With a two-transistor Vbe-multiplier, you can get very low dynamic impedance that's constant from DC all the way up through the audio band, only rising at HF when the transistors' current gain starts falling....passive diode bias or Vbe multiplier...
By adding a zobel, you can keep the impedance low and (more-or-less) constant up into RF.
Values below are just ball-park guesses to show the idea. Note that Q1 (BD139) is on the heatsink, while Q2 can (should?) be at ambient temp.
Cheers - Godfrey
Attachments
I can't believe I just read all pages of this thread.
I was about to bring up my own bootstrap modification but the aim is to reproduce the AKSA in spirit and I'm sure Hugh has tested my idea before... Gets me thinking, if back EMF matters so much for the sonics, then perhaps output impedance has to be factored in with the aforementioned CLG, OLG stuff? AndrewT was "close"...
- keantoken
Hello keantoken
There is another way to made the amp more immune from back EMF.
You keep the negative feedback only for the LTP and the VAS, and no feedback at the drivers and output transistors, I've sim it and it have more distortion but a nice thd spectrum without bad hf thd harmonics.
It would need few mods and works to try that on the Baby Aksa but it can do a nice variation of the Baby Aksa. It would give a warmth tube sound with a very good sound-stage.
Bye
Gaetan
Last edited:
With a two-transistor Vbe-multiplier, you can get very low dynamic impedance that's constant from DC all the way up through the audio band, only rising at HF when the transistors' current gain starts falling.
This circuit also benifits from a B-C cap. Try it! The impedance of the transistor alone is multiplied by the resistor voltage divider, but if we bypass the top resistor we return to the impedance of a single transistor (or two in this case).
Gaetan, I've heard that inter-stage feedback sounds very good. However I'm not sure about better controlling EMF. Without taking feedback from the output, the output impedance becomes that of the emitter resistors and transistors of the output stage. Also the frontend is loaded the same, for the same output power. I'm not sure about controlling higher order harmonics either, unless there are real-life factors involved that I don't see in the simulator.
- keantoken
Last edited:
Member
Joined 2009
Paid Member
Kean, This worries me too, ClassAB output with no feedback to control the cross over distortion and essentially no damping factor either. This is why I've been looking at other options like error correction and variable bias class A.
Anyhow, all of this is off the path for being an AKSA.
Anyhow, all of this is off the path for being an AKSA.
This circuit also benifits from a B-C cap. Try it! The impedance of the transistor alone is multiplied by the resistor voltage divider, but if we bypass the top resistor we return to the impedance of a single transistor (or two in this case).
Gaetan, I've heard that inter-stage feedback sounds very good. However I'm not sure about better controlling EMF. Without taking feedback from the output, the output impedance becomes that of the emitter resistors and transistors of the output stage. Also the frontend is loaded the same, for the same output power. I'm not sure about controlling higher order harmonics either, unless there are real-life factors involved that I don't see in the simulator.
- keantoken
Hello keantoken
I found that simulating an amp with non-nfb at the output are not precise, and offset is hard to fix, a prototype need lot more work but give more realistic result.
Back EMF affect also the global negative feedback, even with a low output impedancew, so no nfb at the output do have some advantage.
B-C cap seem interesting compared to E-C cap, have try it on a prototype ?
Bye
Gaetan
Bye
Gaetan
One last thing about the Vbe multiplier business, this post might be of interest. It's not likely that Vbe multiplier will be unstable enough to affect amplifier operation but it's not bad to cover all your bases:
Vbe multiplier (in)stability ? mythbusting.
- keantoken
Vbe multiplier (in)stability ? mythbusting.
- keantoken
Greg says he'll experiment. Unfortunately I don't have the parts right now, and I wouldn't know what I'm listening to anyways! I've never even heard a tube amp, much less been to any concert of any kind.
Shame!
- keantoken
Hello keantoken
You should go to an Audio Show, those show have lot of high-end SS and tube amps, I think there is one each year in a big city in Texas.
Air-Thight and VTL are very excellent tube amps to listen.
Bye
Gaetan
Last edited:
Kean, This worries me too, ClassAB output with no feedback to control the cross over distortion and essentially no damping factor either. This is why I've been looking at other options like error correction and variable bias class A.
.............
Hi Bigun,
Out of curiosity I tried (read: simmend) just that: one FB loop from the VAS to the LTP and one error feedback loop around the OPS, as discussed here. So no global NFB. Results weren't that bad:
THD-20Hz: 2.7ppm
THD-1kHz: 3.5ppm
THD-20kHz: 47ppm
Zout @ 20kHz: 1.4mOhm
If you like to see the schematic I will put it in the other thread.
Cheers,
E.
Hi Bigun,
Out of curiosity I tried (read: simmend) just that: one FB loop from the VAS to the LTP and one error feedback loop around the OPS, as discussed here. So no global NFB. Results weren't that bad:
THD-20Hz: 2.7ppm
THD-1kHz: 3.5ppm
THD-20kHz: 47ppm
Zout @ 20kHz: 1.4mOhm
If you like to see the schematic I will put it in the other thread.
Cheers,
E.
Hello Edmond
Yes, and give us the link to that schematic.
Thank
Bye
Gaetan
Hi All,
I wasn't going to post this just yet, but seeing the topic has come up, I might as well. I have one of these amps and was interested with the feedback circuit enough to consider it a worthwhile experiment for the B-AKSA down the track.
http://www.overclockers.com.au/wiki/images/9/9c/KD-261_Schematic.png
regards
I wasn't going to post this just yet, but seeing the topic has come up, I might as well. I have one of these amps and was interested with the feedback circuit enough to consider it a worthwhile experiment for the B-AKSA down the track.
http://www.overclockers.com.au/wiki/images/9/9c/KD-261_Schematic.png
regards
Hi Greg,
I mistakenly wired up my KSA100Klone similarly.
I omitted the output to driver link and it worked using feedback from the drivers only.
When I noticed the mistake and put in the missing link, the amp needed more Cdom to make it stable on slightly capacitive loads and sounded terrible as a result.
Yes, it was more tolerant of output reactance with the feedback taken from the drivers only. I did not listen to that, it was still in the testing stage. Maybe I should go back and remove that link, recompensate and listen.
I mistakenly wired up my KSA100Klone similarly.
I omitted the output to driver link and it worked using feedback from the drivers only.
When I noticed the mistake and put in the missing link, the amp needed more Cdom to make it stable on slightly capacitive loads and sounded terrible as a result.
Yes, it was more tolerant of output reactance with the feedback taken from the drivers only. I did not listen to that, it was still in the testing stage. Maybe I should go back and remove that link, recompensate and listen.
Hello Edmond
Yes, and give us the link to that schematic.
Thank
Bye
Gaetan
Here it is: http://www.diyaudio.com/forums/soli...le-operating-biass-output-13.html#post2251493
Cheers,
E.
.... Edmond that is seriously lot of transistors
Anyway:
I have done some extensive study of the picture of the LF55. I think you can actually see so much as you could do a direct copy, from the picture. (Think I have more than 95% down in a LTSpice sim, and I will say it does really well(can´t see the components on the backside .... some more close-ups would held ) ).
I'll also take my words back about sophistication. It actually have some improvements in the input that you don´t see very often. The code word is Sziklai/CFP pair.
It is actually mentioned by Self in one of his books as a means to kill 3. order harmonics in the input stage ... but he obviously didn't think that much about odd or even harmonics, and opted for pure low overall distortion.
It actually does improve the stage alot (in sim at least ), so if you want to improve the baby, I defenetly think that is a good way to go.
If you take a close look at the resistor values chosen in the LF55, you will also not that it does use much lowe resistor values in the feedback circuit to improve noise.
Of course I might be wrong about every thing above, but then it is still a very good way to improve the baby AKSA, and easy to implement on your already made boards.
The LF55 sure looks to be a performer
I would like to hear some comments on the sound produced by the bootstrap ... Has naybody done a direct test of an amp where the only parameter change is the bootstrap vs. CCS??
From the explanation from Hugh, I would gess you also get a "softer" / less controlled bass, which will probably work miracles on smaller speakers ... don't get me wrong here ... to me less controlled don't have to mean bad .....
And from my experience the bass is actually one of the most important parts of creating room and space in the sound.
.... but here I'm purely speculating .... any experience you could share here?
Best regards Baldin
Anyway:
I have done some extensive study of the picture of the LF55. I think you can actually see so much as you could do a direct copy, from the picture. (Think I have more than 95% down in a LTSpice sim, and I will say it does really well(can´t see the components on the backside .... some more close-ups would held ) ).
I'll also take my words back about sophistication. It actually have some improvements in the input that you don´t see very often. The code word is Sziklai/CFP pair.
It is actually mentioned by Self in one of his books as a means to kill 3. order harmonics in the input stage ... but he obviously didn't think that much about odd or even harmonics, and opted for pure low overall distortion.
It actually does improve the stage alot (in sim at least ), so if you want to improve the baby, I defenetly think that is a good way to go.
If you take a close look at the resistor values chosen in the LF55, you will also not that it does use much lowe resistor values in the feedback circuit to improve noise.
Of course I might be wrong about every thing above, but then it is still a very good way to improve the baby AKSA, and easy to implement on your already made boards.
The LF55 sure looks to be a performer
I would like to hear some comments on the sound produced by the bootstrap ... Has naybody done a direct test of an amp where the only parameter change is the bootstrap vs. CCS??
From the explanation from Hugh, I would gess you also get a "softer" / less controlled bass, which will probably work miracles on smaller speakers ... don't get me wrong here ... to me less controlled don't have to mean bad .....
And from my experience the bass is actually one of the most important parts of creating room and space in the sound.
.... but here I'm purely speculating .... any experience you could share here?
Best regards Baldin
The CFP-LTP input stage, as I believe you are implying, is a very good option especially for low-NFB amps because it is very linear and unlikely to result in harmonic "upfolding". One of Bigun's amps uses this.
I found that for my speakers, any amount of series resistance resulted in a large improvement (I tested as low as 4 ohms). So low damping factor is no indication of sound quality as I see it. We consistently see this in reports of good sounding non-FB amps.
Also consider speaker impedance. Around 250Hz is the lowest impedance point usually WRT to frequency, and this really shows when using higher series resistors with my speakers. Too high series resistance cuts this area out, and it really shows. The sound gets lifeless. So this is an important area of the sound, in my experience. Also of note is that the amp is stressed most at this point, probably producing more 3rds than normal. At this frequency bootstrap capacitors are beginning to be stressed more, if you're using values around 470uF and lower (just a guess).
At high frequencies current through the speaker lags voltage, speaking generally, because of the voice coil inductance. Since it is current mostly that determines voice coil field strength (as I understand it) highs are getting phase shifted which should impact imaging. Increasing series resistance helps this too, because it shifts more towards current-drive which cancels the phase shift. It seems to me this would affect imaging, though I wonder if less than .22 ohms series resistance is enough.
Just my thoughts, I may be wrong.
- keantoken
I found that for my speakers, any amount of series resistance resulted in a large improvement (I tested as low as 4 ohms). So low damping factor is no indication of sound quality as I see it. We consistently see this in reports of good sounding non-FB amps.
Also consider speaker impedance. Around 250Hz is the lowest impedance point usually WRT to frequency, and this really shows when using higher series resistors with my speakers. Too high series resistance cuts this area out, and it really shows. The sound gets lifeless. So this is an important area of the sound, in my experience. Also of note is that the amp is stressed most at this point, probably producing more 3rds than normal. At this frequency bootstrap capacitors are beginning to be stressed more, if you're using values around 470uF and lower (just a guess).
At high frequencies current through the speaker lags voltage, speaking generally, because of the voice coil inductance. Since it is current mostly that determines voice coil field strength (as I understand it) highs are getting phase shifted which should impact imaging. Increasing series resistance helps this too, because it shifts more towards current-drive which cancels the phase shift. It seems to me this would affect imaging, though I wonder if less than .22 ohms series resistance is enough.
Just my thoughts, I may be wrong.
- keantoken
Hi Greg,
I mistakenly wired up my KSA100Klone similarly.
I omitted the output to driver link and it worked using feedback from the drivers only.
When I noticed the mistake and put in the missing link, the amp needed more Cdom to make it stable on slightly capacitive loads and sounded terrible as a result.
Yes, it was more tolerant of output reactance with the feedback taken from the drivers only. I did not listen to that, it was still in the testing stage. Maybe I should go back and remove that link, recompensate and listen.
Hi Andrew,
When I first saw the schematic I thought it was a drafting error but I have seen it in a couple of similar schematics now. It could be worthwhile, who knows.
regards
.... Edmond that is seriously lot of transistors
[snip]
Best regards Baldin
Hi Baldin,
Sure, lots of trannies. Performance has its price. However, most of the trannies cost just pennies.
In principle, it's possible to reduce the number of trannies from 50 to 27 by using a simple asymmetric topology, no separate drivers and PSU filtering of the front-end, etc.
BUT... heavily at the expense of the performance.
(btw, the PGP amp uses even more trannies)
Cheers,
E.
Hi Greg, just curious about a few things on the schematic (trying to learn) ...
- R12; Why 2 in series instead of a single 470R?
- What does PE stand for? Power Earth?
- The input cap seems a tad small. Where does the low freq. roll off? It might be worth experimenting with different values while listening to speakers with extended LF response.
- Which schematic package did you use to create the PDF you posted? (nice!)
Last edited:
- Home
- Amplifiers
- Solid State
- Based on Hugh Dean's AKSA 55