- Joined
- Feb 23, 2016
- Messages
- 20,771
- Likes
- 37,637
Why are some of the graphs to 20 khz and some 80 khz. Was the ADC sample rate different for some of them?
-
WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!
Master Thread: Are measurements Everything or Nothing?
- Thread starter amirm
- Start date
Why the differences in frequency scale?
Scaled the graph to better see apart the suspect frequency peaks.
Another problem discovered with the same DAC.
705.6k source 1k tone looks fine:
Switching to 19+20 kHz IMD test tone however:
Why did the noise floor jump up?
Then dropped sampling rate to half:
Competing DAC based on same DAC chip, at 705.6k IMD:
Note, 1 kHz IMD tone is even below -140 dB here.
If you know for example that the return currents produced by the transducers can in some cases represent a problem, you will find a way to measure the effects of the problem. Ok. But you must know, otherwise you will not take that measure.
Looks like different FFT settings. Why would the -60dB signal have a more than 20 dB higher noise floor? Does not make sense.
solderdude
Grand Contributor
Could be verified by someone else measuring the D90SE.
That's the beauty of testing and standards and/or reporting what test conditions were.
That's the beauty of testing and standards and/or reporting what test conditions were.
C. Cook
Senior Member
- Joined
- Sep 21, 2021
- Messages
- 366
- Likes
- 185
It's been a long time since I've heard a TI based DAC. Who's using them these days? I may have an old ROTEL CD player that uses Burr Brown or Wolfson. I know Cambridge Audio was still using Wolfson last time I checked. Cirrus is based in Austin, I think - used to see their building every day.
When you say "less smooth and rounded off" what specific aspects or measurements do you mean? My experience with the TEAC DACs was that they had less 'crisp' treble than ESS and a generally more rolled-off high end response. I'm not saying that's a bad thing, but I - like you - do notice the difference. Of course, it's almost certainly NOT due to the DAC chip itself, but rather the surrounding circuitry and even partnering equipment.
I think it's pretty well established that ESS has that 'bump' and 'in your face' sound in many applications, so I'm in agreement with you there.
For similar reasons noise floor seems to vary based on sampling rate in other measurements I've seen of the same DAC.
I suspect the reason could be different implementation on how the 4 DAC channels are ganged on the output. If there's no suitable mixing resistance between the channels, it can drive feedback loops of the separate I/V stages used on D90SE (but usually not used on other ESS implementations) nuts. Since output of four delta-sigma converters run from four different modulators is NOT the same signal at high frequencies (before analog reconstruction filter)! You need to remember that you are mixing four high frequency (tens of MHz) signals together.
Also, if your (partial) measurement tells you: "everything is ok", but your ear tells you: "everything is not ok here! Check the high frequencies better!", What do you do, try to control better or give the blame it on Madame Dopamina and Mr Placebo?
Control what better?
Measurement far far more likely than DUT. Your as-yet-undisclosed methodology and setup appear to be a bit shaky.
Try to control at all. Unless you're someone who doesn't trust his ears and needs to peek. If it holds up, it's worth chasing.
Narrator: it's unlikely to hold up if we're talking about electronics.
At least iFi and T+A for example. Also in TEAC UD-501 (PCM1795).
Not any measurement, listening. I could nicely measure (analyze) too, given access to the raw data going to conversion stage, from the DEM stage output. Some things are quite challenging to analyze from analog output, especially when it's a multi-element converter.
Peeking.
This is up to the listener's knowledge. He is in charge to measure, if his ear is warning about some unpleasent effect.
Sgt. Ear Ache
Major Contributor
What if his ear is BSing him?
What if the brain is making up this unpleasant effect? Should you not make sure that that does not happen?
How well do you understand the measurements that are typically done?
If there was a clearly identifiable issue, such as you describe, I would presume it was broken. I would use some basic controls to confirm it before I sent it back, as I know first hand how easily I can be fooled.
Again talking solid state devices. Transducers are in a different category to me.
Similar threads
