[gonesurfing]

Hello all,

I recently built my first DAC with TI's pcm2902 USB chip. I basically used the datasheet schematic example with all recommended values for resistors, caps, and voltage regulation. When using the chips digital out over Toslink to my receiver I get full and clean sound. However if I use the DAC function and go analog out to the same receiver via rca cables, the sound is fairly soft and I have to turn the amp volume up an extra 15db to get the same volume. Doing so is obviously having an effect on sound quality. The lows are overpowering with mid and high frequencies lacking. I know this is not a top of the line dac chip, but I expected a little more sound out of the chip than this. Datasheet indicates vout as .6*vcci which should be around 2.25v pp with the REG103 regulator I'm using. I see other designs using this and similar chips without opamp amplification. Any idea where I might be going wrong? Am I interpreting the datasheet correctly?

Thanks in advance.

 

[tomb]

Quote:

Hello all,

I recently built my first DAC with TI's pcm2902 USB chip. I basically used the datasheet schematic example with all recommended values for resistors, caps, and voltage regulation. When using the chips digital out over Toslink to my receiver I get full and clean sound. However if I use the DAC function and go analog out to the same receiver via rca cables, the sound is fairly soft and I have to turn the amp volume up an extra 15db to get the same volume. Doing so is obviously having an effect on sound quality. The lows are overpowering with mid and high frequencies lacking. I know this is not a top of the line dac chip, but I expected a little more sound out of the chip than this. Datasheet indicates vout as .6*vcci which should be around 2.25v pp with the REG103 regulator I'm using. I see other designs using this and similar chips without opamp amplification. Any idea where I might be going wrong? Am I interpreting the datasheet correctly?

Thanks in advance.

The application schematics for the PCM2902 all show "LPF amp" on the analog outputs (Low-Pass Filter amp).
 
It's just a guess ...
 
Edit - was looking at the wrong thing for Vout, where 0.6 x Vcci is Vpp.  I get a 1.98 Vpp, meaning 4.29V top and 2.31V bottom.  That should actually be plenty, so I'm not sure what's going on, either.  Did you use the proper capacitors on the output to block the DC?  It may be pegging your receiver inputs if you didn't.

 

[gonesurfing]

The application schematics for the PCM2902 all show "LPF amp" on the analog outputs (Low-Pass Filter amp).

It's just a guess ...

Edit - was looking at the wrong thing for Vout, where 0.6 x Vcci is Vpp.  I get a 1.98 Vpp, meaning 4.29V top and 2.31V bottom.  That should actually be plenty, so I'm not sure what's going on, either.  Did you use the proper capacitors on the output to block the DC?  It may be pegging your receiver inputs if you didn't.

Yeah, I had a LPF on the output with a -3db cutoff around 33khz but I was worried that was part of the problem and removed it. I am using the 10uf coupling caps that are recommended, and went with Elna electrolytic.

 

[wakibaki]

Quote:

When using the chips digital out over Toslink to my receiver I get full and clean sound. However if I use the DAC function and go analog out to the same receiver via rca cables, the sound is fairly soft and I have to turn the amp volume up an extra 15db to get the same volume. Doing so is obviously having an effect on sound quality. The lows are overpowering with mid and high frequencies lacking.

 
You can't necessarily expect the same volume when using these 2 different signal paths. When all is said and done the path using Toslink has an unknown gain. You say that 'turning up the amp volume 15dB is obviously having an effect on sound quality'. I don't think that this is an obvious effect at all. I would expect that you would hear no difference in sound quality at all except perhaps a difference in background hiss. That's the purpose of an amp if it is truly transparent (which most amps of reasonable quality are nowadays, unless they are 'audiophile' equipment intended to introduce colouration).
 
It's commonly known that a difference in volume is often perceived as a difference in SQ. You cannot be sure that such a difference exists unless you have done a blind comparison with the volumes matched to <1dB (using instrumentation).
 
Have a think about what might be going on in the receiver in terms of amplification and EQ and try to make your observations a bit more objective before you jump to any quick conclusions about a performance deficit in the analog path.
 
w

 

[gonesurfing]

Quote:

 
You can't necessarily expect the same volume when using these 2 different signal paths. When all is said and done the path using Toslink has an unknown gain. You say that 'turning up the amp volume 15dB is obviously having an effect on sound quality'. I don't think that this is an obvious effect at all. I would expect that you would hear no difference in sound quality at all except perhaps a difference in background hiss. 
 
w

 
The observation with volume/SQ was based on a few things. Please forgive my terminology as I'm sure it is both lacking and incorrect...
 
-Observation #1 was that the path with the higher gain (toslink) appeared to have more "range" or "depth" and very clear higher frequencies. The best way I can describe it, is that the analog path sounds almost compressed or flat similar to FM radio vs CD but not to that extent.  
 
-Observation #2 was that the headphone output on the laptop has both higher gain and sounded less flat.. Turning up the volume +15db did not make up for this flatness, but turning the gain on the USB out volume above 100% (alsa linux) seemed to help.
 
Thanks for the feedback so far. I'm still stuck on the coupling capacitors as being a possible issue.  From what I have read on coupling capacitors, they essentially both block DC and act as a high pass filter based on capacitance and load impedance. Actually, the 10uF caps were not specified on the datasheet, but were on pavouk.org.  

 

[wakibaki]

It's widely known by those involved in serious objective audio testing that 'louder' is almost always heard as 'better'.
 
Your DAC may in fact be producing an inferior sound to the Toslink receiver, but verifying that this is truly the case will be extremely difficult. Just turning up an amplifier '15dB' and making a decision by listening to the result doesn't mean that it's truly so, so advising you as to what might be going wrong is well nigh impossible, especially since the problem may just be in your head.
 
If you've built the DAC to the datasheet, and I advise you to follow the datasheet, not something you found on some guy's website, then it should deliver a respectable sound. The chip is used in the Behringer UCA202 which has fared well in comparisons with other codecs, despite its bargain-basement price. If you could borrow one (UCA202) to do some comparisons with your DAC then you might be closer to discovering whether it is performing as it should. Bear in mind though that the UCA202 has some built-in amplification, and both line and headphone outputs.
 
Try reading this:- http://seanolive.blogspot.co.uk/2009/04/dishonesty-of-sighted-audio-product.html
 
w

 

[gonesurfing]

I've had some time to do some more experimentation, so I thought I would follow-up.
 
I replaced the 10uf Elna electrolytic coupling caps with Kemet polyester film 4.7uf. I can't compare A-B obviously but the output level or sound doesn't appear to have changed significantly.
 
I have an ART phono usb ADC that has the same PCM2902 chip, just an earlier version. Only way to get USB-analog is through the headphone output which has an amplified output. The gain can be higher due to the amplification but the sound is a little cleaner on the highs. The mids and low end really lack compared to my version. Score 1 for the homebuilt 2902!
 
Right now I'm listening to my galaxy s3 over USB to my homebuilt 2902 and it sounds pretty decent.  I compared it to the headhpone output and the input gain is just about the same. The sound is fuller over the PCM2902 than the built in DAC/headphone amp. Score 2 for the 2902!
 
The Intel HD soundcard on the Dell laptop still sounds better than the 2902. Higher gain, clearer highs, sound is full. Maybe this is a halfway decent soundcard?
 
In summary, the non-blind listening results are 
PCM2902 AO vs Harmon-Kardon DAC via TOSLINK = Harmon Kardon win without a doubt.
PCM2902 vs ART USB phono plus(2902) = Close, but my PCM2902 wins, probably due to lack of headphone stage.
PCM2902 vs Samsung galaxy S3 headphone out = PCM2902
PCM2902 vs Dell intel HD soundcard = Intel soundcard.
 
So apparently the analog performance isn't that bad afterall. Future plans include talking a look at PCM2704/6 and possibly i2c or spdif out to something a little more extravagant.