Using 1.132 I’m seeing the following using shorting blocks on the inputs. I had the same problem with 0.999.
Hi @DanClark, if I had to guess I’d say one of the input resistors opened up. That makes the input appear as a high impedance and it’ll pick up 60 Hz and exhibit the higher noise floor. You could either contact support and we could send you label if in the US and repair it for you, or if you wanted to have a go at it, I could give you instructions here (the likely culprits for the right channel are R95 and R94 which you can check with a DVM. Just short the inputs to make sure the input Cs are bled down. These are sacrificial resistors that will open up if too much current flows).
Hi Matt, they both measure 468. The right channel is behaving the same as left at the moment and the spurious noise is gone in loopback, I thought maybe it was a cold solder joint and reflowed the resistor, so that may have solved it. That said it’s still acting odd.
I ran a multitone scan in both single and continuous modes and also got a very odd result in loopback.
In single I get this (note the DAC underflow error on both)
But when I run a continuous scan the performance totally degrades both with 0.999 and 1.132…
Any other thoughts or should I return it?
Hi @DanClark , we can send you a label, run it on the test fixture and do an eval and if no trouble is found return it quickly and then we’ll have a common starting point for further evaluation. Can I send you a return label to the address from which you placed the order?
PS. The DAC and ADC underflow/overflow messages are in the “experimental” branch of the code right now and I fear a bit too pessimistic (eg false positives), so, I’d not read too much into the message (yet).
Here’s the unit on my desk, 5 tones per octave, -30 dBV total RMS, with 128K FFT and HANN windowing. Can you replicate this with the same settings? Note that if you go to a smaller FFT then the noise floor will rise. And if you go to a larger FFT the noise floor will drop. So, without knowing the FFT size it’s hard to know if your plot is right or not. But with the right and left showing the same, that’s at least a good sign.
And the plot:
Looks ok then. I think I found the problem, it was a cold solder on the right input resistor. A reflow seems to have fixed it. I do note the graphs are different below 40Hz, not sure how to interpret that. Any concern or a possible software/graphing glitch?
Hi @DanClark, I’d be very surprised if it was a cold solder joint as everything goes through reflow and then it goes through a sophisticated Automatic Optical Inspection which makes a bunch of measurements and builds a 3D image of each board and looks for outliers. But, something was up and I’m glad it’s fixed.
Note you have a message “Input Gain -12 dB” which means we’re not at the same place for the measurement. When time permits, can you re-run but first do a “File->New Settings” and then repeat with setting in prior post (FFT size, multitone level, etc). And then when you share the picture, please include the full screen shot. We should make sure you can get to the nearly identical measurement to make sure something else isn’t up.
Sorry, missed the input gain. Looking like a great match to your reference now, thanks.