Hi
THe JLH circuit uses an "emitter follower" buffer on the VAS to reduce the loading from the VAS transistor. The circuit is very similar to the double bootstrap you mentioned earlier. The collector is bootstrapped to some extent, the idea is to keep its collector to emitter voltage from changing, as this reduces Early effect (where the current gain increases with collector voltage) distortion. But with the components shown it isn't clear how much bootstrapping is applied to the collector. Modern transistors probably would perform better than MJE521 and I would not have recommended a 2N3054 for this position in any case as it is very slow. Far better to use a cascoded VAS even running at higher power to eliminate Early effect distortion, I would have thought.
Also the Zener diode might protect the reverse bias on the MJE521 but it would be better to use a series diode as mentioned earlier as this cuts off the npn drive whereas the Zener will - or could- turn on and blow the Zener and VAS (probably in that order).
John
THe JLH circuit uses an "emitter follower" buffer on the VAS to reduce the loading from the VAS transistor. The circuit is very similar to the double bootstrap you mentioned earlier. The collector is bootstrapped to some extent, the idea is to keep its collector to emitter voltage from changing, as this reduces Early effect (where the current gain increases with collector voltage) distortion. But with the components shown it isn't clear how much bootstrapping is applied to the collector. Modern transistors probably would perform better than MJE521 and I would not have recommended a 2N3054 for this position in any case as it is very slow. Far better to use a cascoded VAS even running at higher power to eliminate Early effect distortion, I would have thought.
Also the Zener diode might protect the reverse bias on the MJE521 but it would be better to use a series diode as mentioned earlier as this cuts off the npn drive whereas the Zener will - or could- turn on and blow the Zener and VAS (probably in that order).
John
Hi Paul
Pros- it provides some protection for the VAS transistor (as long as a protection diode is used) (and I think this its most useful advantage as it limits the VAS current without needing emitter current sense resistors or series resistors in the bases of the driver transistors); it allows the drive voltage to each half of the amplifier to go above (or below) the rail voltages
Cons -it doubles the current in the VAS stage; adds components and power dissipation (but these might be worth trading for the protection) but has the same problem as any bootstrap - it can only move as fast as the output stage "pushes" it.
You can achieve the same effect as a double bootstrap with a double current source VAS, this speeds up the response (or could do) and works at DC but would not be able to push the drive voltage over the rail voltages.
John
Pros- it provides some protection for the VAS transistor (as long as a protection diode is used) (and I think this its most useful advantage as it limits the VAS current without needing emitter current sense resistors or series resistors in the bases of the driver transistors); it allows the drive voltage to each half of the amplifier to go above (or below) the rail voltages
Cons -it doubles the current in the VAS stage; adds components and power dissipation (but these might be worth trading for the protection) but has the same problem as any bootstrap - it can only move as fast as the output stage "pushes" it.
You can achieve the same effect as a double bootstrap with a double current source VAS, this speeds up the response (or could do) and works at DC but would not be able to push the drive voltage over the rail voltages.
John
Bootstrapping the VAS is often revived when it is cheaper than conventional CS designs. I doubt that HT amplifier designers set out to make them sound "nice".
I was interested to check RXV models but those I checked including RXV450, 550 were conventional CS designs. Do you have model numbers for some bootstrapped VAS designs?
Bumped in to this thread ...read all of it here is what i think :
A) Hugh is mastermind one of the very few designers that can translate numbers in to audio impressions
B) Even though i consider Carlos a very nice person and a a charming personality here some of his posts are simply impossible to follow and impossible to make any sense ...
C) There is a number of good people adding really valid opinions and some of them detailed and deeply calculated but still i will second thought if any of these people ever heard the difference in real life conditions
D) finally except theory and especially when we talk about sonic signature ( beyond electrical data ) there is a lot to do with the rest of circuit IE type of output EFP or CFP type of input LTP or diamond and so on and on ...Only this will add way too many variables in a conversation on sonic signature ..
Even though i like the bootstrap on the P3A and always had in mind to construct a well made P68 in order to put it on PA duty to push it around never had the time to do so not only to evaluate the sonic signature but also the safety aspects of such a circuit .
kind regards
Sakis
A) Hugh is mastermind one of the very few designers that can translate numbers in to audio impressions
B) Even though i consider Carlos a very nice person and a a charming personality here some of his posts are simply impossible to follow and impossible to make any sense ...
C) There is a number of good people adding really valid opinions and some of them detailed and deeply calculated but still i will second thought if any of these people ever heard the difference in real life conditions
D) finally except theory and especially when we talk about sonic signature ( beyond electrical data ) there is a lot to do with the rest of circuit IE type of output EFP or CFP type of input LTP or diamond and so on and on ...Only this will add way too many variables in a conversation on sonic signature ..
Even though i like the bootstrap on the P3A and always had in mind to construct a well made P68 in order to put it on PA duty to push it around never had the time to do so not only to evaluate the sonic signature but also the safety aspects of such a circuit .
kind regards
Sakis
I've eliminated "the rest of the circuit".
I tried this before on my modular EF2 , (below) is the old 2008 "BX"
(bootstrap input stage). It was my favorite in 2008 , we will see how it
stacks up against the latest round of PPM (ultra low THD) amps.
PCB will have the option for a "resistive" LTP source , to see if this has any
SQ effect. The 2015 EF3 output stage has robust cap multipliers (if I use the resistor instead of the CCS).
Simple 6 device input stage has one drawback , more "thump" on startup.
Edit - I even thought about "safety" - clip diode .... and if the clip becomes too
asymmetrical , the protection circuit will trip !
OS
Attachments
Last edited:
Not sure at present. I simulated a double-current-source version (rather than bootstrapped) and obtained a lower distortion figure (0.025% compared with 0.03%). However, H2 and H3 were pretty much the same.
This might be due to the feedback arrangements - H3 is usually crossover related and the Miller compensation does not help to reduce this.
John
This might be due to the feedback arrangements - H3 is usually crossover related and the Miller compensation does not help to reduce this.
John
Hi Paul,
I don't have this issue on DBS circuit...... I am inclined to agree with John's comment that your feedback situation and compensation issues might be the reason. You might also revisit the bias issue, and ensure that you have a reasonable driver current, say 15mA upper, and 8mA lower. This gives sufficient current to drive the lower gate, and enough grunt to drive the shebang with the VAS collector.
Hugh
I don't have this issue on DBS circuit...... I am inclined to agree with John's comment that your feedback situation and compensation issues might be the reason. You might also revisit the bias issue, and ensure that you have a reasonable driver current, say 15mA upper, and 8mA lower. This gives sufficient current to drive the lower gate, and enough grunt to drive the shebang with the VAS collector.
Hugh
Member
Joined 2009
Paid Member
really ? - I thought it arose from several places and is quite obvious even in Class A (without cross over) whenever you have a push-pull output.
You are right, Class A can generate "squashed" sinewaves (including Linsley Hood's simple class A) if the output stages run out of gain, which is H3. So better to use some of the modern transistors which have better gain hold-up. Crossover distortion is mostly indicated by odd harmonics but H3 is usually the largest. And Miller compensation prevents the VAS from responding quickly to reduce it.
Hello
An old thread coming back.
I did not follow all the thread, but I can reply about the VAS not responding quickly enough.
In my amps I use a low value Miller compensation capacitor and also a phase lead capacitor to keep the amp stable. So I get a better slew rate and a faster VAS.
So Tessier, for his amp in post #54, should reduce the Miller compensation capacitor to 17 Pf and add a 10 Pf capacitor between Q7 collector and Q4 base (it's a phase lead capacitor), and he also should do as Hugh suggested for the drivers current.
Bye
Gaetan
An old thread coming back.
I did not follow all the thread, but I can reply about the VAS not responding quickly enough.
In my amps I use a low value Miller compensation capacitor and also a phase lead capacitor to keep the amp stable. So I get a better slew rate and a faster VAS.
So Tessier, for his amp in post #54, should reduce the Miller compensation capacitor to 17 Pf and add a 10 Pf capacitor between Q7 collector and Q4 base (it's a phase lead capacitor), and he also should do as Hugh suggested for the drivers current.
Bye
Gaetan