My_Ref Fremen Edition - Collaborative DC-Servo design - diyAudio
Go Back   Home > Forums > Amplifiers > Chip Amps

Chip Amps Amplifiers based on integrated circuits

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 4th June 2014, 06:30 PM   #1
Reality first!
diyAudio Member
 
ClaveFremen's Avatar
 
Join Date: May 2005
Location: Milan, Italy
Default My_Ref Fremen Edition - Collaborative DC-Servo design

While it was planned I never introduced a DC servo in the FE since my electronic knowledge is pretty basic and designing a good sounding/performing DC servo is not a trivial task and I wanted first to have a good sounding/performing new amp as a base.

My_Ref FE Schematic

Black Gates for C9 position of the My_Ref and its variants, like my Fremen Edition, are good enough to mitigate the need for a DC servo but they are every day more difficult to source and prices are rising sky-high.

Now I think it's time to give the My_Ref a DC servo to eliminate such cap but I need help from more prepared forumers, like AndrewT, KSTR, Gotee or whoever is willing to help/teach us a pratical approach to its design.

In the last days I've had an interesting and useful mail exchange with Siva who is willing to contribute to this thread (in fact he already started contributing via mail )

Design goals are good performance, simplicity and a as lower as possible parts count since there's no much space on the PCB.

The first thing to define is where the DC servo has to be inserted in a nested design like this.

IMHO it's not too difficult if we 'model' the LM318 and LM3886 as a single power opamp, like in this simplified schematic:

Click the image to open in full size.

The model opamp is used in a non inverting topology since both stages (LM318 and LM3386) are inverting, so the combined amp is non inverting.

The next step is to add a simple, basic DC servo like the one used in AN-1192 (Figure 14):

Click the image to open in full size.

If I've undestood correctly the RSI1/CS1 combo is a LP filter with 0.16 cutoff freq. (AN-1192 circuit used 2.21M and 0.47uf with fc=0.15Hz) so that the integrator deal only with near DC signal.

The RSO1 resistor, from what I've read, is usually set 10x the feedback resistor of the amp so to lower noise and AC injection to the main amp.

Another approach could be Bob Cordell one's (from his book Designing Audio Power Amplifiers) which uses a single inverting integrator, so to use a single high quality cap, followed by an inverting unity gain integrator to have a non inverting DC-Servo.

Click the image to open in full size.

I've, at leat initially, excluded post DC-Servo filtering since the shunt cap would be another critical (soudwise) part, not less important than the original C9.

If a good enough circuit will be defined I'll need some beta testers to build the amp and possibly test it for stability and objective performance with measuring/diagnostic equipment, subjective testers are obviously welcome too.

Any feedback/suggestion/contribution/revision is welcome.
Attached Images
File Type: jpg FE Simplified.jpg (38.2 KB, 995 views)
File Type: jpg FE DCS AN-1192.jpg (51.6 KB, 983 views)
File Type: jpg FE DCS Cordell.jpg (54.7 KB, 538 views)
__________________
Dario
ClaveFremen "Bailando Salsa en el Sietch"

Last edited by ClaveFremen; 17th June 2014 at 12:26 PM.
  Reply With Quote
Old 4th June 2014, 08:59 PM   #2
bcmbob is offline bcmbob  United States
diyAudio Member
 
Join Date: Dec 2009
Location: Bay City, Michigan
Are you thinking about a module that attaches to the current board or a new PCB?
__________________
Bob M.
"Arrange Whatever Pieces Come Your Way."
  Reply With Quote
Old 4th June 2014, 09:17 PM   #3
Reality first!
diyAudio Member
 
ClaveFremen's Avatar
 
Join Date: May 2005
Location: Milan, Italy
Quote:
Originally Posted by bcmbob View Post
Are you thinking about a module that attaches to the current board or a new PCB?
All is possible but I've thinking more about new PCBs, I don't know if it would be stable with a piggy back module.
__________________
Dario
ClaveFremen "Bailando Salsa en el Sietch"
  Reply With Quote
Old 4th June 2014, 09:46 PM   #4
bcmbob is offline bcmbob  United States
diyAudio Member
 
Join Date: Dec 2009
Location: Bay City, Michigan
__________________
Bob M.
"Arrange Whatever Pieces Come Your Way."
  Reply With Quote
Old 4th June 2014, 10:34 PM   #5
diyAudio Member
 
lehmanhill's Avatar
 
Join Date: Jan 2008
Location: Chelsea, Michigan
Quote:
Originally Posted by ClaveFremen View Post
All is possible but I've thinking more about new PCBs, I don't know if it would be stable with a piggy back module.
Dario,

You and Siva have already given this a fair bit of thought. I am pretty sure that I don't know enough to contribute much, but am willing to help where I can. I also would like to support as a beta tester and will work on how I could test stability. I have some measurement tools, not as good as others, and not enough experience to address the best way to test. I will put my focus to learning about how to test.

Although your idea of a new board is probably the right one, please keep open the idea of a daughter board as it gives existing boards an option.

Ironically, I am planning to try a few alternatives to C9, including the BG NX, Tom's Jensen's, and the low cost Mundorf M-Cap AC. That should make a nice contrast to the DC Servo question.

Jac
  Reply With Quote
Old 4th June 2014, 10:45 PM   #6
diyAudio Member
 
lehmanhill's Avatar
 
Join Date: Jan 2008
Location: Chelsea, Michigan
One other question to think about. R7 has compensation in the form of C32 and possible affects from C10. How do those function and how would a DC Servo working in parallel interact with them? This may be where simulation will help, but I am hoping that an expert has a good theoretical understanding and can help explain.
  Reply With Quote
Old 4th June 2014, 10:51 PM   #7
Reality first!
diyAudio Member
 
ClaveFremen's Avatar
 
Join Date: May 2005
Location: Milan, Italy
Hi Jac,

Quote:
Originally Posted by lehmanhill View Post
I am pretty sure that I don't know enough to contribute much, but am willing to help where I can. I also would like to support as a beta tester and will work on how I could test stability.
Thanks, your support is welcome and appreciated

Quote:
Originally Posted by lehmanhill View Post
Although your idea of a new board is probably the right one, please keep open the idea of a daughter board as it gives existing boards an option.
Before beta PCBs I'll try circuits on veroboards but at the moment I can't be sure about stability but the possibility is open.

Quote:
Originally Posted by lehmanhill View Post
Ironically, I am planning to try a few alternatives to C9, including the BG NX, Tom's Jensen's, and the low cost Mundorf M-Cap AC. That should make a nice contrast to the DC Servo question.
I don't think previous builders will need to upgrade to new boards.

Really, this DC-servo project is to keep the FE future proof, a BG NX in C9 could even be better sounding than the DC-Servo...

Quote:
Originally Posted by lehmanhill View Post
R7 has compensation in the form of C32 and possible affects from C10. How do those function and how would a DC Servo working in parallel interact with them? This may be where simulation will help, but I am hoping that an expert has a good theoretical understanding and can help explain.
Absolutely, this thread will try to address these questions too.
__________________
Dario
ClaveFremen "Bailando Salsa en el Sietch"
  Reply With Quote
Old 5th June 2014, 08:07 AM   #8
diyAudio Member
 
Join Date: Sep 2007
Send a message via Yahoo to linuxguru
Quote:
Originally Posted by ClaveFremen View Post
In the last days I've had an interesting and useful mail exchange with Siva who is willing to contribute to this thread (in fact he already started contributing via mail )
...
Click the image to open in full size.

The model opamp is used in a non inverting topology since both stages (LM318 and LM3386) are inverting, so the combined amp is non inverting.

The next step is to add a simple, basic DC servo like the one used in AN-1192 (Figure 14):

Click the image to open in full size.
Thanks, Dario. To summarize the initial findings:

1) It's possible to inject the correction signal either at the input node or the feedback node (with appropriate changes to the servo polarity). Both topologies are in use in the real world. The AN-1192 circuit (as well as most of the discussion from Gootee on the earlier thread here) injects into the feedback node. The Evolution injects into the input node.

I'm inclined to go with the feedback injection topology, since it simplifies several things. However, since the feedback node has a low impedance to ground (390 || 12k), the pull range of the servo will be fairly small unless a lot of current is injected into the node.

2) As a quick back-of-the envelope calculation, +/- 1 mA gives a pull range of 1 mA times 390 R, or ~390 mV at the feedback node. This seems adequate to correct about +/- 12V at the output. To get +/- 1 mA injection current at +/- 10V swings at the output of the DC servo, we need a series injection resistor (RS01) of about 10k.

3) The Servo has an easier job if the input to the LM318 is AC coupled as shown above - it then only has to correct the offsets introduced by the opamp and chipamp, which are only in the range of 10s of mV. In that case, RS01 can be increased - maybe to somewhere in the 47k to 100k+ region.

4) It seems prudent to to split RS01 into an RCR Low-pass filter T-section, with the C grounded. For instance, if RS01 is ~10k, something like 1k-3.3uF-10k seems to work well in the preliminary simulation (LTSpice seems to have an easier job in converging when this LPF is used, as compared to just using a plain R).

Quote:
I've, at leat initially, excluded post DC-Servo filtering since the shunt cap would be another critical (soudwise) part, not less important than the original C9.
The C is significantly smaller, allowing the use of something like a Wima MKS2XL in the 470nF to 3.3uF range, which is more linear than a BG-PK 220uF. There's also an element of isolation in the form of the second R in the RCR section - if that's > 10k, the contribution of the C to the sonics of the feedback network will be mostly negligible and inaudible.

Last edited by linuxguru; 5th June 2014 at 08:16 AM. Reason: typo
  Reply With Quote
Old 5th June 2014, 01:11 PM   #9
diyAudio Member
 
lehmanhill's Avatar
 
Join Date: Jan 2008
Location: Chelsea, Michigan
Siva,

Very interesting. A lot to think about. I'm out today, but will post what I think is a slightly different version that I found on the internet.

On the subject of the effects of C32, I am coming to the conclusion that the DC Servo will have very little interaction. Please correct me if I am wrong, but it seems that C32 is a bandwidth limiter, that is, at very high frequencies, C32 easily passes current and that reduces gain at high frequencies. This is done to limit the amplification of high frequency noise and to improve stability of the loop at frequencies well above the audible range.

Since the DC Servo is designed to only work at and near DC, it will filter itself to a very low level at high frequencies shouldn't interact with C32.

Between the two designs that Dario showed schematics, I am currently liking the Cordell design. A few reasons that may not be valid. The Cordell design would seem to offer a more consistent output impedance because the second opamp is effectively buffering the varying impedance of the first stage. I suspect that a constant and mostly resistive output impedance would inject less phase distortion to the main feedback circuit.

Also, the second stage of the Cordell design could be at a higher gain, if we need more current injected near DC.

Finally, I like that, in the Cordell design, both opamps are closely referenced to ground. I usually try to avoid using Mega Ohm resistors whenever possible because the are noisy and can pick up RF noise too. The idea that the AN-1192 circuit seems to be isolated between 2 one Meg resistors looks like an opportunity for noise to sneak into the system. Also, any mismatch between the 1 Meg resistors or the caps would show up as an offset voltage at the opamp inputs, wouldn't it? That would make input offset current and quiescent current of the opamp extremely critical, if I am understanding this correctly. I think that the Cordell design would be less sensitive to opamp specs and would probably put it's emphasis on input offset voltage.

Just some thoughts from a very non-expert. Please react with other viewpoints.
  Reply With Quote
Old 5th June 2014, 10:37 PM   #10
KSTR is offline KSTR  Germany
diyAudio Member
 
KSTR's Avatar
 
Join Date: Jul 2007
Location: Central Berlin, Germany
Why not try the easy things first? LM318 has offset balancing pins (pin 1&5) that have not been used in the design so far. Part count is minmal, one 100k trimpot to positive supply.
  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
My_Ref Fremen Edition RC - Build thread ClaveFremen Chip Amps 1250 19th September 2014 10:56 AM
My_Ref Fremen Edition - Build thread and tutorial ClaveFremen Chip Amps 1886 17th September 2014 09:46 PM
My_Ref Fremen Edition GB (Second run) ClaveFremen Group Buys 52 22nd January 2014 08:13 PM
My_Ref Fremen Edition GB (interest for a third run) ClaveFremen Group Buys 107 12th January 2014 11:12 PM
My_Ref Fremen Edition - need help on PCB evaluation ClaveFremen Chip Amps 450 23rd October 2013 03:38 PM


New To Site? Need Help?

All times are GMT. The time now is 07:43 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