Filtering USB power lines

speedcenter · « **on:** 17 Jan 2006, 09:25 pm »

I'm fighting some gremlins in my USB-I2S interface and need to find a somewhat cost efficient way to brute-force filter the 5V power coming from the computer. Does anyone have some suggestions? Anything short of a $169 optical USB extender would be a step forward from the options I am looking at right now. I cannot cut the power line, even thogh the DAC is self-powered. The silly PCM2707 chip needs to see the PC power to be told that it is self powered and to be recognized on the USB bus. Not sure how to work around that (I tried, trust me...), and all my computers seem to introduce a really annoying buzz to the audio except for the laptop when running on batteries.

Will I need a transformer for proper ground isolation? Anyone know how these little gadgets here work: http://www.icsdatacom.com/pdfs/USB-GT_ds.pdf I figure the parts in there can't be all that expensive or high tech, but I don't want to pay $250 for a medical grade device.

Looking for suggestions. I really don't want to start working on cleaning up the big PSU in my computer which isn't even in the same room with the audio system, nor on the same circuit.

Peter

westend · « **Reply #1 on:** 18 Jan 2006, 06:34 am »

Besides rewiring the DAC's 5v. USB input to a dedicated 5v. and dealing with the necesary upstream signal problems, I would look at the actual ground points of the USB jack, the quality of the cable, the motherboard connections (if you are connecting through a pinout to a backplane or case), and the computers PSU and 110v. ground. From your post, it sounds like you have a ground loop in the circuit. You could try a ferrite close to the computer or close to the DAC to see if that would help.

speedcenter · « **Reply #2 on:** 18 Jan 2006, 05:37 pm »

Quote from: westend

Besides rewiring the DAC's 5v. USB input to a dedicated 5v. and dealing with the necesary upstream signal problems, I would look at the actual ground points of the USB jack, the quality of the cable, the motherboard connections (if you are connecting through a pinout to a backplane or case), and the computers PSU and 110v. ground. From your post, it sounds like you have a ground loop in the circuit. You could try a ferrite close to the computer or close to the DAC to see if that would help.

I already did the 5V replacement with a regulated battery supply. That doesn't work, i.e. the DAC won't register with the computer unless the power of the computer is connected, and at that point, it buzzes. Break it after the computer registers the DAC, the computer will report a problem with the USB device. Note that most of the DAC runs off a 3.3V supply, which will cause the same problem if disconnected, i.e. the thing really doesn't run on 5V. The "upstream signal problem" is what? the DAC reporting to the computer that it's there? How can I fake the computer into believing there's a device attached without the power connection being there?

The buzzing happens with three different PCs, while it doesn't happen on the battery powered laptop (plug in the charger and cringe, though...), slightly different flavor each time, probably based on the type of power supply that's used.

Ground loop? The DAC is completely battery powered and all it connects to are the power amps. Putting the laptop on the same circuit, even on a big isolation transformer, causes the buzz. Yes, ground loop, but not one you can get rid of easily. The DAC is dead silent in the same righ using SPDIF from a regular transport, but clearly, that bypasses the whole USB interface board.

I'm doubtful that a ferrite bead on the power feed will do much, given this is highly audible (about -20db or louder), but I'll start with that. Ultimately, there's that optical USB coupler. What bugs me is that the designer of the board uses it with various AC-powered PCs and claims it's dead silent within the same type of battery-powered DAC.

Peter

hanguy · « **Reply #3 on:** 18 Jan 2006, 07:53 pm »

Quote

I already did the 5V replacement with a regulated battery supply. That doesn't work, i.e. the DAC won't register with the computer unless the power of the computer is connected, and at that point, it buzzes.

Peter, I have successfully converted my M-audio Transit to use an external linear power supply. The key is to connect the ground of the power supply to the ground pin of the USB connector. I was getting the same error that the computer did not see the Transit until I have the grounds connected. You may be able to connect the ground through a ferrite bead to further attenuate the HF noise from the computer.

You will probably find the external power supply gives you better sound from your DAC because you keep the noise away from its power supply.

Mike

speedcenter · « **Reply #4 on:** 18 Jan 2006, 10:12 pm »

Quote from: hanguy

Peter, I have successfully converted my M-audio Transit to use an external linear power supply. The key is to connect the ground of the power supply to the ground pin of the USB connector. I was getting the same error that the computer did not see the Transit until I have the grounds connected. You may be able to connect the ground through a ferrite bead to further attenuate the HF noise from the computer.

You will probably find the external power supply gives you better sound from your DAC because you keep the noise away from its power supply.

Mike

"further attenuate" ??

you have no idea how loud this is. The moment I connect ground, the whole thing starts buzzing.

The DAC power is all battery, super regulated, blackgates. No noise. It's all coming from the computer AC supplies. As I said before - I am not using the computer power. The TI PCM2707 USB chip requires the connection to the computer power on one pin to tell it NOT to use it. How stupid is that? But that's what the data sheet states. Pin 3 is labeled "HOST" and it has to see 5V from the computer to tell the DAC that it is actually self-powered. Too bad that the noise form the computer gets in no matter how silent the DAC supply is.

I also have a Transit (totally gutted and rebuilt by now) and a M-Audio USB Audiophile. Neither will do any of this, but then the transit is a toslink unit, and the Audiophile doesn't need to be seen by the computer to be self-powered. You can just turn the power on and it shows up on the computer. The PCM2707 will not show up without being essentially powered by my own supply AND the USB bus.

the reason I am trying so hard with the PCM2707 is that it is direct USB-I2s and I am eliminating the entire SPDIF interface. And as long as I am running the source computer on batteries, everything is working just grand. Plug in the AC power and the show is over...

Peter

hagtech · « **Reply #5 on:** 25 Jan 2006, 05:21 am »

Try putting some ferrite beads around the USB cable. You can buy clamp-on ones. Or loop the USB cable through a big toroid core a few turns.

Now you know why I chose to use S/PDIF output on HagUsb instead of I2S. Requires only one relatively inexpensive output transformer. And it gives really good power and ground isolation.

jh

speedcenter · « **Reply #6 on:** 25 Jan 2006, 05:08 pm »

I ordered an optical USB coupler - got tired of fighitng noise that I can keep isolated on the digital end. Cheaper than trying to fight it in the computers. I should have it today or tomorrow. SPDIF is so 80s

Peter

speedcenter · « **Reply #7 on:** 28 Jan 2006, 03:48 pm »

I received a 10 meter optical USB extender yesterday. Not cheap, but it clearly solves all problems at the root. Given I had to get a long USB link to a remote room server anyway, I can almost justify the price ($177 including shipping...

)

Total silence in my speakers even with the noisy laptop switching charger supply plugged in.

There probably are cheaper ways to do an opto coupling inside the DAC, but that'll have to wait for the next DAC I build.