I just built a 9018 dac with all of Ian's FIFO, isolater, clock etc and although it was very good, after a few days found it just too much like a Sabre dac, very accurate, highly technical but a bit soulless . Tried a nomber of I/V but eventually put an OPA861 on output.
Fixed it, enjoyable to listen to now, very close to my 1541 with stacked 844 in terms of musicality .
These trans conductance devices are something else, Pedja was on the money when he starting using them.
Fixed it, enjoyable to listen to now, very close to my 1541 with stacked 844 in terms of musicality .
These trans conductance devices are something else, Pedja was on the money when he starting using them.
Ditto,
And if you do Abraxalito's bipolar 10uf 50v smd cap mod from + to - rail on the AD844 and on whatever buffer your using, you'll pick up a little more.
http://www.diyaudio.com/forums/digital-source/227677-using-ad844-i-v-86.html#post3965465
Cheers George
After listening now for a while to Abraxalito's mod above, I can say it's very good in a subtle way. There is now a cleaner sense to the music, and the mirco detail is the most improved area.
Abraxalito can you give some techincal explanation on what happens, when your mod is in place, as I've look at it on the bench tests and there is nothing I can pick up on???
Cheers George
Abraxalito can you give some techincal explanation on what happens, when your mod is in place, as I've look at it on the bench tests and there is nothing I can pick up on???
Cheers George
I haven't any technical measurements to back it up but it seems that its reducing the noise generated by the AD844 itself on its power rails. The self-generated noise is an issue for all circuits not running in pure classA in that there's a current taken which varies with the signal. The output stage is the biggest generaor of noise and this noise will depend on the load - the current drawn by the IC under load looks like a half-wave rectified version of the signal. So its frequency content is higher than the original signal, hence the need for higher frequency decoupling, beyond the audio band.
Newbie Questions
Hello world
I've read a good part of this threat, but not every post yet....
So maybe i'll post some already answered candid questions, sorry about that.
1/ Did you compare stacked AD844 over OPA860 with and without internal buffers ?
Which one sounded best ? As I see that Pedja moved from AD844 to OTA.
2/ What should be done to use AD844 with AD1955 differential current sink ?
Would it be a good combo ?
3/ What do you think about using AD825 as "buffer with gain" out of AD844 TZ pin ?
Thank you for your answers.
Jean
Hello world
I've read a good part of this threat, but not every post yet....
So maybe i'll post some already answered candid questions, sorry about that.
1/ Did you compare stacked AD844 over OPA860 with and without internal buffers ?
Which one sounded best ? As I see that Pedja moved from AD844 to OTA.
2/ What should be done to use AD844 with AD1955 differential current sink ?
Would it be a good combo ?
3/ What do you think about using AD825 as "buffer with gain" out of AD844 TZ pin ?
Thank you for your answers.
Jean
and i just tried earlier today snubber (+bypass) for 1387 as per abraxalito's suggestion and also got a good improvement. hail, abraxalito.
Hi,
I'm also planning a tda1387 dac with ad844 active I/V. Can you share your implementation details please? Did you make any adjustment according to 1387's specs? One ad844 per Iout is going to be viable I think.
I somehow lost in thread and couldn't find the actual schematic with external buffer (BUF03) implemented.
Keantoken's buffer seems very good candidate for buffer.
Last edited:
A little off topic, and for the measurement gurus.
What sort of noise/glitches come out of a dac's output, eg: frequency, duration, peak to peak levels etc.. Has anyone ever measured these? I think it maybe hard to measure as they happen so quick, I've been told in some cases in a nano second sometimes.
As I strongly suspect that any I/V stage that has feedback around it maybe gets confused maybe even goes into sporadic oscillations when they are bombarded by them, and this is why a non feedback I/V such as the AD844 sounds so good as it is able to handle these better and remain more stable. Just like amps that have lots of feedback are far less stable than amps than ones have no feedback or very little.
Cheers George
What sort of noise/glitches come out of a dac's output, eg: frequency, duration, peak to peak levels etc.. Has anyone ever measured these? I think it maybe hard to measure as they happen so quick, I've been told in some cases in a nano second sometimes.
As I strongly suspect that any I/V stage that has feedback around it maybe gets confused maybe even goes into sporadic oscillations when they are bombarded by them, and this is why a non feedback I/V such as the AD844 sounds so good as it is able to handle these better and remain more stable. Just like amps that have lots of feedback are far less stable than amps than ones have no feedback or very little.
Cheers George
I'm about as far from being a measurement guru as its possible to get but glitching in DACs is one of my pet topics. I agree putting glitches into feedback amps is a fairly bad idea from the pov of SQ. As regards what the glitches look like, the DSs of non-audio DACs oftentimes show glitch waveforms and tell us something about the total area under the glitch (in nV-s). No audio DAC DS though has ever shown such data to my knowledge.
Hi,
amount and shape of glitches depend very much on the type and layout of the DAC.
A classic R2R DAC being probabely the most critical.
Manufactureres have done a lot of survey to lower or omit with glitches.
BBs PCM63 featuring a variant of an R2R DAC (named Colinear), already specs 0 glitch energy and a rise time of 200ns fullscale.
Quite often You find a shunting cap at the IV-input stage (with the typical inverting OPAmp-IV its the cap in parallel to the feedback resistor from In- to out) that serves two purposes.
One is to shunt glitches awy from the input and as such keep them from entering the IV.
The second is to reduce raise time of the signal, as that may easily overdrive most but extemely fast video-OPAmps.
If the signals raise time is too fast the OPAmps output doesn´t react fast enough in ´opposing´ the rise via the feedback network.
This puts the OPAmp under openloop conditions with accordingly high THD.
amount and shape of glitches depend very much on the type and layout of the DAC.
A classic R2R DAC being probabely the most critical.
Manufactureres have done a lot of survey to lower or omit with glitches.
BBs PCM63 featuring a variant of an R2R DAC (named Colinear), already specs 0 glitch energy and a rise time of 200ns fullscale.
Quite often You find a shunting cap at the IV-input stage (with the typical inverting OPAmp-IV its the cap in parallel to the feedback resistor from In- to out) that serves two purposes.
One is to shunt glitches awy from the input and as such keep them from entering the IV.
The second is to reduce raise time of the signal, as that may easily overdrive most but extemely fast video-OPAmps.
If the signals raise time is too fast the OPAmps output doesn´t react fast enough in ´opposing´ the rise via the feedback network.
This puts the OPAmp under openloop conditions with accordingly high THD.
Did someone try to feed +-15V to the AD844 stack by DC-DC converters? Soon I will test 3xAD844 on a 1541 dac and will follow that PSU path.
I used these converters in the past in different dacs and always had great results. At least. SUPER dynamics and great tone. Even if it might not look as a audiophile solution.
I guess these are not used in commercial audio as they're expensive. But on the bay it's possible to find them for little money.
Choose a 2 to 5W one, Tracopower or Xp brand are very good. They've a single voltage input and dual output on some models.
I used these converters in the past in different dacs and always had great results. At least. SUPER dynamics and great tone. Even if it might not look as a audiophile solution.
I guess these are not used in commercial audio as they're expensive. But on the bay it's possible to find them for little money.
Choose a 2 to 5W one, Tracopower or Xp brand are very good. They've a single voltage input and dual output on some models.
Last edited:
I have measured the temp from one to two to three stacked, and it/they did not get any temp change worth talking about, but the dac itself will most probably get warmer, for memory my PCM1704 did but only slightly.
As for stacking the I/V stage only, it was sim'ed, and is on these pages by ?? a long way back and it was definitely for the better.
As distortion went down, potential current limiting was removed, and the big bonus the input impedance was reduced without reducing the input level from the dac and creating more noise because of more amplification that was needed later.
And everyone who's stacked these has stated vast improvments, even with up to 4 stack for the TDA1541. I think there's one guy who has done 6!!! going for the Guiness Book of Records maybe.
Cheers George
75ohm on TZ will do the low gain for you Fabio. But I don't know if the I/V stage is unity gain stable and you will be close to it, also the cap needed for filtering across the 75ohm TZ resistor will have to be around .047uf this could cause the I/V stage to ocillate.
Only watching this on the scope will give you your answer.
Cheers George