Experience with this DIY DAC ?

[Kevin Wood]

I am torn between doing the mod posted and modifying the output stage to the configuration used on the CS4398 eval board, which also does away with the coupling caps (bar one on the line output). I'm guessing Crystal will have built an eval board that will do their chip justice so it is probably worth a look. For now I'm just going to enjoy it though.

The CS4398 does upsample as part of the digital filter that's built into the chip, and the filter performs best if you run the chip in the mode that gives maximum upsampling of your source data. There's also a switched capacitor filter after the converter.

This means that the analogue output is free from crud close in to the audio passband so you can use a gentle low pass filter in the analogue output stage (-3db @ 38KHz on the eval board) that doesn't roll-off the treble. This is where it wins over NOS designs, of course.

I mainly feed mine from a Squeezebox 3 playing FLAC files ripped using EAC (this DAC makes a huge improvement over the SB3's analogue output!). Interestingly my Alpha One CD player didn't sound too hot initially. When I looked at the RF waveform from the laser it was obvious why so that now has new optics and is sounding good again, otherwise a Shigaclone would have been the next project for me!

Kevin

[ah-luo-ko-ko]

To answer Puffin's question, the oversampling is as follow:

Single speed mode: I/P sampling rate up to 50K - 128x oversampling
Double speed mode: I/P sampling rate up to 100K - 64x oversampling
Quad speed mode: I/P sampling rate up to 200K - 32x oversampling

I think Kevin is right on in saying that the best configuration is the single speed mode and I am not sure why it works when I configure the DIP switches with double speed mode when the I/P sampling is at 44.1K though?

Currently I am using the OPPO970 as the 'transport' to feed into this DAC. I've done some 'capacitor upgrade' to the OPPO970 in both the analog O/P circuit and also, the PS. I have also added bypass capacitors at various 'points' of digital chip of the OPPO.

I remain presently surprised that this DAC sounds 'subjectively better ;-) than the mod OPPO970 analog out. Currently, I am using this DAC to feed into a Yain tube buffer (also modded using input from other threads in this forum), than to a tube amp.

I will try out the LM4562 chip when I receive them.

BTW, the OPPO970 appears to be capable of outputting at 192K with its HDMI connector. I am not sure yet if it is straight forward to tap of the digital signal from the HDMI connector to feed into this DAC?



Kevin - Do appreciate if you will post your experience with this DAC when you start 'modding' it.

Do encourage others who have experience with modding this DAC to support this thread.

One thing for sure though - it does sound good as is.

One other note - I did change the capacitor values for the PLL for this DAC. It appears that the stock capacitor values are the ones recommended for CS4397. If you look at the spec for the CS4398, the PLL values recommended are different from those that are mounted in the stock board.

[gabrielbecheanu]

Hi
It is possible to make this DAC to decode HDCD?
It will be nice to put some led to show us what kind of signal goes inside. Ex 44khz 96 kHz 192 kHz or HDCD signal.
Thanks

[Kevin Wood]

I don't know a lot about HDMI. My feeling is that, if the appropriate mode is selected, the DAC is operating at or near its' highest sample rate and the oversampling is carried out internally in a digital filter whose characteristics are carefully matched with the switched capacitor filter that follows the DAC.

Upping the rate elsewhere in the chain is not likely to be advantageous, IMHO, but that's purely a gut feeling that I can't back up with any listening. I've not fed anything other than 44.1 into it yet.

The CS8416 outputs a "96K" signal that indicates if the incoming sample rate is above an arbitrary threshold that's somewhere between 48k and 88k but to determine the input rate exactly would require some additional hardware as far as I can see. This would be required to switch the DAC into Quad speed mode for inputs above 100K.

Sounds like a microcontroller driving a display on the front panel, counting the sample rate via one of its' timer channels and configuring the CS8416 and CS4398 according to what it sees.

Kevin

[luxury54]

take a look at post #88 here,how it can be attached a tube output stage to this dac which i'm very pleased with
DAC project completed

[Kevin Wood]

Well, a tube output stage is something else that was in the back of my mind to try. Very interesting.

Interesting also that you changed the PLL mode of the 8416 to minimise outband jitter. I noticed this setting appeared to be non-ideal. Looks like I'm going to be busy with this DAC but given that I'm already very happy with the sound I have high hopes for what can be achieved with a little work.

Kevin

[johnm]

Kevin, have you decided which mod to implement yet? I can't help feeling that the op-amp mod, with no capacitors in the way to muddy the sound will be very impressive indeed.

What other options are there... can the output from the first op-amp be taken directly from the op-amp leg and just put through a 2.2uF film cap - for example - and directly to phono sockets? Would that work? Probably easier than trying to squeeze a 14K7 resistor in there as per the first mod I posted.

- John

[Kevin Wood]

It should do. The 14K7 resistor is only an attempt to cure the offset anyway, at the expense of injecting power supply noise.

I haven't obtained / figured out the whole circuit yet so I'm not 100% sure what the purpose of the 2nd opamp is. I suspect it's just a buffer stage and totally unnecessary.

It might also be interesting to take a look at what Cirrus do on their Eval board Here (Page 11). Pretty much what you describe. DC coupling from the DAC to the first OpAmp and just a single decoupling cap on the output. If you really get keen you could use an integrator to feed back the DC error and cancel the offset.

When I can prise myself away from listening to it I think I'll try both the valve route and the Op Amp. See what sounds best.

Kevin

[johnm]

Super link - thanks for that

Could you e-mail me off-board please Kevin, I may be able to help with something: private23 AT hotmail.co.uk

- John

[chriscox]

Hello,

Without trying to hijack this thread, can I ask what the current requirements are for these boards on the 15v and 8v supplies, please?

I've purchased one of these without the transformer and have no specs for this.

Thanks,

Chris