suspect the switching power supply my USB hub uses may be the culprit. Can’t seem to get rid of it in any location or with any fluorescent lighting turned off, etc.
And… is the purpose of the “use right channel as reference” in the automatic frequency response test dialog box to partially compensate for the bandwidth limitations of the device, i.e. it adds the gain loss of the right channel to the left channel to better capture the true frequency response of the amp being tested?
Many thanks,
Greg
Hi @swordfish, can you do a File->New Settings, short the inputs and show what the spectrum looks like at 0 dBV max input? Turn off right channel, bump up FFT to 128k. The unit on my desk shows the following.
In your plot, the 50 Hz is creeping in someplace, and everything you see is either a harmonic of the 50 Hz OR a mixing product of the 50 Hz (eg 1000 - 50 = 950). If you short the inputs and the spectrum cleans up, then it means the 50 Hz is getting in via the test setup. Sometimes it can be beneficial to take a ground wire and touch it to the various ground points of your circuit (see safety note below) . A USB cable plugged into your PC (not laptop, but PC plugged into a wall with 3 prong plug) can work as a ground wire in a pinch. Just plug the USB cable into a USB plug on your computer, and then touch the shell of the cable part that normally plugs into the USB device to your circuit ground and see if that changes things. SAFETY NOTE: Only touch the ground to externally available connections on your DUT and no not touch to any power entry points or amp power exit points, etc. Let me know if you need more clarification because if you are working on an open amp, and poke around inside with a ground wire it could be dangerous. But in general, your plot suggests your DUT ground is floating.
s the purpose of the “use right channel as reference” in the automatic frequency response test dialog box to partially compensate for the bandwidth limitations of the device
The compensation isn’t just a brute-force addition. Instead, the technique does a complex divide of the the left input and the right input. Thus, things that are the same for the left and right channel cancel out. And things that are different stay. So, the left and right channel anti-alias filters, the ADC filters, etc, are all “erased” using this technique. It’s a pretty old technique (decades at least) but still works very well.
1 Like
Thanks Matt,
My FFT with the inputs shorted looks like yours, I just wasn’t sure about the peak up near 100kHz. Would it affect any distortion measurements done at higher frequencies e.g 20kHz?
I’m not too concerned about the LF stuff, as you say it’s likely a grounding or power supply issue, I’ll sort it out.
Thanks for clearing up the freq response question. It does indeed work very well!
It’s a very cool device, keep it at an accessible price point for the masses and you will have a winner. So good to see how my amps actually measure after building them blind for all these years.
1 Like
That little rise you see is intrinsic to the ADC/DAC and specificially with sigma-delta converters. It’s usually called the noise bump or similar. But with sigma delta, the converter designers take the in-band noise and push it up to higher frequencies. So, what you see up there is normal and expected!
Glad it’s working for you and thanks for the kind words!
Ok sorry I’m back again but I thought you might find my findings interesting. Thanks for answering my questions!
It really seems my issue is related to the cheap(ish) USB hub I have been using. It has a SMPS wall wart which seems to create a lot of RFI. Plugging the 402 into my laptop directly fixes everything. Fortunately my laptop seems to be able to power it.
With the hub connected it has a clean fft with the inputs shorted. However, as soon as I connect cables they seem to pick up the RFI. Plugging the 402 directly into the laptop or turning off the hub supply removes the noise.
Below is an FFT of my amp at 10W with the hub powered off. I’m getting a little 50hz hum and associated harmonics, but all the noise I was getting around the 1khz fundamental and up at 40khz is now gone (see old pic above).
Do you have any recommendations for a suitable hub? Presumably no one else is having this issue so maybe even just a more expensive one would be better.
1 Like
These linked below are in our test bays. Are you sure it’s not something radiated? How does your noise appear at 0 dBV input with inputs on the QA402 all open? On my desktop unit it shows as follows. Note the RMS increases about 15 dB, which is mostly due to the 100kohm input z.
Thanks Matt,
Sorry for the slow reply.
Yep, I think it is radiated from the SMPS in the USB hub supply. It gets interference as soon as I open the input test leads. I’ll try another one…or even build a linear supply for it.
Thanks for your help!
Greg
I’ll try another one…or even build a linear supply for it.
Hi @swordfish, you shouldn’t need a linear supply, there are lots of switchers that are yielding good results (such as the Sabrent shown above). I’d give that a try first.
For what it is worth, I got this from amazon, and it seems to be working fine w/out any extra noise that I could tell:
I also purchased a “heavy duty” usb cable was was recommended
Thanks guys, I’ll give one of the recommended hubs a go first.
Cheers,
Greg
Good news, I got a new hub and it has solved the interference problem. Definitely seems like the SMPS on the old hub was causing the issue.
Can I please confirm I am safe to test an amp with +/- 35V rails (70V p-p / 24Vrms) directly in differential mode?
My understanding is that I can do up to 40V RMS (112V p-p) with the QA402 without issues - which is really quite awesome. I have an 8R dummy load with 50% attenuation (4R tap) I can use if that is preferable.
Sorry I know this is stating the obvious but just being super paranoid in case I am somehow misinterpreting the spec sheet.
Great news on the hub, thanks for reporting back. Yes, test away on your amp at those levels! Remember the ADC will deliver best THD performance when roughly 20 dB below peak. So, if you have an exceptional amp, and you are running at 40Vrms = 32 dBV, then at 42 dBV full scale you are 10 dB away from max.
If you want to get an idea of how the analyzer performs at 10 dB below full scale, then pick 18 dBV full scale and set the gen1 to 8 dBV and note that THD in loopback. And then, set gen1 to 2 dBV (6 dB less) and make the same measurement in loopback. That improvement (which will probably be 13 dB or so) will give you some clue as to the win you’d get from using the built-in attenuator on your load in terms of the THD limit. My guess is that the amp is much worse than the analyzer in both ranges and it won’t matter much, but it’s just something to be aware of.
Thanks for that Matt.
You’ve been super helpful and patient and it’s much appreciated!
Greg
Is there a way to fix the graph extents? every time I add a new plot they reset to the default. Also as shown in the original set of plots the loopback high frequency response leaves a lot to be desired. Is there a way to correct for flat loopback response? (Ver 1.12)
dave
Hi @dave, the workflow is such that you should add all the curves you want, and once you have curves, then pretty up the graph as needed. Some of the plug-ins need to manage the axis so that the THD dB and % curves line up and as you are seeing, those examples (and others) will fight you the entire way.
You can get a very flat response if you use the “right channel as reference” technique.
There will still be some limitations based on channel-to-channel variations at higher sample rates. Here is the loopback response for 192K sample rates. The marker shows 0.05 dB at 90 kHz.
And if you want flatter still, you can export the loopback response, and delete the single line:
so that it appears as follows:
And then input that as a user weighting file:
And then apply that:
And then your magnitude error is about +/0.01 dB flat to 90k or so.
1 Like