[Miguel]

Quote:

Originally Posted by SnoopyRocks 16nV/rtHz translates to an SNR of 112.9 dB for a 1Vrms source over the audio band. The input referred noise of the AD823 should not be a limiting factor in this case.

Nice, thanks!

I've just ordered these parts (AD823, 4x caps for line-out, cap for PSU) hope to mod it soon!

I am not still sure about the PSU cap (the one next to the 3.3V regulator). Anyway, I will try that mod as well to see how it performs.

 

[Garbz]

IIRC the 3.3v regulator ran the Codec chip, the actual opamp measured 5v single ended from the multimetre. At least I remember originally evaluating 5v opamps before I decided to build a dac.

 

[amb]

Since I have a Transit, this piqued my interest; so I opened it up and took a measurement. The voltage across the V+ and V- pins of the opamp was 4.47V. This is about half a volt lower than the operating range specified in the AD8066 datasheet. Bummer.

As for noise, since I use the Transit for RMAA measurements, the noise floor performance is of great importance to me. Included below is the noise performance of the stock Transit in loopback as measured by RMAA. Note that through most of the audio range the noise floor is well below -125dB, which is excellent (the two spikes at 8KHz and 10KHz are artifacts, probably interference). If I replace the opamp with inferior noise performance I fear that the noise floor would be raised.

 

[Garbz]

intreesting that thoes spikes affect 2 different channels. Did you re-run the test. ALso how did you measure? Ouput on max, input volume adjusted to compensate?

 

[Miguel]

Quote:

Originally Posted by Garbz IIRC the 3.3v regulator ran the Codec chip, the actual opamp measured 5v single ended from the multimetre. At least I remember originally evaluating 5v opamps before I decided to build a dac.

Hmmm interesting, maybe I could give a try to a 5v opamp like the AD8066 (I guess I could even try the AD8620 if the opamp is powered at 5V

) The only problem is that I will need to break an opamp each time I try; I don't know how to de-solder them properly. The "frying" technique that Garbz suggested me is the only way I know

Edit: sorry, not the AD8620 - it's rated for +-5V (10V single supply)

 

[Miguel]

Quote:

Originally Posted by amb As for noise, since I use the Transit for RMAA measurements, the noise floor performance is of great importance to me.

I agree, in your case the use of a low noise opamp like the stock one is a priority. I have a few spikes as well in my RMAA tests, but I have noticed this is common in RMAA tests. They must be some kind of interference, as you said.

 

[amb]

Quote:

Originally Posted by Garbz intreesting that thoes spikes affect 2 different channels. Did you re-run the test. ALso how did you measure? Ouput on max, input volume adjusted to compensate?

Yeah I re-ran the test and the spikes were still there. IIRC I set the output and wave volume to max, and input volume adjusted until the RMAA level "meter" shows -1dB or so.

 

[pburke]

Sunday I pulled a lot of parts off my Transit PCB - the Toslink sender is gone, all the electrolytics, the 3.3v regulator as well, the LED and the USB jack.

The reason for removing the USB jack is to install a new onethat doesn't connect the 5V from the bus to the board (two pins get clipped). Instead I'll be feeding it 5V from a very clean regulated battery supply. I also will feed 3.3V directly from a leparate regulator. The 10uF caps I'll replace with BG 22uF NX caps, while I am still looking for a good replacement of the 1uF cap near the output and a 2.2uF in another location. Much of what I have in my parts bin has leads that do not fit the tiny holes in the PCB. I'm a little reluctant to drill them out, but I guess I will have to do that. I won't really need a replacement for the 100uF cap, since that is on my external 3.3V supply that I will connect to the Transit in exactly the location of the 100uF cap.

The opamp and caps around it will probably be completely removed, since I only care about S/PDIF out. I grabbed all the datasheets and saw there's three S/PDIF outputs on the codec chip - gotta figure out where to tap in, but I guess the best place is the input pin of the now removed toslink connector. I just will have to build some sort of output stage, with or without transformer. The goal is to feed the output to a CS8412, which will be right in the same enclosure. Anyone ever do something like that before?

Peter

 

[Garbz]

If your putting the Transit into the DAC enclosure I recomend the following method which is what i'm working with. Use an ISO150 digital isolator to completely isolate the computer from the DAC. It needs the computer's ground, computer's 5v, and the input SPDIF signal, and it will spit it out the other side referenced to the DAC's 5v an DGND.

Failing that look for a 1:1 pulse transformer which has sufficient bandwidth to cover the S/PDIF signal (6ghz I believe.)

If your not doing anything analogue I seriously doubt that a battery supply will make any noticable or measurable improvement.

 

[pburke]

Quote:

Originally Posted by Garbz If your putting the Transit into the DAC enclosure I recomend the following method which is what i'm working with. Use an ISO150 digital isolator to completely isolate the computer from the DAC. It needs the computer's ground, computer's 5v, and the input SPDIF signal, and it will spit it out the other side referenced to the DAC's 5v an DGND. Failing that look for a 1:1 pulse transformer which has sufficient bandwidth to cover the S/PDIF signal (6ghz I believe.) If your not doing anything analogue I seriously doubt that a battery supply will make any noticable or measurable improvement.

I disagree on the power supply, mostly because I have heard the Transit directly against the M-Audio Audiophile, and the biggest difference there was the dedicated power supply in the Audiophile. I have both but I'd like to get the USB right into the DAC, which is why I am messing with the Transit right now.

Since I already gutted the thing and destroyed the 3.3 regulator, well, I gotta power it with what I got available, and that's a battery and some custom regulators, which are already built.

The transformer is an option (6 gigahertz?) - trying it without first. Here's the transformer I am looking at in case I can't do it with just some resistors: http://www.scientificonversion.com/digital_audio.html - I was going to get one for my CD transport anyway.

Peter

 

[Garbz]

My mistake the bandwidth occupation for a S/PDIF signal is 200khz to 6Mhz not Ghz. The point is the same you will need a transformer capable of such a speed.

http://www.epanorama.net/documents/audio/spdif.html

Should give some lovely insight.

If you are coupling it direclty I highly recomend isolating it via transformer or ISO150.

 

[Miguel]

I just modded the M-Audio Transit caps - PS cap changed by BG STD 470uF 16V (it almost didn't fit, this cap is huge) and output caps changed by BG NX 22uF 6.3V (just one per channel, I was going to try the super-E-caps config but it was too expensive, both caps costs me £8.50 in UK, around $15.50 USD). Burning it right now

I will mod the opamp later, as I am not still sure about it. I think I will try the AD823, I checked the opamp supply voltage in my Transit and it's 4.52V (similar to the 4.47V measured by amb) so 5V ones will not work.

Could you please give me your comments on this two opamps, or how they could compare to AD823? (if you know them, or understand their specs, I can't understand most of them): SSM2135, OP284

Both are rated to 4V, so they should work. The 1st one is specific for audio.

 

[Miguel]

BTW, the ssm2135 seems to be the one used in the Echo Indigo. I wonder how it performs against the AD823.

 

[SnoopyRocks]

The MAX4477 is worth considering. It was designed with this type of application in mind the and the specs reflect it: 2.7-5.5V single supply, rail to rail output, low power, low noise, high linearity and good speed.

 

[Miguel]

Thanks

I will have a look to MAX4477 specs (althought I will not understand half of them

) and then decide between AD823, SSM2135 and this one. Those are the only choices I have so far.

I am curious about SSM2135, as it is used in the superb Echo Indigo I/O, but some people have reported a not too good performance when using low impedance cans + echo indigo i/o (e.g. 26ohms) and it must be due to the op-amp (maybe the same happens with AD823 or MAX4477, no idea on this). As my main cans are the UM2 IEMs I am a bit concerned about this, as I plan to use the Transit sometimes without an external headphone amp.