QA403 generator spike

I just want to notice that when you increase QA403 generator from 7dBV to 8dBV there is a huge spike on output signal.

Best regards.

Hi @bmsorin, thanks very much for reporting this. This is when generating only in idle mode, is that right?

I think what is happening is this: Anyplace you have relay boundary in the output (8 dBV, -2 dBV, -12 dBV) the sine amplitude sent to the DAC needs to be in sync with the DAC attenuator setting. So, you are at 7 dBV and you are sending to the DAC an (almost) 0 dBFS sine. That is, the sine is at ~0 dBFS and the attenuator is at 10 dB.

And then you switch to 8 dBV output, and the relay attenuator goes to 0 dB attenuation, and the sine level goes to roughly -10 dBFS. What happens in the code is the relays always switch immediately (to maximize the settling time). And so, at 7 dBV you have a 0 dBFS sine in the various DAC buffers (USB PC side, USB firmware side, I2S buffers). And then the relay switches to 0 dB, and suddenly that 0 dBFS sine is playing with a 0 dB attenuation, getting you very close to the full scale output of the DAC.

There are probably a few different approaches to take here:

  1. Keep as is. But the momentary full scale for 200 mS isn’t good.
  2. Don’t change the relay until the old data has cleared. This will give a momentary 10 dB dip in output. But a momentary dip is better than momentary full scale.
  3. In idle mode, keep the output attenuator at 0 dBFS all the time. This will give perfect transition between steps, but delivers excess noise and lower levels.
  4. Stop the idle generation when transitioning from 7 to 8 dBV and then quickly restart. This is similar to 2), but instead of a 10 dB dip in amplitude, you get a full period of silence.

I think #2 might be the best. What do you think?

Looking at this more, I think the next release will do #3. This will give seamless amplitude transitions across the full range by locking the DAC output attenuator at 0 dB.

Hi Matt, yes, idle mode. I test an auto ranging device, and at 7-8dBV step the device switching relay to more that it should, testing with scope reveal what i post.
I think #3 will be best option. One think i have in mind right now, testing RIAA premp will be at limit.