Experience with this DIY DAC ?

Think I misunderstood your previous instructions Kevin? My DAC also has the two 22k resistors in place. But I thought these had to be replaced by 36k items you said earlier? :

"The filter stage posted on lampizator.eu didn't look right to me and has just occurred to me that the reason there is a DC offset at the output of the first Op-Amp stage after removal of the coupling capacitors is that the value of R14 (22k on the circuit diagram) is incorrect. To achieve common mode rejection down to DC it needs to be the same value as R8 (36K). The 14.7K resistor is then no longer required."

If that's not the case then great - not a fan of de-soldering in general ;)
 
Hi John,

It was the fact that the schematic showed R8 to be 36k and R14 to be 22K that rang alarm bells. If these resistors are matched in value there's no problem. There's a bit more gain if they're higher values and, in fact, on my DAC they have increased C32 to 100pf accordingly(to keep the filter cutoff at about 41 KHz).

Best thing to do is to remove the caps and see if you have a DC offset.

Kevin
 
Just got the LM4562s today, plugged the chip in replacing the NE5532 without any mod. Do observe a hugh difference between the two set of chips.

LM4562 - more bass, mid range vocal more forward. However, the LM4562 seems to have lost the transparent high though when compare to the NE5532!

Swapped between the two chip sets and listened to my favorite test CDs - still the same impression!

Being not a electronic handy as both Kevin and John, will have to see what their opinions are after having their mod experiment complete ;-)
 
LM4562 take a few days if running in to sound their best. Before then they can sound cold and clinical. After that they sort of disappear and leave the rest of the DAC to do the 'talking' ;)

If you like a bit more warmth then the OPA627s (2 on a Brown Dog adapter) are also fantastic.

- John
 
OK I have just modded my 'BIG' Dac and...... all I can say is if you have one you MUST try this. Sitting here with a BIG grin on my face!

All I did was remove op-amp nearest the edge of the circuit board, and replaced the one nearest the center with a single LM4562.

I then replaced all the four caps nearest that chip (marked L+ / L- / R+ / R- ) with wire links.

I then soldered a wire from pin 1 of the op-amp to the left output socket, and pin 7 from LM4562 to right output socket.

So if you don't like this tweak it is easily reversable and no harm done... I somehow think you will keep it like this though ;)

Seldom heard digital sound so good..... superb! No obvious problems with the lack of a muting circuit either thus far.

** NO resistors needed to be removed on my board, like Kevin's. But you may need to check your particular board as there are some variations **
 
Kevin, glad to hear you're enjoying yours. I'm looking forward to being able to wick the volume up tomorrow and give it a real workout :)

Just one thing, should I have taken the feed to the phonos from after that 1K resistor that you mentioned earlier, or is taking it directly from pins 1 and 7 ok? If so which of those two of the four caps is it best to take the output from please?

Thanks,

- John
 
Ideally a little resistance would be a good idea in case the OpAmp takes exception to a difficult load and oscillates. 1K is probably a bit on the high side for that purpose anyway. A couple of hundred ohms would be better. Either way, if you're only driving a reasonably short length of coax it should be OK.

Kevin
 
1).i sugest replacing first opamp with AD826 because of it's high slew rate and frequency response so critical in this position after the dac (it is a video opamp linear up to a few GHZ and has 350V/uS slew rate) just try it,you'll never switch back to LM4562...also AD826 can drive heavy loads and you can add just 22 ohm resistor on it's outputs directly to RCA socket without any coupling cap
2).bypassing the second opamp is a good idea
3).also i suggest removing all the feedback capacitors near the first opamp just like in the picture...they are big values and they have a cut off filter above 13khz so you'll experience briliant highs after taking them all out - and it WON'T oscillate without them because CS4398 has a digital filter inside filtering all the junk
4).also consider local decoupling of the +/-12v rails of that AD826 with one 10uF tantalum bypassed with 100n Film

An externally hosted image should be here but it was not working when we last tested it.


this is the frequency response after taking out all that capacitors that drive HF to ground and leave only that 100 pico green capacitor that is in the opamp's feedback (in the schematic it is written that it has 56pF value but in reality it is 100pF - it might be changed also to 56pF - it is for the opamp not to oscillate)

An externally hosted image should be here but it was not working when we last tested it.
 
It's interesting that you found the first op-amp stage to roll off from 13KHz:bigeyes: . I got this the first time I plugged the values into a simulation tool, but not in practice.

However, on the circuit diagram on www.lampizator.eu the value of C34 is wrong. This should be the same value as C32, and the value of C32 is also wrong. Both C32 and C34 should be 100pF, and are, in practice, on my board. This is probably what caused the early rolloff.

So now we have 3 errors in that part of the circuit diagram. These appear to have been corrected on my board.

The following values are what I have, and I believe them to be correct:

R8 = R14 = R15 = R16 = 22k
R10 = R11 = 6K8
C36 = C37 = .001uF (102)
C32 = C34 = 100pF

This gives a -3db point at about 41 KHz and no DC offset. I would advise everyone with these boards to check they have the right values in these locations, for starters!

However, since the output of the DAC can drive a low impedance, with an Op Amp such as the AD826 with decent output drive, as you point out, I don't see any reason to have such high impedances around this section, so I will probably redesign this filter circuit around lower impedances when I upgrade the Op Amp.

I want to have some filtering because the output goes via a volume control straight into my power amp (loosely based on a Mullard 5-20) and it doesn't take kindly to HF on the signal. (you will get some artifacts from the CS4398's switched capacitor filter but they are pushed far enough above the audio spectrum for a nice gentle filter to handle them).

I agree that the second Op Amp stage is a total waste of time, along with all the inter-stage electrolytics.

I was also thinking that supply decoupling is a little sparse on this board so decoupling the supply as you suggest would be a good plan.

I notice that you have modded pin 20 of the CS8416 too, to change the mode of the PLL. I did this on mine last night but a warning to those who want to try this: pin 20 on the chip is not connected to pin 20 on the DIL adaptor board on which it sits so you will need to make a connection directly to the pin of the CS8416 to achieve this (steady hand and good eyes required!).

Kevin
 
luxury54: if I stick with the LM4562 (It's sounding superb at the moment in this DAC after my mods, and running a short length of co-ax without a resistors after output seems fine in practice) is it still wise to take out the six capacitors you mention? I presume they are just removed and no wire links used? (unlike the electrolytics)

My board appears to be the same as Kevin's - so a lot of the errors have been corrected.

Thanks,

- John
 
yes,just pull them all out and that's all - after that compare the sound without them...my impression was of greater and accurate HF with good focus and imaging
the electrolytes are replaced by small wires;
if you have a chance to borrow an AD826 you could try also with that (don't forget local decoupling direct on pins,since AD826 is pretty demanding for this) - i threw away all my LM4562s after this together with a small sortiment of OPAs

but all this put together could just hide away in front of a simple anode follower tube output stage that bypasses all this analog stage and outputs a crisp clear,natural, very dynamic sound with the magical midrange warmth of a vacuum tube

I have also found CS4397 to sound better to my tastes compared to it's "better" brother the CS4398
you can build a small adapter board with direct pin to pin connection for this diy dac since it was first made for CS4397 direct pinout and after that came the CS4398 on that small board
with different pin connection only on the small board itself...
here is the picture,also shown in another thread
An externally hosted image should be here but it was not working when we last tested it.
 
...pin 20 goes directly to +5v via a 47k resistor...

That remiinds me of another difference. Not very significant but both Pins 23 and 21 of the CS8416 are powered by +3.3v on my board.

If running without the filtering, to be strictly correct, you should replace the two 6k8 resistors with wire links and change C32 to a much lower value (10pf perhaps?) just to keep things stable.

C32 and R10 from one of the poles of the filter in the inverting input and without a corresponding pole in the non-inverting input the circuit will be unbalanced at high frequencies and could result in common mode high frequency noise not being rejected.

Kevin
 
Would does this PLL modification achieve? (the wire between pin 20 and adapter board)

Secondly, I don't suppose anyone here had a spare AD826 they'd be willing to swap for either a LM4562 or a double OPA627 on Brown Dog converter? Cannot afford a new one at the moment :(

I'd also like to find a spare CS4397 already on an adapter board.

cheers,

- John


P.S. If anyone has the necessary photoshop skills it may be an idea to post a visual representation of the areas to mod on this DAC, with the component numbers superimposed over the top? I'd do it myself if I knew how :cannotbe:
 
P.S. If anyone has the necessary photoshop skills it may be an idea to post a visual representation of the areas to mod on this DAC, with the component numbers superimposed over the top? I'd do it myself if I knew how

Yep. There are getting to be quite a few to keep on top of now! I will take some photos at the weekend and see what I can do.

The Pin 20 mod is to put the CS8416 clock recovery PLL into a different mode. Basically, it has two modes of the PLL. One aims to reduce jitter in the wide band and the other in-band. Depending on the type of DAC connected to the device, the sensitivity to jitter is different and it's optimal for delta-sigma converters to have this signal pulled high to select low in-band jitter.

Have a look at PDUR on page 36 of the CS8416 data sheet.

Kevin
 
That would be great Kevin - mucho thanks :)

I think the problem is there's several board revisions by the sounds of it, plus a circuit diagram which is pretty inaccurate, to say the least! By the looks of it my DAC is exactly the same as yours, but I think there's at least two other revisions judging by what I've read over the past couple days.

Luxury54: Just removed those 6 caps and the sound seems to have gained more 'air' and a little more sparkle, though that COULD be Placebo effect ;)

I don't think there's much else I can do to my DAC now as I am already running the LM4562 direct to the output phonos. I guess I can now look into different rectifiers for the PSU, and perhaps trying an AD826 if I can get my hands on one.

It really does sound good though, better overall than my Audiosector NOS DAC and DPA PDM One - series 2... in fact it seems to combine the best of both! Open, natural, smooth but detailed, non-fatiguing, great dynamics (startling sometimes). The bass is fantastic/solid. Instruments appear to have correct 'weight' if tha makes sense? In short it all sounds very analogue in the best sense of the word. Thanks for all these mods guys :)
 
interesting that you are so surprised with LM4562's bass reproduction,my oppinion is that bass is it's weakest point...
then you would be amazed with AD826's bass reproduction :D

but,all these comparisons must be done during A/B tests,so in the lack of a same existing unmodified dac to compare it to,the best thing is to use another dac or cd player with the same test audio-cd that is tied up to another input of the amplifier.
then is easy to make an A/B test from the input selector comparing modded dac with LM to that other dac's sound and after that the AD opamp from the moded dac also to that other dac/cd player...
this way you can point out the differences pretty well because you have only one reference to refer both modifications to...

maybe you are also an analog sound lover and a vinyl listener ,and maybe you also have these UB40 tracks to compare to these 2 tracks are made after changing the opamps from an ordinary NE5532 to AD826 in a phono preamplifier
maybe you will get a taste of AD's bass reproduction and focus...it's a huge improvement over the NE5532 that also
comes with this DAC KIT - tracks are 24bit/96khz FLAC losseless format, so Foobar2000 would be the best choice for playing them into a soundcard or to an external 24/96 DAC like this CS4398

track nr. 1 FLAC (42 MB) : http://www.zshare.net/download/56254309fbc37d6e/

track nr. 2 FLAC (31 MB) : http://www.zshare.net/download/5625509361f271b2/

I forgot to mention that would be a real improvement in this dac to bypass the whole CS8416 and take directly the I2S signal out of the transport with the help of a cheap buffer and 50 ohm cables (Belden RG58)
or,option nr. 2, to build Elso's asyncronous reclocker to reclock the I2S signals generated by CS8416 getting rid of most of the jitter transmited on coax cable and recovered by the receiver...