Hello,
When testing a DAC based on the PCM5102A I faced to something new (for me) I'd like to share.
I do not hold a degree in electronics but in electricity and mechanics. So excuse me if this will be too basic for you.
I carried out some measurement on a chip PCM5102A, and I discovered a slight difference in noise when injecting a digital silence. I used NF noise meter M2177. I like these old fashioned meter with pointer. I noticed a 300% in noise level form output L to output R.
Then I used my Audio Precision P1DD to check few basic parameters (level THD+N) on both channels everything was consistent with few tenth of a dB off.
Then I carried out a 0dB FS output level back with the NF and the P1DD and noticed 6 % difference in level. Buh?
Because everything is integrated in the PCM5201A I'm use to more accuracy.
So I started my very old Panasonic analyzer, I hate the Panasonics fans so despite it's very quality I do not use it. I measured a 20dB THD+N difference at the DAC output. Interesting.
I confirmed the P1DD THD+N was matched with a second measurement.
Then I traced both output signals with a scope at 20kHz and guess what the chip is defective. One output is filtered, the other not.
Conclusion:
That for me a multiple surprise, I never face to a partially defective chip. The PCM 5102A is a stereo chip you cannot select filter for one channel only (maybe with China counterfeits).
Then THD+N measurement show it's limits when identify a problem, one analyzer is "blind" the other show a huge difference.
PFB
When testing a DAC based on the PCM5102A I faced to something new (for me) I'd like to share.
I do not hold a degree in electronics but in electricity and mechanics. So excuse me if this will be too basic for you.
I carried out some measurement on a chip PCM5102A, and I discovered a slight difference in noise when injecting a digital silence. I used NF noise meter M2177. I like these old fashioned meter with pointer. I noticed a 300% in noise level form output L to output R.
Then I used my Audio Precision P1DD to check few basic parameters (level THD+N) on both channels everything was consistent with few tenth of a dB off.
Then I carried out a 0dB FS output level back with the NF and the P1DD and noticed 6 % difference in level. Buh?
Because everything is integrated in the PCM5201A I'm use to more accuracy.
So I started my very old Panasonic analyzer, I hate the Panasonics fans so despite it's very quality I do not use it. I measured a 20dB THD+N difference at the DAC output. Interesting.
I confirmed the P1DD THD+N was matched with a second measurement.
Then I traced both output signals with a scope at 20kHz and guess what the chip is defective. One output is filtered, the other not.
An externally hosted image should be here but it was not working when we last tested it.
Conclusion:
That for me a multiple surprise, I never face to a partially defective chip. The PCM 5102A is a stereo chip you cannot select filter for one channel only (maybe with China counterfeits).
Then THD+N measurement show it's limits when identify a problem, one analyzer is "blind" the other show a huge difference.
PFB
Ouch, that's really strange.
http://www.ti.com/lit/ds/symlink/pcm5101.pdf
What sample rate and bit depths were used for these measurements?
Is there some other possible explanation? What is the source of the signal for all measurements shown? Is it the P1DD, and is it possible that something is wrong with it?
You need to find an alternate digital source. (Even a CD player with a SPDIF output playing test tones downloaded off of the internet and burned to a CDR might provide adequate evidence of a real problem and validate that the P1DD/source used is OK. A known good DAC is another way to validate that source)
TI I suspect would be very interested in seeing this if true, and possibly the chip if not genuine.
http://www.ti.com/lit/ds/symlink/pcm5101.pdf
What sample rate and bit depths were used for these measurements?
Is there some other possible explanation? What is the source of the signal for all measurements shown? Is it the P1DD, and is it possible that something is wrong with it?
You need to find an alternate digital source. (Even a CD player with a SPDIF output playing test tones downloaded off of the internet and burned to a CDR might provide adequate evidence of a real problem and validate that the P1DD/source used is OK. A known good DAC is another way to validate that source)
TI I suspect would be very interested in seeing this if true, and possibly the chip if not genuine.
For sure missing the anti-alias filter on one channel. Odd indeed. It' snot outboard is it, as Jens mentions?
Looks like 88.2K 8X over-sampled. Nice image of what the filter actually does.
Hello,
The DAC is encapsulated so I have no access to RC output filter. The actual PCM5102A datasheet is here http://www.ti.com/lit/ds/symlink/pcm5102a.pdf
Application shows an RC filter.
The application generator I use is "ToneGenerator" form NCH I had numerous problems with my MAC with various generator apps generating a continuous stable tone. I use this app to stream via USB a signal.
I set various sample rates all creates the same effect, from 44.1 and up to 192. AP P1DD connected to digital out (SPDIF) shows the active bits at close to FS and 20kHz bit 24 to 8 are active. P1DD max sampling rate is 96kHz.
The curious DAC mesure as follow:
THD+N, Panasonic VP7722A LPF30kHz, averraged, 0dBFS 1000kHz
R -84,5dB (0,00609%)
L -60,15dB (0,0967%)
THD+N Panasonic VP7722A LPF30kHz, averraged,-1dBFS 1000kHz
R -85dB (0,00554%)
L -62dB (0,0790%)
THD+N avec Audio Precision P1DD 22Hz-22kHz, averraged, -1dBFS 1000kHz
R -85,4dB (0,0053%)
L -85,5dB (0,0052%)
The trace are the output level L and R 1:1 no probe direct feed at RCA connector. I have no access to chip because of encapsulation. I dis it with probing because of LF involved (20kHz).
I swapped almost every things results come from DAC (maybe RC filter with a C out).
But what is really curious is AP P1DD tells every things is fine....measurement is analog in.
P1DD is a work horse, pass all self test, I was very confident in it's results (until now) it's stable runs a 300kHz analyzer, I never add a problem. My panasonic is also super fine.
PFB
The DAC is encapsulated so I have no access to RC output filter. The actual PCM5102A datasheet is here http://www.ti.com/lit/ds/symlink/pcm5102a.pdf
Application shows an RC filter.
The application generator I use is "ToneGenerator" form NCH I had numerous problems with my MAC with various generator apps generating a continuous stable tone. I use this app to stream via USB a signal.
I set various sample rates all creates the same effect, from 44.1 and up to 192. AP P1DD connected to digital out (SPDIF) shows the active bits at close to FS and 20kHz bit 24 to 8 are active. P1DD max sampling rate is 96kHz.
The curious DAC mesure as follow:
THD+N, Panasonic VP7722A LPF30kHz, averraged, 0dBFS 1000kHz
R -84,5dB (0,00609%)
L -60,15dB (0,0967%)
THD+N Panasonic VP7722A LPF30kHz, averraged,-1dBFS 1000kHz
R -85dB (0,00554%)
L -62dB (0,0790%)
THD+N avec Audio Precision P1DD 22Hz-22kHz, averraged, -1dBFS 1000kHz
R -85,4dB (0,0053%)
L -85,5dB (0,0052%)
The trace are the output level L and R 1:1 no probe direct feed at RCA connector. I have no access to chip because of encapsulation. I dis it with probing because of LF involved (20kHz).
I swapped almost every things results come from DAC (maybe RC filter with a C out).
But what is really curious is AP P1DD tells every things is fine....measurement is analog in.
P1DD is a work horse, pass all self test, I was very confident in it's results (until now) it's stable runs a 300kHz analyzer, I never add a problem. My panasonic is also super fine.
PFB
Hello,
I played various frequencies, at various sampling rate. Despite I'm not sure how exactly this PCM5102A handle various sampling rate I discovered that the distortion vary by a conjonction of frequency and sampling rate is not stable.
I use the monitor output of the Panasonic to trace on my scope the distorsion figure. At HF 20kHz or so the THD+N is high and pulsed at 10kHz or so one channel looks unfiltered like my previous picture. But the truth is the wave switch from filtered to unfiltered at 10kHz. If I lower the frequency the wave start to literally pulse on the defective channel 500msec with THD+N, 500msec "normal".
This unstable pulsed distortion may explain the behavior of my analyzers. One average more "quickly" and detect the trouble, the P1DD is slower and does not see the "pulsed THD+N. At certain frequency both analyzers cannot latch and mesure a valid THD, P1DD indicate 999dB and the Panasonic display switch off.
I have a very good contact with the DAC manufacturer, he told me some early design suffered form curiosities and since new lots were tested ok since years he just performs basic test with a digital analyzer.
PFB
I played various frequencies, at various sampling rate. Despite I'm not sure how exactly this PCM5102A handle various sampling rate I discovered that the distortion vary by a conjonction of frequency and sampling rate is not stable.
I use the monitor output of the Panasonic to trace on my scope the distorsion figure. At HF 20kHz or so the THD+N is high and pulsed at 10kHz or so one channel looks unfiltered like my previous picture. But the truth is the wave switch from filtered to unfiltered at 10kHz. If I lower the frequency the wave start to literally pulse on the defective channel 500msec with THD+N, 500msec "normal".
This unstable pulsed distortion may explain the behavior of my analyzers. One average more "quickly" and detect the trouble, the P1DD is slower and does not see the "pulsed THD+N. At certain frequency both analyzers cannot latch and mesure a valid THD, P1DD indicate 999dB and the Panasonic display switch off.
I have a very good contact with the DAC manufacturer, he told me some early design suffered form curiosities and since new lots were tested ok since years he just performs basic test with a digital analyzer.
PFB
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.