Hi,
I need to place a Cdom in the Vas stage of a new amplifier I'm building. I am wanting to use a boosted design so I'm using 2 transistors. I believe the local feedback needs to be around the whole lot but I don't know where to connect it.
If you can help, please show me how.
Diagram attached.
Cheers,
I need to place a Cdom in the Vas stage of a new amplifier I'm building. I am wanting to use a boosted design so I'm using 2 transistors. I believe the local feedback needs to be around the whole lot but I don't know where to connect it.
If you can help, please show me how.
Diagram attached.
Cheers,
You answered your own question. Round the whole lot, i.e. move it from the collector of the first transistor to the emitter of the second transistor where the resistor is.
I have ask if you stumbled on this should you be trying to design an amp? Especially as this one looks as though it might be a little complicated and difficult to stabilise. Maybe try starting with something simpler.
I have ask if you stumbled on this should you be trying to design an amp? Especially as this one looks as though it might be a little complicated and difficult to stabilise. Maybe try starting with something simpler.
Thanks Richie,
I have seen it written that boosting the Vas transistor has benefits but I've seen mainly standard Darlington designs. The trouble with the Darlington is that it is biased at 1.2V. This would mean large resistors in the current mirror that precedes this part of the circuit. If I go with this Sziklai combination it allows me a 0.6V bias.
I'm hoping it will work out!
Cheers,
I have seen it written that boosting the Vas transistor has benefits but I've seen mainly standard Darlington designs. The trouble with the Darlington is that it is biased at 1.2V. This would mean large resistors in the current mirror that precedes this part of the circuit. If I go with this Sziklai combination it allows me a 0.6V bias.
I'm hoping it will work out!
Cheers,
Hi,
there are alternatives.
Cdom can return HF feedback from the emitter of the compound pair to the input base of the pair. But there is another out of phase point that the HF feed back can be returned to, the inverting input of the long tail pair.
There are others, but these two are the most commonly used.
The Cdom is the easy and very effective solution that virtually guarantees stable operation of a three stage "LIN" topology. But the downside is reputedly poor sound quality. Try the non-inverting input for improved sound quality. And use a very low capacitance transistor for VAS to reduce the current that gets sucked out of the LTP.
there are alternatives.
Cdom can return HF feedback from the emitter of the compound pair to the input base of the pair. But there is another out of phase point that the HF feed back can be returned to, the inverting input of the long tail pair.
There are others, but these two are the most commonly used.
The Cdom is the easy and very effective solution that virtually guarantees stable operation of a three stage "LIN" topology. But the downside is reputedly poor sound quality. Try the non-inverting input for improved sound quality. And use a very low capacitance transistor for VAS to reduce the current that gets sucked out of the LTP.
I hesitate to reinforce what may be an old wives tale, but when I stabilized an old Tiger amp with a Cdom on the vas stage, the sound quality appeared to deteriorate noticeably. I went back to a rolloff cap across the global feedback resistor, and everything went back to normal. IMO, though the response and distortion numbers were similar, an SA of the distortion probably wouldn't be. Since the amp has rather high distortion, it makes sense. For a design with fundamentally lower distortion, it might not matter. As for the actual question, you really need to order up Doug Self's book on audio power amps. It will answer your questions and bring up issues you'd take a decade to come across on your own.
An increased Cdom shifts the whole Aol to the left, reducing the gain available for global feedback at all frequencies above the pole. Whereas a global roll-off cap only reduces Acl above the corresponding pole (to get a rate-of-closure of ~20dB/decade at the Aol 1/beta intersect where additional Aol poles raise their heads), way above audio frequencies. Hence with the Cdom compensation one quite often gets higher distortion in the audio band for a given closed-loop gain.
Mr.Self's website, in particular his excellent distortion article, might suffice for Steve's question.
http://www.dself.dsl.pipex.com/ampins/dipa/dipa.htm
Regards, Klaus
Mr.Self's website, in particular his excellent distortion article, might suffice for Steve's question.
http://www.dself.dsl.pipex.com/ampins/dipa/dipa.htm
Regards, Klaus
Cheers guys,
I do have the latest version of Mr Self's book but I couldn't see any examples of my configuration. My reading is now steering me to the discussion of matching the slew-rate performance which I haven't paid great attention to. I wonder how important it is to match the positive and negative slew rates ...
I shall try to get further from the Self site concerning distortion too thanks KSTR.
Andrew, the transistor downwind from the differential pair is a 2SA970 or 2SC2240. Both of these were selected for their reputedly linear response and their low capacitance. All the low power transistors are either of these.
I assume the 'non-inverting' input is the feedback side of the diff pair? Or is it depending on the configuration? I wonder where I could find some good examples of the Cdom configuration that taps into this 'inverting-input'?
Thanks for your thoughts guys.
Rgds,
Steve
I do have the latest version of Mr Self's book but I couldn't see any examples of my configuration. My reading is now steering me to the discussion of matching the slew-rate performance which I haven't paid great attention to. I wonder how important it is to match the positive and negative slew rates ...
I shall try to get further from the Self site concerning distortion too thanks KSTR.
Andrew, the transistor downwind from the differential pair is a 2SA970 or 2SC2240. Both of these were selected for their reputedly linear response and their low capacitance. All the low power transistors are either of these.
I assume the 'non-inverting' input is the feedback side of the diff pair? Or is it depending on the configuration? I wonder where I could find some good examples of the Cdom configuration that taps into this 'inverting-input'?
Thanks for your thoughts guys.
Rgds,
Steve
Hi,
all these don't use the Cdom method.
JLH 80W mosfet,
Crimson series,
Sugden p128
Is it by coincidence they are all British?
Someone here once suggested that JLH may have been consulting to various manufacturers.
The JLH and Sugden are posted here.
Self does not discuss the alternatives to Cdom in detail.
He became blinkered with 0.0006% distortion target.
He completely messed up his treatment/solution/failure to properly analyse his asymmetric slew rate problem.
Read someone else for balance.
all these don't use the Cdom method.
JLH 80W mosfet,
Crimson series,
Sugden p128
Is it by coincidence they are all British?
Someone here once suggested that JLH may have been consulting to various manufacturers.
The JLH and Sugden are posted here.
Self does not discuss the alternatives to Cdom in detail.
He became blinkered with 0.0006% distortion target.
He completely messed up his treatment/solution/failure to properly analyse his asymmetric slew rate problem.
Read someone else for balance.
the two bases either side of the LTP are anti-phase (differential) inputs.
Does assymetrical slew rate really matter? If the slew limits of either side are not approached then I can't see how it does.
With the example VAS shown the cfp configuration will amplify the feedback signal and will lower the output impedanceof the VAS by a factor equal to the total hfe. If the VAS is fed from a standard LTP then the only pre requisite is that the LTP quiescent current is sufficient to sink the VAS feedback current for the designed voltage swing. I don't think this was messed up at all by Doug Self and his explanation of the LTP and VAS was completely satisfactory. Anyway what's wrong with 0.0006% THD. I've heard some amplifiers that employ a single ended VAS with cDom sound exceptionally good with those sorts of THD figures.
With the example VAS shown the cfp configuration will amplify the feedback signal and will lower the output impedanceof the VAS by a factor equal to the total hfe. If the VAS is fed from a standard LTP then the only pre requisite is that the LTP quiescent current is sufficient to sink the VAS feedback current for the designed voltage swing. I don't think this was messed up at all by Doug Self and his explanation of the LTP and VAS was completely satisfactory. Anyway what's wrong with 0.0006% THD. I've heard some amplifiers that employ a single ended VAS with cDom sound exceptionally good with those sorts of THD figures.
There's nothing wrong with setting a 0.0006% target. What is flawed is his total exclusion of other topologies that may miss that target and concentrate on Cdom exclusively.
In my opinion the current mirror on the LTP also falls into this target setting trap and alledgedly blinkered view.
I have another opinion, these two topologies, Mirror loaded LTP and Cdom, may be why the blameless, trimodal and load invariant amplifiers are often cristicised for less than perfect subjective sound quality. Is it a coincidence that all three variants incorporate ONLY these two topologies in the first two stages?
Now compare the Audiolab/Tag power ampliers often described as "sterile" or "clinical" despite being technically accurate. They too use the same two front end stages.
In my opinion the current mirror on the LTP also falls into this target setting trap and alledgedly blinkered view.
I have another opinion, these two topologies, Mirror loaded LTP and Cdom, may be why the blameless, trimodal and load invariant amplifiers are often cristicised for less than perfect subjective sound quality. Is it a coincidence that all three variants incorporate ONLY these two topologies in the first two stages?
Now compare the Audiolab/Tag power ampliers often described as "sterile" or "clinical" despite being technically accurate. They too use the same two front end stages.
AndrewT said:
OK I confess I have built variations of the blamless amp and also found it to be a little on the sterile side. However, when you analyse the residual thd you will find that it's mostly odd order, particularly if you go to some lengths to balance the LTP current mirror. I suspect that the odd order components are the reason rather than some of the almost fanciful explanations I've seen which relate to potential slewing problems. If you degenerate the LTP emitters and reduce Cdom you will hear a noticeable improvement in that regard, but I think this is purely the reduction of HD3 from the LTP. Cdom will reduce the VAS hf distortion to vanishingly low levels so it's only the VAS and output stage that contribute significantly to THD, and that means it's nearly all odd order.
BTW, according to Bruce Candy (Halcro) his DM58/68 use current mirred LTP's and Cdom VAS compensation and are current source fed. His amplifiers have been well received world wide and are regarded as the best by many. So there are obviously some amplifiers that sound very good with this LTP / VAS design. I suspect that the implementation is the key and that starts with device selection, board layout and power supply isolation and regulation etc.
I agree with what David Lewis is saying. This "Blameless and sterile" sounding topology can be made to really "sing" by careful matching of the LTP and Current Mirror devices for both HFE and VBE. Both perform audibly better with suitable degeneration
resistors. Some years ago there was a discussion about Current Mirrors, and the overall consensus was that the SQ was noticeably improved by using matched , high HFE devices in the current mirror, with typically 100R emitter resistors.Best results were found with devices having an HFE around 800 (!!)
In my case,Cdom has been reduced to 68pf polystyrene,as well.
Matching the LTP devices for VBE and HFE, and then IDENTICAL current flow gives even better results. You need to simulate identical loading of the VAS , on the unloaded LTP side.
A close approximation in many cases, could be as simple as a
1% 22Megohm resistor from the tied base and collector of the current mirror, to the -VE rail. I am not using this method,but my
modified Silicon Chip 15W/Ch. Amplifier has a HUGE soundstage and pin point location of instruments and voices. The front end, however, uses a very low noise , very low impedance, PSU addon designed by the late John Linsley Hood.
SandyK
resistors. Some years ago there was a discussion about Current Mirrors, and the overall consensus was that the SQ was noticeably improved by using matched , high HFE devices in the current mirror, with typically 100R emitter resistors.Best results were found with devices having an HFE around 800 (!!)
In my case,Cdom has been reduced to 68pf polystyrene,as well.
Matching the LTP devices for VBE and HFE, and then IDENTICAL current flow gives even better results. You need to simulate identical loading of the VAS , on the unloaded LTP side.
A close approximation in many cases, could be as simple as a
1% 22Megohm resistor from the tied base and collector of the current mirror, to the -VE rail. I am not using this method,but my
modified Silicon Chip 15W/Ch. Amplifier has a HUGE soundstage and pin point location of instruments and voices. The front end, however, uses a very low noise , very low impedance, PSU addon designed by the late John Linsley Hood.
SandyK
Hi Andrew,
Some have said this, yes, but there is no question that these amps are extremely clean. Could it be that your notion of 'musical enjoyment' means a little fudging at the edges of the waveform, dare I say it, a little DISTORTION. (There, I've said it!!)
I'm not sure that these are the dearest models, but heck, Andrew, you are more than entitled to a POV, even from a strictly subjective stance, and there is no doubt that you understand the electronics to considerable, objective level. Why wouldn't you have an opinion, and a valid one at that?
BTW, you can even have compensation across the output collectors of the LTP, and the gain of the VAS can be weighted down brutishly with a simple capacitor and/or resistor to ground!
Cheers,
Hugh
Some have said this, yes, but there is no question that these amps are extremely clean. Could it be that your notion of 'musical enjoyment' means a little fudging at the edges of the waveform, dare I say it, a little DISTORTION. (There, I've said it!!)
I'm not sure that these are the dearest models, but heck, Andrew, you are more than entitled to a POV, even from a strictly subjective stance, and there is no doubt that you understand the electronics to considerable, objective level. Why wouldn't you have an opinion, and a valid one at that?
BTW, you can even have compensation across the output collectors of the LTP, and the gain of the VAS can be weighted down brutishly with a simple capacitor and/or resistor to ground!
Cheers,
Hugh
Each to his own.
The "bright" remarks of Jeremy Clarkson are constantly quoted nowadays, and the program has been appointed the "best" automobile tv-show title.
Some people forgot that 10-15 years ago Clarkson was the turd of the Top Gear series, none of the commercial tv-stations were interested in the big guy.
My abfav car program, untill the fat sloppy guy became the No1 honcho when the BBC was forced to produce a follow-up, never watched it since.
Overhere we had Jan Kool, known for his late '80s statement : Wonder Cable or a pile of new CD's ? An easy choice for me !
Highly skilled bloke, also highly biased.
Reading Kessler resembles eyeballing an article in The Sun, the newspaper at least has nice boobies to watch.
(update, a top of the line STC cable is 20 $-grand for 2m)
The "bright" remarks of Jeremy Clarkson are constantly quoted nowadays, and the program has been appointed the "best" automobile tv-show title.
Some people forgot that 10-15 years ago Clarkson was the turd of the Top Gear series, none of the commercial tv-stations were interested in the big guy.
My abfav car program, untill the fat sloppy guy became the No1 honcho when the BBC was forced to produce a follow-up, never watched it since.
Overhere we had Jan Kool, known for his late '80s statement : Wonder Cable or a pile of new CD's ? An easy choice for me !
Highly skilled bloke, also highly biased.
Reading Kessler resembles eyeballing an article in The Sun, the newspaper at least has nice boobies to watch.
(update, a top of the line STC cable is 20 $-grand for 2m)
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.