dac I/V convertion with very low distortion - Page 20 - diyAudio
Go Back   Home > Forums > Source & Line > Digital Source

Digital Source Digital Players and Recorders: CD , SACD , Tape, Memory Card, etc.

Please consider donating to help us continue to serve you.

Ads on/off / Custom Title / More PMs / More album space / Advanced printing & mass image saving
Reply
 
Thread Tools Search this Thread
Old 15th August 2012, 12:48 PM   #191
diyAudio Member
 
Join Date: Apr 2003
Location: Eastern Pennsylvania
An issue with using DAC chips such as the AD1955 or AD1853 with a grounded-base I/V circuit is that the differential current output pins on these DACs want to see a common-mode input voltage that's offset above D.C. ground. For those ADI DACs, the figure is near +2.8VDC. The PCM179x series, on the other hand, wants to see a common-mode input voltage that's at D.C. ground. This issue can be addressed by level shifting the grounded-base input bias reference level from ground to the required D.C. offset, with a resistor biasing network and A.C. ground coupling capacitors.
__________________
Ken

Last edited by Ken Newton; 15th August 2012 at 01:18 PM.
  Reply With Quote
Old 15th August 2012, 01:04 PM   #192
diyAudio Member
 
Join Date: Apr 2003
Location: Eastern Pennsylvania
Quote:
Originally Posted by dirkwright View Post
So, are you saying that we should use 15.6mA for the current generator in our simulations when it is connected in differential mode as shown for example in the attached figure?
How the simulators handles differential signal source generator modeling might make giving an simple answer problematic. The most clear advice I can think of is to use whichever of the two current values for the source gen. which produces a 7.8mA peak and 15.6mA peak-peak signal current swing through the I/V resistor of the I/V stage being simulated - assuming, of course, that both output phase currents are being differentially utilized.
__________________
Ken

Last edited by Ken Newton; 15th August 2012 at 01:26 PM.
  Reply With Quote
Old 15th August 2012, 02:41 PM   #193
diyAudio Member
 
Join Date: Apr 2003
Location: Eastern Pennsylvania
Quote:
Originally Posted by dirkwright View Post
So, are you saying that we should use 15.6mA for the current generator in our simulations when it is connected in differential mode as shown for example in the attached figure?
One other point. When modeling a two-terminal differential signal source you should set the internal source impedance equal to twice that of a single DAC output pin impedance. For example, if the impedance of a signle DAC output pin were 1K, then set the modeled generator internal impedance to 2K. That is because the DAC differential outputs are effectively in series with each other with respect to a common, differentially driven load. Which means their individual output impedances are also in series, again with respect to a common, differentially driven load.

An increased source impedance has beneficial THD implications for the I/V stage. Although, the doubling of the signal current magnitude then degrades the I/V stage THD.
__________________
Ken

Last edited by Ken Newton; 15th August 2012 at 02:48 PM.
  Reply With Quote
Old 15th August 2012, 02:50 PM   #194
diyAudio Member
 
dirkwright's Avatar
 
Join Date: May 2011
Location: Virginia
Quote:
Originally Posted by Ken Newton View Post
One other point. When modeling a two-terminal differential signal source you should set the internal source impedance equal to twice that of a single DAC output pin impedance. For example, if the impedance of a signle DAC output pin were 1K, then set the modeled generator internal impedance to 2K.

That is because the DAC differential outputs are effectively in series with each other with respect to a common, differentially driven load. Which means their individual output impedances are also in series, again with respect to a common, differentially driven load.
OK, thanks.

The outputs from the current generator in the image I posted above are 180 degrees out of phase and half the value of the generator, so it appears to be functioning as a differential current source. I've done this with the voltage generator in other circuits in Tina also with the same effect.

(edit: whoops, made a mistake. Setting the current generator to 7.8mA produces that amount on BOTH pins of the generator, not half like I thought)

Last edited by dirkwright; 15th August 2012 at 03:15 PM.
  Reply With Quote
Old 16th August 2012, 06:37 AM   #195
Calvin is offline Calvin  Germany
diyAudio Member
 
Calvin's Avatar
 
Join Date: Nov 2004
Location: close to Basel
Hi,

see the LTspice Sim model for the PCM-DACs.
Values of R and C are what I use, but may be different.
Seemingly nobody at TI is willing or able to supply for real figures.

jauu
Calvin
Attached Images
File Type: jpg Idac source.jpg (68.1 KB, 267 views)
  Reply With Quote
Old 16th August 2012, 06:38 AM   #196
xslavic is offline xslavic  Moldova
diyAudio Member
 
Join Date: Feb 2011
hi
can you try with dual mono pcm1794?
  Reply With Quote
Old 16th August 2012, 08:19 AM   #197
Calvin is offline Calvin  Germany
diyAudio Member
 
Calvin's Avatar
 
Join Date: Nov 2004
Location: close to Basel
Hi,

in Mono mode the DACs currents just double to -12.4mA | +-7.8mApp for the PCM1794a, or -7mA | +-4mApp for the PCM1795.
Now the outputs of L+ and R- must be connected together as noninv current summing point. Same applies L- and R+ for the inv sum. The DAC must be set to mono mode and the appropriate channel Left or Right selected via setting of Pin1 and Pin2 of the Chip.

As mentioned before low THD requires that the signal current stroke remains small against the current conveying transistor´s bias current. As such a PCM1795 in Mono mode has similar current values and this results in similar IV-distortion figures as a PCM1794a in stereo mode. On the other hand will a PCM1794a in mono mode lead to increased THD of the IV-stage.
Higher DAC output currents are not neccessarily a improvement.
If for example one bypasses the digital filter section of the DAC chips a PCM1795 in Mono mode may be better than a PCM1794a in stereo mode.

jauu
Calvin

Last edited by Calvin; 16th August 2012 at 08:32 AM.
  Reply With Quote
Old 16th August 2012, 08:53 AM   #198
MiiB is offline MiiB  Denmark
diyAudio Member
 
Join Date: May 2005
Location: Denmark
Could this be a valid model for the balanced DAC current output..??
Attached Images
File Type: jpg current source.JPG (14.9 KB, 240 views)
  Reply With Quote
Old 16th August 2012, 09:24 AM   #199
smms73 is offline smms73  Portugal
diyAudio Member
 
smms73's Avatar
 
Join Date: Jul 2005
Location: Lisboa, Portugal
Quote:
Originally Posted by Terry Demol View Post
Good circuit

I have built quite a few circuits based on these principles using mirrors to
sum the +and- phases then that current modulating cascoded OP stage.
DAC's , front ends of open loop SS power amplifiers and even a mic pre that
have a bal IP and SE OP (fully complimentary).

I have also implemented the same type of OP stage with those cross
coupled darlingtons. They do work very well to linearise non linear base currents and capacitances.

It's really a design twist of the 'Baxandall pair' connected BJT's, using the
'error' current from the bases of the main OP BJT's as + feedback. Your sim
numbers tell the story.

It all works just like it should in real life too with one gotcha - stability.
You may or may not have stability issues. It depends on layout etc. C1 will
certainly help but you may need additional base stoppers in either the main
OP BJT's or the 'helpers'.

Once you get the whole thing on a board and prototyped, it really is worth
having a good listen with and without all the Baxandall collector connections
to up/down labelled points. With these connected, the distortion will
certainly measure a lot better, however it may not actually sound better.

IME, with all these small + feedback loops, the sound can be worse. Without
them it is often more natural, but you won't get the super low dist numbers.

I'll be interested if you have similar results with this design. Keep us posted.

BTW, your choice of 327 / 337 in most locations is a pretty good one for 2
reasons - who can guess what they are?


T
Nice to hear that someone has already test the concept, and it work.

I have thought of this idea a long time ago when i needed a high voltage amplifier for a valve circuit (the input was a valve), but it was never build and it remain forgotten until last week, when I remembered that could be a good idea for a i/v converter. I never heard of 'Baxandall pair' but i will investigate.

STABILITY:
stability is one of the biggest problems in audio circuits, and for me is a very important matter.
Is perhaps the number one reason as otherwise very good concepts sound bad,. And is also one of the reason people tend to prefer a simple circuit versus a more complex one, as normaly, in many schemas there is no attempt to use stability measures

I do not know which methods you use to stabilize the circuit but rely only in base stoppers and good pcb layout is important but not enough.
this topology is very fast, so there are many possibilities for instability to build up. the
solution is to slow it down.
Today i have spend some time analysing possible nodes where stability can be a problem, and i will post soon the results.
After the circuit is rock solid stabilized there will be no reasons to sound bad.
  Reply With Quote
Old 16th August 2012, 09:58 AM   #200
smms73 is offline smms73  Portugal
diyAudio Member
 
smms73's Avatar
 
Join Date: Jul 2005
Location: Lisboa, Portugal
Quote:
Originally Posted by Calvin View Post
Hi,

the current mirror reminds of an ´full Wilson´. See for example sboa046 "IMPLEMENTATION AND APPLICATIONS OF CURRENT SOURCES AND CURRENT RECEIVERS" of TI/BB.
A ´complementary full Wilson´ is the Zetex part ZDS1009 in SM-8 casing, which might be a possible replacement part for the discrete BC327/337, because of the matching issues/oscillation issues of these cascoded current mirrors.

Well You can of course imbalance the whole stage by increasing the lower ccs bias current, but what´s the deal in designing a symmetrical stage just to spoil its symmetry right away to accommodate for the DAC? Yes, a dedicated CCS would mean more complexity and a slight noise penalty. But a simple JFET ccs is the simplest and the least noisy ccs You can have here, just two parts, a JFET and an R. Noise figures will be slightly higher (1-2dB), but still be well below -110dBV.
In the case of a current sourcing DAC with an offset current a SE-circuit, basically using just the upper half and a ccs towards the negative supply, would be more elegant and less complex.
I agree with Terry, that the multiple loops and cascoded mirrors may give stability problems and may sonically not be as overwhelming as the distortion figures imply (which have yet to be proofed in reality).
In my tests a simple SE-grounded base stage (rem. "Jocko" or SEN) certainly sounded better than a similar, but more complex and lower THD Hawksford current steering I/V stage.
As Zinsula correctly remarked in #159, THD improves with rising Ibias/Idac relationship. But there´s a sweet-point of optimum bias depending on the chosen transistors. The PCM1795/1796 or 1798 DACs have lower currents (-3.5mA+-2mA) than the 1792/1794A and allow for lower power requirements. Sourcing of matched transistors may be easier if SMD casings can be used, due to lower currents and power requirements. You just need to increase the value of the Riv for same level output.

The probabely best match to the PCMs is the TI/BB SRC4392 asynchronous sample rate converter and DIX.

You may linearize input impedance with the Zobel, but You put a burden on the DAC that needs to charge/discharge the Cap. It slows and rounds the current steps of the DAC (lower bandwidth) and is often used with OPamp-IVs featuring ´too slow´ OPamps. I typically found that such caps were no improvement sonically, but quite the opposite.

jauu
Calvin
Thanks for the tip about ZDS1009 and the interesting literature about current mirrors , i will see if i arrange time to read it.

In the last circuit i have post the distortion is smaler with the -6.2ma off ofset that without them.

about the stability, i have talk about that in my last post, and the final circuit will not have stability problems.
I have been posting the evolution of this project, i start to doubt if this is the best thing to do in this forum, but ok.

the idea of using a small current output dac as pcm1795/96 is a very good one, i will see the datasheet on this dac.

The input snuber will have very low interaction in the audio frequency, and is not a burden for the dac.
  Reply With Quote

Reply


Hide this!Advertise here!
Thread Tools Search this Thread
Search this Thread:

Advanced Search

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off


Similar Threads
Thread Thread Starter Forum Replies Last Post
ultra low noise low distortion preamplifier needed neazoi Solid State 13 11th October 2011 01:03 PM
IRF610 I/V convertion from Pass Labs D1 dac - single ended schiller Digital Line Level 5 15th October 2009 04:33 PM
Distortion+low volume.. GeirW Tubes / Valves 5 25th November 2004 12:04 PM
Best low-noise low-distortion buffer? borges Solid State 7 9th December 2003 12:40 PM


New To Site? Need Help?

All times are GMT. The time now is 04:07 PM.


vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2014 DragonByte Technologies Ltd.
Copyright ©1999-2014 diyAudio

Content Relevant URLs by vBSEO 3.3.2