[ble0t]

Putting small inductors in those traces might help things...

 

[Francis_Vaughan]

We have to allow rise times that allow the USB signal to work.

But the USB noise is a problem. The USB signals are differential - and this probably helps quite a a bit, but there will be comon mode noise as well. This will come in common on the USB signals and across the power or ground. Since there is no signal in either of these reasonably agressive ferrites on those legs would seem to be a good idea. Indeed one wonders about significant low pass filtering.

Other than that, galvanic isolation is a good idea, but does indeed need a transformer. Still not a total panacea.

 

[00940]

on my usb2.0 add-on pci card, there are ferrites on the ground and power usb lines just before the usb jacks. It probably cannot hurt to add two more at the dac side.

 

[Francis_Vaughan]

Another nasty thought - well potentially mitigating thought. Your test setup, I assume, uses the same computer for both input and output. So there is very likey a common ground led through the USB ground. This could easily cause significant problems. Problems that are never seen in normal operation of the DAC. Hard to know how to fix, but there may be some benefit in playing with dressing the cables, shorter USB cable, playing sith signal ground on the input to you sound card (such as tying it to the USB ground.)

 

[Alf]

I found the problem. A nut on a RCA connector got loose causing poor electrical connection to the ground. This caused excessive noise. The noise level changed every time I touched the board. I tightened up the nut and finally I started getting consistent results. Well, the difference between individual results now is within 1% for noise and DR.

Also I bought an expensive USB cable (Profigold 1.5m) to check whether it makes any difference. According to the description, the cable features some superior Interference Absorption Technology (yeah sure) eliminating EMI. It seems that the cable does make some difference. However it is within 1.5db which is very close to the 1% margin.

With the new cable and the fixed connector I get figures around 85-86 for noise/DR if USB powered. If a battery is used, the figures go up a bit to 87-88. I cannot get 90-91 I got during my very first tests. I will continue my investigation tomorrow. Subjectively I cannot tell much difference between USB and battery. Maybe USB power sounds a bit better but I cannot tell for sure and I cannot explain why.

We need alternative RMAA measurements to find out whether the figures above are specific to my test environment and my substandard soldering skills or it is indeed a feature of the new revision. Can someone help with this?

 

[Sinbios]

Hm, from the measurements, it sounds like Rev A is actually better when powered with batteries?

Then again, the Rev A tests were done with a 1212 - might that have contributed to better numbers?

 

[Alf]

Quote:

Originally Posted by Sinbios Hm, from the measurements, it sounds like Rev A is actually better when powered with batteries? Then again, the Rev A tests were done with a 1212 - might that have contributed to better numbers?

Magsy did tests with EMU0404. The best figures were when connected to a laptop. It did not matter whether it was USB or battery. Then goes PC/battery and the worst one is PC/USB.

There is no information whether ITZBITZ used a laptop or a PC.

 

[DaKi][er]

Hmm, just thinking aloud here

Transformer for the D+ and D-, something that is used in plain digital or networking type, and optocoupler for the power, as the pcm2702 only really needs a voltage here to say that the USB is plugged in, then run it off battery/external from there
I would think that they will need to be coupled to the same ground still

Just thinking

 

[Alf]

I have been playing with the DAC yesterday trying different powering options. And then I asked myself a question. PCM2702 requires only 4.5V to run properly. What would happen if I connect USB VBUS directly the 9V pad? How would REG102 react if only 5V is supplied?

I decided to check it out. I connected S1 to AV+. My DMM showed 4.9V at the 5V pad and perfect 3.3V at the 3V3 pad. It was close enough and the DAC was working!

Then I decided the check how (bad) it performs. RMAA produced results that indeed surprised me. Bypassing both the switch and LM317 improved the performance significantly. The figures went from mere 85db up to 92-93db. I repeated my tests a few times trying different configurations but the results were consistent.

USB
Battery
Comparison

Well, it seems that Revision B is an over-engineered design. It is about time to take away what is superfluous. The switch section must be removed. It was not needed there in all revisions. I am not sure about the LM317 section yet – it may be useful for some people.

I noticed quite a few factors that affect the DAC’s performance I cannot say it is audible, but RMAA shows some difference (1-2db each). These factors include but not limited to:
* USB cable.
* USB port.
* HDD activity.
* Output capacitors.

 

[guzzler]

Interesting Alf, how reproducable are the results on the cable and port? I'm tempted to think a 1dB difference is sensitive to the setup, and certainly not audible. A 7dB reduction by skipping the switching regulator is good, and definately remove it if it gives such a good improvement (although still probably not hearable

)

 

[MASantos]

Alf I would like to apologise for not having a review or some coments yet.
I have tons of college work now but I will try to take some time to make some listening tests and comparisons between revision A and B.

Manuel

 

[00940]

Alf : if you consider going that road, you could perhaps change the REG102 for the TPS793** serie. Those regulators are cheaper, almost as good and there's one that outputs 4.75V if you feed him with 5V. It will allow you to cope better with varying usb voltages (some usb ports don't output the full 5V). The PCM2702 can live according to the datasheet with as low as 4.5V.

 

[Mod_Evil]

Hi,

Me and a Friend have mounted this a DAC with the PCM2704.

But in now we will balance my dac. How I make this?

Can I make my balanced DAC with the DRV134?

Thanks.

 

[ATAT]

IIRC.. you need DACs for each signal (ie 4) since the PCM2704 is single-ended.

 

[00940]

The pcm2704 is a stereo chip.

Yes, the DRV134 will give you a balance output. Considering the quality of those chips, no need to go to anything more complicated.