Here's a link to Samuel Groner's article on new amplifier topology...It originally appeared in Linear Audio, and now has showed up in EE Times Design Section:
A new audio amplifier topology with push-pull transimpedance stage - Part 1: Introduction
Akitika GT-101
A new audio amplifier topology with push-pull transimpedance stage - Part 1: Introduction
Akitika GT-101
@djoffe
The link to the online article only shows the first figure, the rest are blank so I dug out the
Linear Audio article. There sure are a lot of folded cascodes in that design but since they
are grounded base and unity current gain they should not harm stability too much. A lot
of "extra" transistors in that design but they are cheap. Even with the front end on full
rail voltages many of the transistors only see one rail which allows for the use of lower
voltage devices.
It seems strange that the goal seems to be for the basic design to have much better PSRR
which he achieves but then he goes ahead and regulates the front end supplies. Interesting
that the front end can run on lower supplies further reducing the need for high voltage
transistors. This has the advantage of allowing a JFET diff pair and if the regulation is to
+/-15V that supply can also be used for a DC servo OP amp.
I thought that one of his goals was high slew rate but the final design was in fact very slightly
worse than a Blameless topology. He used CDOM compensation and I suppose that several
of the more advanced types could be employed.
The link to the online article only shows the first figure, the rest are blank so I dug out the
Linear Audio article. There sure are a lot of folded cascodes in that design but since they
are grounded base and unity current gain they should not harm stability too much. A lot
of "extra" transistors in that design but they are cheap. Even with the front end on full
rail voltages many of the transistors only see one rail which allows for the use of lower
voltage devices.
It seems strange that the goal seems to be for the basic design to have much better PSRR
which he achieves but then he goes ahead and regulates the front end supplies. Interesting
that the front end can run on lower supplies further reducing the need for high voltage
transistors. This has the advantage of allowing a JFET diff pair and if the regulation is to
+/-15V that supply can also be used for a DC servo OP amp.
I thought that one of his goals was high slew rate but the final design was in fact very slightly
worse than a Blameless topology. He used CDOM compensation and I suppose that several
of the more advanced types could be employed.
Did not read the whole article, but I am not wow-ed by the topology. The circuit is a current mirror loaded folded cascode input stage followed by a complementary folded cascode 2nd stage. While the 2nd stage is indeed complementary, the output currents are limited due to the folded cascodes, so I don't see the purpose for the complication.
@PB2: Do you see something special in this circuit?
Finally this is totally different from the AD797. The AD797 is a single stage amp, i.e. there is no VAS. The AD797 achieves high gain by using a very clever current mirror.
@PB2: Do you see something special in this circuit?
Finally this is totally different from the AD797. The AD797 is a single stage amp, i.e. there is no VAS. The AD797 achieves high gain by using a very clever current mirror.
The advantage that I was looking for was a push-pull VAS, it turns out that Groner was
looking for improved PSRR which he achieved. His front end works at lower voltages
which is another advantage providing for regulation without more taps on the transformer
and the use of low voltage diff pair transistors or FET without a cascode.
Do you have a favorite single diff pair front end with a push-pull VAS topology?
looking for improved PSRR which he achieved. His front end works at lower voltages
which is another advantage providing for regulation without more taps on the transformer
and the use of low voltage diff pair transistors or FET without a cascode.
Do you have a favorite single diff pair front end with a push-pull VAS topology?
Last year I saw an APT-1 in a friends shop. This was what one might call cosmetically blemished but if one ignored that it was a good find.
I bought equipment from this friend and he has written done the prices asked more than once.
He had no knowledge of the APT-1 so I told him it was a premium product in its' day and he sold it for a good profit. Apparently the cosmetic blemishes were not indicative of problems in the circuitry due to age.
I remember this amplifier from around 1980 and found a detailed service manual on Hi-Fi Engine. This was a design that looks ahead of its' time and it seems to be an unheralded Classic although there have been a couple of recent threads.
It is worth downloading the service manual which is more instructive as to the working than most manuals. This does have a single LTP driving complementary VAS stages.
Hi Pete, so I read the whole article, and I am changing my stance from not wow-ed to there is something there. As you point out, the main advantage that I also see is that you can run the input stage at lower power supply voltages with his topology so you can use regulators without issues as he shows in figure 7. He also shows improved distortion in figure 12, but IMO that is a side effect from some design choices he made rather than inherent to the topology.
The best single diff pair front end with a push-pull VAS topologies that I have seen are implemented in IC op-amps. The problem with these topologies is that they are difficult to implement in discrete. When building a push-pull VAS the problem you will always run into is how to bias the complementary gain transistors. All schemes in IC op-amps rely heavily on matching to do proper biasing. While matching is relatively easy in IC's, not so easy in discretes. I used a topology that worked quite well in my master's thesis which I wrote eons ago. You can read it in the following link, section 2.2 talks about the input stage and section 2.3 about the second stage.
https://dspace.mit.edu/bitstream/handle/1721.1/36765/78918759-MIT.pdf?sequence=2&isAllowed=y
For discretes, so far I have not seen something that I have been impressed so far. Even for the current design I am doing in my video series, the SW-VFA-01, I'll be using a non-complementary VAS design, since I have not come up either with a discrete design that would wow anyone. On the other hand, I'll be adding some improvements to the classic VAS that I have not seen anywhere (stay tuned for those).
The topology is ok. Biasing is done by loading the input stage with resistors and then using emitter resistors on the second stage transistors. Also to the get the complementarity, on leg uses the a current mirror for sense shifting.
I am sure it works, but it is not novel in anyway.
Forgot to mention, the topology I used in my thesis, was later used in the AD8029, so it is proven
.
Hi Pete, I just stumbled across this circuit also from Mr. Groner, and I like it quite a bit. No matched transistors needed, only matched resistors.
Regarding op-amp implementations, look up patents by Moshe Gerstenhaber in google patents. Specifically, Rail to Rail output stages. There are a few that are interesting and can be adapted to a push pull VAS (Moshe used to be one of my mentors at ADI along with Scott Wurcer).
Attachments
Hi Pete, I know that. I put it up because I thought it was interesting and to give you some inspiration. Here is my inspiration from Groner's circuit to fit your needs. What do you think?
The box labeled Groner's bias circuit is the same as the one in the schematic of my previous post: R13-R19, C3, U1, U2, and V1.
Enjoy, Sandro
P.S. Not sure why in the world the forum uploader keeps rotating the picture, in my computer the file is the right orientation
The box labeled Groner's bias circuit is the same as the one in the schematic of my previous post: R13-R19, C3, U1, U2, and V1.
Enjoy, Sandro
P.S. Not sure why in the world the forum uploader keeps rotating the picture, in my computer the file is the right orientation
Attachments
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.