If you are talking about the input offset of 82mv, it is caused by the inbalance caused by the ccs of 5ma, I do not know if this is problematic for the pcm1704 , I know for shure that in DAC like the pcm1794 or es9018 there is no problem it that low d.c. input offset.
Hi Joe, I am using a similar technique for a pré-amplifier and a phono pré , but with discrete componentes.
with the AD844, do you have any problema with the noise? Have you measure it? what is the dac that you use with this circuit?
Last edited:
Thanks Sergio, missed that, E input impedance will be 5ohm the way we want to use it. The wording on the graph threw me, as it's using E as the output.
So obsessed am I, I forget there are other ways to use it. C will be our output and that will be 25kohm output impedance, this is why I think a fet input buffer (like the BUF03) with it's far higher input impedance than the internal buffer will be the better match.
Cheers George
Attachments
Last edited:
One thing that I don't like in the OPA660, is the output impedance so low, 25k ohms is very low and it probably means that the mirrors inside the chip are very rudimentary.
comparatively, the output impedance of AD844 is 3Mega ohms, a much better value. This is the problem of using chips , to much compromises
, that's the reason I prefer doing my one discrete circuits .
comparatively, the output impedance of AD844 is 3Mega ohms, a much better value. This is the problem of using chips , to much compromises
, that's the reason I prefer doing my one discrete circuits .
Last edited:
Noise is quite low provided AD844 sees a low source impedance and gain no more than about 50 is used (which is about the max I have). Mind you, not all 844's have equally low noise, there can definitely be samples that are lower noise than others. I know this because I have also used it as a microphone preamp, for which it works very well too.
Used it with a variety of DAC's. Best to use with DAC's that have no offset voltage, the ES9018 is an exception as it does have 1.65V offset, but can be (don't try it with others) grounded.
Yes, I can measure noise via Clio FW10 setup from Audiomatica.
Cheers, Joe
.
Last edited:
Triple stack 844
Hi George, I ordered some BUF03AJ's for my 844 buffer. Waiting parts..... Since I have 2 near identical 1541A dac's. My plan is to try a BUF03 buffer on both the discrete Pedja circuit and the triple stack 844. I'll report my findings.... This thread is getting rather interesting.... I took the triple stack dac to my Cousin's. He has a Bride of Zen Pre, Class D amp and Alon` 4's. No smear was heard. It sounded natural although the 844's buffer does have a sound quality of it's own. Fast, liquid and organic is the best description I can come up with. I can't say I have any issue with it. Best digital playback I have heard to date. I am listening to the discrete Pedja at the moment and I tell you they are very close. 3- 844's VS discrete Pedja
Hi George, I ordered some BUF03AJ's for my 844 buffer. Waiting parts..... Since I have 2 near identical 1541A dac's. My plan is to try a BUF03 buffer on both the discrete Pedja circuit and the triple stack 844. I'll report my findings.... This thread is getting rather interesting.... I took the triple stack dac to my Cousin's. He has a Bride of Zen Pre, Class D amp and Alon` 4's. No smear was heard. It sounded natural although the 844's buffer does have a sound quality of it's own. Fast, liquid and organic is the best description I can come up with. I can't say I have any issue with it. Best digital playback I have heard to date. I am listening to the discrete Pedja at the moment and I tell you they are very close. 3- 844's VS discrete Pedja
Hi Dave, I thought you still had some old stock PMI BUF03?
Over the next week or so we'll be able to compare on the same units with PCM1704K's, double vs tripple vs quad AD844's i/v stage only, into BUF03 buffers, all dc coupled throughout, no nasty caps in the signal path anywhere to colour the sound. I'll post up the findings.
And we'll do it not only on my big esl based system, but because the BUF03 can drive anything, we'll also do the same comparision with Sennheiser HD650 driven strait from the BUF03 which sounds very much like system does, except with the system you see the sound stage with your eyes and feel the bass, but with the phones you need to close your eyes (as they decieve you) for your brain to see it. This works if you have not tried it with phones
Cheers George
Over the next week or so we'll be able to compare on the same units with PCM1704K's, double vs tripple vs quad AD844's i/v stage only, into BUF03 buffers, all dc coupled throughout, no nasty caps in the signal path anywhere to colour the sound. I'll post up the findings.
And we'll do it not only on my big esl based system, but because the BUF03 can drive anything, we'll also do the same comparision with Sennheiser HD650 driven strait from the BUF03 which sounds very much like system does, except with the system you see the sound stage with your eyes and feel the bass, but with the phones you need to close your eyes (as they decieve you) for your brain to see it. This works if you have not tried it with phones
Cheers George
Last edited:
Hi George, I still have a few PMI BUF03's. All are mounted on adapters. So decided to go with a clean install. On my discrete Pedja I have + 11 mV offset to deal with. I think there is enough adjustment with the null on the buffer to go DC coupled. If I use a cap, it could be quite small with the high input impedance of the BUF03. I guess the discrete transistors need to be matched better. Not an issue with the AD844.
Hi George, I still have a few PMI BUF03's. All are mounted on adapters. So decided to go with a clean install. On my discrete Pedja I have + 11 mV offset to deal with. I think there is enough adjustment with the null on the buffer to go DC coupled. If I use a cap, it could be quite small with the high input impedance of the BUF03. I guess the discrete transistors need to be matched better. Not an issue with the AD844.
The PMI's EJ are good also, the ones I got off you were the exact heat to touch 50c each, the first Analog Devices pair I got one was 4c cooler than the other and they were the mil spec AJ.
With TO-99 fined heat sink at +-15v they sit on 50c. And 11mV cd offset adjustment is no trouble as they have +- 80mV nulling range. Go dc coupled Dave the best cap is no cap.
Cheers George
Hi again Sergio, just wondering did you do just a single AD844 ltspice without your mod, so I can directly compare it to your quad stack that you posted?
Cheers George
This is interesting then Sergio, it looks like it may confirm the big advantages of stacking even with a very low +-1.2mV input dac, because what you posted comparing the 4 stack up against your single modded one is close anyway.
We did 4 x stack on my friends pcm 1704K cdp he took it home to listen on his system (Wilson watt/puppy 8, Halcro 68 Lightspeed) he was very enthusiastic when he rang me back, I'll take my identical unit which now has only two 844's to compare next week.
Cheers George
We did 4 x stack on my friends pcm 1704K cdp he took it home to listen on his system (Wilson watt/puppy 8, Halcro 68 Lightspeed) he was very enthusiastic when he rang me back, I'll take my identical unit which now has only two 844's to compare next week.
Cheers George
When you can Sergio, no rush.
When you setup to do a single AD844 from the PCM1704, is it not too much trouble to do a double as well while your setup?
Because on my system (esl's) and headphones (HD650's) my hearing and head are telling me that from the PCM1704 a double 844 stack is the optimum.
Cheers George
Yes your PCM63 is also low output 2mA. not quite as low as 1704 1.2mA but I think the amount of stacking is relative to which dac' output current. As the TDA1541 (@ 4 mA) seems to like 3 or 4 stack. You know what may sound very good is a dac that has very low output around .5mA then this maybe be perfect with one 844?
I will know this when I compare identical cdp's with1704K's with 2 and 4 stack next week.
And if Sergio can do the 1704 measurements with single and the double this may also give some indication, as he has done a 4 stack already.
Very interesting this, I will never go back to an i/v opamp with feedback again. Only the AD844 for me and maybe if I ever get around to it a OPA660 or OPA861 but the last two will be a pain for me to get going, in this cdp.
Cheers George
Pcm63 has 4mA for full scale, same like TDA1541. Differencies is at bipolar zero, PCM63 0mA, TDA 2mA. Because with i/v with opamp TDA has offset voltage but PCM has not. The same thing like PCM63 has PCM1704 but less current at output. Because i/v resistor must have greater resistance for the equal voltage output.
George, the first image is the fft for one ad844 with 0,6mA input simulating a -6dbfs of pcm1704, the second is for the 2 stacked, the 3º image is for 3 ad844 stacked, ofcourse this is only simulations , so ... use with a grain of salt
Attachments
George, the first image is the fft for one ad844 with 0,6mA input simulating a -6dbfs of pcm1704, the second is for the 2 stacked, the 3º image is for 3 ad844 stacked, ofcourse this is only simulations , so ... use with a grain of salt
Thank you very much for going to all this trouble Sergio
. You say "this is only simulations , so ... use with a grain of salt".
But to me it seems that with each consecutive stacking the distortion is around 10db lower for each one added. And this is maybe what we are hearing. .25% thd for 1 stack vs .027% thd for 4 stack
As the gain at TZ is the same regarless of one or four stack, the reduction in distortion to me says maybe it's because of the lowering of input impedance with each one stacked 40-50ohm for 1, and down to 16ohm for 4??>
Your sim network around a single 844 ( #636 ) looks the best with .001% thd, and has the lowest Z in at 5.3ohm, anyone tried it yet???
Cheers George
Last edited:
Thank you very much for going to all this trouble Sergio
You say "this is only simulations , so ... use with a grain of salt".
But to me it seems that with each consecutive stacking the distortion is around 10db lower for each one added. And this is maybe what we are hearing. .25% thd for 1 stack vs .027% thd for 4 stack
As the gain at TZ is the same regarless of one or four stack, the reduction in distortion to me says maybe it's because of the lowering of input impedance with each one stacked 40-50ohm for 1, and down to 16ohm for 4??>
Your sim network around a single 844 ( #636 ) looks the best with .001% thd, and has the lowest Z in at 5.3ohm, anyone tried it yet???
Cheers George
You might get a better approximation if you build the 844 circuit from
discrete transistors so it will take non linear capacitance into account.
As such then check 1kHz and 10kHz.
Once you start stacking, the LF distortion goes down but at some point the
HF distortion will start rising due to higher capacitance of paralleled devices.
Regardless, if it sounds good well stack away.
As I have discussed with George these open loop 844 I-V's, even stacked
are not really ideal open loop circuits and still have many compromises.
Refer below a simulation of my open loop tube / SS hybrid, no local FB,
No CFP connected transistors, quite a simple circuit actually.
This is driven with 1k DAC at +-0.001mA as per 1704. OP voltage is +-2V,
frequency is 20kHz.
Much cleaner spectrum, some 2nd at 0.0003, tiny bit of 3rd and nothing else
to speak of. That's linearity!
Even at hundreds of kHz this circuit is still really clean.
For an example of how extreme performance it is capable of I simmed it at
1MHz ref plot 2, 2V pk OP. Yes that is 1MHz
The sound is very very good but obviously it's more complex to build than
simple 844 having tube related supplies etc.
