Hi All,
I'm new in here, but found a lot of seemingly competent brains chatting along... maybe one of you is able to give some inspitration?
I have made a voltage gain stage, with a single output (voltage), but unable to deliver a lot of current. I now need that to drive an - if possible- simple, non-feedback class A/B power stage with approx. */- 40V rails supply.
I always tend to have my output stage more complicaterd than I like it, although I know that some bias is needed along with thermal stability.
Does anyone have a good idea for a fairly simple output stage with unity gain? (I guess it's basically I power-capable voltage follower...)
Best Regards,
Jens "Jennice" Sylvest (Jens was taken
as a user name)
I'm new in here, but found a lot of seemingly competent brains chatting along... maybe one of you is able to give some inspitration?
I have made a voltage gain stage, with a single output (voltage), but unable to deliver a lot of current. I now need that to drive an - if possible- simple, non-feedback class A/B power stage with approx. */- 40V rails supply.
I always tend to have my output stage more complicaterd than I like it, although I know that some bias is needed along with thermal stability.
Does anyone have a good idea for a fairly simple output stage with unity gain? (I guess it's basically I power-capable voltage follower...)
Best Regards,
Jens "Jennice" Sylvest (Jens was taken
as a user name)How about a Class A?
www.sound.au.com project 83
or
http://www.audiodesignguide.com/my/index.html#MY_FOLL
www.sound.au.com project 83
or
http://www.audiodesignguide.com/my/index.html#MY_FOLL
Hi Paulb,
Thanks for your reply ;-)
First, I forgot to post it in solid state section (just noticed that)
Secondly, I have two criteria which I forgot to mention:
1)
The output should be with BJT's
2)
It sould (hey, didn't I mention that? *S*) be a class A/B design, not requiring the massive bias power...
I know I'm picky...
Regards,
Jens
Thanks for your reply ;-)
First, I forgot to post it in solid state section (just noticed that)
Secondly, I have two criteria which I forgot to mention:
1)
The output should be with BJT's
2)
It sould (hey, didn't I mention that? *S*) be a class A/B design, not requiring the massive bias power...
I know I'm picky...
Regards,
Jens
I think the best choice would be the Threshold S300 output stage.
http://www.diyvideo.com/forums/attachment.php?postid=38107
No global feedback, and you may bias it any way you want. A simplified version of this design was also licensed to Nakamichi for their receivers a while back.
I have heard both the Threshold S150 and the Nakamichi SR4A, and they are both very good.
I would use the MJE15030/31 as drivers and the MJ21193/4 as outputs. This will have enough current gain to run directly off your tube gain stage.
If you keep the bias low enough you could use only one pair of outputs on ±40V, more would allow you to increase the bias.
I can recommend a DC servo design if you need help with that.
http://www.diyvideo.com/forums/attachment.php?postid=38107
No global feedback, and you may bias it any way you want. A simplified version of this design was also licensed to Nakamichi for their receivers a while back.
I have heard both the Threshold S150 and the Nakamichi SR4A, and they are both very good.
I would use the MJE15030/31 as drivers and the MJ21193/4 as outputs. This will have enough current gain to run directly off your tube gain stage.
If you keep the bias low enough you could use only one pair of outputs on ±40V, more would allow you to increase the bias.
I can recommend a DC servo design if you need help with that.
djk said:
Thank you djk for the schematic. Interesting to see. I don't know if I'll copy it directly, but it has certainly given inspiration
Looks as if the driver stage is designed for a considerable ammount of bias?
Hmm... is the circuit subject to offset problems, since you comment on the DC servo?
Thanks -
Jens
"Looks as if the driver stage is designed for a considerable ammount of bias?"
Pass ran it in class A in the Threshold gear. Nakamichi ran it in a low bias AB mode.
"Hmm... is the circuit subject to offset problems, since you comment on the DC servo?"
Not really, all you need is a bias circuit and the driver and output stages. Two constant current sources, one to the positive rail, and one to the negative rail will provide the bias. Make one of these so you can trim the DC off-set to zero and you should be fine.
Hawk Audio has a 50W FET hybrid that has a real nice DC servo that adjusts both constant current sources at the same time, but they took down that page from the www.
The End Millenium from LC Audio is a no feedback amp that has a nice DC servo design too, you can download the schema on their site as a pdf.
Pass ran it in class A in the Threshold gear. Nakamichi ran it in a low bias AB mode.
"Hmm... is the circuit subject to offset problems, since you comment on the DC servo?"
Not really, all you need is a bias circuit and the driver and output stages. Two constant current sources, one to the positive rail, and one to the negative rail will provide the bias. Make one of these so you can trim the DC off-set to zero and you should be fine.
Hawk Audio has a 50W FET hybrid that has a real nice DC servo that adjusts both constant current sources at the same time, but they took down that page from the www.
The End Millenium from LC Audio is a no feedback amp that has a nice DC servo design too, you can download the schema on their site as a pdf.
Hi Jens,
Speaking broad brush, here's my suggestions:
1. Run your voltage amplifier at around 8mA. This gives it reasonable drive. Use a 250MHz device as both voltage amp and current source.
2. Use a conventional Class AB output stage, bipolar, with fast devices, no less than 30MHz on the outputs and at least 50MHz on the drivers. Both should have working gains of around 60.
3. Use the Double emitter follower Type II from Doug Self.
4. With a combined current gain of 3,600, a healthy output current of say 6 amps will require a current injection into the drivers of 1.66mA. This is just under 5 times the stage current of the voltage amplifier, and quite acceptable notwithstanding the difficult crossover area where load alteration is variable and non-linear. However, at crossover, the load current is tiny, so this change in input impedance of the output stage is manageable.
5. Concentrate on amp stability, using the smallest lag comp cap possible. Start at 68pF and work down.
6. Try to eliminate current mirrors where possible. They guarantee diff pair imbalance, and seem to damage the sonics.
This is all conventional topology, but you'd be surprised how the sound can be tailored by scrupulous care with the device choices and working environment of the VAS and OS.
Cheers,
Hugh
Speaking broad brush, here's my suggestions:
1. Run your voltage amplifier at around 8mA. This gives it reasonable drive. Use a 250MHz device as both voltage amp and current source.
2. Use a conventional Class AB output stage, bipolar, with fast devices, no less than 30MHz on the outputs and at least 50MHz on the drivers. Both should have working gains of around 60.
3. Use the Double emitter follower Type II from Doug Self.
4. With a combined current gain of 3,600, a healthy output current of say 6 amps will require a current injection into the drivers of 1.66mA. This is just under 5 times the stage current of the voltage amplifier, and quite acceptable notwithstanding the difficult crossover area where load alteration is variable and non-linear. However, at crossover, the load current is tiny, so this change in input impedance of the output stage is manageable.
5. Concentrate on amp stability, using the smallest lag comp cap possible. Start at 68pF and work down.
6. Try to eliminate current mirrors where possible. They guarantee diff pair imbalance, and seem to damage the sonics.
This is all conventional topology, but you'd be surprised how the sound can be tailored by scrupulous care with the device choices and working environment of the VAS and OS.
Cheers,
Hugh
Hi Jens,
Sure, here it is:
http://www.dself.dsl.pipex.com/ampins/dipa/dipa.htm
Half way down the page, EF Type II. There are a number of output stages discussed; this is the second. I've tried them all.
Cheers,
Hugh
Sure, here it is:
http://www.dself.dsl.pipex.com/ampins/dipa/dipa.htm
Half way down the page, EF Type II. There are a number of output stages discussed; this is the second. I've tried them all.
Cheers,
Hugh
Hi Hugh,
Thank you very much for the link.
It seems pretty conventional, except that I don't see the need for the one or two resistors between the driver transistors.
Whst is it/they good for?
Thanks,
Jens
Btw.: I'm planning to visit eatern Australia next spring. Do you know any good resources for travel info on Sidnay & Brisbane? (spelling???)
If so, please send it directly to my home address:
jens@sylvest.net
Thank you very much for the link.
It seems pretty conventional, except that I don't see the need for the one or two resistors between the driver transistors.
Whst is it/they good for?
Thanks,
Jens
Btw.: I'm planning to visit eatern Australia next spring. Do you know any good resources for travel info on Sidnay & Brisbane? (spelling???)
If so, please send it directly to my home address:
jens@sylvest.net
Hi Jens,
Read the article carefully and you will see that a resistor between the drivers greatly enhances clean switching of the driver and the outputs. It actually makes a palpable difference to the sound, and is foolproof. Simple, works well, sounds terrific. The idea has been around for a long time but is not much used because of simple prejudice. Try it.
For travel information on Sydney and Brisbane I'd simply plug the words into Google and go for it. Excellent examples of this are http://www.sydney.com.au/ http://www.cityofsydney.nsw.gov.au/
where there is a huge amount of travel, tourist and accommodation information. For Brisbane, try Lonely Planet, a Melbourne company here: http://www.lonelyplanet.com/destinations/australasia/brisbane/
and try this one: http://www.maxlink.com.au/bcl/default.htm
This is just a start. The country is large (within 10% of continental USA) but the population is low, so outside the major cities (Sydney 4.5M, Melbourne 3.6M, Brisbane 1.3M) you will find a lot of nature. Visit during the southern hemisphere autumn and spring; avoid summer in Australia because temperatures are often over 40C. Winter is marvellous in Northern Australia, like Cairns, but too cold down south.
Joe, thanks for your response!
Hope this is helpful,
Cheers,
Hugh
Read the article carefully and you will see that a resistor between the drivers greatly enhances clean switching of the driver and the outputs. It actually makes a palpable difference to the sound, and is foolproof. Simple, works well, sounds terrific. The idea has been around for a long time but is not much used because of simple prejudice. Try it.
For travel information on Sydney and Brisbane I'd simply plug the words into Google and go for it. Excellent examples of this are http://www.sydney.com.au/ http://www.cityofsydney.nsw.gov.au/
where there is a huge amount of travel, tourist and accommodation information. For Brisbane, try Lonely Planet, a Melbourne company here: http://www.lonelyplanet.com/destinations/australasia/brisbane/
and try this one: http://www.maxlink.com.au/bcl/default.htm
This is just a start. The country is large (within 10% of continental USA) but the population is low, so outside the major cities (Sydney 4.5M, Melbourne 3.6M, Brisbane 1.3M) you will find a lot of nature. Visit during the southern hemisphere autumn and spring; avoid summer in Australia because temperatures are often over 40C. Winter is marvellous in Northern Australia, like Cairns, but too cold down south.
Joe, thanks for your response!
Hope this is helpful,
Cheers,
Hugh
Thanks for the travel info links
As for the southern hemosphere (spelling!?) summer to be avoided, I was thinking of late March - early April.
Regarding the audio:
I've run through the article, but noticed one common thing about the output stages.
The stages all seem to generate an off-set if comparing the input to the output, and then compensate in some way or another (in some cases I don't see any compensation at all.)
Wouldn't it be possible (as my source soesn't have any off-set), to split the bias-voltage generator in two, and supply the source signal in the middle?
In stead of the clasic Vbe * 4 setup with a transistor and two resistors (one of them wariable), I was thinking of making two seperate Vbe-doublers. The offset would theoretically be gone, and if any remains, I can adjust the sources uneven to compensate. The thermal compensation factor would have the same effect as long as both bias-generators are mounted on the heat sinks. The resistor between the drivers would not be affected by it either.
Is there something critical I missed, or is it as straight-forward as it seems?
Best regards,
Jens
As for the southern hemosphere (spelling!?) summer to be avoided, I was thinking of late March - early April.
Regarding the audio:
I've run through the article, but noticed one common thing about the output stages.
The stages all seem to generate an off-set if comparing the input to the output, and then compensate in some way or another (in some cases I don't see any compensation at all.)
Wouldn't it be possible (as my source soesn't have any off-set), to split the bias-voltage generator in two, and supply the source signal in the middle?
In stead of the clasic Vbe * 4 setup with a transistor and two resistors (one of them wariable), I was thinking of making two seperate Vbe-doublers. The offset would theoretically be gone, and if any remains, I can adjust the sources uneven to compensate. The thermal compensation factor would have the same effect as long as both bias-generators are mounted on the heat sinks. The resistor between the drivers would not be affected by it either.
Is there something critical I missed, or is it as straight-forward as it seems?
Best regards,
Jens
Re: Re: power output stage
Well, the name "Jennice" was originally given to me by an elderly secretary at a company in the USA. A colleague and myself went there to be trained (being employed at their danish subsidiary) for two weeks. She never figured out the "Jens", but translated my pronounciation of it into "Jennice" when she wrote out employee name tags. Since then, it kinda' stuck among colleagues.
(I'm lucky... the other guy ended up being called "Heineken" after the beer as he likes it, and isn't exactly skinny )
I consider myself too calm (generally) to adopt "Jennice the Mennace", but I'm open for woting about the "Nordic Nutter" (nutcase), as a penpal called me. Wonder why...
As for the DC and off-set control, the idea with a split Vbe source (bias) was created to obtain a DC coupling while avoiding the DC-offset as seen in D. Self's designs.
I'll post a picture of the schematic as soon as I figure out how
peranders said:
Well, the name "Jennice" was originally given to me by an elderly secretary at a company in the USA. A colleague and myself went there to be trained (being employed at their danish subsidiary) for two weeks. She never figured out the "Jens", but translated my pronounciation of it into "Jennice" when she wrote out employee name tags. Since then, it kinda' stuck among colleagues.
(I'm lucky... the other guy ended up being called "Heineken" after the beer as he likes it, and isn't exactly skinny
)I consider myself too calm (generally) to adopt "Jennice the Mennace", but I'm open for woting about the "Nordic Nutter" (nutcase), as a penpal called me. Wonder why...
As for the DC and off-set control, the idea with a split Vbe source (bias) was created to obtain a DC coupling while avoiding the DC-offset as seen in D. Self's designs.
I'll post a picture of the schematic as soon as I figure out how
The original requirement was no global feedback, ±40V class AB.
"3. Use the Double emitter follower Type II from Doug Self."
Large error in this circuit with no global feedback, unless high bias.
http://www.dself.dsl.pipex.com/ampins/dipa/dpafig13.gif
When driven at full output a 4R there is an error of 2V from the 0R22 emitter resistors (fig B).
If we use the CFP (fig C) with the emitter resistors moved to the emitters (as Nelson Pass does in the Stasis design), we can run with no global feedback.
My comments as regards the bias and DC off-set:
"Not really, all you need is a bias circuit and the driver and output stages. Two constant current sources, one to the positive rail, and one to the negative rail will provide the bias. Make one of these so you can trim the DC off-set to zero and you should be fine. "
To help you visualize:
http://www.homestead.com/whaan/files/hyb_6463.jpg
In its simplest form:
http://www.homestead.com/whaan/files/5676_circ.jpg
"3. Use the Double emitter follower Type II from Doug Self."
Large error in this circuit with no global feedback, unless high bias.
http://www.dself.dsl.pipex.com/ampins/dipa/dpafig13.gif
When driven at full output a 4R there is an error of 2V from the 0R22 emitter resistors (fig B).
If we use the CFP (fig C) with the emitter resistors moved to the emitters (as Nelson Pass does in the Stasis design), we can run with no global feedback.
My comments as regards the bias and DC off-set:
"Not really, all you need is a bias circuit and the driver and output stages. Two constant current sources, one to the positive rail, and one to the negative rail will provide the bias. Make one of these so you can trim the DC off-set to zero and you should be fine. "
To help you visualize:
http://www.homestead.com/whaan/files/hyb_6463.jpg
In its simplest form:
http://www.homestead.com/whaan/files/5676_circ.jpg
Have a look at my circuit for a power follower here:
http://www.diyaudio.com/forums/showthread.php?s=&threadid=14320
It uses MOSFET's, however it is good enough to work without overall feedback and provides low distortion and output impedance even in AB.
Cheers
x-pro
http://www.diyaudio.com/forums/showthread.php?s=&threadid=14320
It uses MOSFET's, however it is good enough to work without overall feedback and provides low distortion and output impedance even in AB.
Cheers
x-pro
Thanks for the schematics (artwork? ) and comments/suggestions.
I have tried to simulate the output stage of the stasis design but it tends to oscillate.
Otherwise it could be interesting as it has a neat kind of local feedback.
As for the mosfet stage, I'll have a look at it, although I am most familiar with BJT's and hance have a pile of good Motorola devices at hand.
Jens
) and comments/suggestions.I have tried to simulate the output stage of the stasis design but it tends to oscillate.
Otherwise it could be interesting as it has a neat kind of local feedback.
As for the mosfet stage, I'll have a look at it, although I am most familiar with BJT's and hance have a pile of good Motorola devices at hand.
Jens
Jennice,
I've tried all the CFP configurations. They work well in Class A, but not in AB. The EF II from Self may have an error at crossover, but it's MUCH better sounding in Class AB than the CFP - I built both with identical front ends and compared them.
The de Haan circuit looks good, and with those rather large output devices gives good performance. Without global feedback, however, the Zout will only be around 0R5, good enough for most applications, but not as good as the average SS amp with global feedback for bass control.
Cheers,
Hugh
I've tried all the CFP configurations. They work well in Class A, but not in AB. The EF II from Self may have an error at crossover, but it's MUCH better sounding in Class AB than the CFP - I built both with identical front ends and compared them.
The de Haan circuit looks good, and with those rather large output devices gives good performance. Without global feedback, however, the Zout will only be around 0R5, good enough for most applications, but not as good as the average SS amp with global feedback for bass control.
Cheers,
Hugh
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.