I have to apologize forum folks and Doctor Self,
I feel sorry..... really sorry, this is the third Blameless i have constructed, and seems that sounded fantastic.... i do not know what happened before...but i will discover soon and will tell you, as this interests me a lot.
The first time sounded awfull.... stérile, without dinamics, compressed...a battery portable radio.
And i made it once more when Colin re-started with that Blameless conversation.....and i was silent to publish once again that was awfull.
But this time, with the same speaker and almost the same parts (supply is different....hummm) i had a very good results.
Sounded very good, very precise, controled basses, nice and detailed non harshing trebles...nice sound stage.
I have removed the mirror and the compressive sounded disappeared and i have tweaked a little the VAS, increasing the IK resistance and reducing the capacitor to 27 picofarads..very stable it was, even without the capacitor...VAS used BC547 and BD139.
The CCS, that is substituting the colector loads into the differential had removed the emitter resistances.
This way sounded the way i like...so good that i may use Doctor Self VAS and this CCS....i have compared using resistances, and despite more compressive sonics, i perceived that details were more precise, more brigth and more focused.
So.... complicated solutions sometimes sounds fine.... Dx was wrong..... Carlos was wrong and saying sorry because his damn mouth.
Not confortable to come here to say that...but i think i have to...i must do that.
Yes, Andrew T.... complicated solutions, Doctor Self ideas can sound fine.
I could use the Blameless thread...i think i have opened that one....and others opened others about the same subject or small variation around the same theme.
Of course, moderators can join those threads if they feel this need...no problem to me.
The intention to open the new one is to show, clearly, without any doubts, that i have changed my opinion and i am saying sorry to every forum folks that have readed my test.
Sorry to you too Douglas.
regards,
Carlos
I feel sorry..... really sorry, this is the third Blameless i have constructed, and seems that sounded fantastic.... i do not know what happened before...but i will discover soon and will tell you, as this interests me a lot.
The first time sounded awfull.... stérile, without dinamics, compressed...a battery portable radio.
And i made it once more when Colin re-started with that Blameless conversation.....and i was silent to publish once again that was awfull.
But this time, with the same speaker and almost the same parts (supply is different....hummm) i had a very good results.
Sounded very good, very precise, controled basses, nice and detailed non harshing trebles...nice sound stage.
I have removed the mirror and the compressive sounded disappeared and i have tweaked a little the VAS, increasing the IK resistance and reducing the capacitor to 27 picofarads..very stable it was, even without the capacitor...VAS used BC547 and BD139.
The CCS, that is substituting the colector loads into the differential had removed the emitter resistances.
This way sounded the way i like...so good that i may use Doctor Self VAS and this CCS....i have compared using resistances, and despite more compressive sonics, i perceived that details were more precise, more brigth and more focused.
So.... complicated solutions sometimes sounds fine.... Dx was wrong..... Carlos was wrong and saying sorry because his damn mouth.
Not confortable to come here to say that...but i think i have to...i must do that.
Yes, Andrew T.... complicated solutions, Doctor Self ideas can sound fine.
I could use the Blameless thread...i think i have opened that one....and others opened others about the same subject or small variation around the same theme.
Of course, moderators can join those threads if they feel this need...no problem to me.
The intention to open the new one is to show, clearly, without any doubts, that i have changed my opinion and i am saying sorry to every forum folks that have readed my test.
Sorry to you too Douglas.
regards,
Carlos
Do you see it? I never mistrusted Dr. Self. I just pointed out that many of his schematics on the net are plainly wrong 
(not subtle errors, but gross, it takes little amp experience to debug them).
I'm designing a D.Self inspired amp (a sort of ) now, maybe I can post the schematic to gather suggestions
So we can make this the "Oh Self Amps and Self Style Sounds So Good" thead 
Btw, dear Carlos, that's very interesting about your amplifier. Well, if you read all over DIYAudio, you'll find many times that experienced amp builders get better sound with resistor loading of the input stage instead of the current mirror, and many others "improvements" (such as CCS, etc) are detrimental to the sonics (for their opinion, instead THD always increases - this confuses me, in my experience lowering THD over the whole audio band is related to good sound if properly done).
Hope the discussion will take a great start
(not subtle errors, but gross, it takes little amp experience to debug them).
I'm designing a D.Self inspired amp (a sort of
) now, maybe I can post the schematic to gather suggestions So we can make this the "Oh Self Amps and Self Style Sounds So Good" thead
Btw, dear Carlos, that's very interesting about your amplifier. Well, if you read all over DIYAudio, you'll find many times that experienced amp builders get better sound with resistor loading of the input stage instead of the current mirror, and many others "improvements" (such as CCS, etc) are detrimental to the sonics (for their opinion, instead THD always increases - this confuses me, in my experience lowering THD over the whole audio band is related to good sound if properly done).
Hope the discussion will take a great start
Yes.... not so important this load to the differential, i felt a little bit more
precise, more real, but sonics was not that warm i like.... i made it as a more precise, more modern unit, but was good to support that worry i have about those transistors as input differential load.
regards to you Lumanauw.
Carlos
......................................................................................................
Yes Giaime.... you told me...i remember that
One more turn the blameless.
I hope discussions will provide us some interesting ideas, some adjustments into the blameless.
Will be good for us.
regards,
Carlos
....................................................................................................
I do not intend to go feeding this thread... was more an apologise to a opening of another round of discussions...but of course, take your seat and tell whatever you want about this subject... Doctor Self is very important for us and will be a good idea to listen people talking nice things about him too... i will feel better with myself watching nice meetings talking about Self ideas...as i have bombed him a lot and it is time to correct those things.
regards,
Carlos
precise, more real, but sonics was not that warm i like.... i made it as a more precise, more modern unit, but was good to support that worry i have about those transistors as input differential load.
regards to you Lumanauw.
Carlos
......................................................................................................
Yes Giaime.... you told me...i remember that
One more turn the blameless.
I hope discussions will provide us some interesting ideas, some adjustments into the blameless.
Will be good for us.
regards,
Carlos
....................................................................................................
I do not intend to go feeding this thread... was more an apologise to a opening of another round of discussions...but of course, take your seat and tell whatever you want about this subject... Doctor Self is very important for us and will be a good idea to listen people talking nice things about him too... i will feel better with myself watching nice meetings talking about Self ideas...as i have bombed him a lot and it is time to correct those things.
regards,
Carlos
.......It is a very good amplifier
Lumanauw
I agree 100% with what you are saying.
Some years ago, there was quite a bit of information about current mirrors. The general consensus at the time,was that for best results, the current mirror transistors should be matched, and with as high an HFE as possible. An HFE of around 800 was even suggested.
A friend and myself have found these suggestions enhance performance, as does minimal loading of the current mirror by the VAS stage. MPSA18 transistors are often suitable for this task, due to their high HFE, and low cost. (provided their voltage limitations and polarity are taken into consideration.)
SandyK
Lumanauw
I agree 100% with what you are saying.
Some years ago, there was quite a bit of information about current mirrors. The general consensus at the time,was that for best results, the current mirror transistors should be matched, and with as high an HFE as possible. An HFE of around 800 was even suggested.
A friend and myself have found these suggestions enhance performance, as does minimal loading of the current mirror by the VAS stage. MPSA18 transistors are often suitable for this task, due to their high HFE, and low cost. (provided their voltage limitations and polarity are taken into consideration.)
SandyK
Hi GiaimeGiaime said:
IMHO the current mirror it is usefull in practice. Because it balances the collector currents of the difference amplifier transistor pair then the offset in output becames almost zero (about 5mV usually). Thus there is no need for trimming the difference amp. transistors with the use of trimmer. As for the sound quality, if it is improved, i don't know and i am not interested. I am sure in any case that it is not aggravated by a current mirror. Well why so many noise about?
Fotios
Hi, Carlos,
There's a little trick to make DSelf "blameless" sounds much better, but still a "blameless" design, not changing much of the basic concept.
DSelf himself actually mentioned this on his handbook, but didn't apply it in his "blameless" design.
The trick is to move the Cdom input point. Usually this Cdom is put between Collector of VAS to base of VAS (or B-enhancement transistor that preceeds this VAS transistor).
But to make this, you will have to cascode the input differential first.
Cdom cap has 2 legs. Keep one leg which is on the VAS collector, but move the other, originally in the VAS base, and move this capacitor leg to the input differential's collector (=emitor of cascode's).
Besides in DSelf's handbook, this method also can be read in Sackinger paper (JCX mentioned this several times).
There's a little trick to make DSelf "blameless" sounds much better, but still a "blameless" design, not changing much of the basic concept.
DSelf himself actually mentioned this on his handbook, but didn't apply it in his "blameless" design.
The trick is to move the Cdom input point. Usually this Cdom is put between Collector of VAS to base of VAS (or B-enhancement transistor that preceeds this VAS transistor).
But to make this, you will have to cascode the input differential first.
Cdom cap has 2 legs. Keep one leg which is on the VAS collector, but move the other, originally in the VAS base, and move this capacitor leg to the input differential's collector (=emitor of cascode's).
Besides in DSelf's handbook, this method also can be read in Sackinger paper (JCX mentioned this several times).
....it is a very good amplifier
Carlos
Douglas Self also recommends that when the input mirror is made from discrete transistors,it needs it's own emitter degrneration for good current matching (30-60mV voltage drop will be enough ....) If degeneration is omitted, there is significant variation in HF distortion performance with different specimens of the same transistor type.
Regards
SandyK
Carlos
Douglas Self also recommends that when the input mirror is made from discrete transistors,it needs it's own emitter degrneration for good current matching (30-60mV voltage drop will be enough ....) If degeneration is omitted, there is significant variation in HF distortion performance with different specimens of the same transistor type.
Regards
SandyK
...it is a very good amplifier
Hienrich
I posted a partial schematic and other suggestions in
Compensation capacitor Cdom
Sandyk
http://www.diyaudio.com/forums/showthread.php?s=&postid=1279758#post1279758
Hienrich
I posted a partial schematic and other suggestions in
Compensation capacitor Cdom
Sandyk
http://www.diyaudio.com/forums/showthread.php?s=&postid=1279758#post1279758
Re: ....it is a very good amplifier
Hello all!!
Hello SandyK, I was mainly concerned by the VceO of those devices. I've got 40V rails and a 45V device doesn't look promising (maybe on startup, they could be subjected to the full difference of the rails? I don't know). I'm using BC546B now.
Hello fotios,
I absolutely agree, on paper it's a great thing. I was asking myself if it gives too much instability problems, that's why (I thought) people says it's worse sounding. Or maybe they forgot to use high beta devices, and match them
Hello lumanauw,
I remember I read this in the book. But he also said (rightly) that by cascoding the differential amp there isn't great improvement, since Vce of the devices is pretty constant (since the stage gives a current output, not voltage). But that (cascoding) does increase PSRR. I may think about it, now I'm using CFP input pair devices (yes, the inputs are made of a Sziklai super-transistor).
Hello SandyK,
yes I've done that very very important. The emitter resistors increase the output resistance of the current mirror, also stabilizing it thermally. But I think that ALL the differential amplifier transistors should be very closely matched and put very near on the PCB, touching each other to minimize thermal differencies.
Hello all!!
sandyK said:
Hello SandyK, I was mainly concerned by the VceO of those devices. I've got 40V rails and a 45V device doesn't look promising (maybe on startup, they could be subjected to the full difference of the rails? I don't know). I'm using BC546B now.
fotios said:
Hello fotios,
I absolutely agree, on paper it's a great thing. I was asking myself if it gives too much instability problems, that's why (I thought) people says it's worse sounding. Or maybe they forgot to use high beta devices, and match them
lumanauw said:
Hello lumanauw,
I remember I read this in the book. But he also said (rightly) that by cascoding the differential amp there isn't great improvement, since Vce of the devices is pretty constant (since the stage gives a current output, not voltage). But that (cascoding) does increase PSRR. I may think about it, now I'm using CFP input pair devices (yes, the inputs are made of a Sziklai super-transistor).
sandyK said:
Hello SandyK,
yes I've done that
very very important. The emitter resistors increase the output resistance of the current mirror, also stabilizing it thermally. But I think that ALL the differential amplifier transistors should be very closely matched and put very near on the PCB, touching each other to minimize thermal differencies.
Hi, Giaime
You're right, cascoding the input differential makes small change in sound (infact, without cascoding, the % of ripple is smaller, because Vce is alot bigger). But it is needed to change the input point of the Cdom, from rail referenced point (current mirror referenced to rails) to ground referenced point (emitor of cascodes, to which the voltage divider for cascodes is referenced). Ground is relatively cleaner from audio artifacts compared to power rails. This makes alot of different in DSelf "blameless" design, because the gain in front end is very big, so attention to PSRR in these front end sections may change the perceived sonics alot.
You're right, cascoding the input differential makes small change in sound (infact, without cascoding, the % of ripple is smaller, because Vce is alot bigger). But it is needed to change the input point of the Cdom, from rail referenced point (current mirror referenced to rails) to ground referenced point (emitor of cascodes, to which the voltage divider for cascodes is referenced). Ground is relatively cleaner from audio artifacts compared to power rails. This makes alot of different in DSelf "blameless" design, because the gain in front end is very big, so attention to PSRR in these front end sections may change the perceived sonics alot.
it is a very good amplifier
Giaime
Re the ratings of the MPSA18, if you saw the partial schematic at the link I gave, you would see that the loaded side of the current mirror is only around 1.2V above the -VE rail due to the base of the emitter follower and it's associated emitter resistor . The other side of the current mirror is in essence a "super diode" in it's conducting direction, with current limited by the differential transistor. Unless I am missing something ?
I always thermally couple both the differential pair transistors to each other, and the current mirror transistors to each other. In fact, my own Class A amplifier uses a 2N3811A high spec. dual metal can transistor.The current mirror transistors have a small piece of tinplate wrapped around them, and then a small amount of epoxy over them .
SandyK
Giaime
Re the ratings of the MPSA18, if you saw the partial schematic at the link I gave, you would see that the loaded side of the current mirror is only around 1.2V above the -VE rail due to the base of the emitter follower and it's associated emitter resistor . The other side of the current mirror is in essence a "super diode" in it's conducting direction, with current limited by the differential transistor. Unless I am missing something ?
I always thermally couple both the differential pair transistors to each other, and the current mirror transistors to each other. In fact, my own Class A amplifier uses a 2N3811A high spec. dual metal can transistor.The current mirror transistors have a small piece of tinplate wrapped around them, and then a small amount of epoxy over them .
SandyK
it is a very good amplifier
Giaime and Lumanauw
You do do not need to go to all the complexity of a cascode in the differential, or CFP s. Just use a differential pair where the transistors are very well matched for HFE and VBE, as well as thermally coupled. All you need to do is load the unloaded side of the current mirror with the exact equivalent of the base current on the other side, then ensure that part of this "dummy load" is a diode, or diode connected transistor for thermal tracking purposes. When the voltage difference between the collectors of the differential pair is reduced to <5mV (!) you will notice a marked improvement in soundstage and an apparent lowering of the noise floor.If you also feed the front end with a low impedance , low noise power supply , you should notice on good source material, a 3 dimensional sound stage. Perhaps even holographic !
Even DTV audio can have an excellent soundstage.
Yes , even soapies such as "Days of Our Lives" etc.
There is extra information at the previously mentioned link.
SandyK
Giaime and Lumanauw
You do do not need to go to all the complexity of a cascode in the differential, or CFP s. Just use a differential pair where the transistors are very well matched for HFE and VBE, as well as thermally coupled. All you need to do is load the unloaded side of the current mirror with the exact equivalent of the base current on the other side, then ensure that part of this "dummy load" is a diode, or diode connected transistor for thermal tracking purposes. When the voltage difference between the collectors of the differential pair is reduced to <5mV (!) you will notice a marked improvement in soundstage and an apparent lowering of the noise floor.If you also feed the front end with a low impedance , low noise power supply , you should notice on good source material, a 3 dimensional sound stage. Perhaps even holographic !
Even DTV audio can have an excellent soundstage.
Yes , even soapies such as "Days of Our Lives" etc.
There is extra information at the previously mentioned link.
SandyK
Hello SandyK,
thank you very much. I tried MPSA18 for current mirror (in SPICE) and have little better results. I will try what you say about diffamp loading and equal voltages, but see, I have very close voltages. See the attached schematic:
Maybe a small R on the collector of the loaded side of the current mirror could help making voltages equal?
Also what voltages should I match, Vce of the "main" differential transistors (Q41 - Q42) or the absoulte voltage at the current mirror output?
Thank you very much in advance!
thank you very much. I tried MPSA18 for current mirror (in SPICE) and have little better results. I will try what you say about diffamp loading and equal voltages, but see, I have very close voltages. See the attached schematic:
Maybe a small R on the collector of the loaded side of the current mirror could help making voltages equal?
Also what voltages should I match, Vce of the "main" differential transistors (Q41 - Q42) or the absoulte voltage at the current mirror output?
Thank you very much in advance!
Attachments
it is a very good amplifier
Giaime
Looks like you are fairly close already. You have double trouble there !
I would say Q41 and Q42 initially.Try and get difference between their collectors closer by a resistor from the unloaded side of the differential pair collector to -VE rail, which gives equivalent to the loading of the BC547. (try something like 22Megohms to see if it makes an audible improvement, and take it from there) I would imagine that you already have a better than average sound stage if those pairs of transistors are well matched .
SandyK
Giaime
Looks like you are fairly close already. You have double trouble there !
I would say Q41 and Q42 initially.Try and get difference between their collectors closer by a resistor from the unloaded side of the differential pair collector to -VE rail, which gives equivalent to the loading of the BC547. (try something like 22Megohms to see if it makes an audible improvement, and take it from there) I would imagine that you already have a better than average sound stage if those pairs of transistors are well matched .
SandyK
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.