Using the AD488 as common base zero feedback I/V stage, for current output dac's

[statman]

Hers something of interest to you 16 bit TDA1541 guys. Its a TDA1541 board with no spdif, just an I2S input using an Amanero USB-I2S board. Works fine up to 24 bit/96Khz, but of course it outputs in 16 bit. 

I built this out of bits in my junk box to test the concept, nothing fancy, although it does have a trans-impedance active I/V buffer stage, much better than resitive passive I/V imo. 

Its an old Pedja Rodgic board from 2004 from whence virtually all of the TDA1541 diy dacs originated from. You can see it has very little components on it- nice and simple. I have another spdif TDA1541 dac with all the fancy bits, but I'm a USB guy and dont have much time for spdif.

The output stage I used is an old Micrex valve pre modified for this purpose- gas regulators, e280f  trans coupled output, and one of the new high voltage Schottky diodes- they sound very nice.

[statman]

Thought so, I have found that the PCM1704 (@1.2mA) output, has too much for a single AD844!! It needs two. And it's less than half the TDA1541's (4mA) For the TDA1541 I think 3 or 4 piggy backed. They're only $5 each on ebay. 

If you want to really make that TDA1541 sing, read my last few posts over at DIYForum. On my thread "Using the AD844 as an I/V stage)

http://www.diyaudio.com/forums/digital-source/227677-using-ad844-i-v.html

And only use the pin 5 TZ output as the internal pin 6 buffer is not that good, I used one of these unity gain zero feeback fet output buffers (all the fet's have been matched already) which killed it and for the money is outstanding and measures great. Or you could use one of the many tube buffers on ebay, if that floats your boat, or build your own.

http://www.ebay.com.au/itm/400458112615?ssPageName=STRK:MEWAX:IT&_trksid=p3984.m1423.l2649

 

Cheers George

A tube buffer? I'd rather eat a goats turd. And an ebay goats turd-i'll pass on that too (:

[georgehifi]

A tube buffer? I'd rather eat a goats turd. And an ebay goats turd-i'll pass on that too (:

Glad to hear it, I though you may have been one of the many here (heaven forbid) with their head up their **** and preach tube buffers.

 

BTW that AD844 I/V stage and buffer I use is all direct coupled and no feedback used anywhere, from dac chip output all the way through to the output RCA's (no dirty caps in the signal path).

And the fet output buffer on ebay got some great comments from the son of god (nelson Pass) John Curl "A very compedent design" said he. 

 

Cheers George

Edited April 28, 2013 by georgehifi

[georgehifi]

Looking at you board you posted again, this looks as though it is the early original Pedja design, those fets and trimpot you have I believe are not for an output buffer, but rather for adjustment of the input to the AD844 from the TDA5141. It looks to me that the output to the RCA's come from pin 6 which is the internal output buffer of the AD844, it's not very good. This needs to be coverted to Pedja's last reincarnation of the AD844 with TDA1541 which is far far better. I have that here I will post it up for you on SNA because I can't do attachments with PM's . But you will then need an output buffer like the one I use.

 

Cheers George

[damohpi]

statman, using a similar setup here, even down to the plywood but using a later Pedja board (AYA II), sounds quite good but want to experiment further with isolating & reclocking the amanero outputs, when/if time permits

 

 

interesting thread over at diyaudio, haven't seen any reference to taking the output from pin5 before.. I still have a few of the older AYA boards somewhere that used the ad844 so will dig them out and have a play.

 

This was the output stage as used in the AYA (2.0c), bf245a current source used to null the tda1541a offsets as you suggested. Later AYA II switched to a similar circuit but using OPA861 - have you tried that?

 

Edited April 29, 2013 by damohpi

[statman]

Looking at you board you posted again, this looks as though it is the early original Pedja design, those fets and trimpot you have I believe are not for an output buffer, but rather for adjustment of the input to the AD844 from the TDA5141. It looks to me that the output to the RCA's come from pin 6 which is the internal output buffer of the AD844, it's not very good. This needs to be coverted to Pedja's last reincarnation of the AD844 with TDA1541 which is far far better. I have that here I will post it up for you on SNA because I can't do attachments with PM's . But you will then need an output buffer like the one I use.

 

Cheers George

The board also has an "optional" fet buffer using 2SK170 BL's , which takes the output from pin 5 of 844. Very similar to your buffer, I'm going to use it in conjunction with your stacked 844 idea. Will let you know how it goes.

[statman]

statman, using a similar setup here, even down to the plywood but using a later Pedja board (AYA II), sounds quite good but want to experiment further with isolating & reclocking the amanero outputs, when/if time permits

 

IMG_20130429_081545_187.jpg

 

interesting thread over at diyaudio, haven't seen any reference to taking the output from pin5 before.. I still have a few of the older AYA boards somewhere that used the ad844 so will dig them out and have a play.

 

This was the output stage as used in the AYA (2.0c), bf245a current source used to null the tda1541a offsets as you suggested. Later AYA II switched to a similar circuit but using OPA861 - have you tried that?

 

aya_ad844.png

No I haven't tried that one, but as George says the 844/output buffer stage is probably much more audible, I doubt the offset nulling has much significance. Georges idea about stacking the 844 seems very valid, and pretty easy to do.

Yeah I've got one of the Acko reclocker/isolater but the smd stuff is too small for my old eyes, and I'm yet to see any reports of them actually making a difference.

[georgehifi]

statman, using a similar setup here, even down to the plywood but using a later Pedja board (AYA II), sounds quite good but want to experiment further with isolating & reclocking the amanero outputs, when/if time permits

 

IMG_20130429_081545_187.jpg

 

interesting thread over at diyaudio, haven't seen any reference to taking the output from pin5 before.. I still have a few of the older AYA boards somewhere that used the ad844 so will dig them out and have a play.

 

This was the output stage as used in the AYA (2.0c), bf245a current source used to null the tda1541a offsets as you suggested. Later AYA II switched to a similar circuit but using OPA861 - have you tried that?

 

aya_ad844.png

Hi, Damophi, your far better of just using the common base I/V input stage and take the signal from pin 5 TZ to a better ouput buffer, as the ouput buffer at the end of  the AD844 is very ordinary, I've done the comparisions and it's not much to write home about. And definately stack for each channel the AD844's, 2 x for PCM1704 dacs and 4 x for TDA1541 dacs this way the input impedance is lowered but the output stays the same, and then they are not current starved as they share it, there is no noise penalty. So it's a win win with stacking these no current starvation on transients and closer to ground input impedance that the current output Multibit dacs love to see. PS only need to solder pins 2 3 4 5 & 7 of the stacked ones, bend the other pins out.

BTW: Charles Hansen the designer Ayre cdp's used 8 stacked AD844's, I'd say 4 per channel in a player, probably a TDA1541 based unit.

Cheers George

Edited April 29, 2013 by georgehifi

[tuyen]

hey guys!     I know spec and listening wise, the tda1541 chip is superior to a single tda1543 chip, but I'd be curious of your thoughts about when when one stacks multiple tda1543 chips?     Benefit also being that the output of the tda1543 is strong enough that it can output direct to the RCA sockets after IV resistor and a output cap.   No need to worry about designing a 'good' output stage.

 

thoughts?

[statman]

hey guys!     I know spec and listening wise, the tda1541 chip is superior to a single tda1543 chip, but I'd be curious of your thoughts about when when one stacks multiple tda1543 chips?     Benefit also being that the output of the tda1543 is strong enough that it can output direct to the RCA sockets after IV resistor and a output cap.   No need to worry about designing a 'good' output stage.

 

thoughts?

Thats an old concept, there were lots of TDA1543 with 8 chips in parallel- the AH Lite Dac was one of them, but to my ears it wasn't as good as a well designed 1541 implementation, even with modified output stages.  And not even remotely close to the stacked 844/1541/1704 concept.

There were a few paralleled 1541 too, but generely gains equalled losses. 

[georgehifi]

This is a new thread on how to use the AD844 as a common base zero feedback transimpedance amp for I/V duties, it is miles ahead of any I/V I have listened to.

Moderator "proftournesol" (Michael) said he would either shift or link the posts (18 of them) that have been already done on this subject from another thread that is a different subject (and is getting messy mixed up with it)  to what we will be discussing here.

 

Cheers George 

Edited May 2, 2013 by georgehifi

[Nada]

Im fascinated.

 

For DIY beginners like me please can you spell some basics out?

 

• My DAC has a resistor for IV duties for TDA1541 and then goes into a tube stage.
• Do I just tap the IV resistor in and outs and put in the stacked four AD844 or do I still need a buffer stage after the AD844 and the tube input?
• Can I power the pair of stacked  4 x AD844 from a 5V lithium battery?
• Where is a reliable place to buy AD844  for Aussies as some of the eBay AD844 look very dodgy and might be frauds? For TDA1541 do I buy eight  AD844 ie four per channel?

Thanks George.

Edited May 2, 2013 by Nada

[statman]

This is a new thread on how to use the AD844 as a common base zero feedback transimpedance amp for I/V duties, it is miles ahead of any I/V I have listened to.

Moderator "proftournesol" (Michael) said he would either shift or link the posts (18 of them) that have been already done on this subject from another thread that is a different subject (and is getting messy mixed up with it)  to what we will be discussing here.

 

Cheers George 

It should include posts 318,319 and 321 to complete the picture. The point is lost without them.

[georgehifi]

This is going to get messy, but here goes, copy and pasteing from the other thread post: 318, 319, 321 and 322.

 

Cheers George

 

 

 

 

 

 

statman

statman

Member

• 
• Members
• 184 posts
• State: WA

Posted Yesterday, 07:29 AM

statman, on 28 Apr 2013 - 17:33, said:

A tube buffer? I'd rather eat a goats turd. And an ebay goats turd-i'll pass on that too (:

 

 

georgehifi, on 28 Apr 2013 - 17:03, said:

Thought so, I have found that the PCM1704 (@1.2mA) output, has too much for a single AD844!! It needs two. And it's less than half the TDA1541's (4mA) For the TDA1541 I think 3 or 4 piggy backed. They're only $5 each on ebay.

If you want to really make that TDA1541 sing, read my last few posts over at DIYForum. On my thread "Using the AD844 as an I/V stage)

http://www.diyaudio....-ad844-i-v.html

And only use the pin 5 TZ output as the internal pin 6 buffer is not that good, I used one of these unity gain zero feeback fet output buffers (all the fet's have been matched already) which killed it and for the money is outstanding and measures great. Or you could use one of the many tube buffers on ebay, if that floats your boat, or build your own.

http://www.ebay.com....984.m1423.l2649

 

Cheers George

Hats off to you George, your idea of stacking the 844's gives an outstanding result. I can now see why you were so enthusiastic about it over at diyaudio. Best 1541 implementation I've ever heard and has somewhat made me re-evaluate my opinion of 16 bit , I still think my very best 24/32 bit dac has better definition , but theres very little in it, and I cant deny it sounds "hi-fi" in comparism. The 24 bit dac has the edge with hi-resolution files as the 1541 can only go up to 96Khz, but I cant deny the 1541 with the stacked 844's is bloody nice to listen to, without that silly "euphonic" sound so common to modern 1541 dacs. I'll stick with the 32/24 bit dac as I have a lot of 192Khz and DSD files, but if I just listened to <96Khz I would be very happy with the 1541/stacked 844 dac. Kudos George, it took a leap of faith to solder those extra 844's on top of the original, nice bit of thinking outside of the box.

• Nada likes this

 

 

 

 

• georgehifi
  georgehifi

  Member

  • 
  • Members
  • 683 posts
  • 1 warning points
  • State: NSW

  Posted Yesterday, 07:46 AM

  Good to see someone else has finaly done it as well, how many did you use for the 1541, as I think it would have to be 4 minimum, as it has more than double the output current than the PCM1704 has, and I used 2 on that. I cannot tell whether the AD844 I/V common base stage is not being current starved now or the fact that it input impedance is being drasticly lowered from a single 844 30-50ohm that is causing it to be so much better, I will work it out one day. Maybe simming can tell

   

  Statman you should post thtat up on my AD844 thread over at diyforum as there are some heavy hitters over there involved in that thread who have been watching what I was doing and helping me simulate some of the stuff I was comming up with with the 844, who have also not taken up the bait I gave on stacking these 844's on 1704's or 1541's

   

  PS: You should try to get hold of a good old cdp or dac that uses PCM1704's or PCM1702's as the da conversion as I believe that it's even better than the TDA1541 then you'll only need 2 AD844's. Just remember to cut out the feedback loop of the old I/V opamp when you sub it for the AD844's. My CDP that I've done is the Cary 303/200 uses the PCM1704 "K" chips which are even better again

   

  Cheers George

  
  

  Edited by georgehifi, Yesterday, 08:08 AM.

  • Nada likes this

   

   

   

  • statman
  • statman

    Member

    • 
    • Members
    • 184 posts
    • State: WA

    Posted Yesterday, 09:11 AM

    georgehifi, on 02 May 2013 - 07:46, said:

    Good to see someone else has finaly done it as well, how many did you use for the 1541, as I think it would have to be 4 minimum, as it has more than double the output current than the PCM1704 has, and I used 2 on that. I cannot tell whether the AD844 I/V common base stage is not being current starved now or the fact that it input impedance is being drasticly lowered from a single 844 30-50ohm that is causing it to be so much better, I will work it out one day. Maybe simming can tell

     

    Statman you should post thtat up on my AD844 thread over at diyforum as there are some heavy hitters over there involved in that thread who have been watching what I was doing and helping me simulate some of the stuff I was comming up with with the 844, who have also not taken up the bait I gave on stacking these 844's on 1704's or 1541's

     

    PS: You should try to get hold of a good old cdp or dac that uses PCM1704's or PCM1702's as the da conversion as I believe that it's even better than the TDA1541 then you'll only need 2 AD844's. Just remember to cut out the feedback loop of the old I/V opamp when you sub it for the AD844's. My CDP that I've done is the Cary 303/200 uses the PCM1704 "K" chips which are even better again

     

    Cheers George

     

    George

    I used 3x 844 , ordered some more to go to 4, BTW I soldered all pins, no problems. Yeah I'll do a post on diyaudio.

    Too many things to do to play around with 1704, and as I dont use CDP's only interested in USB implementations. I'm more interested in pushing the new designs than playing around with old dacs, but it was interesting to hear a 16 bit dac sound so good. But its a quandry- the 24 bit dacs have more "air", definition and preciseness around the music, but the 16 bit 844/1541 sounds "nicer", more natural. Is that because the 844/1541 is better technically, or is it inferior technically, but inferior (less definition) sounds better? I dont know.

    I'd like a 32/24 bit dac that has all the definition, so you can hear the recording as it is, but with the "tone" (not sure if that is the right word) of the 16 bit. You could cheat with a coloured output stage, but that's not my choice. I suspect the real issue is the quality of the digital files, maybe the latest dacs are fine, and the recording quality needs to catch up, DSD done properly is certainly at a different level. There's so many variables, who really knows? Anyway the stacked 844 is a real "killer", congrats again.

    • Nada likes this

     

     

     

     

    • georgehifi
      georgehifi

      Member

      • 
      • Members
      • 683 posts
      • 1 warning points
      • State: NSW

      Posted Yesterday, 10:01 AM

      Yeah I don't know if Peja knows how good this, as he has never thought of stacking the 844's. And I could imagine with the 1541 stacking is even more important.

      Looking at your pic of the 1541 dac board the your 844's are socketted, do yourself a favour and solder them in "really" I/V's don't like socketts (ok for buffers) but not I/V's Jocko Homo told me this and he is right they are crud for I/V's this will make your SQ jump again.

      Looking forward to reading your post on diy forum as this will make them jump into action, on stacking, I had a feeling they were waiting for me to come to an end of the experimentation with the 844.

      Also play around with the 1nf cap across the TZ resistor, I am now using none, little bit of very very high HF crap (about 2 or 3mV) on the output that would be imposible to hear because it's up in the 10's of megahertz region but better sound, I think the output stage of the I/V @TZ doesn't like to be capacitive loaded.

       

       

      Cheers George

      • Nada and Ozcall like this

Edited May 2, 2013 by georgehifi

[georgehifi]

Im fascinated.

 

For DIY beginners like me please can you spell some basics out?

 

• My DAC has a resistor for IV duties for TDA1541 and then goes into a tube stage.
• Do I just tap the IV resistor in and outs and put in the stacked four AD844 or do I still need a buffer stage after the AD844 and the tube input?
• Can I power the pair of stacked  4 x AD844 from a 5V lithium battery?
• Where is a reliable place to buy AD844  for Aussies as some of the eBay AD844 look very dodgy and might be frauds? For TDA1541 do I buy eight  AD844 ie four per channel?

Thanks George.

Hi Nada, Just take out the i/v resistor all together, and put in this circuit http://www.stereo.net.au/forums/index.php?app=core&module=attach&section=attach&attach_rel_module=post&attach_id=54286 but with three more AD844's stacked with only pins 2,3,4,5, & 7 need to be solder onto the first one, bend the other pins out. then take the singnal from the TZ pin 5 to you input of your tube buffer, if you must use a tube for a buffer, I must add here that pin five has very high output impedance and needs the buffer to have a very high input impedance, I noticed the 100kohm loading resistor on my fet buffer http://www.stereo.net.au/forums/index.php?app=core&module=attach&section=attach&attach_rel_module=post&attach_id=54289 drop the output of the signal at TZ pin 5 by about 10% so I removed the buffers 100kohm loading altogether which leaves just the fets impedance itself, whatever that may be?

You need +&- 15vdc for the AD844 which nearly all sock I/V's use in cdp's and dac's so it's inside them already.

AD844's can be bought from Element 14 for around $10 you can get them on eBay from Asia, but who knows what your getting, I ordered some but the first delivery got lost so they are re-sending maybe I'll get these but who knows, I ll let you know if they'er any good if they arrive, you don't save much anyway, but I'm just a tight arse and have to try. At least the one from Element 14 are genuine from Analog Devices.

 

Cheers George  

[Nada]

Thanks George.

 

RS has the AD844 too: http://australia.rs-online.com/web/p/operational-amplifiers/5229880/

[damohpi]

Is this in your "killerdac" Nada? if so then the above circuit isn't really suitable for insertion between your tda & tube gain stage, IIRC the killerdac uses around a 50R I/V resistor, the above circuit has a 1k5 I/V resistor (following the ad844), which gives typical CD level output (@4mA = 6v p-p ~ 2.1v RMS)

[georgehifi]

Is this in your "killerdac" Nada? if so then the above circuit isn't really suitable for insertion between your tda & tube gain stage, IIRC the killerdac uses around a 50R I/V resistor, the above circuit has a 1k5 I/V resistor (following the ad844), which gives typical CD level output (@4mA = 6v p-p ~ 2.1v RMS)

Damohpi hi, when the AD844 is used as I have outlined with a single AD844 on the TDA5141 there is enough (the same) output with a unity gain buffer or even when using the internal unity gain buffer of the AD844. And this gain btw can be varied with the TZ resistor to ground. 

Then when they are stacked yes the loading as seen the TDA1541 drops from a single AD844 at around 50ohms to 25ohms for 2 x AD844, then to 12.5ohms for 3 AD844's, and then 6ohms for 4 x AD844's, but the output from pin 5 (TZ) remains the same so the unity gain buffer will also have the same output as a single AD844 or as he has it now 50ohm I/V resistor.

 

For instance I have a PCM1704 which is 1.2mA (more that 3x less than the TDA1541) and I am using a 4.7kohm TZ resistor then that goes into my unity gain buffer which has more than enough output to drive two ME850's into 86db efficient Martin Logan ESL's

 

Cheers George    

Edited May 3, 2013 by georgehifi

[damohpi]

you're missing my/the point. 

 

if his DAC is what I think it is, then his tube stage is not a buffer, rather a gain stage and hence the small I/V resistor. Using the linked circuit (AD844 with 1k5 I/V resistor - giving full cd level output), prior to his existing tube stage will not work. so the advice of simply removing his i/v resistor and replacing it with the circuit, stacked or not, is incorrect, unless he changes the i/v resistor post ad844, or removes/changes his tube stage.

 

edit: what you refer to as the TZ resistor is the I/V resistor in this circuit, is it not?

 

edit 2: 1.2mA with 4k7 I/V = 5.6V p-p = 2V RMS

Edited May 3, 2013 by damohpi

[georgehifi]

Well if the tube buffer as Nada has said, is a gain stage, there is even less of a problem. His I/V resistor is 50ohm correct it gets removed and the AD844 goes in it's place stacked or not stacked, and the output from TZ goes his "gain" tube stage buffer or it can be reconfigured  if he wants to be a unity gain tube buffer.

Statman is having no problem with his TDA1541 with a fet unity gain buffer, why should Nada with a tube buffer which has gain.

If you still insistant on this post a circuit of Nada's TDA1541 output to the output rca's so we can see what you are concered about.  

 

Cheers George   

[damohpi]

ok..

 

existing: tda1541a = 4ma p-p output current, 50R I/V = 0.2V p-p into ~30x gain from ecc88 stage = 6V p-p = 2V RMS ~ CD Level output

 

proposed: tda1541a = 4ma p-p output current, ad844 with 1k5 I/V = 6V p-p into ~30x gain stage ~ 180V p-p ~ 63V RMS.. see the problem? (ignoring that the tube stage wouldn't be able to swing that voltage, but lets just assume it can)

 

I wouldn't expect statman to have any issue with a unity (or less than) gain buffer following the ad844 circuit as linked (as that was Pedja's intention for the circuit in the original instance)

Edited May 3, 2013 by damohpi

[georgehifi]

The load as the TDA5141 see is still 50ohms with a single AD844 it is not 1.5k as you have stated, The TZ resistor has no bearing on the input load as seen by the TDA1541 this has been proved at diy forum,  and this 1.5k TZ resistor can be varied up and down to produce more or less gain from the common base stage of the AD844.

As I said post a circuit of Nada's output from pin 25 or 6 of the TDA1541 to the output rca's for me to see if it has a problem anywhere. I cannot see any without the circuit to tell me different. 

 

 

Cheers George

Edited May 3, 2013 by georgehifi

[damohpi]

hmm.. from Pedja's original discription of the circuit:

 

The TDA1541A’s output current flows into the AD844’s array marked with blue and is mirrored to another array (marked in green). In the classic opamp connection the output voltage would be set by the feedback but since AD844’s transconductance node (TZ) is externally available we can use it to set the output voltage.

 

 

With 1k5 I/V resistors, the output voltage is about 2.1V RMS (output voltage is equal to TDA's Iout (4mA) multiplied by I/V resistor’ value, so 6V; this is the peak-to-peak value so RMS is 6V/ (2*sqrt2) = 2.1V). 

 

and the complete circuit as I posted earlier in the other thread

 

 

and from his description of the later and similar OPA861 circuit:

 

The same current is then mirrored to the OPA861 “collectorâ€. At this point I/V resistors are low impedance path to the ground, so this current flows through these resistors, developing the voltage which is the product of this current and the value of these resistors. So, with 4mA peak-to-peak output current, typical CD/DAC output voltage of 2V is achieved by 1k5 resistors 

 

 

 

Edited May 3, 2013 by damohpi

[Nada]

My modded Audiocentric TDA1541A DAC is like all electronics a complete mystery to me.

Ive read the tubes are amplifying the signal?  Apparently the tube is thought by some to be one of the Russian super triodes with very low distortion but has narrow operating specs.

Edited May 3, 2013 by Nada

[georgehifi]

And if you stack these AD844's input impedance goes down accordingly to how many you use (which current output dac love to see) but the output from TZ remains the same if the TZ resisitor is not changed.

I believe Charles Hanson developer and designer of the top Ayre cdp's and dac's used 8 stacked Ad844's now that could mean 4 per channel.

 

Cheers George