lol sorry its a while since I designed any class AB amps and got the VAS and Vbe mulitpliers mixed up.
Yes, the whole point of it is to provide bidirectional current limiting in conjunction with D1 and D3. In practice, there's a couple of gotchas. The trick is to make sure that neither Q1 nor Q3 can turn off. IIRC, the main things to be aware of are Q2's collector current being added in at Q3's emitter, and that the full input stage current flows through D1/D3 during clipping.
I guess it does, but that wasn't a design consideration. I expect that allowing voltage swing on Q1's collector actually makes distortion a bit worse, especially as Vce is so low to start with. I considered that a small price to pay in exchange for civilized clipping behavior though.
Not really, it just lowers distortion a bit and gives higher gain. A more normal approach might be to connect the bottom of R9 to the -30V rail, but if you do that, then the voltage across R9 = Vbe of Q1 + Vbe of Q2. The problem is that the signal voltage across Q1's base-emitter junction is much larger and more distorted than that of Q2. With a current mirror, it's not an issue as there's no input resistor.
I'd want to clip the output a few volts before the output stage reaches it's limits. My approach would probably be to run the front end from the same rail voltage as the output stage but with extra filtering, and then set the clipping level 5 volts or so below the front-end supply voltage.
IMHO it's a mistake to use higher voltage rails for the VAS than the output stage to maximize output power. Sure you get a couple of extra watts, but clipping behavior is horrible.
Attachments
godfrey,
Makes sense. I bet where the higher rails are used, additional protection circuitry to make it more beguine is employed. That's one thing I noticed between top amps of the 80's and now. New ones have a LOT of parts for management.
Why such high VAS current? The other amps I have looked at are more like 6 or 8 mA.
Makes sense. I bet where the higher rails are used, additional protection circuitry to make it more beguine is employed. That's one thing I noticed between top amps of the 80's and now. New ones have a LOT of parts for management.
Why such high VAS current? The other amps I have looked at are more like 6 or 8 mA.
jcx, I am using a beta enhanced VAS that feeds a cascode. And the whole amp has TMC and it appears to be quite stable and I have done a lot of testing on the physical amp. I prefer to see TMC as 'transitioning' the output stage (triple in my case) out of the global feedback loop at HF rather than positive feedback, or are you perhaps talking about a different mechanism?
I think in a classic Darlington connection to the VAS you talk about Homeodder then Cob indeed is an issue, but the VAS feeding the cascode is a little different. Sims show a very big improvement at HF (both with and without TMC). The whole idea of fixing the collector voltages is so that changes in Vce cannot take place and changes in Vcb are minimal anyway (a few mV) However, I'd be interested to see you findings - I might be missing something here.
I think in a classic Darlington connection to the VAS you talk about Homeodder then Cob indeed is an issue, but the VAS feeding the cascode is a little different. Sims show a very big improvement at HF (both with and without TMC). The whole idea of fixing the collector voltages is so that changes in Vce cannot take place and changes in Vcb are minimal anyway (a few mV) However, I'd be interested to see you findings - I might be missing something here.
It is a bit over the top. I was using a double EF output stage and wanted a VAS with very low output impedance and high output current capability, to drive it with a "fist of iron". (The low output impedance happens with local feedback connected, but requires high current gain, hence the Darlington)
Thanks. Long learning process.
I have seen two pole Miller compensation, called TMC correctly or not, implemented several ways. The leg returned to ground, to a rail ( which makes no sense to me), or to the output. Sometimes they seem to have huge effects, sometimes nothing in basic sim. Where do I get a more complete read on the subject?
I have seen two pole Miller compensation, called TMC correctly or not, implemented several ways. The leg returned to ground, to a rail ( which makes no sense to me), or to the output. Sometimes they seem to have huge effects, sometimes nothing in basic sim. Where do I get a more complete read on the subject?
See Douglas Self's Audio Power Amplifier Design Handbook. In the 5th edition, he covers it some on pgs. 218-222, and 362 (look under compensation in the index).
Douglas puts a bit more information on TMC (transitional miller compensation)
at this link, the "Baxandall Papers".
The Baxandall Papers: Transitional Miller compensation
Jan Didden has available more of that here:
Linear Audio | your tech audio resource
Douglas puts a bit more information on TMC (transitional miller compensation)
at this link, the "Baxandall Papers".
The Baxandall Papers: Transitional Miller compensation
Jan Didden has available more of that here:
Linear Audio | your tech audio resource
There's been a lot of discussion of TPC (two pole compensation) and TMC (transitional Miller compensation) in Bob Cordell's threads, especially this one: http://www.diyaudio.com/forums/solid-state/94676-bob-cordell-interview-negative-feedback.html, but also these two: http://www.diyaudio.com/forums/solid-state/171159-bob-cordells-power-amplifier-book.html and http://www.diyaudio.com/forums/solid-state/89023-bob-cordell-interview-error-correction.html.
Long threads, but worth reading if you ever have the time.
Agreed it would, but there is a difference though between beta enhanced and the circuit shown by godfrey. Beta enhanced means collector of first transistor is connected to either ground or opposing voltage supply. In a Darlington the two collectors are joined then fed to a cascode. For fair comparison though youd need to keep the gain the same or try a beta enhanced cascoded vas and then compare to a darlington cascoded vas. It has to be apples compaired to apples. I recall Self does talk a little about this in his amplifier book although he only touches on why beta enhancement is better than darlington. I can think that the darlington would eventually render better results at some point as some darlingtons can have hfe s of upto 40 000 however one could use super beta transistors in beta enhanced cascoded vas.
What's the best Vas stage in power amp? Simple one or two transistor? Darlington? Cascode? Complementary sysmetry?...
Of course, ''the best'' are: sounding, gain, bandwidth, speed, distortion...
For me this is the best VAS:
Memory Distortion Philosophies - Part 5 : Circuits, continued
check also out this threads
http://www.diyaudio.com/forums/solid-state/336-lavaradin-amp-memory-distortion.html
http://www.diyaudio.com/forums/solid-state/25428-cancelling-memory-distortion.html
http://www.diyaudio.com/forums/everything-else/29976-memory-distortion-crap-serious.html
http://www.diyaudio.com/forums/solid-state/100810-memory-distortion-some-new-beginnings.html
http://www.diyaudio.com/forums/solid-state/30819-new-renardson-amp.html
http://www.diyaudio.com/forums/solid-state/125650-mjr-7-mosfet-amplifier-m-renardson.html
http://www.diyaudio.com/forums/soli...lifiers-feedback.html?perpage=25&pagenumber=1
An other VAS stage without input-LTP is to find by post #4 about
http://www.diyaudio.com/forums/soli...ody-burmester-avm-schematics.html#post2598494
Last edited:
I was under the impression that you were opposed to high feedback amps.
I'd certainly encourage anyone using TMC to try the equivalent 2-pole with matched high frequency gain, stability (not the same component values; search for user: megajocke posts in the Cordell book thread for equations)
I suspect some would reject the 2-pole comp for appearing unstable - not realizing their TMC is "the same" on stability
as long as people fully understand the stability equivalence then they can decide on the performance differences in their amp's particular circuit topology
my sims shows matching gain intercept with TMC, conventional Miller, measured in the outer, global loop gives a higher "true" gain intercept for TMC as measured around the output Q, and less phase margin: http://www.diyaudio.com/forums/soli...lls-power-amplifier-book-134.html#post2420438 - and next page
judging TMC stability from analogy with conventional Miller, or based on step response damping will be overly optimistic
I suspect some would reject the 2-pole comp for appearing unstable - not realizing their TMC is "the same" on stability
as long as people fully understand the stability equivalence then they can decide on the performance differences in their amp's particular circuit topology
my sims shows matching gain intercept with TMC, conventional Miller, measured in the outer, global loop gives a higher "true" gain intercept for TMC as measured around the output Q, and less phase margin: http://www.diyaudio.com/forums/soli...lls-power-amplifier-book-134.html#post2420438 - and next page
judging TMC stability from analogy with conventional Miller, or based on step response damping will be overly optimistic
Last edited:
Edmond/jcx, what cross over frequency do you normally shoot for with TMC assuming 1281/1382 output stage and an equivalent MC ULG of say 1MHz? I have tried from 500KHz throght to nealy 1MHz without too many problems. However, I will admit I always err well on the side of caution with loop gain and ULG. Interested to know your views.
It is not always possible to consider the so-called VAS (Voltage Amplification Stage) as independent of the Input Stage (IS).
Impressive results have been obtained from a very simple topology where the IS also works as a VAS (see the link to Renardson's MRJ7 given by Tiefbassuebertr).
Currently, I would like to built an amp using the push-pull VAS (differential pair loaded by a current mirror) shown by Samuel Groner fig. 56, page 48 of his "Comments on Audio Power Amplifier Design Handbook by Douglas Self". There is a common mode form of feedback from the second stage to the first stage which allows equality of current in the differential input devices. This topology allows a very high PSRR which may be a feature as important as low distortion.
Impressive results have been obtained from a very simple topology where the IS also works as a VAS (see the link to Renardson's MRJ7 given by Tiefbassuebertr).
Currently, I would like to built an amp using the push-pull VAS (differential pair loaded by a current mirror) shown by Samuel Groner fig. 56, page 48 of his "Comments on Audio Power Amplifier Design Handbook by Douglas Self". There is a common mode form of feedback from the second stage to the first stage which allows equality of current in the differential input devices. This topology allows a very high PSRR which may be a feature as important as low distortion.
I like to add that TMC gives greater improvement of the TIS (formerly called VAS). That's why TPC and TMC give largely equivalent results (sometimes TPC is slightly better, sometimes the other way around, depending on circuit details)
As for linearizing the input stage together with TMC, this topic has been covered here: (Super TIS)
Cheers,
E
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.