Hi Lumanauw,
Yes, it is the Sakinger's idea I thought of, I even have his paper.
Giaime 11 kOhm, 180 pF low-pass negative feedback circuit :
this enhances the closed loop gain above 80 kHz and the distorsion as well as there is less and less negative feedback. But there is still some feedback above 80 Khz until the open loop gain has been entirely exhausted. I think the Linsley-Hood 's 75W amplifier (HiFi News 1973) had something similar.
Yes, it is the Sakinger's idea I thought of, I even have his paper.
Giaime 11 kOhm, 180 pF low-pass negative feedback circuit :
this enhances the closed loop gain above 80 kHz and the distorsion as well as there is less and less negative feedback. But there is still some feedback above 80 Khz until the open loop gain has been entirely exhausted. I think the Linsley-Hood 's 75W amplifier (HiFi News 1973) had something similar.
Hello all!!! 
Hello forr,
thank you very much for your suggestions. About C12, you're right, but the improvement in PSRR is extremely small. But this enables me to use a lower voltage cap, that's always good ( = bigger cap).
Only with real testing I can determine if the small capacitor you indicate is necessary...
The resistor on the base of Q30 was my mistake! Thank you very much, the power efficiency of the amp is great improved, keeping the same rails I got 20W more on 8ohm (95W instead of 75W). The output node swings to 100mV from the rails, without the resistor I was scratching my head, I couldn't see where the asimmetrical clipping happenend. I put 470ohm in there.
About R65. Self's suggestion is very good, if you connect the capacitor to the supply rail (the same of the output bjts). However, with the inclusion of C17 and R67, the PSRR increase doesn't happen...
Hello lumanauw,
look at the Leach amp. It's almost the same compensation scheme:
http://users.ece.gatech.edu/~mleach/lowtim/fdbk.html
In pratice you can, doing this way, wrap the output stage with gNFB only in the audio band, where is needed. Putting gNFB in a stage with an uknown output load (maybe an electrostatic speaker?), big transistors (big capacitances), over the range of frequencies of interest, is not necessary. Prof. Leach explains better than me in the article
Updated schematic:
forr said:
Hello forr,
thank you very much for your suggestions. About C12, you're right, but the improvement in PSRR is extremely small. But this enables me to use a lower voltage cap, that's always good ( = bigger cap).
Only with real testing I can determine if the small capacitor you indicate is necessary...
The resistor on the base of Q30 was my mistake! Thank you very much, the power efficiency of the amp is great improved, keeping the same rails I got 20W more on 8ohm (95W instead of 75W). The output node swings to 100mV from the rails, without the resistor I was scratching my head, I couldn't see where the asimmetrical clipping happenend. I put 470ohm in there.
About R65. Self's suggestion is very good, if you connect the capacitor to the supply rail (the same of the output bjts). However, with the inclusion of C17 and R67, the PSRR increase doesn't happen...
lumanauw said:
Hello lumanauw,
look at the Leach amp. It's almost the same compensation scheme:
http://users.ece.gatech.edu/~mleach/lowtim/fdbk.html
In pratice you can, doing this way, wrap the output stage with gNFB only in the audio band, where is needed. Putting gNFB in a stage with an uknown output load (maybe an electrostatic speaker?), big transistors (big capacitances), over the range of frequencies of interest, is not necessary. Prof. Leach explains better than me in the article
Updated schematic:
it is a very good amplifier
Other than Fotios, have any of you taken these results and
ideas out from the simulator and text books, and used them in REAL amplifiers? I know for example,that CFP can be a very hard beast to tame in a real amplifier. A friend of mine uses a CFP VAS, and other compensation methods. He had trouble with VHF instability, and had to resort to using ferrite beads etc to tame the amplifier.You also need a very wideband CRO to faultfind these problems e.g. a 100MHZ CRO.
SandyK
Other than Fotios, have any of you taken these results and
ideas out from the simulator and text books, and used them in REAL amplifiers? I know for example,that CFP can be a very hard beast to tame in a real amplifier. A friend of mine uses a CFP VAS, and other compensation methods. He had trouble with VHF instability, and had to resort to using ferrite beads etc to tame the amplifier.You also need a very wideband CRO to faultfind these problems e.g. a 100MHZ CRO.
SandyK
Re: it is a very good amplifier
From that you wrote about the CFP, i remembered one event before 25 years when i was in position only for copying circuits. I had bought a power amplifier of 100Wrms/8Ohms for selling only purpose and i had asked from the agent the plans of it for servicing purposes. Because the amplifier sounded nicelly, i decided to make a copy for personal use. In the plans the amplifier presented with one pair of transistors in his output (MJ15003-MJ15004) and in the device indeed thus they was. The arrangement of output presented in the plan was CFP (then - in 1985 - i still did not know the EF and the CFP and much more but i was verry good in coppies). After the completion of the coppy, and during the audition tests with a pair of proffesional speakers of 250W each, the output transistors where burned or shorthed continuously! To my big luck, i had copied the speaker protection circuit rightly! After this, i examined carefully the original amplifier and to my big surprisse i discovered that the plans was wrong! Because in reality the output of the original model was EF type!
Then i changed simply the routings in PCB, i placed a new pair of transistors in output and all went well then. Today i can see all the previews under a different view. I remember when i had read for first time the D.Self book before 6 years, when i saw the part with the CFP outputs, woke up in my nightmares of past!
Fotios
Hi SandyK and allsandyK said:
From that you wrote about the CFP, i remembered one event before 25 years when i was in position only for copying circuits. I had bought a power amplifier of 100Wrms/8Ohms for selling only purpose and i had asked from the agent the plans of it for servicing purposes. Because the amplifier sounded nicelly, i decided to make a copy for personal use. In the plans the amplifier presented with one pair of transistors in his output (MJ15003-MJ15004) and in the device indeed thus they was. The arrangement of output presented in the plan was CFP (then - in 1985 - i still did not know the EF and the CFP and much more but i was verry good in coppies). After the completion of the coppy, and during the audition tests with a pair of proffesional speakers of 250W each, the output transistors where burned or shorthed continuously! To my big luck, i had copied the speaker protection circuit rightly! After this, i examined carefully the original amplifier and to my big surprisse i discovered that the plans was wrong! Because in reality the output of the original model was EF type!
Then i changed simply the routings in PCB, i placed a new pair of transistors in output and all went well then. Today i can see all the previews under a different view. I remember when i had read for first time the D.Self book before 6 years, when i saw the part with the CFP outputs, woke up in my nightmares of past!
Fotios
Just to assure folks that it is possible, I have built a couple of amps with CFP (CFB?) output configurations. It did take three trys on the layout to get them to work although one iteration was just to correct a dumb positioning of the output inductor. Output devices were 2SA1943/2SC5200, drivers were 2DS669/2SB649. The choice of drivers seems to be important; with MJE340/350 there were oscillations. I did not need to slap on caps to get it to work.
However, this is for a single set of output devices. I could not get parallel devices be stable.
I've been told but can't confirm that "slower" output devices, older MJ15xxxx are easier to handle.
However, this is for a single set of output devices. I could not get parallel devices be stable.
I've been told but can't confirm that "slower" output devices, older MJ15xxxx are easier to handle.
CFP
A friend of mine uses CFP output with 2 pairs of Sanken 50MHZ output devices , AND a CFP VAS. As mentioned earlier, he had to sort out instability problems. It obviously makes a big difference if you are able to construct around a working design with a proven layout.Our Class A and Class AB designs were based around Silicon Chip CFP designs. I wonder if Silicon Chip magazine, as official resellers of Douglas Self books, had access to further layout information etc. from the man, himself ? I do know that several of their designs had a long developement period.
Incidentally, AKSA has heard reports of the 100W/Ch. Class AB amplifier previously mentioned. Those extremely good results were BEFORE the original design's front end regulated supplies were implemented. It has an even higher performance now.
DiyAudio's "vhfman" deserves a lot of credit for taking the performance of this design to such a very high level.
With these type of Class AB designs, DiYAudio member awpagan , has been able to extract even higher performance from the original SC 100W ULD, by incorporating a John Linsley Hood power supply addon between the front end regulated supply and the front end. He spent some considerable time making the addon PSU work with + and - 55V rails, which was way higher than the original design published in ETI many years ago.
SandyK
A friend of mine uses CFP output with 2 pairs of Sanken 50MHZ output devices , AND a CFP VAS. As mentioned earlier, he had to sort out instability problems. It obviously makes a big difference if you are able to construct around a working design with a proven layout.Our Class A and Class AB designs were based around Silicon Chip CFP designs. I wonder if Silicon Chip magazine, as official resellers of Douglas Self books, had access to further layout information etc. from the man, himself ? I do know that several of their designs had a long developement period.
Incidentally, AKSA has heard reports of the 100W/Ch. Class AB amplifier previously mentioned. Those extremely good results were BEFORE the original design's front end regulated supplies were implemented. It has an even higher performance now.
DiyAudio's "vhfman" deserves a lot of credit for taking the performance of this design to such a very high level.
With these type of Class AB designs, DiYAudio member awpagan , has been able to extract even higher performance from the original SC 100W ULD, by incorporating a John Linsley Hood power supply addon between the front end regulated supply and the front end. He spent some considerable time making the addon PSU work with + and - 55V rails, which was way higher than the original design published in ETI many years ago.
SandyK
Giaime, fratello, i will try your input and VAS in the future.
I will be waiting people construct some samples of my HR II...than, around December, i will try your input (only the input and VAS).... i hope no copyrigth, as i have intentions to use your ideas if they proved to sound fine.
regards,
Carlos
I will be waiting people construct some samples of my HR II...than, around December, i will try your input (only the input and VAS).... i hope no copyrigth, as i have intentions to use your ideas if they proved to sound fine.
regards,
Carlos
Re: Giaime, fratello, i will try your input and VAS in the future.
Hello friends!
Thank you Carlos! I almost finished my preamplifier, and will post the schematic when it's tested. After that, I plan on breadboarding this amplifier, and we'll see how it works. No copyright problems for Carlos, my dear friend. Just mention who is the original author
Hi Greg,
I think R55, the 22ohm resistor in the Vbe multiplier, is in the right place, as Self's diagrams. It's there to compensate the increase of VbeMult current with rail voltage increase. When the VbeMult current increases, the bias voltage will go up, but with the addition of R55 an increase of VbeMult current will give a decrease of bias due the voltage drop on R55, thus balancing the two actions. I never tried approfondite tests on that, I just put the suggested value. See figure 12.27b in the third edition of his book.
An up to date schematic:
Hello friends!
destroyer X said:
Thank you Carlos! I almost finished my preamplifier, and will post the schematic when it's tested. After that, I plan on breadboarding this amplifier, and we'll see how it works. No copyright problems for Carlos, my dear friend. Just mention who is the original author
Greg Erskine said:
Hi Greg,
I think R55, the 22ohm resistor in the Vbe multiplier, is in the right place, as Self's diagrams. It's there to compensate the increase of VbeMult current with rail voltage increase. When the VbeMult current increases, the bias voltage will go up, but with the addition of R55 an increase of VbeMult current will give a decrease of bias due the voltage drop on R55, thus balancing the two actions. I never tried approfondite tests on that, I just put the suggested value. See figure 12.27b in the third edition of his book.
An up to date schematic:
R55 is in the wrong location.
It should be in the horizontal feed to the top of the multipliying transistor, not in the vertical feed to the resistor divider string.
R55=22r is the compensation value for a multiplier current of about 5.5mA, when using Self's transistor type. More common approximate values are
18r for 6.5mA,
16r for 7mA,
15r for 7.5mA,
12r for 9mA,
10r for 10mA.
Remember to adjust the resistor string current if you change the string values to find the total quiescent current through the VAS and CCS.
If the multiplying transistor is changed to another type then it's likely that these resistors will need some adjustment.
This simple constant voltage circuit is very simple to breadboard and test for your chosen transistor/resistors.
It should be in the horizontal feed to the top of the multipliying transistor, not in the vertical feed to the resistor divider string.
R55=22r is the compensation value for a multiplier current of about 5.5mA, when using Self's transistor type. More common approximate values are
18r for 6.5mA,
16r for 7mA,
15r for 7.5mA,
12r for 9mA,
10r for 10mA.
Remember to adjust the resistor string current if you change the string values to find the total quiescent current through the VAS and CCS.
If the multiplying transistor is changed to another type then it's likely that these resistors will need some adjustment.
This simple constant voltage circuit is very simple to breadboard and test for your chosen transistor/resistors.
Mr. Andrew T...friends use to tell me you use to criticise me.
Well man... was already published, and in a very clear form, that i do not like you.
And also is clear that you also do not like me.
This is something that is up to ourselves..our problems...not forum problems.
To avoid to read your texts, that seems to me text books repeated, no practice, not diy schematics or constructions, and above all those things, you invested yourself with the charge to be the forum adviser.... and no one invited you to do this job...invitations were clear and posted as permanent pages...so...you are figthing to have some place under the sun....I hope you realise that without enter everythread to explains some calculations or theories...this is only a part of this Diy forum knowledge, but even beeing a part..you are posting everywhere, even trying to be under the spot arguing with our big brains..
I decided to let you be free, just avoiding your posts.... you are in the ignore list...but friends tells me you insist to provocate, to criticise.
Well man...this is a forum...our personal problems should be decided man to man in the "field of honor"..ahahahahah...but too much distant to realise our happy meeting.
I will ask you , to let me in peace, the same way i use to let you in peace.... you are provocating me...this is clear and people run to tell me that, because i do not use to read your posts...so..provocations to not reach me.
I will ask moderators, to suggest you, not to provocate.
I hope i will not need anymore to discuss this subject with you.
Be happy..... my preference.... very distant from me.
About Copyrigth, to you, as i am sure Giaime has this understanding.... schematic published, into an open forum, and International forum, normally have no copyrigth.
Knowledge, now a days, are reaching the entire world...no one can be the owner, as the schematic discoveries are beeing published and transmitted into the WEB.... no one is more"original", as we are a product, a result, or our environment, our intellectual chain that has the Media to inform everybody about everything.
Also, your book knowledge is something that do not belongs to you.... that knowledge existed, in big ammount, in advance related the day you have born.... no copyrigth to books contents..we can read, learn and use.... our obligation is to respect the ones had intelectual property.... we need only to say they have made that, developed that or created that.
And also, those things suffer modifications, upgrades, updates, some adaptation or corrections, loosing part of the characteristics that will connect some diagram to some owner.
Patent is very clear...if you change a little...not more inside those rules.
I hope i will not need to discuss with you.
regards,
Carlos
Well man... was already published, and in a very clear form, that i do not like you.
And also is clear that you also do not like me.
This is something that is up to ourselves..our problems...not forum problems.
To avoid to read your texts, that seems to me text books repeated, no practice, not diy schematics or constructions, and above all those things, you invested yourself with the charge to be the forum adviser.... and no one invited you to do this job...invitations were clear and posted as permanent pages...so...you are figthing to have some place under the sun....I hope you realise that without enter everythread to explains some calculations or theories...this is only a part of this Diy forum knowledge, but even beeing a part..you are posting everywhere, even trying to be under the spot arguing with our big brains..
I decided to let you be free, just avoiding your posts.... you are in the ignore list...but friends tells me you insist to provocate, to criticise.
Well man...this is a forum...our personal problems should be decided man to man in the "field of honor"..ahahahahah...but too much distant to realise our happy meeting.
I will ask you , to let me in peace, the same way i use to let you in peace.... you are provocating me...this is clear and people run to tell me that, because i do not use to read your posts...so..provocations to not reach me.
I will ask moderators, to suggest you, not to provocate.
I hope i will not need anymore to discuss this subject with you.
Be happy..... my preference.... very distant from me.
About Copyrigth, to you, as i am sure Giaime has this understanding.... schematic published, into an open forum, and International forum, normally have no copyrigth.
Knowledge, now a days, are reaching the entire world...no one can be the owner, as the schematic discoveries are beeing published and transmitted into the WEB.... no one is more"original", as we are a product, a result, or our environment, our intellectual chain that has the Media to inform everybody about everything.
Also, your book knowledge is something that do not belongs to you.... that knowledge existed, in big ammount, in advance related the day you have born.... no copyrigth to books contents..we can read, learn and use.... our obligation is to respect the ones had intelectual property.... we need only to say they have made that, developed that or created that.
And also, those things suffer modifications, upgrades, updates, some adaptation or corrections, loosing part of the characteristics that will connect some diagram to some owner.
Patent is very clear...if you change a little...not more inside those rules.
I hope i will not need to discuss with you.
regards,
Carlos
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.