Experience with this DIY DAC ?

[legarem]

For those of you who wants to use only one op amp (LM4562)

Here is a schem which doesn't roll off the highs and is direct coupled to the dac. There's no cap at the output and no output dc.

This is the best I used with an op amp on this dac

[littlechicken]

Hey Mister M.. (alias James57) can this be done on any kind of circuit or does it need to be tailored. (I can`t PM you directly as I am too new on this board but will find a way to connect). Just thought this might interest a few.

[Kevin Wood]

Well, I made the changes to the filter components yesterday to the values from the Cirrus application note:

R8, R14, R15, R16 = 13k7
R10, R11 = 3k32
C36,C37 = 1nF
C32, C34 = 220pF

I also changed the Op Amp to an AD826 and added some decoupling to its' power supply.

Took a few photos but haven't had time to post them up yet and also haven't had time for any serious listening. Those boards are indeed a pain to rework. Very difficult to solder in places because the heat is conducted straight into the ground plane.

Frequency response now measures absolutely flat from DC to 20KHz and about 4 mV of DC offset on the worst of the outputs.

Kevin

[legarem]

Originally posted by littlechicken
Hey Mister M.. (alias James57) can this be done on any kind of circuit or does it need to be tailored. (I can`t PM you directly as I am too new on this board but will find a way to connect). Just thought this might interest a few.

Use it as is. If you place it in the H-K CS4398 dac, you only have to change the resistors and caps and it's great.

This was the output stage Counterpoint used with their rapture card which had a CS4396.

They used a OPA627 op amp. I used a LM4562 with great results. Better is use a metal can LM4562 they have better sound. You have to solder it on a socket

Even better is the use of two LM (4562 singles which I don't remember the number)

[luxury54]

[i]I also changed the Op Amp to an AD826 and added some decoupling to its' power supply.

Kevin [/B]

Hi,how do you like the sound characteristic of AD826?

it is a very fast opamp with 350V/uS slew rate compared to only 20 of LM4562 ,and in this place after DAC it is imperative to have such a good slew rate and wide frequency capability, that's why also tubes sound great because they have a very high slew rate never masking the microdetails (it is a video opamp linear up to a few GHZ,so is perfect in audio and will not oscillate easily)
i don't like transformers,i have the feeling they are more mellower and laid back than tubes kind of compressing and leaving only macrodinamycs ,masking the small micro details, it's about lazy rise times,so another slew rate thing...

[Kevin Wood]

I haven't had time for much listening with the AD826. I only fired it up on the nasty system in my "electronics room" last night. I will do some more listening over the weekend and report.

I must admit I tend to avoid transformers where possible although that's a view formed more from an engineering background than from critical listening.

I would tend towards offering the DAC the easiest load possible (an op-amp stage) but it's all just conjecture without a good comparative listening session.

Kevin

[Segran]

Hi legarem,
in the "lampizator" web page you have written "OP AMP BIASED MORE IN CLASS A", I guess you are referring to the 14k7 resistor from +12VDC to the output, after the 1k resistor. Please explain how this added resistor affects the biasing of the op amp. In my view, it's merely compensating for the (in your case obviously negative) offset at the output of the opamp by adding a small positive voltage to the output after the 1k.

You have the choice to leave it as is or completly redone the circuit arount this op amp.

Well, this is what a few of us have done by re-designing the low pass filter, and IMHO with excellent results!

I used a pair of Lundahl LL1690 amorphous core and NOTHING beats this topology. It takes AT LEAST 100 + hours to be broken in but after that, you're in another league.

I have no reason to doubt your first statement about the lundahl transformer, but as an engineer I truly doubt that transformers need "breaking in"! Can you substantiate that claim with any scientific facts or theory?

The LM4562 is one of the best IF IT IS BROKEN IN.

I have read about the LM4562 so I decided to make an adapter so I could compare LM4562 in one channel and NE5532 in the other. I haven't got to actually do the comparison yet. I don't really want to debate about "breaking in" op-amps, as I haven't seen any scientific rationale for how this would or could change the sonic impression. I would be more than grateful if anyone can show som evidence for this.

Replace the Sumlink pulse transformer

In what way does the Sumlink affect the sound?

Try a DIR9001 module to replace the CS8416 module

I have asked in an earlier post if anyone tried it, and I guess you have. Can you describe in what way the sound improves as a result of the lower kitter in DIR9001?

[Segran]

Hi legarem, you wrote in post #101:

Here is a schem which doesn't roll off the highs

I have simulated your proposal and agree that it doesn't roll off any highs, in fact it doesn't work as a filter at all. If anything, omitting C32/C33 makes it prone to oscillate at around 0,5 Mhz.

I still put my money to the filter designed by Crystal and discussed in a few of the previous posts by me and Kevin, it does exactly what it is supposed to.

[SoNic_real_one]

Originally posted by Segran

I have no reason to doubt your first statement about the lundahl transformer, but as an engineer I truly doubt that transformers need "breaking in"!
....
I don't really want to debate about "breaking in" op-amps, as I haven't seen any scientific rationale for how this would or could change the sonic impression. I would be more than grateful if anyone can show som evidence for this...

I subscribe to this. My only explanation is that the "new" transformers and "OA" are in fact surrounded by new electrolytic capacitors on those boards that hold them and people assume that the OA is the one that "breaks-in" when actually is the initial chemical electrolytic reaction in the capacitors.

[Kevin Wood]

I must admit I'm uneasy when the concept of "breaking in" components arises. The cynic (and engineer, incidentally) in me says that it's a process of getting used to something that you didn't like as much the first time you heard it and less to do with the components themselves.

Perhaps I'm wrong. I keep an open mind on such things but I've not noticed components improving with age myself.

Kevin

EDIT: I suppose with transformers there's a possibility of the core becoming magnetised during production / shipping. I guess it would gradually "degauss" itself with use?