The internal buffer for me was a bit ho-hum, a OPA627 attached to pin 5 was better it in every area.
Cheers George
I had 13.6V for the BUF03. Stacked 3x AD844 but the sound was same, boring in long run. Then took output from 1 buffer of AD - and the music opened, better resolution etc., this is it. The BUF03 adds some dynamics but also smoothens the sound. Not a high fidelity for me. I'd be satisfied with present sound but wil try also 3 parallel buffers and direct from pin 5.
I found the same thing. The output buffer was likely designed with feedback correction in mind, this for high slew rate conditions and not of priority for parameters that include non-linearity, power supply rejection ratios, etc.
The non-linearity of the output buffer can be measured by grounding the inputs and driving pin 5 of the AD844 with a signal generator and then measuring the non-linearity. The power supply rejection ratio can be measured as well.
Cheers,
Gerrit
There's an idea Gerrit, because we have access to pin5, use the internal buffer as the output, but just put some feedback (pin 6 back to pin 5) around it only, it "could turn out good" instead of ho hum.
BTW I found the BUF-03 very strong, but a bit dirty, even with the smoothest power supply for it, much prefer the OPA627
Cheers George
Last edited:
Same here. After trying almost 2 dozens of op amps, I finally settled with discrete op amp from Sparkos Labs. BTW, my IV is based on OPA861 instead of AD844. I preferred OPA861 for its simplicity. They both sound great to me though.
I tried some other options with 3 stacked AD844s: output direct from pins 5, output from 3 parallel buffers, output from 1 buffer. Only the last one does it for me. The former 2 had merely o.k. sound which I'd abandon soon. It seems that discrete solutions have higher probability to show the "magic" than opamps although in this case it is not classical opamp use. This time I was probably lucky.
I have one more question to Pedja circuit - how smooth is the pin 5 offset nulling? In my case the zero is not very stable and I keep under- or overshooting when setting the trimmer (500 Ohm). There are stil 2-3mV present, plus or minus. Also the setting has certain hysteresis. Is it normal?
I have one more question to Pedja circuit - how smooth is the pin 5 offset nulling? In my case the zero is not very stable and I keep under- or overshooting when setting the trimmer (500 Ohm). There are stil 2-3mV present, plus or minus. Also the setting has certain hysteresis. Is it normal?
George,
This doesn't surprise me at all.
The open loop OP resistance of AD844 is listed as 15 ohms, so this will =
approx 30 ohms per OP device. There are no OP resistors shown on simplified
schematic, even if they did exist, everything points to the OP stage working at
very low bias current.
IME for an open loop follower to sound decent, there needs to be significant
bias (idle) current.
T
Hi Terry
The 844's output buffer was very ordinary, I even tried stacking them with isolation resistors, better but still ordinary.
What about what I asked before with no takers, is there a way of using feedback just around the 844's buffer to give it some better qualities?
Cheers George
Sure, only this requires another device comparing the source signal input to the output and making corrections to the input of the AD844 output buffer. This is normally done by incorporating the whole of the AD844 device. Since the input stage of the AD844 already has a reputation for being very good it seems reasonable to use it.
The AD844 permits placing a resistor at the Tz junction (pin 5) to ground. This permits the open loop gain of the complete AD844 to be reduced by paralleling the intrinsic 2MegOhm resistance at the Tz node junction. As an example a 2KiloOhm resistor can reduce open loop gain by 60dB. This also increases the bandwidth because this lower resistor is in parallel with the intrinsic 5pF of Tz node junction.
By controlling the Tz node resistance this permits an ability to tune the output buffer as a function of open to closed loop feedback. In conducting experiments it has proven useful to add something in the order of a miniature 1 Ohm 15uHenry or so coil in series with the output as to limit parasitic capacitances from the feedback path.
Cheers,
Gerrit
Hi George
I was thinking the same thing about Pin 5 Tz and I have been using these ADs on and off for many years, even got some from a Chinese source lately - and if you use it open loop and set Tz resistor to ground and it works, then I can't see it being a fake - like OPA627 faked as TL071 is just relabelling. What do you think, and others too? Because if that is true, then I haven't encountered any fakes yet. So if Tz gain works in open loop, then it should be OK, right?
Cheers, Joe
Hi Joe, yes that's should be real, but they'er only $10 with (free delivery for any price purchase) from RS Components, and we know they are authorised AD resellers.
I not playing around at the moment with the AD844's but it would be nice for someone who is, to try going from pin 6 back to pin 5 with a feedback loop to see if this then can equal a OPA627 in sonic performance.
Cheers George
I not playing around at the moment with the AD844's but it would be nice for someone who is, to try going from pin 6 back to pin 5 with a feedback loop to see if this then can equal a OPA627 in sonic performance.
Cheers George
Thanks George. Re Pin 6 back to 5, the AC signal should be the same and in phase through unity gain buffer, so I don't think that will achieve anything. I would have thought stacking 844s and using its own buffer x 3 and triple the current, that should work better than a 627 with 100% feedback. But then again I am by nature inclined against using opamps as buffer. My admission.
Cheers, Joe
Agreed. A legitimate AD844 also has laser trimmed input offset in the 150 uVolt region. The input buffer has a characteristic 50 Ohm output impedance at the inverting terminal, hence with inputs grounded the error current is 150uVolt/ 50 Ohm. The error current is mirrored into the Tz node. Placing a 50K Ohm resistor from the Tz node to ground will magnify the input offset voltage at the output by about 1000 (50K Ohm/50 Ohm). if the output is in the order of 150 mVolt this indicates that the device under test has an active pin 5 Tz node and that the device has low input offset as likely laser trimmed.
I can't say that all the AD844's purchased are legitimate, only that they behave as reasonably expected.
Cheers,
Gerrit