I've been playing with PCM2900's for a long time, improving my skills of PCB design. A stubborn issue is how I often have a clicking sound on my headphones, whenever playback starts or stops. When I set the volume to zero in a media player, and start or stop playback, the click is distinctive. A few boards didn't have the issue, but I can't reproduce when it does or doesn't happen.
The PCB is double-sided with a good layout, almost all SMD parts, good quality capacitors with capacities according to the PCM2900 datasheet. There's an opamp to amplify the signal, with a switchmode supply. There's an USB common mode filter on the 5v and GND, and an IC to limit inrush current.
I had a look with a digital oscilloscope at the current board. With one channel playing a sine wave and the other audio channel muted. The clicking sound seems to be a bump in the DAC output, right around when the databus starts running.
Power supplies look clean, except for VCOM and VBUS that are a bit irregular a bit before playback starts, that's when the USB databus starts running. There's a big capacitor where USB power enters the board, then a 2R2 resistor, and a 1uf capacitor close to the board. VCOM has a 10uf and 100nf capacitor. All of these are according to the datasheet. Most power supply decoupling capacitors are specced at 1uf in the datasheet, it says 'max 2uf'.
I might go and increase capacitors, I don't know. I'm looking for some help.
zooming in on the bump:
Vbus:
Vcom:
The PCB is double-sided with a good layout, almost all SMD parts, good quality capacitors with capacities according to the PCM2900 datasheet. There's an opamp to amplify the signal, with a switchmode supply. There's an USB common mode filter on the 5v and GND, and an IC to limit inrush current.
I had a look with a digital oscilloscope at the current board. With one channel playing a sine wave and the other audio channel muted. The clicking sound seems to be a bump in the DAC output, right around when the databus starts running.
Power supplies look clean, except for VCOM and VBUS that are a bit irregular a bit before playback starts, that's when the USB databus starts running. There's a big capacitor where USB power enters the board, then a 2R2 resistor, and a 1uf capacitor close to the board. VCOM has a 10uf and 100nf capacitor. All of these are according to the datasheet. Most power supply decoupling capacitors are specced at 1uf in the datasheet, it says 'max 2uf'.
I might go and increase capacitors, I don't know. I'm looking for some help.
zooming in on the bump:
Vbus:
Vcom:
Might help if we knew more about the dac. It is USB based? How are you starting and stopping playback when the problem occurs? Etc.
EDIT: Maybe also helpful if we could see clear hi res pics of the top and bottom of the board.
EDIT: Maybe also helpful if we could see clear hi res pics of the top and bottom of the board.
Last edited:
yes, it's an USB DAC. playback starts and stops simply on a media player in windows. no other software accessing the audio. maybe I should see if the issue persists in Linux
I'll take some photos tomorrow
USB board in use, your own XMOS implementation, other? If USB board, does it have a MUTE output that is being used?
the DAC has HID0 for mute. it's active-high, 3.3-V CMOS-level input with internal pull-down. right now not connected to anything.
I don't know what you mean with that
I used to hear click when i listened pandora. At the start of every song. It was not dac related, it was laptop software generating those clicks. Once i desabled windows system sounds, all was good. I used external dac.
Okay, I see. PCM2900 data sheet says it has a built-in USB 1.1 interface.
In contrast to that most modern DACs use an external USB 2 interface. In that case the USB interface MUTE signal may need to be interfaced to the DAC chip.
Anyway getting back to PCM2900, sometimes USB interfaces don't mute a DAC when it should be muted, and sometimes computer audio applications send clicks or pops that the USB interface doesn't know about. In other words, it means the problem you want to solve might be in hardware or software. It could even be that sometimes hardware causes it and other times its the software.
dang!
I already have all windows sounds disabled.
Hmm. Sometimes it doesn't happen. I'll test the mute function and see if it makes a clicking sound.
Perhaps I should look at another chip anyway.
Mute shouldn't make a clicking sound. Rather, mute could be used to hide clicks that would otherwise be audible.
Interesting I have the same issue with DIY AD based dac , the chicks are not that often but still i can hear them when musing is not playing , is the only cure for now using Linux , any ASIO tested ?
Does the audio go thru a dc blocking caps? If so, check what happens with the dc level before the he caps.
What is the transfer technology from PC to this DAC, USB audio? If so, what USB controller chip is used, what firmware?
Could it be the software driver putting the DAC into some sort of standby or power saving mode when there is no audio playing?
I bring that up because I had this exact problem with two different commercial USB audio interfaces from Focusrite, a Scarlett 6i6 and a Clarett 2Pre. This happened under Linux. I did not test the Windows behavior. I tried all sorts of OS and audio subsystem tweaks but could not get it to stop. When audio data ceased to flow, after about 15 seconds the unit went into some sort of sleep mode and reset the audio format and rate values. Whenever I later restarted playback there was a pause and a faint pop. It seemed that the sleepy time audio format and rate were not the same as the audio I was sending to the DAC, so it had to change over the clock internally (44.1 vs 48kHz) or something.
In the end I used a "dumb" approach: I wrote a script that used aplay to send /dev/zero data to the DAC using the same format and rate as the audio data I would otherwise send. When I stopped sending my own audio data (via an app I wrote) I fired up the zero-data script to keep the interface alive and running. When I want to send audio data to it again, I just kill the script before doing so (all automated). Works like a charm. Not a universal solution, however!
You could test for this same behavior using some other Windows audio app for some period of time, then killing that and starting to send your audio data again.
I bring that up because I had this exact problem with two different commercial USB audio interfaces from Focusrite, a Scarlett 6i6 and a Clarett 2Pre. This happened under Linux. I did not test the Windows behavior. I tried all sorts of OS and audio subsystem tweaks but could not get it to stop. When audio data ceased to flow, after about 15 seconds the unit went into some sort of sleep mode and reset the audio format and rate values. Whenever I later restarted playback there was a pause and a faint pop. It seemed that the sleepy time audio format and rate were not the same as the audio I was sending to the DAC, so it had to change over the clock internally (44.1 vs 48kHz) or something.
In the end I used a "dumb" approach: I wrote a script that used aplay to send /dev/zero data to the DAC using the same format and rate as the audio data I would otherwise send. When I stopped sending my own audio data (via an app I wrote) I fired up the zero-data script to keep the interface alive and running. When I want to send audio data to it again, I just kill the script before doing so (all automated). Works like a charm. Not a universal solution, however!
You could test for this same behavior using some other Windows audio app for some period of time, then killing that and starting to send your audio data again.
Yes USB is based on Anamero transport - Atmel chip and i hear clicks only when using it , there is caox in with AKM and there are not clicks .
Are these clicks only when full silence is played/playback stopped, or also during very quiet passages (i.e. the clicks are not loud and appear also during playback, just not audible)?
They appear when there is silence in the music , for example at the end of drum solo , or when singer not singing but the track is playing .
Do you have a schematic you can share? If so, maybe that would help shed some light on how clicks might occur.