Mikeks,
I have read Gunderson's patent. I am not sure whether the documented reduction in distortion is indeed due to the isolation of the capacitances between output stage and intermediate stage.
The proposed OFICC seems to effect this isolation, but it also adds quite some loop gain, which in itself would lead to considerable reduction in distortion.
It's not Hawksford error correction, as the loop gain for the OFICC is undefined quantitively.
What's your take on this?
Jan Didden
I have read Gunderson's patent. I am not sure whether the documented reduction in distortion is indeed due to the isolation of the capacitances between output stage and intermediate stage.
The proposed OFICC seems to effect this isolation, but it also adds quite some loop gain, which in itself would lead to considerable reduction in distortion.
It's not Hawksford error correction, as the loop gain for the OFICC is undefined quantitively.
What's your take on this?
Jan Didden
Gunderson's topology is actually an EC, whether it is intentionally built to be an EC or it is side effect of Gunderson's main intention.
I think this is why the distortion reduction can be so dramatical.
The AC voltage at base of Q24 will follow the output node 2.
The key is Q24.
Imagine output node 2 is too positive than it should be. Then the base of Q24 will be too positive too. With the same V-emitor of Q24, too positive of base voltage will make Q24 not-conducting.
If Q24 is not conducting, then Q25+Q26 will be pulled down (negative) by S24. Until balanced condition achieved.
The opposite condition. Output node is too negative than it should be. Then base of Q24 will be too negative towards its emitor, making Q24 has big VBE, making Q24 on. If Q24 on, then current will flow down towards base of Q25-26 and pull these transistor up until equilibrum.
But this EC mechanisme done by Q24 has limited current, that is pre-determined by Q23, cannot give more than that. (Unlike other EC that has other CCS to power its EC action).
A question. How big is the drop of D20? Is it 2V ? Is it 30V?
The Rotel schematic will be less effective in this EC mechanism. Rotel does not use Zener (fixed drop), but uses voltage divider (R679-R681) for determining cascode point towards output node. This will make the EC mechanism less effective than if using zener diode.
I think this is why the distortion reduction can be so dramatical.
The AC voltage at base of Q24 will follow the output node 2.
The key is Q24.
Imagine output node 2 is too positive than it should be. Then the base of Q24 will be too positive too. With the same V-emitor of Q24, too positive of base voltage will make Q24 not-conducting.
If Q24 is not conducting, then Q25+Q26 will be pulled down (negative) by S24. Until balanced condition achieved.
The opposite condition. Output node is too negative than it should be. Then base of Q24 will be too negative towards its emitor, making Q24 has big VBE, making Q24 on. If Q24 on, then current will flow down towards base of Q25-26 and pull these transistor up until equilibrum.
But this EC mechanisme done by Q24 has limited current, that is pre-determined by Q23, cannot give more than that. (Unlike other EC that has other CCS to power its EC action).
A question. How big is the drop of D20? Is it 2V ? Is it 30V?
The Rotel schematic will be less effective in this EC mechanism. Rotel does not use Zener (fixed drop), but uses voltage divider (R679-R681) for determining cascode point towards output node. This will make the EC mechanism less effective than if using zener diode.
lumanauw said:
The only Slone amp I have seen that uses some variant of this is the Optimos, and it also uses a resistive voltage divider instead of a fixed level shifter with zener diode or similar. I wonder if they think this is better or if it is intended as a way to circumvent Gundersons patent? The patent says that the base should be connected to the output, but what does that mean techically? Patents always tend to be unclear and cryptic, so is a level shifter a direct connection but a resistive divider not? (I admit already here that I only read the patent claims, not the whole patent).
Anyway, this is interesting since all the cases where Slone uses cascoded VAS's in his book use ordinary cascodes referenced to the rail, while in the later Optimos he suddenly switched to this approach. This has been up for discussion and speculation before on the forum, but then nobody had any better suggestion for its purpose than that it was intended as some kind of soft clipping circuit. The Gunderson patent certainly sheds some new light upon it.
So, are there any commercial amps using the level-shifting approach (if so presumably with a license from Gunderson)?
In the patent, the main intention is "isolation" between point 1 and 3. The patent seems don't care about "EC" effect of the schematic. (column 7 section 50-55)
But the pic.1 and pic.6 to me seems it is "EC" that lowers the distortion dramatically, not the "isolation"
If we want "isolation", then the cascode can be made by voltage divider towards output node (or even a classic static VAS cascode, I think).
Charles Hansen uses folded cascode towards output node too in his AyreV3proto.
But if we want "EC", zener would be better.
What is the value of Zener D20? What voltage range?
But the pic.1 and pic.6 to me seems it is "EC" that lowers the distortion dramatically, not the "isolation"
If we want "isolation", then the cascode can be made by voltage divider towards output node (or even a classic static VAS cascode, I think).
Charles Hansen uses folded cascode towards output node too in his AyreV3proto.
But if we want "EC", zener would be better.
What is the value of Zener D20? What voltage range?
scott wurcer said:
Yes, it didn't occur to me that so long time had passed that the patent might have expired already, so your comment was appreciated. With 17 years, I guess the Optimos was published very shortly after the patent expired, so that might explain why Slone suddenly used this technique while not having mentioned it in his book, which was published earlier.
Still I think this is alot to do about nothing. The "isolation" transistor can be seen as a simple additional gain stage with local feedback. You can call it EC, why not, but it is not Hawksford-style EC.
I am pretty sure that someone could come up with a similar circuit, call it "intermediate+" gain stage, and get the same distortion reduction without mentioning or suggesting anything about "isolation".
The resistiive divider instead of the zener level shift in some designs could well be to limit the additional loop gain so as to prevent stability problems, as this scheme does add loop gain to the forward path.
Jan Didden
I am pretty sure that someone could come up with a similar circuit, call it "intermediate+" gain stage, and get the same distortion reduction without mentioning or suggesting anything about "isolation".
The resistiive divider instead of the zener level shift in some designs could well be to limit the additional loop gain so as to prevent stability problems, as this scheme does add loop gain to the forward path.
Jan Didden
Hi, Janneman,
How are you
To make use of this cct's capability of EC, the level shifter is still better using zener (D20) instead of voltage divider (towards rails).
To maintain stability, I think we can put a resistor between Q23's collector and Q24's emitor (something like base stopper), and put resistor (about 22k maybe?) from Q23's collector to ground (position // with Co).
This way the EC still can obtained, but the stability is also good.
PS : How's your trip plan?
How are you
To make use of this cct's capability of EC, the level shifter is still better using zener (D20) instead of voltage divider (towards rails).
To maintain stability, I think we can put a resistor between Q23's collector and Q24's emitor (something like base stopper), and put resistor (about 22k maybe?) from Q23's collector to ground (position // with Co).
This way the EC still can obtained, but the stability is also good.
PS : How's your trip plan?
lumanauw said:Hi, Janneman,
How are you
To make use of this cct's capability of EC, the level shifter is still better using zener (D20) instead of voltage divider (towards rails).
To maintain stability, I think we can put a resistor between Q23's collector and Q24's emitor (something like base stopper), and put resistor (about 22k maybe?) from Q23's collector to ground (position // with Co).
This way the EC still can obtained, but the stability is also good.
PS : How's your trip plan?
Well, I still think calling it EC is misleading, this way you can call ALL feedback circuits EC (which they are in a sense, it's just that the uncommon use of the term is confusing).
To use a term you like: the circuit can be seen as a black box with a differential input (E and B) and a current output (C). That would make it clear it is just an added gain stage. This is a mix-up of isolation and added gain. Therefor, there is no way to know whether the increased performance is really due to isolation. It would be, if the black box had a gain of 1 with negligable reverse transmission. replacing the black box with a unity gain V to C converter (VCCS) would make that clear. Anybody wants to do the sims?
My trip plan: How is the situation in Panagaran and Bandung after yesterdays' tsunami? Panangaran is on my itenerary.
Can you give me any inside info on the situation, if necessary by PM?
Jan Didden
Hi, Janneman,
Maybe it will be misleading to call it EC All feedback amp is EC itself, since they compare output to input and try to fix the difference I agree, to advoid misinterpretation, EC is better used for very local fixing action, if covering some stages better call it feedback.
In other EC (Hawksford or NP-PMA), the EC cct has other power source to power the EC action (additional CCS or reserve of current/voltage). In here there is no other such energy source to fix things, all full dependant of what Q23 gives, cannot have more than that.
What is interesting to me, if the energy of fixing things comes only from differential pair+VAS transistor (Q23), how come adding Q24 gives quite a better performance, while there is no other energy source? Where the energy to fix things coming from? Fig 6 shows that it is happening with additional of Q24.
Pangandaran? I just got back from there 4 days ago (I come back to Bandung on last Friday). Lucky me, thanks God. I saw in TV, the hotels are quite flat to ground. The hotel that I was in (Nyiur Indah Hotel) maybe is not there anymore. In TV they become wrecks and messy. In Nyiur Indah Hotel, when I was there, there was 1 bus of German Tourist coming from Jogjakarta. They already checked out on Friday. There is a good seafood restaurant in Pangandaran, "Rasa Sayang" restaurant. Pangandaran is a very small place, unlike Bali. You can walk circle all Pangandaran in some hours.
I think your tour organizer will advoid Pangandaran on your forecoming trip. Before the tsunami, there are 3 tectonic earthquakes. If the earth slabs are still moving towards each other in these days, maybe better to advoid beaches.
But other than beaches, like here in Bandung, nothing is happening. (Bandung is 200km from Pangandaran).
These eartquake and tsunami are local accident. But better be safe.
Maybe it will be misleading to call it EC
All feedback amp is EC itself, since they compare output to input and try to fix the difference I agree, to advoid misinterpretation, EC is better used for very local fixing action, if covering some stages better call it feedback. In other EC (Hawksford or NP-PMA), the EC cct has other power source to power the EC action (additional CCS or reserve of current/voltage). In here there is no other such energy source to fix things, all full dependant of what Q23 gives, cannot have more than that.
What is interesting to me, if the energy of fixing things comes only from differential pair+VAS transistor (Q23), how come adding Q24 gives quite a better performance, while there is no other energy source? Where the energy to fix things coming from? Fig 6 shows that it is happening with additional of Q24.
Pangandaran? I just got back from there 4 days ago (I come back to Bandung on last Friday). Lucky me, thanks God. I saw in TV, the hotels are quite flat to ground. The hotel that I was in (Nyiur Indah Hotel) maybe is not there anymore. In TV they become wrecks and messy. In Nyiur Indah Hotel, when I was there, there was 1 bus of German Tourist coming from Jogjakarta. They already checked out on Friday. There is a good seafood restaurant in Pangandaran, "Rasa Sayang" restaurant. Pangandaran is a very small place, unlike Bali. You can walk circle all Pangandaran in some hours.
I think your tour organizer will advoid Pangandaran on your forecoming trip. Before the tsunami, there are 3 tectonic earthquakes. If the earth slabs are still moving towards each other in these days, maybe better to advoid beaches.
But other than beaches, like here in Bandung, nothing is happening. (Bandung is 200km from Pangandaran).
These eartquake and tsunami are local accident. But better be safe.
lumanauw said:
But there is a lot of energy sources, see fig 4 & 5. Current sources all over the place. The "isolation" transistor is modulating these and in that way providing the gain.
And no, I don't call the NP-PMA an EC amp, it has local feedeback loop inside the overall loop
Jan Didden
Don't worry, you can email me anytime. If there is anything here happening I will email you too
Indonesia is a huge-huge place. Many of islands. Like Bali is only a small-small part of Indonesia. If in TV news, the Pangandaran tsunami is like a big accident, it doesn't mean all Indonesia (or all Java island) is hit by tsunami. It's a local accident.
lumanauw said:
Don't worry, you can email me anytime. If there is anything here happening I will email you too
David, the hotel we were supposed to go to in Pangandaran is (was?) the Sandaan hotel. Do you know if it is still standing?
Jan Didden
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.