Well, here is a major revamp.
I have been thinking about all comments, both on this thread and in private e-mail. Also, after a lot of reading, the following revamp has resulted along with the associated questions.
I hope you all are not dismayed by all my questions. But I am learning and hopefully so are the silent majority that read this thread.
I have redrawn the design for a PNP DA and stripped it to the basics so that the schematic is more familiar and less cluttered. I have also implemented Constant Current Sources. As usual please check my work.
The parts you see floating and not connected are options and will be addressed below.
The 475R on the VAS CCS is a contribution from the Pass A40 article. He said it will improve negative clipping recovery. Is this desireable here?
The 10R stopper on the bases of the outputs is contribution from the Leach Low TIM amp. It is a belt and suspenders option to supress pasitic oscillations. Is this desireable here?
The CCS is a contribution from the Pass A40. I used a resistor to feed the diodes because the fet option requires selection and is a real pain to implement. The spec sheet for 1N4148 says that at 2ma the diode will drop about 0.65V. I am not sure how to deal with the two CCS transistor base currents. Can someone please check to be sure the 16K2 resistor is the right value to keep 2ma flowing through the diodes?
Operating points are a concern for me. To that issue:
As drawn (without the driver emitter resistor option) the VAS will be biased at about 0.68V. Is this sufficient to keep the VAS in Class A up to clipping? What should the Bias be to keep the VAS in Class A to clipping?
If the VAS emitter resistor (49R9) option is used, will the bias requirement for the VAS change? In other words will the DA collector resistor (500R) need to be doubled to keep the VAS in Class A? Or can the VAS emitter resistor be experimented with (say from 60R to 0R) without changing the DA collector resistor?
The VAS in the amp that started this thread (Digi125) is a TO92 device. It has been reported to be somwhat fragile so in this design the VAS is a TO220 device. Should there be any concern about TO92 device life for the CCS that feeds the VAS? How about the Bias circuit?
For those who suggested floating the driver emitter resistors, is the option what you had in mind? It will be very easy to incorporate into the PCB.
Is are the resistor and pot in the bias circuit optimal for the current state of the circuit? Any discussion?
And last question. Now that CCSs are implimented, are there better values to use for current into the DA and VAS?
I surely hop all the questions I have posed will be answered. Silence is truely deafening.
Well, here it is:
I have been thinking about all comments, both on this thread and in private e-mail. Also, after a lot of reading, the following revamp has resulted along with the associated questions.
I hope you all are not dismayed by all my questions. But I am learning and hopefully so are the silent majority that read this thread.
I have redrawn the design for a PNP DA and stripped it to the basics so that the schematic is more familiar and less cluttered. I have also implemented Constant Current Sources. As usual please check my work.
The parts you see floating and not connected are options and will be addressed below.
The 475R on the VAS CCS is a contribution from the Pass A40 article. He said it will improve negative clipping recovery. Is this desireable here?
The 10R stopper on the bases of the outputs is contribution from the Leach Low TIM amp. It is a belt and suspenders option to supress pasitic oscillations. Is this desireable here?
The CCS is a contribution from the Pass A40. I used a resistor to feed the diodes because the fet option requires selection and is a real pain to implement. The spec sheet for 1N4148 says that at 2ma the diode will drop about 0.65V. I am not sure how to deal with the two CCS transistor base currents. Can someone please check to be sure the 16K2 resistor is the right value to keep 2ma flowing through the diodes?
Operating points are a concern for me. To that issue:
As drawn (without the driver emitter resistor option) the VAS will be biased at about 0.68V. Is this sufficient to keep the VAS in Class A up to clipping? What should the Bias be to keep the VAS in Class A to clipping?
If the VAS emitter resistor (49R9) option is used, will the bias requirement for the VAS change? In other words will the DA collector resistor (500R) need to be doubled to keep the VAS in Class A? Or can the VAS emitter resistor be experimented with (say from 60R to 0R) without changing the DA collector resistor?
The VAS in the amp that started this thread (Digi125) is a TO92 device. It has been reported to be somwhat fragile so in this design the VAS is a TO220 device. Should there be any concern about TO92 device life for the CCS that feeds the VAS? How about the Bias circuit?
For those who suggested floating the driver emitter resistors, is the option what you had in mind? It will be very easy to incorporate into the PCB.
Is are the resistor and pot in the bias circuit optimal for the current state of the circuit? Any discussion?
And last question. Now that CCSs are implimented, are there better values to use for current into the DA and VAS?
I surely hop all the questions I have posed will be answered. Silence is truely deafening.
Well, here it is:
Attachments
Revision Version looks good
maybe one resistor from 390pF cap to base of input trans.
some say 470 to 2200R
this is like gate resistor in MOS, put very near the base/gate
You have now higher current in input pair.
maybe 200R need to be lower,
the drop there is now 0.272 volt
for 0.5mA and 220= 0.110 volt
for 0.3mA and 200= 0.060 volt
I am almost sure this amp will work well.
But there might be something, the other guys have seen,
that I am missing.
This is some reading - Many have read it!
Distortion In Power Amplifiers - an investigation by Douglas Self
Same Site:
The Amplifier Institute
This can be complement to what other amplifier designers have found.
We all have heard namnes like Elliot, Pass and Linsley-Hood.
We can learn something new from anyone with some experience.
And that experience we once got, we hopefully can pass along
to newborn generations, when these olden goldies' lives are terminated.
maybe one resistor from 390pF cap to base of input trans.
some say 470 to 2200R
this is like gate resistor in MOS, put very near the base/gate
You have now higher current in input pair.
maybe 200R need to be lower,
the drop there is now 0.272 volt
for 0.5mA and 220= 0.110 volt
for 0.3mA and 200= 0.060 volt
I am almost sure this amp will work well.
But there might be something, the other guys have seen,
that I am missing.
This is some reading - Many have read it!
Distortion In Power Amplifiers - an investigation by Douglas SelfSame Site:
The Amplifier Institute
This can be complement to what other amplifier designers have found.
We all have heard namnes like Elliot, Pass and Linsley-Hood.
We can learn something new from anyone with some experience.
And that experience we once got, we hopefully can pass along
to newborn generations, when these olden goldies' lives are terminated.
Aborza,
You ask good questions and so answering them is no burden.
CCSs look ok to me.
I don't understand the benefit of the 10 ohm base resistors on the power trannies. If you are sure you do then include them otherwise I'd drop 'em. They reduce the transconductance of the darlingtons and slow them down a little.
The VAS will not clip in the way you think. The loop feedback will adjust the current through the 500 ohm resistor to keep the VAS conducting. If the 500 ohm resistor is the wrong value you'll see this as a voltage imbalance between the + and - inputs. The 500 ohm should be adjusted until the input offset is balanced. The value will need to be higher if you use the 49R9 emitter resistor in the VAS. These two need to change at the same time.
In general you should use the lowest power transistors you can get away with in each part of the circuit. The bigger the transistor the worse many of it's properties tend to become. So for the VAS I'd use the pnp equivalent of a BC546. The Miller cap is a little high at 150pF (especially as you are reducing the loop gain at the LTP with the 200R resistors). 50pF to 100pF to start with.
BD139/140s are pretty ancient transistors. Have you access to the newer Motorola types (eg: MJE15030/31) for the drivers?
The bias circuit values look ok, although I'd make the bypass cap >10uF.
10mA is plenty of bias current for the VAS output. 1 or 2mA for the input is fine.
You ask good questions and so answering them is no burden.
CCSs look ok to me.
I don't understand the benefit of the 10 ohm base resistors on the power trannies. If you are sure you do then include them otherwise I'd drop 'em. They reduce the transconductance of the darlingtons and slow them down a little.
The VAS will not clip in the way you think. The loop feedback will adjust the current through the 500 ohm resistor to keep the VAS conducting. If the 500 ohm resistor is the wrong value you'll see this as a voltage imbalance between the + and - inputs. The 500 ohm should be adjusted until the input offset is balanced. The value will need to be higher if you use the 49R9 emitter resistor in the VAS. These two need to change at the same time.
In general you should use the lowest power transistors you can get away with in each part of the circuit. The bigger the transistor the worse many of it's properties tend to become. So for the VAS I'd use the pnp equivalent of a BC546. The Miller cap is a little high at 150pF (especially as you are reducing the loop gain at the LTP with the 200R resistors). 50pF to 100pF to start with.
BD139/140s are pretty ancient transistors. Have you access to the newer Motorola types (eg: MJE15030/31) for the drivers?
The bias circuit values look ok, although I'd make the bypass cap >10uF.
10mA is plenty of bias current for the VAS output. 1 or 2mA for the input is fine.
The revised circuit with CCS for both the diff pair and the VAS can be built on the $6 Pass A40 board http://www.audioxpress.com/bksprods/pcbs/nelsonpass.htm
The output stage is simple enough for point-to point wiring.
I have built amplifiers as large as 400W with the A40 board.
Those that like the cross coupled dual diff inputs (GAS, SAE, Leach, etc) should look at the $12 EB60 Borberly board
http://www.audioxpress.com/bksprods/pcbs/solidstate.htm
The output stage is simple enough for point-to point wiring.
I have built amplifiers as large as 400W with the A40 board.
Those that like the cross coupled dual diff inputs (GAS, SAE, Leach, etc) should look at the $12 EB60 Borberly board
http://www.audioxpress.com/bksprods/pcbs/solidstate.htm
PNP vs. NPN input pairs
What I have guessed
the use of PNP input pair is due to
what transistors was/is available.
If you use PNP input you have NPN in the VAS-stage
NPN had/has often higher bandwith, Ft, MHZ
If PNP had/has more often lower noise,
I do not know.
I guess these days you can always find good transistors
of both PNP and NPN techniques.
The sometimes used thinking, that symmetrical amplifiers
will give symmetrical output, can be wrong.
To me this is a kinda odd association.
To find transistors with perfect match,
is a lot more difficult using PNP/NPN
then if you use NPN/NPN of same technique and denomination.
What I have guessed
the use of PNP input pair is due to
what transistors was/is available.
If you use PNP input you have NPN in the VAS-stage
NPN had/has often higher bandwith, Ft, MHZ
If PNP had/has more often lower noise,
I do not know.
I guess these days you can always find good transistors
of both PNP and NPN techniques.
The sometimes used thinking, that symmetrical amplifiers
will give symmetrical output, can be wrong.
To me this is a kinda odd association.
To find transistors with perfect match,
is a lot more difficult using PNP/NPN
then if you use NPN/NPN of same technique and denomination.
Re: Revision Version looks good
Thanks for this link, Halojoy. It was some great reading about amplifier theory. As a Leach Amp proponent, it validates nearly everything that was done with that design, though Leach went a step further (IMHO) to make his design fully symmetric.
I highly recommend this link to anyone interested in amplfier theory.
Jeff R.
This is some reading - Many have read it!
Same Site:
The Amplifier Institute
Thanks for this link, Halojoy. It was some great reading about amplifier theory. As a Leach Amp proponent, it validates nearly everything that was done with that design, though Leach went a step further (IMHO) to make his design fully symmetric.
I highly recommend this link to anyone interested in amplfier theory.
Jeff R.
Modern Transistors
Traderban and halojoy have both raised the issue of transistors.
The most recent iteration of the design (you will see it shortly) will deliver about 4ma through a CCS to the DA and about 7ma through a CCS to the VAS. The Drivers and Output will be relatively unchanged except the Output collector resistors will be reduced to about 0R1 for better AB linearity.
So the questions become, what are the current readily available transistors (with really good linear response in the operating regions I am using) that the posters reccomend for:
1. TO92 transistors for the CCSs, DA, VAS and Bias circuits?
2. TO220 transistors for the Drivers?
3. Flatpack and TO3 transistors for the Outputs.
I live in the US so I can purchase anyting sold by DigiKey, Newark, MCM, Futurelec, Mouser, etc.
BTW I chose the transistors shown on the schematics because I understood them to be very good for audio use, with nice linearity for the purpose. And I understand the outputs are modern sturdy devices with exceptionally flat audio response.
Traderban and halojoy have both raised the issue of transistors.
The most recent iteration of the design (you will see it shortly
) will deliver about 4ma through a CCS to the DA and about 7ma through a CCS to the VAS. The Drivers and Output will be relatively unchanged except the Output collector resistors will be reduced to about 0R1 for better AB linearity.So the questions become, what are the current readily available transistors (with really good linear response in the operating regions I am using) that the posters reccomend for:
1. TO92 transistors for the CCSs, DA, VAS and Bias circuits?
2. TO220 transistors for the Drivers?
3. Flatpack and TO3 transistors for the Outputs.
I live in the US so I can purchase anyting sold by DigiKey, Newark, MCM, Futurelec, Mouser, etc.
BTW I chose the transistors shown on the schematics because I understood them to be very good for audio use, with nice linearity for the purpose. And I understand the outputs are modern sturdy devices with exceptionally flat audio response.
Re: Modern Transistors
They have the big advantage of being readily available.
That is a quality I put very HIGH.
Without that, the other qualities are more or less useless,
if you know how I think.
-----------------------------------------
I have only one DATA-book. As I have to do some economical priority I have so far only bought one: tdv1.
I would also like to have C..to Z. It covers among others IRFxxxx.
As I live in Europe I bought tdv1.
It covers most BCxxx and BDxxx, and also have groups that match those like 2Nxxxx and MJxxx and 2Sxxx.
For example there is one group with most of the low noise TO92:s.
Each A...BUZ transistor have some data of their performance,
regarding noise, speed and what they can take, volt current power.
My edition is the one from 1993. Latest is 1998/99.
I have almost read it into pieces. I have to be careful not to get the lose pages all out over the floor.
I have stated before that it is important to "know your components" (see that post)
If you have knowledge about how a special component behaves
in situations, it can be better, than to use components
that are newer and better, which you have no experience of.
----------------------------------------------
tdv volume 1 EUR:30,17
Series tdv: Transistors under control
Data Directories and Equivalents Tables giving the maximum and characteristic ratings of transistors and field-effect transistors along with their pin assignments. These books also including the dimensioned outline drawings and selector tables for quickly identifying equivalent types.
tdv 1 (A...BUZ)
Multilingual:
German, English, French, Spanish, Italian
6nd updated edition 1998/1999
416 pages, DIN A 4
ISBN 3-88109-057-6
--------------------------------------------------------------
There are tdv1 to tdv4. Covering most of transistors in the world.
You can also get them on CD-Rom.
See more info at this site:
ECA Electronic GmbH - OnlineShop
There is absolutely nothing wrong with your choice of transistors.aborza said:
They have the big advantage of being readily available.
That is a quality I put very HIGH.
Without that, the other qualities are more or less useless,
if you know how I think.
-----------------------------------------
I have only one DATA-book. As I have to do some economical priority I have so far only bought one: tdv1.
I would also like to have C..to Z. It covers among others IRFxxxx.
As I live in Europe I bought tdv1.
It covers most BCxxx and BDxxx, and also have groups that match those like 2Nxxxx and MJxxx and 2Sxxx.
For example there is one group with most of the low noise TO92:s.
Each A...BUZ transistor have some data of their performance,
regarding noise, speed and what they can take, volt current power.
My edition is the one from 1993. Latest is 1998/99.
I have almost read it into pieces. I have to be careful not to get the lose pages all out over the floor.
I have stated before that it is important to "know your components" (see that post)
If you have knowledge about how a special component behaves
in situations, it can be better, than to use components
that are newer and better, which you have no experience of.
----------------------------------------------
tdv volume 1 EUR:30,17
Series tdv: Transistors under control
Data Directories and Equivalents Tables giving the maximum and characteristic ratings of transistors and field-effect transistors along with their pin assignments. These books also including the dimensioned outline drawings and selector tables for quickly identifying equivalent types.
tdv 1 (A...BUZ)
Multilingual:
German, English, French, Spanish, Italian
6nd updated edition 1998/1999
416 pages, DIN A 4
ISBN 3-88109-057-6
--------------------------------------------------------------
There are tdv1 to tdv4. Covering most of transistors in the world.
You can also get them on CD-Rom.
See more info at this site:
ECA Electronic GmbH - OnlineShop
Aborza,
Your amplifier is looking good.
Traderbam gives his opinion on boostraps and distortion.
Now the dust is clearing, let me give mine
Bootstraps are just fine for two reasons.
First, they allow the positive signal drive voltage (when supported from the positive rail) to swing well beyond the rail.
This means positive clipping is solely determined by the saturation voltage of the output stage and the voltage drop across the emitter resistors of the output stage.
Second, and this is subtle, bootraps use an electrolytic cap to couple the output to the midpoint of the VAS feed resistor. Since the impedance of an electro starts to skyrocket at high frequencies, it begins to lose effectiveness, and this means the collector of the VAS sees a steadily dropping impedance as frequency rises. This has a significant effect.
As collector loading increases, gain drops, and with it, open loop gain. Thus, as the phase shift increases with frequency, eventually causing negative feedback to become positive and make a lovely oscillator, the bootstrap pulls back OLG, which helps the amp reduce it's overall gain to below unity before the pole frequency. This is necessary to avoid oscillation, and normally this effect is created by lag compensation across the VAS. The bootstrap means that you need less lag compensation to pull the amp within the Bode-Nyquist criteria, and this in turn improves the sonics since the slew rate is increased.
Distortion is a many splendored thing. It is much maligned, but in truth, I believe the recording process does terrible things to music, steadily stripping away the higher harmonics. Add a little low order distortion and the music sounds fuller, more natural. Perhaps this explains something of the joy of the SET!!
A little H2, and even a little H3, is no bad thing. I mean something around 0.1%, not exceeding 0.5%. The slavish adherence to a 'NO DISTORTION' regime throws the baby out with the bathwater, and this is NOT what audiophiles like. The engineers like low distortion, and yet I don't know a tube amp yet which delivers less than 0.5% at full tilt.
I believe your amp will sound better if you use 100R base stoppers on the drivers, and 10R base stoppers on the outputs. And I believe Nelson is right when he suggests a 1mA stage current through the diff stage.
I also believe that the amp will image far better if the diff pair and the VAS are fed from the same node. This means resistive feeds, NOT a current source.
Cheers,
Hugh
www.aksaonline.com
Your amplifier is looking good.
Traderbam gives his opinion on boostraps and distortion.
Now the dust is clearing, let me give mine
Bootstraps are just fine for two reasons.
First, they allow the positive signal drive voltage (when supported from the positive rail) to swing well beyond the rail.
This means positive clipping is solely determined by the saturation voltage of the output stage and the voltage drop across the emitter resistors of the output stage.
Second, and this is subtle, bootraps use an electrolytic cap to couple the output to the midpoint of the VAS feed resistor. Since the impedance of an electro starts to skyrocket at high frequencies, it begins to lose effectiveness, and this means the collector of the VAS sees a steadily dropping impedance as frequency rises. This has a significant effect.
As collector loading increases, gain drops, and with it, open loop gain. Thus, as the phase shift increases with frequency, eventually causing negative feedback to become positive and make a lovely oscillator, the bootstrap pulls back OLG, which helps the amp reduce it's overall gain to below unity before the pole frequency. This is necessary to avoid oscillation, and normally this effect is created by lag compensation across the VAS. The bootstrap means that you need less lag compensation to pull the amp within the Bode-Nyquist criteria, and this in turn improves the sonics since the slew rate is increased.
Distortion is a many splendored thing. It is much maligned, but in truth, I believe the recording process does terrible things to music, steadily stripping away the higher harmonics. Add a little low order distortion and the music sounds fuller, more natural. Perhaps this explains something of the joy of the SET!!
A little H2, and even a little H3, is no bad thing. I mean something around 0.1%, not exceeding 0.5%. The slavish adherence to a 'NO DISTORTION' regime throws the baby out with the bathwater, and this is NOT what audiophiles like. The engineers like low distortion, and yet I don't know a tube amp yet which delivers less than 0.5% at full tilt.
I believe your amp will sound better if you use 100R base stoppers on the drivers, and 10R base stoppers on the outputs. And I believe Nelson is right when he suggests a 1mA stage current through the diff stage.
I also believe that the amp will image far better if the diff pair and the VAS are fed from the same node. This means resistive feeds, NOT a current source.
Cheers,
Hugh
www.aksaonline.com
Hugh writes:
"First, they allow the positive signal drive voltage (when supported from the positive rail) to swing well beyond the rail."
I agree.
"...and this means the collector of the VAS sees a steadily dropping impedance as frequency rises."
It does anyway due to the output stage and the miller capacitance.
"The bootstrap means that you need less lag compensation to pull the amp within the Bode-Nyquist criteria, and this in turn improves the sonics since the slew rate is increased."
The bootstrap may help a little bit. What do you measure the impedance of your bootstrap cap to be, say, at 1MHz? I am skeptical of the significance of this effect and I feel it is less important than the negative impact of the bootstrap at lower frequencies. Have you any data on the phase response impact?
"The slavish adherence to a 'NO DISTORTION' regime throws the baby out with the bathwater, and this is NOT what audiophiles like."
I am an audiophile and I don't like distortion. I understand what you are saying about deliberate introduction of some harmonics, especially even harmonics, that make something imperfect sound a little better. But I don't agree that most recorded media requires more distortion to sound more real. I also think that one is on a slipperly slope if one is not committed to reproducing the source material as accurately as possible. I think it is an easy trap to fall into to start thinking the source is at fault rather than the amplifier but in my experience it is invariably the case that the CD/amplifier is causing the problems that the designer then "compensates" for by introducing more distortion. That is my experience.
I should say that I think the THD measure is pants and is too crude a way to judge real distortion. So this is why an 0.001% THD amp can sound worse than a 0.1% THD amp. I definitely agree that slavish pursuit of THD figures will end in tears.
"I believe your amp will sound better if you use 100R base stoppers on the drivers, and 10R base stoppers on the outputs. And I believe Nelson is right when he suggests a 1mA stage current through the diff stage. I also believe that the amp will image far better if the diff pair and the VAS are fed from the same node. This means resistive feeds, NOT a current source."
Why?
Cheers,
BAM
"First, they allow the positive signal drive voltage (when supported from the positive rail) to swing well beyond the rail."
I agree.
"...and this means the collector of the VAS sees a steadily dropping impedance as frequency rises."
It does anyway due to the output stage and the miller capacitance.
"The bootstrap means that you need less lag compensation to pull the amp within the Bode-Nyquist criteria, and this in turn improves the sonics since the slew rate is increased."
The bootstrap may help a little bit. What do you measure the impedance of your bootstrap cap to be, say, at 1MHz? I am skeptical of the significance of this effect and I feel it is less important than the negative impact of the bootstrap at lower frequencies. Have you any data on the phase response impact?
"The slavish adherence to a 'NO DISTORTION' regime throws the baby out with the bathwater, and this is NOT what audiophiles like."
I am an audiophile and I don't like distortion. I understand what you are saying about deliberate introduction of some harmonics, especially even harmonics, that make something imperfect sound a little better. But I don't agree that most recorded media requires more distortion to sound more real. I also think that one is on a slipperly slope if one is not committed to reproducing the source material as accurately as possible. I think it is an easy trap to fall into to start thinking the source is at fault rather than the amplifier but in my experience it is invariably the case that the CD/amplifier is causing the problems that the designer then "compensates" for by introducing more distortion. That is my experience.
I should say that I think the THD measure is pants and is too crude a way to judge real distortion. So this is why an 0.001% THD amp can sound worse than a 0.1% THD amp. I definitely agree that slavish pursuit of THD figures will end in tears.
"I believe your amp will sound better if you use 100R base stoppers on the drivers, and 10R base stoppers on the outputs. And I believe Nelson is right when he suggests a 1mA stage current through the diff stage. I also believe that the amp will image far better if the diff pair and the VAS are fed from the same node. This means resistive feeds, NOT a current source."
Why?
Cheers,
BAM
More bootstrapping even.
How about if you also bootstrapped the supplies to the driver transistor collectors too? Say 100 ohms (or less?) from each collector to rail, then a big electrolytic from output to each driver collector. Watch the polarities! The drivers would have a nearly constant CE voltage all the time even though the signal is going up and down.This should be good for linearity and distortion I think, but I have never tried it.
How about if you also bootstrapped the supplies to the driver transistor collectors too? Say 100 ohms (or less?) from each collector to rail, then a big electrolytic from output to each driver collector. Watch the polarities! The drivers would have a nearly constant CE voltage all the time even though the signal is going up and down.This should be good for linearity and distortion I think, but I have never tried it.
Have you tried it yet? huh? huh? huh?
I probably saw the idea somewhere and then fell asleep, and when I woke up I felt like I thought of it myself. It should be quite easy to retrofit to an existing amp. I haven't got any suitable ones to try it on at the moment. I'm just dying to hear if it works ok.
I probably saw the idea somewhere and then fell asleep, and when I woke up I felt like I thought of it myself.
It should be quite easy to retrofit to an existing amp. I haven't got any suitable ones to try it on at the moment. I'm just dying to hear if it works ok.
Bam,
I am not sure.
You want the answer to everything; this is not always possible.
I'm not actually trying to give the answers here, because they sometimes go outside the known engineering, I suspect because of the psycho-acoustic interface. But I'm offering information which I know works, because I've built them, listened for hundreds of hours, come up with amps which work well, and know what does not work.
The market demands a good sounding amp, and does not much care why. I'm obliged to deliver the goods, for whatever reason. I'm interested in why, certainly, but can't spend too much time musing the philosopher's stone.
I'm actually quite surprised you are a Pom; your use of the language is blunt, rather unrefined.
Cheers,
Hugh
I am not sure.
You want the answer to everything; this is not always possible.
I'm not actually trying to give the answers here, because they sometimes go outside the known engineering, I suspect because of the psycho-acoustic interface. But I'm offering information which I know works, because I've built them, listened for hundreds of hours, come up with amps which work well, and know what does not work.
The market demands a good sounding amp, and does not much care why. I'm obliged to deliver the goods, for whatever reason. I'm interested in why, certainly, but can't spend too much time musing the philosopher's stone.
I'm actually quite surprised you are a Pom; your use of the language is blunt, rather unrefined.
Cheers,
Hugh
Well, we can't all be poets like you, Hugh.
How about concentrating on the audio discussion? Design critique is not meant to be taken personally. I know you are trying to maintain a brand reputation but you can't have everything in a public forum: you get exposure and you also get challenging questions.
How about concentrating on the audio discussion? Design critique is not meant to be taken personally. I know you are trying to maintain a brand reputation but you can't have everything in a public forum: you get exposure and you also get challenging questions.
Bam,
Poet, huh? I hate poetry.........!
Thank you for your post, and balanced comment.
First, I'll be happy to talk about what I've found works, but I'm only half prepared to say why. There are two reasons; one, I'm sometimes uncertain why, but the other is more to do with IP protection. There is no effective protection of IP in this world unless you are a large corporation; already there are some interesting developments in that area I need to address. If I explain the effect, and then the cause, it's all there for anyone to use, and I need to keep a few secrets because I've put in a lot of time and effort to find them. There is also the matter of time. I can only allocate half and hour a day to this forum, since I'm pretty busy in my business.
You ask for impedance measurements on a typical boostrap. This information may be calculated from manufacturer data, and yes, the impedance reduces at HF for a spectrum of reasons. But I won't divulge my detailed work on this stuff here. I normally work from measured frequency response, operating currents, impedances, etc - but I use my ears and those of others to assess the final product. To take this approach further and minutely examine the reasons why is time consuming, and not directly productive in my position.
OTOH, I'm delighted to describe the benefits and the topology, which is enough for most interested people.
My suggestion, Bam, is to build the thing, take it on faith, and do the measurements yourself. Then, you might be able to examine the phenomenon carefully and tell us!
I believe Circlotron's idea of using a boostrap on the drivers is excellent, since it will definitely remove a further source of distortion which, however, is capably compensated by the global nfb loop. However, it would have to be a rather large cap; if the feed resistors to the driver were kept to say 47R apiece, you'd need several thousand uF to maintain voltage across the collector/emitter.
Cheers,
Hugh
www.aksaonline.com
Poet, huh? I hate poetry.........!
Thank you for your post, and balanced comment.
First, I'll be happy to talk about what I've found works, but I'm only half prepared to say why. There are two reasons; one, I'm sometimes uncertain why, but the other is more to do with IP protection. There is no effective protection of IP in this world unless you are a large corporation; already there are some interesting developments in that area I need to address. If I explain the effect, and then the cause, it's all there for anyone to use, and I need to keep a few secrets because I've put in a lot of time and effort to find them. There is also the matter of time. I can only allocate half and hour a day to this forum, since I'm pretty busy in my business.
You ask for impedance measurements on a typical boostrap. This information may be calculated from manufacturer data, and yes, the impedance reduces at HF for a spectrum of reasons. But I won't divulge my detailed work on this stuff here. I normally work from measured frequency response, operating currents, impedances, etc - but I use my ears and those of others to assess the final product. To take this approach further and minutely examine the reasons why is time consuming, and not directly productive in my position.
OTOH, I'm delighted to describe the benefits and the topology, which is enough for most interested people.
My suggestion, Bam, is to build the thing, take it on faith, and do the measurements yourself. Then, you might be able to examine the phenomenon carefully and tell us!
I believe Circlotron's idea of using a boostrap on the drivers is excellent, since it will definitely remove a further source of distortion which, however, is capably compensated by the global nfb loop. However, it would have to be a rather large cap; if the feed resistors to the driver were kept to say 47R apiece, you'd need several thousand uF to maintain voltage across the collector/emitter.
Cheers,
Hugh
www.aksaonline.com
"but the other is more to do with IP protection. "
This is fair enough and I feel the same way. I am also not prepared to give away everything I've struggled to learn in a trice. But I hope myself and others may be excused for asking probing questions and I don't always expect a complete answer.
"My suggestion, Bam, is to build the thing"
Fair enough.
BAM
This is fair enough and I feel the same way. I am also not prepared to give away everything I've struggled to learn in a trice. But I hope myself and others may be excused for asking probing questions and I don't always expect a complete answer.
"My suggestion, Bam, is to build the thing"
Fair enough.
BAM
Re: Have you tried it yet? huh? huh? huh?
It does have some similarities to Nelson Pass' patent 5,343,166 where a pair of output stage fets are fed from source followers who's gates are fed from the output signal. i.e. the supply to the output fet drains is actively bootstrapped from the output signal.Circlotron said:I probably saw the idea somewhere and then fell asleep, and when I woke up I felt like I thought of it myself.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.