DIYHiFi.org

Carlos/Mike,

Could you post a picture of the opamp and the way you have placed the bypasses around them?

I think that this part is prone to instability as it is not unity gain stable. Used as a I/V the gain is about -1, depending on output resistance of the dac. Also, as it is a very fast opamp, extreme consideration must be given to parasitic capacitance.

It is my bet that, when this opamp is perceived to be lacking in dynamics/bass, is is actually oscillating at a high frequency.

regards,

ABo

It is actually the noise gain that matters here and it can drop below unity.

Pedja

How is that?

That would mean that the output resistance of the DAC is negative or badly out of phase?

Please enlighten me.

ABo

Huh?
That would mean opening the CDP, remove the main PCB and unsolder some wires... I'm not willing to do that right now, sorry.
It's sounding great, btw.



It is unity gain stable, in inverting mode (-1).

_________________
Carlos Filipe

"Any intelligent fool can make things bigger, more complex, and more violent. It takes a touch of genius - and a lot of courage - to move in the opposite direction." Albert Einstein

ABO, this is not exactly what you wanted to see, but it serves to show how picky I am in getting the (low ESR) electrolytic caps near the opamp's PSU pins.
The SMD opamp in on the bottom layer. Smaller bypass caps are used in that same side, but no specific board space is provided (or needed) for them.

_________________
Carlos Filipe

"Any intelligent fool can make things bigger, more complex, and more violent. It takes a touch of genius - and a lot of courage - to move in the opposite direction." Albert Einstein

No, but there is a difference between the signal gain and noise gain of the opamp. For classic non-inverting opamp configuration signal gain is equal to noise gain, while for classic inverting configuration noise gain is equal to signal gain + 1. When it comes to the stability it is the noise gain that counts and this can't drop below one. Signal gain, on the other hand, do can be set below one if opamp is used in the inverting mode.

Pedja,

I'm sorry, I thought that you said that the noise gain could drop below one.

I am used to analyse the loopgain of circuits. The loopgain is indepent of configuration and is the sole determant of stability. Terms like gain >2 , or noise gain only seem to introduce misunderstanding.

In any case, the loopgain consists of the openloop gain and the feedback factor. In case of a standard I/V, the feedback factor is 1 (assuming perfect current output from DAC) and the THS would be unstable.
Only if the output resistance is lower than the feedback resistor, stable feedback is attained (the feedback factor is in this case 0,5, equal to a normal non inerting amplifier with gain=2).
I'm quite sure we all know what we are talking about and that we do agree on this.

For my PCM1704 the output resistance is specified as 1k, so a 2k5 I/V resistor would in theory be good. With this opamp I'm still worried, because of the phase displacement in the feedback, due to parasitic capacitance, lowering phase margin at frequencies of about 20-30 Mhz.

This is not something I have already analysed, but it just seems that stability cannot be taken for granted here.

Best,

ABo

I am sorry, I've just seen typo in my first post above (hence it looks like I claim the opposite things in two posts)... It should read "noise gain can NOT drop below unity".

I recently acquired four AD844ANs from DigiKey, BrownDog'd them, and popped them into the MF A324 georgehifi noted in the original message for this thread.

I can't get these guys to work ... period.

They become VERY warm at power-up, and sounded extremely distorted. I immediately powered down, double-, triple- and quadruple-checked everything: mounting on BrownDog, BrownDog circuit itself, etc. With my limited test equipment (Fluke DMM), the chips "Ohm out" ok, too, wrt pins.

Did I get four bad AD844s from DigiKey? Or is it something else? Frankly, I don't care as this is kind of a time-sink tweak, AFAIC. The handful of chips I already have on hand -- AD825, OPA627, AD843, etc. -- work well enough. Besides, I've got this never-started-upon D1 I/V kit sitting on my shelf (for about 6 mos. now!) ... so either it or one of the Borbely projects is going to be the next (re)evolution for the analog section of this DAC.

It was already said before (Chad in post 16, etc...), but it worths checking anyway:
Did you remove those 3.3n across the feedback resistors?

Sidiy:

You're right! In my overzealousess, I jumped the gun here, and didn't carefully check this thread.

While I haven't listened to DAC, I can say that the AD844s are now only slightly warm to touch. This is comparitively warmer than any other IC on this PCB, but no way near the temps encountered before removing those caps.

Will report back if/when I hear anything interesting from this mod.

Thx!

I put an RTD probe on the AD844s, after the unit was on for about 1/2 hr and while a burn-in sweep from a test CD was being played. The temperature on the top surface of the IC settled to ~103 F (39.4 C). Although this falls within the manuf's operating range (–40 C to +85 C), other opamps I've tried (AD825, AD843, OPA627, etc. all SOIC, BTW) run cool (~ room temp). So, I'm not sure these op-amps are "optimal" for this circuit (as it stands).

That said, I was surprised that even with almost no break-in time invested, these new op-amps sound quite good. To characterize: very similar to AD825. Who knows how they'll perform after broken in?

After burning in the four AD844s for +70hrs, and re-listening/evaluating, I'm going to recant my earlier, somewhat-positive comments about the AD844s. Specifically, after putting the AD825s back in with the 3.3n caps for a cross-comparison, I noted a fairly substantial improvement (return) of fidelity. I don't like "audiophile" jargon (e.g. emotional adjectives, "synesthesia") so I'll leave the sonic description at that.

I think I more-favorably assessed the AD825's earlier because of one or more of the following:

- Sub-optimal (comparison-geared) listening experience I had with the AD825's just before the AD844 mod (I was fighting off jet lag); and/or
- Changing weather/barometric press (as it has been here in So Cal beach communities lately) which causes ear drums to do funny things; and/or
- "Excitement" about the mod: which can psychologically/emotionally "flavor"/"prejudice"/contaminate listening experiences

This experience almost makes me wish I could ABX-test certain mods. But that would be sacrilege (!) because I don't feel A/B or ABX tests are suitable for the "subtle" details audiophiles/DIYers often want to discern.

Perhaps one should just forget about current-feedback I/V in this DAC, as JH suggests. OTOH, Pedja's report on AD844s is interesting. Anyone know to implement those I/Vs on diff. summing opamp pairs?

What do you guys think about the Borbely ready made discrete current I/V stage, good for me that it is ready made, stereo unit, direct coupled and has power supply on board.
http://www.borbelyaudio.com/eb604417.asp

Cheers George

_________________
Can't play any more, I think I hear my mother calling me.

I managed to get one of these very cheap with complete shunt regulated power supply.
What do you guys think of this circuit for my PCM1738E, input has been set up for a balanced current output dac, pin 4 on the input doesn't seem to go anywhere, maybe a circuit typo.

Cheers George

_________________
Can't play any more, I think I hear my mother calling me.