So I am really new to this and wanted to know if I am approaching this the wrong way. I have done a ton of reading and havn’t been able to find anything covering this or I am just dumb :). So I did the loop calibration on my Q403 which worked great and produced expected results. I then tried connecting the cables I made along with the dummy load (8ohm/4ohm) I created to test vintage audio equipment. So I am trying to do a base test on my env with out any amp connected to get an idea of what it looks like? So here is the newbie part not sure but I played around with the settings (Full Scale Input (dbV), sample rate, input signal voltage,Freq 1khz) to see what combination of settings gives me the lowest THD and SNR ratio? So the best THD % I could get was .010-.009% on both channels and my SNR dB was 65 to 66. with a input sig running about 1v RMS and the Full Scale Input was set to 0. ( reading up to understand the SNR db stuff) . So am I just testing how noisey the resistors and cables are and is it really a good test? I copied the setup from “Vintage Audio Review” on youtube ( note have noting to do with the channel at all) But I liked his approach and setup. here is what I used for my dummy load :
Hi @fmajewsk, you can always compare your test setup to what you get in loopback. And once you know the limits of the system in loopback, you can start to apply that to your DUT and learn how your DUT influences things.
For example, connect in single ended loopback (L+ OUT to L+ IN, L- IN shorted)
- File->New Settings
- Turn off right channel
- Enable Gen1
- Increase FFT to 16k
- Increase averaging to 10
- Push SNR button
- Set 0 dBV full scal input
And then run. I get the following plot:
And that suggests that if you stick a DUT in the loop between L+ OUT and L- IN, you can readily detect degradation of SNR if the DUT has an SNR worse than 106 dB.
Now, I very much doubt you’ll be able to see the performance difference between a well made $1 cable a $300 cable in a 6" loopback test. Because there isn’t a measurable difference there.
Resistor noise below 1k is tough to measure because there’s not much noise there. A 1k resistor at room temp in a 20k bandwidth is about -124 dBV. The QA403 has a noise floor around -117 dBV. You need to be up to around 5k to start to measure the resistor noise with the QA403 naked inputs. But if you have a super low noise preamp like the QA472 JFET pre, you can measure resistor noise starting around 100 ohms or so.
A dummy load must be driven by a power amplifier - the QA403 is a line level device. This is you are seeing distortion and poor performance. Rather test it (with your cables if you want) into a 1k load.
@fmajewsk- Indeed you need to be measuring an amplifier to use the load you showed here. I am glad you found the material useful (it is my channel 
) Since you will be doing a lot of measurements on vintage gear, I am more than happy to share what I have learned using the QA40x. Bob Cordell’s manual mentioned in there forums is a great resource as well.
@VAR HEY! love your content!!! very useful for getting setup on my bench! many thanks again!!
Thanks and I do have a video coming up at some point showing how I measure a preamp with the QA403.
Hello,
apologies if this has been asked before and this is going to be a long post! Appreciate any help I can get.
I just got my qa403 2wks ago and I have been struggling with noise. I get NOWHERE near the clean baseline as you show here. I am using a usbc dongle to power the laptop and the instrument and I noticed the noise is ridiculous. I realized that a big LED light and some instrumentation I had plugged in the same outlet as the laptop were creating huge noise spikes maybe 15/20dB of noise and things got better when I powered everything off still but still at least 5/10dB of noise remains. What is the proper way to power this thing? I tried the loopback thd vs frq measurement and I get between 0.001 and 0.0001% THD with freq but the + and - channels don’t look the same. I mean they are within the same range but not the same. The distortion seems higher at LF and then kind of slopes down throughout the audio range. I feel like I need to take a day course on how to setup and run this!
thank you!
Hi @pr1324, can you try this and report back:
- Unplug QA403 hardware
- File->New Settings in the QA403 app
- Select 0 dBV full scale input
- Plug in QA403
- Short all four QA403 inputs
- Press the RMS button in the MEASUREMENTS section
- CTRL+SPACE to run a single cycle
- Edit->Copy Bitmap->Copy Image 800 pixels wide
Paste that here. Here’s the unit on my desk:
Hmm, the scale says about -135dBV, the measurement L and R about -116dBV?
I repeated the measurement several dozen times the L channel is consistently more noisy than the right and the noise jumps within 3 dB that seems unstable. Can go as low as 111 and as high as 114. The right channel is within 0.5dB. Is this imbalance normal?
Hi @pr1324, not that imbalance isn’t normal. Can you swap the shorting blocks and see if you can get the problem to follow the shorting blocks? And gently rock them back and forth and see if you can make the jumping worse. And then please report back. And just to check, you are using 0 or 50 or 75 ohm shorting blocks is that right?
Hi @cfortner, yes, the RMS measurement is the root-mean-square sum of all the freqs from 20 to 20 kHz. The noise floor you see depends on your FFT size. And so, reporting the noise floor (-135 dBV) doesn’t really mean anything unless you have also reported the FFT size. But even then it’s difficult to tease out what it means.
You’ll note as you increase the FFT size, the noise floor drops. But the RMS will stay the same. Now, if you switch to noise density (rthz in DBV context menu) then the noise floor will stay the same regardless of FFT size. BUT, your peak amplitudes will change depending on FFT size.
Whew!
But the RMS hides all that for you. It doesn’t care the size of the FFT. It just sums the noise, regardless of bin size.
@matt Yes, I have swapped the blocks already, problem stays on left channel. Yes they are 0 ohms. You think this need to get sent in for repair? I just got it from selig. I would be less worried if the the noise was more constant but it can vary by 2-3dB from scan to scan.
Speaking of channel imbalances, I sometimes get the left channel with a significant noise-floor across the spectrum imbalance compared to the right. Exiting and restarting the program doesn’t help, but unplugging the USB cable and replugging it after a while, then restarting the program fixes the issue. I figure it’s just a glitch to do with the attenuator as it only does this on the four ranges where the relay attenuator is active- it’s done it since day one.
Here’s an example just now, shorted inputs both channels:
And here it is after several (15) minutes being unplugged. My theory is temperature related. The QA runs pretty warm.
Hi @restorer-john and @pr1324, yes, if you have an imbalance you can send in for repair. The usual culprit is one of four front-end resistors shown in the drawing below:
You can measure these four resistors with a DVM in-circuit and compare them to see if you see a difference. These are very special resistors called “pulse withstanding”. The part number is CRGP0805F560R.
What is special about these resistors is they can withstand huge amounts of power for short periods while the capacitor’s RC time-constant is playing out and the atten is switching in. From the data sheet, you can see the 0805 resistor can handle 10W for 10mS and 5W for 100mS. The QA403 ADC will sense the overload condition and respond almost immediately, and then 1-2 mS later the front-end attenuator will click in.
So, when the QA403 is subjected a momentary overload, these resistors will absorb much of the pain. 10W in a 560 ohm means about 130 mA flowed.
As the short-term power limits are exceeded, the resistor value will shift higher and that higher source impedance is usually what causes the rise in noise.
We get a few of these failures a year. But often when we get the unit on the bench, there’s no trouble found and the issue is customer shorting blocks.
Contact the support alias if you’d like to send in your unit for elevated noise. It’s a quick check and quick repair.
I think you’ve hit the nail on the head Matt and the shorting plugs are the most likely cause. Notice in my plot above, there’s a 50Hz mains spike. For many of us who use SE inputs most of the time, the shorting BNC on the inverting input rarely gets moved.
I just removed the inverting shorting plug and guess what? Same plot! So the shorting BNC periodically (when warm) makes intermittent contact in the socket itself.
Moral of the story is clean and check the BNC plugs before sending your QA back…
My problem persists no matter the plug.
I have a question: why does my graph I posted earlier looks weird and wavy at the low end while yours looks like normal noise? Also, my application window and graph seem very low res, how do you produce such nice graphs? Is there a setting for the app to be high res? I am a mac guy..sry. 
Increase the size of your FFT and drop the sampling rate to what you need for the required bandwidth. Also, there is averaging you can use depending on what you are measuring.
Here is mine at 192k SR and 32k FFT, Not enough resolution for a decent display.
Not sure about the app resolution on a mac, but your graph looks fine to me- is that a screen grab or a saved bitmap?