george have you tried messing around with the bias settings for 844? i just went from 15 to 16 then 18 on my salas shunt. the sound became smoother and richer every time, and now at 18 it's euphoric like a good class a amp. the soundstage is also wider with better 3d imaging.
also have you tried paralleling the internal buffer?
also have you tried paralleling the internal buffer?
No I haven't, as for the internal buffer, it's ok, but a BUF03 is better, makes the sound a touch more relaxed and bigger without any loss of detail or punch and as a bonus it's 2ohm output impedance will drive anything even my Sennheiser HD650 head phones.
Cheers George
Just A/B'ed on two identical CDP's with PCM1704k dacs using Wilson Watt Puppy 8 Halcro DM68 mono blocks and Lightspeed Attenuator passive volume control.
One CDP had double stacked AD844 I/V fed from TZ 4.7khom + 470pf to a BUF03 output buffer.
The other was the identical but for tripple stack AD844.
Both units are direct coupled all the way and calibrated for +- .1mV on the output rca's. The output levels were checked and are identical in level and frequency response (-1db at 20khz), for both cdp's.
The tripple stack was a bit better again, giving bigger sound stage more bass power, just as tight though, more body to the midrage and just as clean and detailed highs yet more relaxed sounding.
Next is to try 4 stack. (one day)
Cheers George
One CDP had double stacked AD844 I/V fed from TZ 4.7khom + 470pf to a BUF03 output buffer.
The other was the identical but for tripple stack AD844.
Both units are direct coupled all the way and calibrated for +- .1mV on the output rca's. The output levels were checked and are identical in level and frequency response (-1db at 20khz), for both cdp's.
The tripple stack was a bit better again, giving bigger sound stage more bass power, just as tight though, more body to the midrage and just as clean and detailed highs yet more relaxed sounding.
Next is to try 4 stack. (one day)
Cheers George
i can also attest that stacking does improve things. i even connected buffer pin together after a listen and this strengthened the sound a notch more with added gain. with stacks there's better dynamic, separation, detail, ect. but transients aren't as focused as it is with single chip. this is the same effect that i've encountered with stacking 1543 but as more chips are added the head and the tail of a note becomes less defined, lending to a perceived 'thickness' of sound. call me romantic but i still have a soft spot for passive output with it's sweet 2h distortion. yet i could happily live with this hard-to-fault sound.
i do have a ground loop hum, though.... laying my finger on certain components quietens
it. from the circuit of the first post, i omitted c1, the output filter including 100k to ground.
have i omitted something that should kill the hum?
i do have a ground loop hum, though.... laying my finger on certain components quietens
it. from the circuit of the first post, i omitted c1, the output filter including 100k to ground.
have i omitted something that should kill the hum?
Last edited:
It might be related to stability as well. Such circuits can be more or less nice behaving, now obsolete MAX436 being the least, some discrete circuits I built being the most stable, and AD844 being somewhere in between. But since my MAX436 effort I routinely use this cap, and some polystyrene here is a safe bet. It is also a chance to form simple 1st order low pass, with -3dB point say at about 100 kHz, which is also all it usually in fact takes regarding filtering.
And "hi-fish sound" indeed too often has something to do with stability issues, or HF mess, and alike.
Hi Pedja, nice of you to join in, after all the AD844 as I/V and open loop was your baby from the start.
If you could I have a couple of questions on stacking, you may or maynot be able to answer, maybe someone could sim it.
Can I ask your opinion on with the PCM1704 (+-1.2mA @ 1Kohm) why stacking the I/V sections of the AD844 sounds progressivly better, so far up to 3 stack.
Could it be the lowering of the input impedance with each extra one?
Or the current starvation issue the AD844 can have is lessened with each extra one?
And with stacking, does the tz point change it's specs?
Cheers George
Thanks George, and it is nice to see the interest in this stuff is alive.
This would be only a guess, but I believe the benefits with paralleling come from relatively low quiescent current of AD844. It is hard to prove this, as this current can not be adjusted from outside, but when you consider total quiescent current for the chip, number of transistor arrays, and inverting input ("emitter") impedance, the quiescent current through the input array is probably less than 1 mA, which is lower than the current that comes from/to the DAC. And, as you point out, this "emitter" has to "drive" both the DAC output current and impedance. Everything about emitter followers and their driving capabilities applies here, the only difference is that the signal goes the opposite way.
I don't believe that the AD844 input impedance is more important than this.
TZ node is very high impedance, just as it has to be, and even paralleling a couple of chips is not an issue in that regard.
BTW, PCM1704 is not that tight load for I/V, relatively speaking. It is easier than say TDA1541(A), so the same I/Vs normally show somewhat better performance with PCM1704 than with TDA1541(A).
This would be only a guess, but I believe the benefits with paralleling come from relatively low quiescent current of AD844. It is hard to prove this, as this current can not be adjusted from outside, but when you consider total quiescent current for the chip, number of transistor arrays, and inverting input ("emitter") impedance, the quiescent current through the input array is probably less than 1 mA, which is lower than the current that comes from/to the DAC. And, as you point out, this "emitter" has to "drive" both the DAC output current and impedance. Everything about emitter followers and their driving capabilities applies here, the only difference is that the signal goes the opposite way.
I don't believe that the AD844 input impedance is more important than this.
TZ node is very high impedance, just as it has to be, and even paralleling a couple of chips is not an issue in that regard.
BTW, PCM1704 is not that tight load for I/V, relatively speaking. It is easier than say TDA1541(A), so the same I/Vs normally show somewhat better performance with PCM1704 than with TDA1541(A).
When you sim the AD844 you get quite same distortion as when you sim its internal schematics.
Then if you either decrease the input current or rise bias currents in the "internal schematics", you get better distortion figures.
Same goes with other discrete i/v common base topologies. They like bias current, but get hot
By paraleling them you halve the input current seen by each AD chip thus halve the distortion.
Hm, its interesting to try to overclock the AD by going with much higher supply voltages plus forced cooling. Have anyone tried it? The maximum voltage rating is of transistor breakdown voltage or just overheating issue? These current mirrors like the high voltage thing.
Then if you either decrease the input current or rise bias currents in the "internal schematics", you get better distortion figures.
Same goes with other discrete i/v common base topologies. They like bias current, but get hot
By paraleling them you halve the input current seen by each AD chip thus halve the distortion.
Hm, its interesting to try to overclock the AD by going with much higher supply voltages plus forced cooling. Have anyone tried it? The maximum voltage rating is of transistor breakdown voltage or just overheating issue? These current mirrors like the high voltage thing.
signal from tz to b1 buffer sounds heavenly. unbelievable transparency and detail, and soundstage to die for. oh my god this is the best sound i have ever heard from 1543, and i never imagined it could be this good.
switching back and forth between 844's internal buffer and b1, the latter sure lack the dynamics, though. if i could have a buffer that could sound as refined as b1 but 'thicker' like 844's, it would be nirvana....
i'd really like to try a discrete solution instead of buf03 or like, because opamps are much harder to come by than transistors here in korea.... any suggestions?
switching back and forth between 844's internal buffer and b1, the latter sure lack the dynamics, though. if i could have a buffer that could sound as refined as b1 but 'thicker' like 844's, it would be nirvana....
i'd really like to try a discrete solution instead of buf03 or like, because opamps are much harder to come by than transistors here in korea.... any suggestions?
Last edited:
[QUOTE i'd really like to try a discrete solution instead of buf03 or like, because opamps are much harder to come by than transistors here in korea.... any suggestions?[/QUOTE]
It's a shame they're not easily avalible, as I think they are almost the perfect buffer.
The BUF03 is very good sounding, as it sound like the TZ output but with dynamics/drive, so it is very transparent.
No feedback, very high class A biased, direct coupled (with dc trimpot), 2 ohm output impedance, very high fet input impedance which is a must for the TZ output to see/drive, and it can drive low impedance headphones if needed.
Cheers George
It's a shame they're not easily avalible, as I think they are almost the perfect buffer.
The BUF03 is very good sounding, as it sound like the TZ output but with dynamics/drive, so it is very transparent.
No feedback, very high class A biased, direct coupled (with dc trimpot), 2 ohm output impedance, very high fet input impedance which is a must for the TZ output to see/drive, and it can drive low impedance headphones if needed.
Cheers George
Post http://www.diyaudio.com/forums/digital-source/227677-using-ad844-i-v-38.html#post3491577
Quote:
Originally Posted by georgehifi
...even Pedja Rogic said with his discrete clone of 844, it was hard to say if it sounded better than the AD844 itself.
Audiolapdance: Well, he actually said it was close enough to not worry about it too much:
bear in mind that it will not be easy to beat the AD844. The layout and supply have to be done carefully and it worth to check if the output buffer will improve something or not. Ultimately this circuit has less grain and can achieve a better resolution but this does not mean that you will notice the grain or lack of the resolution listening to the common base AD844 I/V, you will not. Other than known shortcomings the monolithic solutions also have some advantages which are not only practical by nature.
from: Pedja Rogic, Discrete Diamond Non-Feedback I/V Stage for TDA1541(A) DAC
Quote:
Originally Posted by georgehifi
And he has not heard the advantage of stacking the 844.
Audiolapdance: Well, he fully investigated the lack of heavy bias:
Interesting fact is that the AD844's internal circuit (as is) also could achieve superb distortion performance used as a common base stage, but it needs a higher bias current. Some other variation on this theme can perform well too, and some can perform even better. So, the bias used in the AD844 is below what we need here ...
from: Pedja Rogic, Discrete Diamond Non-Feedback I/V Stage for TDA1541(A) DAC
Audiolapdance: And his discrete circuit addressed it so in a sense he heard higher bias 844.
Last edited:
"Pedja: BTW, PCM1704 is not that tight load for I/V, relatively speaking. It is easier than say TDA1541(A), so the same I/Vs normally show somewhat better performance with PCM1704 than with TDA1541(A). Pedja"
Pedja's comments were about a single AD844 I/V on TDA1541 (4mA).
Using the PCM1704 (1.2mA) it is clearly heard that two 844 I/V's stacked is far better than one, and three better again. Pedja thinks it maybe the current sharing, but I doubt that, as the 1704 only has a less than a third of the output of the 1541.
I still "think" it maybe the fact the input impedance is comming down with each stack, and the closer to 0ohms you get the lower the distortion of the 1704 becomes, from what I seen said. And as s3tup has said "By paraleling them you halve the input current seen by each AD chip thus halve the distortion" he has built discrete 844 I/V, so he could inside and measure things.
AD844/PA630A knockoff: Discrette.
Cheers George
Pedja's comments were about a single AD844 I/V on TDA1541 (4mA).
Using the PCM1704 (1.2mA) it is clearly heard that two 844 I/V's stacked is far better than one, and three better again. Pedja thinks it maybe the current sharing, but I doubt that, as the 1704 only has a less than a third of the output of the 1541.
I still "think" it maybe the fact the input impedance is comming down with each stack, and the closer to 0ohms you get the lower the distortion of the 1704 becomes, from what I seen said. And as s3tup has said "By paraleling them you halve the input current seen by each AD chip thus halve the distortion" he has built discrete 844 I/V, so he could inside and measure things.
AD844/PA630A knockoff: Discrette.
Cheers George
I am very slowly building up in my spare time a OPA861 setup, it will be interesting to see what A/B the differences will be on the same machine/s Jocko Homo once said to me slew rate is very important for an i/v
These are my amateurish observation of the pro's v cons (data measurments) of it v the 844 pro's, please feel free to correct me if I'm wrong.
OPA861 pro's
Lower input impedance for a single unit 13ohms vs 40ohms
Iq adjustment pin.
Faster 80mhz vs 60mhz
AD844 pro's
Higher slew rate 2000v/us vs 900v/us
Lower distortion I think, .005% (with buffer) vs -57db
Cheers George
These are my amateurish observation of the pro's v cons (data measurments) of it v the 844 pro's, please feel free to correct me if I'm wrong.
OPA861 pro's
Lower input impedance for a single unit 13ohms vs 40ohms
Iq adjustment pin.
Faster 80mhz vs 60mhz
AD844 pro's
Higher slew rate 2000v/us vs 900v/us
Lower distortion I think, .005% (with buffer) vs -57db
Cheers George
I have been listening for a couple of days now the 3-stack (both channels) and i can say it sounds better than 2-stack. I also paralleled pin6 and i think it "sweetened" the sound a bit. Lastly i bumped the voltage from 12v to 16.5v, and that was a huge improvement. Very pleasing sound, indeed. Next step is going to 18v...and after that i must get some buf03's to try pin5 output
I'm planning for a while to try out either Pedja's DDNF I/V or an AD844 based I/V with my AD1865 DAC.
Currently I'm using a Pass D1 I/V clone powered by Salas regs and I'm very pleased with it, though I would give a try for either DDNF or AD844.
Have any of you guys tried DDNF or D1 against an AD844 I/V in the same system?
Currently I'm using a Pass D1 I/V clone powered by Salas regs and I'm very pleased with it, though I would give a try for either DDNF or AD844.
Have any of you guys tried DDNF or D1 against an AD844 I/V in the same system?
aw, thank you for the idea, dirk.
i just did 10 parallel fets per side for 20 total in what-is-not-b1-anymore buffer and the sound has gained much needed drive at the expense of once again blurred transients.
maybe buf03 is better still, but this is a setup that i could live with for a looong time
thanks everyone.
i just did 10 parallel fets per side for 20 total in what-is-not-b1-anymore buffer and the sound has gained much needed drive at the expense of once again blurred transients.
maybe buf03 is better still, but this is a setup that i could live with for a looong time
thanks everyone.
I take it you are coming off pin 21 Iout of the AD1865, this is even lower in output current than the 1704! it's only 1mA.
This is giving me even more belief that it's the lower input impedance of the 3 stack that is the reason for progressively better sound due to stacking 844's, rather than current sharing, anyone else seeing this or am I the only one?
Cheers George
Last edited: