Self-biasing class AB input stage - diyAudio
Go Back   Home > Forums > Amplifiers > Solid State

Solid State Talk all about solid state amplification.

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 21st April 2009, 12:53 AM   #1
Mr Evil is offline Mr Evil  United Kingdom
diyAudio Member
 
Join Date: Aug 2004
Location: Behind you
Default Self-biasing class AB input stage

Having recently been considering class AB input stages, I wondered how it might be possible to reduce the added complexity of multiple current sources that the long tail pair type often use. One possibility is to use the LTPs' own current mirrors as current sources.

Now, class AB LTP input stages are symmetrical, so you can't just put current sources in them because the VAS quiescent current becomes undefined. So the first thing to do is apply some current feedback from the VAS. This yields the circuit on the left in the attached picture.

Changing this to class AB yields the circuit on the left. The LTPs are connected at the tails and their tail current sources removed so that each LTP provides the current for the other, and the bases are biased on with diodes. The current can increase past the quiescent current, hence the class B part. The constant currents that feed the diodes are simply provided by adding extra outputs from the current mirrors (Q9/11/13/14). Note that the emitter resistors for these extra current source transistors can be larger than those in the main current mirror to reduce the current through the diodes, for efficiency (Widlar current mirror). It would also be possible to use a "diamond buffer" arrangement to provide the bias voltage instead of diodes, giving higher input impedance.

The whole point of class AB input stages is to increase slew rate, so I simulated that and got 60V/us, compared to 30V/us for the class A version. This is with relatively large 100pF Miller compensation capacitors, LTP tail current at 4.7mA (I1 and I2 in the class A circuit, the current through R5 in the class AB circuit) and VAS current also at a conservative 4.7mA. GBP, phase margin and distortion are all approximately the same for both circuits, so the main compromise is between slew rate and complexity.

Higher slew rates would be possible by increasing LTP current and decreasing the value of R5 and/or the emitter degeneration resistors (which are inextricably linked to the quiescent current, unless the bias scheme is changed). Insanely high slew rates are possible.

So what do you do with all this slew rate? I dunno, but it's interesting anyway.

Note that the capacitors C3/4 were needed to prevent overshoot in the sims, but might not be needed in a real circuit.

EDIT: The image was too big, so I chopped the supply voltages off.
Attached Images
File Type: png classa+ab.png (7.6 KB, 1021 views)
__________________
https://mrevil.asvachin.eu/
  Reply With Quote
Old 21st April 2009, 01:25 AM   #2
diyAudio Member
 
Wavebourn's Avatar
 
Join Date: Sep 2006
Location: Pleasant Hill, CA
Send a message via Skype™ to Wavebourn
Good work!

Here is what I used when wanted as simple as possible complementary class A opamp with gain = 1.
A first, I replaced transistors in current mirrors by diodes.
Second, I replaced transistors in LTP pairs by diodes. And third, I used resistors in tails so currents depend on voltage. LTP currents were adjusted by resistors in tails to get no more than max needed current swing on the real output load, so in case of accidentally lower load resistance it protected itself limiting the current. Surprisingly, the sound was vacuum tube-like!
Attached Images
File Type: jpg gift.jpg (17.2 KB, 963 views)
__________________
"Our youth [...] have bad manners, contempt for authority; they show disrespect for their elders, contradict their parents, [...] and tyrannize their teachers. -- Plato, 447-367 BCE
  Reply With Quote
Old 21st April 2009, 01:31 AM   #3
andy_c is offline andy_c  United States
Banned
 
Join Date: Apr 2003
Just curious - have you seen the class AB input stage from the Smith, Koen and Witulski paper? I think that paper is on OS's site. Here's a pic.
Attached Images
File Type: png classabinputstage.png (5.6 KB, 828 views)
  Reply With Quote
Old 21st April 2009, 01:39 AM   #4
syn08 is offline syn08  Canada
Account disabled at member's request
 
Join Date: Aug 2005
Location: Toronto
Quote:
Originally posted by andy_c
Just curious - have you seen the class AB input stage from the Smith, Koen and Witulski paper? I think that paper is on OS's site. Here's a pic.
On top of my head, before any analysis (manual or Spice) this schematic looks very depending on matching and resistor ratios. If I am correct, then there's little chance to make it working with discretes (assuming it would make any sense to use it). Such schematics were extremely interesting about 30 years ago, when the lateral pnp in standard linear processes was a PITA for any analog designer.
  Reply With Quote
Old 21st April 2009, 01:46 AM   #5
diyAudio Member
 
Join Date: Dec 2007
I designed a high-performance opamp for driving low-Z headphones (50R and under) a while back and have been refining it for a while. The idea was from the Smith et al paper referenced earlier (Figure 13, specifically). You might be interested in it. Performance is absolutely excellent (I have an AP 2722 to play with intermittently so I can get some real numbers).

It is fully class-AB, including the input stage, which is a dual diamond buffer arranged as a fully-differential V-to-I converter. A simple current-mirror gain stage slews a gang of output MOSFETs. All devices are TO-92; my stereo prototype is on a 4.5x4.5" protoboard along with a DC servo. The gate-drain capacitance of the FETs is enough to provide good compensation. It's picky about decoupling around the input stage, so the schematic is a pseudo-layout to show what is needed.

Basic specs:
* Unity gain bandwidth: 35MHz (measured; unity gain stable; 60 deg phase margin)
* Open-loop bandwidth: ~4KHz (approx. 80db OLG)
* Slew rate: 1000v/uS (!!!) (measured; 10v step on 24v rails)
* THD/IMD: under -110dB (<20khz sine at around 2v into 50R), 2nd harmonic
* Idle current: about 12mA for input and gain stages, output set at 20mA
Attached Files
File Type: pdf cfb2.pdf (47.6 KB, 218 views)
  Reply With Quote
Old 21st April 2009, 01:49 AM   #6
andy_c is offline andy_c  United States
Banned
 
Join Date: Apr 2003
Quote:
Originally posted by syn08
On top of my head, before any analysis (manual or Spice) this schematic looks very depending on matching and resistor ratios.
I think you'd want all four resistors equal in this thing, and all NPNs and PNPs to be the same type. If you go around any loop from -Vin to +Vin, there's two resistors (same as a normal diff amp) and four Vbe's. So I'd expect it to be somewhat worse than a normal diff amp for offset behavior. I wouldn't dismiss it as unusable in a discrete design without some more careful investigation though.
  Reply With Quote
Old 21st April 2009, 01:59 AM   #7
Mr Evil is offline Mr Evil  United Kingdom
diyAudio Member
 
Join Date: Aug 2004
Location: Behind you
Quote:
Originally posted by andy_c
Just curious - have you seen the class AB input stage from the Smith, Koen and Witulski paper? I think that paper is on OS's site. Here's a pic.
I haven't seen that before. It looks like it might be a solution to a problem I was contemplating some time ago for using the attached input stage topology in an audio amp. It's functionally identical to an class AB LTP but uses a complementary pair and both outputs are non-inverting. I've used it for things like voltage regulators and other places where the 2*Vbe offset between the inputs doesn't matter or is even useful. Like the one you posted, it's interesting because you get the output from both ends, opening up novel VAS possibilities.



Quote:
Originally posted by sparcnut
...It's picky about decoupling around the input stage, so the schematic is a pseudo-layout to show what is needed...
That's one of the problems with these things - as they get faster, they become very good not only at amplifying audio, but also anything else that comes near, like mobile phones, switching power supplies, and alien mind control rays.
Attached Images
File Type: png ltp.png (2.1 KB, 721 views)
__________________
https://mrevil.asvachin.eu/
  Reply With Quote
Old 21st April 2009, 02:07 AM   #8
syn08 is offline syn08  Canada
Account disabled at member's request
 
Join Date: Aug 2005
Location: Toronto
Quote:
Originally posted by andy_c


I think you'd want all four resistors equal in this thing, and all NPNs and PNPs to be the same type. If you go around any loop from -Vin to +Vin, there's two resistors (same as a normal diff amp) and four Vbe's. So I'd expect it to be somewhat worse than a normal diff amp for offset behavior. I wouldn't dismiss it as unusable in a discrete design without some more careful investigation though.
Offset is only one side of the story. I would be more concerned about poor CMRR (in a discrete implementation).

What would be the use of this configuration in the discrete world? We already have all the PNP's we need, beta, Ft, you name it.
  Reply With Quote
Old 21st April 2009, 02:15 AM   #9
andy_c is offline andy_c  United States
Banned
 
Join Date: Apr 2003
Quote:
Originally posted by syn08
What would be the use of this configuration in the discrete world?
Well, the thread is about class AB input stages, and that's what it is. So it would be for high slew-rate applications.
  Reply With Quote
Old 21st April 2009, 02:27 AM   #10
syn08 is offline syn08  Canada
Account disabled at member's request
 
Join Date: Aug 2005
Location: Toronto
Quote:
Originally posted by andy_c


Well, the thread is about class AB input stages, and that's what it is. So it would be for high slew-rate applications.
Ah, ok
  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
LED cathode biasing in push-pull input stage exeric Tubes / Valves 14 18th August 2008 01:35 AM
Input stage biasing - BJT steve_mak Solid State 24 11th July 2007 03:38 AM
low-biasing or high biasing in Class-AB amps Workhorse Solid State 124 18th October 2005 04:45 AM


New To Site? Need Help?

All times are GMT. The time now is 02:09 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