Tim de Paravichini answered the phone talking to me
In the first place i have to thank him since he actually spent over 5 min on the phone with me .
Then his opinion was that 680pf is not "that much " for an output stage and generally speaking he didn't actually listened much of what i said . He took our conversation as a judgement of his design while i made clear to him that based only in the fact the amp has a world wide unique output stage may the rest of the amp deserves a revision ..
His answer was that since there is no money for him in this he will pass unless i pay him for consulting ...
That is ok i would personally treat my children in a better way
So its up to our hands to find out if this can work without so much miller on it and if finally the all effort will make a better amplifier .
Kind regards
Sakis
In the first place i have to thank him since he actually spent over 5 min on the phone with me .
Then his opinion was that 680pf is not "that much " for an output stage and generally speaking he didn't actually listened much of what i said . He took our conversation as a judgement of his design while i made clear to him that based only in the fact the amp has a world wide unique output stage may the rest of the amp deserves a revision ..
His answer was that since there is no money for him in this he will pass unless i pay him for consulting ...
That is ok i would personally treat my children in a better way
So its up to our hands to find out if this can work without so much miller on it and if finally the all effort will make a better amplifier .
Kind regards
Sakis
Sakis,
It kind of makes it more apparent how grateful we should be to have guys like Nelson Pass, John Curl and the many others that have a financial interest with the industry yet still share a wealth of knowledge to help others better even their own designs from yesteryear. Typically I would say 47pf is approaching too much and and other aspects should be looked at in the open loop gain if you have to exceed this in a power amplifier. It can be done, with still enough negative feedback and phase margin.
Colin
It kind of makes it more apparent how grateful we should be to have guys like Nelson Pass, John Curl and the many others that have a financial interest with the industry yet still share a wealth of knowledge to help others better even their own designs from yesteryear. Typically I would say 47pf is approaching too much and and other aspects should be looked at in the open loop gain if you have to exceed this in a power amplifier. It can be done, with still enough negative feedback and phase margin.
Colin
I will Agree ...Even Tim's tone was absolutely fine with me ( i am trying to say that i meet far less important people in the industry that had far worst attitude)
As about Nelson Pass, John Curl, Hugh from Aksa and other less glamorous names that are around us helping and correcting our mistakes and generally share parts of their life we should be very grateful !
Near the forum you will find very willing and very fine people ...this is so nice Also i think it has a lot to do with the object we are working ...I think it brings close nice souls .
Even people that had nothing to do with diy ( like Bob Carver for example) I found my self very privileged to have a personal answer and long discussions with Bob about one idea and one issue i had some years ago ...Really nice guy ...
Wonderful community ...very proud to be a member ..
Kind regards
Sakis
As about Nelson Pass, John Curl, Hugh from Aksa and other less glamorous names that are around us helping and correcting our mistakes and generally share parts of their life we should be very grateful !
Near the forum you will find very willing and very fine people ...this is so nice Also i think it has a lot to do with the object we are working ...I think it brings close nice souls .
Even people that had nothing to do with diy ( like Bob Carver for example) I found my self very privileged to have a personal answer and long discussions with Bob about one idea and one issue i had some years ago ...Really nice guy ...
Wonderful community ...very proud to be a member ..
Kind regards
Sakis
@Colin
At this point i will not argue between 680- 220-or 47 pf To me the first step will be to find out where this is coming from ...solve it before miller caps is used at the very end
Lets wait a while to see if valery manages to trace it by altering the "simulator thermaly and betaly and hfealy perfect drivers "
I am designing a pcb for that to be ready to make real life tests but it will take a while to get there ....
Kind regards
sakis
At this point i will not argue between 680- 220-or 47 pf To me the first step will be to find out where this is coming from ...solve it before miller caps is used at the very end
Lets wait a while to see if valery manages to trace it by altering the "simulator thermaly and betaly and hfealy perfect drivers "
I am designing a pcb for that to be ready to make real life tests but it will take a while to get there ....
Kind regards
sakis
Experiments with tolerance
Well, played with hfe tolerance. I used the second option (2-nd attachment to post #9, as it runs quite well). Most of the results are pretty much as expected.
Let say, with average (or even lower than average), but matched hfes, with 1KHz, 15V RMS @ 8R output we've got 0.01% THD mostly consisting of 2-nd and 3-rd harmonics (roughly equal levels).
Dis-balancing the drivers or OPs between the shoulders for 50% increases the 2-nd harmonic 2.5-3 times.
Dis-balancing between the first and second pair (but matched between the shoulders) provides less damage - 2-nd harmonic increases 1.5-2 times.
Same effect, but also with 3-rd harmonic increase takes place if I drop hfe for all the OPs significantly (same 50% for example).
Interesting, that if there is significant dis-balance of any kind in power section, VAS hfe decrease raises 3-rd harmonic (and 2-nd a little bit), making things worth. However, if power section is matched, same VAS hfe decrease results in slight 2-nd harmonic drop (making the whole circuit balance better), so overall THD slightly improves. In other words, it could seem that the higher hfe is the better for VAS, however it appears that there is an optimum value, which is not the highest one.
In any case, the amp remained stable, no signs of oscillation whatsoever (with 47 pF caps used for the drivers).
It runs smoothly even without those caps at all, if I also remove the lead compensation cap (15pF parallel with GNFB resistor). If I put this cap back with still no caps at the drivers - the amp oscillates at around 7.5MHz with small amplitude. So, to be on a safe side, I would leave those 47pF at the drivers - at this stage such a small value is not going to damage sonic qualities, but ensures stability.
That's it so far. I will try to run more experiments, if you have ideas on some setups to be checked - let me know
Thermal model would be interesting to play with, however I have to look deeper in the parameters - never tried it before, will take some time.
Cheers,
Valery
Well, played with hfe tolerance. I used the second option (2-nd attachment to post #9, as it runs quite well). Most of the results are pretty much as expected.
Let say, with average (or even lower than average), but matched hfes, with 1KHz, 15V RMS @ 8R output we've got 0.01% THD mostly consisting of 2-nd and 3-rd harmonics (roughly equal levels).
Dis-balancing the drivers or OPs between the shoulders for 50% increases the 2-nd harmonic 2.5-3 times.
Dis-balancing between the first and second pair (but matched between the shoulders) provides less damage - 2-nd harmonic increases 1.5-2 times.
Same effect, but also with 3-rd harmonic increase takes place if I drop hfe for all the OPs significantly (same 50% for example).
Interesting, that if there is significant dis-balance of any kind in power section, VAS hfe decrease raises 3-rd harmonic (and 2-nd a little bit), making things worth. However, if power section is matched, same VAS hfe decrease results in slight 2-nd harmonic drop (making the whole circuit balance better), so overall THD slightly improves. In other words, it could seem that the higher hfe is the better for VAS, however it appears that there is an optimum value, which is not the highest one.
In any case, the amp remained stable, no signs of oscillation whatsoever (with 47 pF caps used for the drivers).
It runs smoothly even without those caps at all, if I also remove the lead compensation cap (15pF parallel with GNFB resistor). If I put this cap back with still no caps at the drivers - the amp oscillates at around 7.5MHz with small amplitude. So, to be on a safe side, I would leave those 47pF at the drivers - at this stage such a small value is not going to damage sonic qualities, but ensures stability.
That's it so far. I will try to run more experiments, if you have ideas on some setups to be checked - let me know
Thermal model would be interesting to play with, however I have to look deeper in the parameters - never tried it before, will take some time.
Cheers,
Valery
Sakis,
Interesting comment about Tim de P. I am aware he is an eccentric, clever fellow, and since most designers live in their heads he sees the world only through his experiences.
I think a really good designer - of anything - tends to mistrust anyone's ideas, display a little hubris, yet the paradox is that they learn their craft from lots of other people, like JLH and Arthur Bailey.
Perhaps new ideas come from people who are contemptuous of the conventional ideas? You understand the political system well, I'm sure you know this is philosophy!!
Great work from Valery, huh? BTW, I've found a bit of H2 is no bad thing, long as it is lower than -75dB it's fine.
Cheers,
Hugh
Interesting comment about Tim de P. I am aware he is an eccentric, clever fellow, and since most designers live in their heads he sees the world only through his experiences.
I think a really good designer - of anything - tends to mistrust anyone's ideas, display a little hubris, yet the paradox is that they learn their craft from lots of other people, like JLH and Arthur Bailey.
Perhaps new ideas come from people who are contemptuous of the conventional ideas? You understand the political system well, I'm sure you know this is philosophy!!
Great work from Valery, huh? BTW, I've found a bit of H2 is no bad thing, long as it is lower than -75dB it's fine.
Cheers,
Hugh
Hugh ... you know that History is part of my interest I always like to find out details about circuits and designs that i like . Truly i think that the circuit belongs to Glen ( the other part of Alchemist ) I think Tim added his touch after the original circuit of the amplifier was created by Glen
I think that if this amplifier was made some other way more commercial, and less orderly in his internal construction, with smaller supply and so on it could be a fail .
Sziklai circuit has the known problems , but after all a musical sound , better aid by well designed pcb power supply , and structure ...I think that the latest was Tim's contribution on the product .
Glen's whereabouts are still a mystery .
Beyond historical data is good to see you Hugh ... It is always a privilege to listen to your comments on my thoughts
Kind regards
Sakis
I think that if this amplifier was made some other way more commercial, and less orderly in his internal construction, with smaller supply and so on it could be a fail .
Sziklai circuit has the known problems , but after all a musical sound , better aid by well designed pcb power supply , and structure ...I think that the latest was Tim's contribution on the product .
Glen's whereabouts are still a mystery .
Beyond historical data is good to see you Hugh ... It is always a privilege to listen to your comments on my thoughts
Kind regards
Sakis
Here is a circuit test for THD by harmonic:
Fundamental 1W 1KHz into 8 ohms resistive:
2 -72.00
3 -75dB
4 -87dB
5 -94dB
6 Immeasurable
7 Immeasurable
THD % by voltage 0.03116
You could say this is a high distortion level, a poor THD.
BUT, it is slightly warm, and extremely musical, almost like a very good tube amp.
Phase shift is less than 3 degrees at 20KHz.
I find that THD is an aggregate of all the artefacts, but when you discount H2, H3 and H4, which are all found in nature, the best indications of sonic purity are speed, and H5 and beyond. Another indication is the ratio of H2 to H5, H5 should be at least 10 times lower (20dB) than H2. Too much H2 is bad, since it usually occurs with H3 and H4, and there is obvious loss of resolution with complex signals such massed choirs. Significantly, SET amps do not do their best work with choral works.
Cheers,
Hugh
Fundamental 1W 1KHz into 8 ohms resistive:
2 -72.00
3 -75dB
4 -87dB
5 -94dB
6 Immeasurable
7 Immeasurable
THD % by voltage 0.03116
You could say this is a high distortion level, a poor THD.
BUT, it is slightly warm, and extremely musical, almost like a very good tube amp.
Phase shift is less than 3 degrees at 20KHz.
I find that THD is an aggregate of all the artefacts, but when you discount H2, H3 and H4, which are all found in nature, the best indications of sonic purity are speed, and H5 and beyond. Another indication is the ratio of H2 to H5, H5 should be at least 10 times lower (20dB) than H2. Too much H2 is bad, since it usually occurs with H3 and H4, and there is obvious loss of resolution with complex signals such massed choirs. Significantly, SET amps do not do their best work with choral works.
Cheers,
Hugh
Last edited:
A few friends bought these when they were also sold complete on small but useful heatinks, ~48 USD for a pair. They are well built and most components appear originals. The drivers may well be fake. I noted P grade Sankens O/P transistors though if you see Y grade, (a mixed gain category) I suspect it means something is wrong.Prices on e-bay make me a bit suspicious
I think the value is still many times that of other kits from Ebay on sound quality alone and the construction is first rate, even if the layout is questionable. - See the linked thread about performance, value and comments from Calvin and ATAudio.
Thank you Ian - ordered a pair of assembled boards with heatsinks from one of the recommended suppliers (exactly $48 usd as you say). They will come in about 6 weeks (yes, I'm living behind the fence
) We'll see. Anyway, I can always replace the transistors with genuine ones, no problem
Having fun people !!!!
Another Alchemist come in for a visit in our shop with all shorts of problems blown outputs and so on
I performed as described above a full set including a new set of faster outputs capacitors blown resistors and so on
Who was the person who said that this amplifier may work with out miller caps of 680pf ....NO WAY HOSE !!!!
From experience i know that if you use better quality caps like mika you can actually fool the circuit by a few pf less ...so i used 500pf 500v silver mika ....oscillation oscillation oscillation oscillation oscillation and nothing else
So once more real life results will vary from simulator ...by far in this case
Kind regards
Sakis
Another Alchemist come in for a visit in our shop with all shorts of problems blown outputs and so on
I performed as described above a full set including a new set of faster outputs capacitors blown resistors and so on
Who was the person who said that this amplifier may work with out miller caps of 680pf ....NO WAY HOSE !!!!
From experience i know that if you use better quality caps like mika you can actually fool the circuit by a few pf less ...so i used 500pf 500v silver mika ....oscillation oscillation oscillation oscillation oscillation and nothing else
So once more real life results will vary from simulator ...by far in this case
Kind regards
Sakis
Hi Sakis,
Interesting experience
Confirming there is a design issue in this circuit.By the way, my simulation (post #9) showed exactly what you are saying - with original active devices it oscillated even with 680pf caps in place (left picture).
It only worked stable with 47pf caps at the drivers with better transistors used (right picture).
Cheers,
Valery
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.