The whole point is that there does not seem to be much compromise - CE outputs theoretically offer better distortion performance and more efficiency.
Seems too good to be true, which is why I want to study it.
Either I will learn why the theoretical analysis is flawed or I will have a circuit for a set of amps to active crossover my speakers.
It is a stroke of luck that IC manufacturers have been so seriously motivated by requirements for low rails/low power.
I don't have to have exactly the same constraints as they do to find their work very helpful.
Boosted rails is a fallback option.
I have quite a few amps to build, 5.1 home theatre x 2 or 3 amps per active speaker.
So it's practical to order a small batch of transformers with an extra tap.
But it's a typically smart idea from Winfield.
I just wish he'd focus on the unfinished X chapters of his book.
Best wishes
David
Hi Dave
Doesn't boosted rails represent a waste of headroom according to your first post?
You can have boosted rails using a charge pump - fancy name for a capacitively-coupled rectifier adding voltage onto the split rails. No extra transformer taps needed, just two caps, a diode and resistor for each polarity.
I used to be enamoured with rail-to-rail swing and used collector-output output stages for that reason. It eliminates having the BE and biasing in series with the output. However, linearity is generally improved with higher rails than what you use and I prefer to reduce THD as much as possible. Besides, if you are building a power amp, the PT and chassis are the big costs and it makes little difference if the PT is a few watts bigger or smaller.
Doesn't boosted rails represent a waste of headroom according to your first post?
You can have boosted rails using a charge pump - fancy name for a capacitively-coupled rectifier adding voltage onto the split rails. No extra transformer taps needed, just two caps, a diode and resistor for each polarity.
I used to be enamoured with rail-to-rail swing and used collector-output output stages for that reason. It eliminates having the BE and biasing in series with the output. However, linearity is generally improved with higher rails than what you use and I prefer to reduce THD as much as possible. Besides, if you are building a power amp, the PT and chassis are the big costs and it makes little difference if the PT is a few watts bigger or smaller.
Boost the VAS stage voltage rails and regulate them for best results.
VAS stage only has to be 7 volts higher then the power rail under full load to deliver full power with mosfets.
My rebuilt SX780 now has 42v VAS stage regulator ( 53v supply ) and output supply sags from 40v to 35v under load, so there is 7v extra to drive the fets.
VAS stage only has to be 7 volts higher then the power rail under full load to deliver full power with mosfets.
My rebuilt SX780 now has 42v VAS stage regulator ( 53v supply ) and output supply sags from 40v to 35v under load, so there is 7v extra to drive the fets.
The current in the VAS and IPS is in the tens of milliamps, a boost of a few volts makes no practical difference to the power dissipation.
Current in the OPS is obviously far more, lower rails here will reduce power dissipation proportionately.
Excellent, please show what you came up with.
Best wishes
David
Getting close to the rail is expensive in terms of distortion, thermal stability and speed. As udok suggested:
The driver and the VA, but you should clamp them and/or not give them much room past clipping or you will get clipping recovery delay.
Another trick is to bootstrap the driver and VA.
The driver and the VA, but you should clamp them and/or not give them much room past clipping or you will get clipping recovery delay.
Another trick is to bootstrap the driver and VA.
I suppose smaller current angle in the really reactive load (higher than its mechanical resonanse) will reduce power dissipation much better...
Obviously we can't drive the output arbitrarily close to rail, either for a CE or EF.
But it should be better with a so-called rail to-rail OPS than an EF.
Modern r-to-r op amps achieve very low distortion, the OPA1641 for instance, so it's clearly possible to make it sufficiently low that it ceases to be a concern.
The thermal stability claim seems incorrect.
For equal output an r-to-r OPS should permit a lower rail , this improves thermal stability, the details are in Bob Cordell's book.
Best wishes
David
Hi, nice to see you still here, you've been quiet on DIY for a while.
So, since you worked with the man, do you have any specific circuit ideas?
Best wishes
David
Hi Damir
Did you ever finish this?
I eventually decided the benefits of Monticelli weren't worth the extra complexity, for an audio amp at least.
So I have started to look at a switched power supply, same as you, and Harry, and Labgruppen, and Bob Carver, and ...
There's a lot of ways to do this!
Best wishes
David
Hi Dave
You wrote: "Excellent, please show what you came up with."
One thing i didn't come up with but which works well is the topology Bryston uses for their amps, most simply represented by the 2B-LP. Here, a fully complementary diff-stage driving VAS drives a CFP output stage with gain using discreet Darlingtons for driver and output. The front-end operates at 30V while the output is at 55V. Only the outputs and drivers see the rail-to-rail swing.
Any CFP output stage combined with an appropriate VAS can achieve rail output levels - minus the usual RE drop for protection circuits and/or current-sharing of parallel outputs.
Because music is dynamic and it is best to avoid hitting the rails, it is inherently best to have excess voltage for headroom. I used to try to count every potential watt from the PSU but eventually realised that is a silly parameter to put before low-THD and other things that are much more critical to a pleasant listening experience.
Rail-to-rail opamps are designed for low-voltage environments, particularly battery-sourced products. It is typical in such products that the signals within are of known and limited values and the efficient use of available voltage IS critical.
Is there a voltage shortage where you live?
Even though my power needs for my listening space do not exceed about 4W total, and my amps are therefore quite low-power (still comparatively over-rated at maybe 20W per channel), I do enjoy the exercise of designing very high power amps with low-THD. In all of the "super amps", there is a hit on how completely the rail voltage is used, yet one can achieve kilowatt levels at good efficiency and ppm-20k THD using a hybrid class-G approach.
Regarding the rail-to-rail output: Has anyone mentioned bootstrapping the driver or VAS? Old and trusty and simple.
You wrote: "Excellent, please show what you came up with."
One thing i didn't come up with but which works well is the topology Bryston uses for their amps, most simply represented by the 2B-LP. Here, a fully complementary diff-stage driving VAS drives a CFP output stage with gain using discreet Darlingtons for driver and output. The front-end operates at 30V while the output is at 55V. Only the outputs and drivers see the rail-to-rail swing.
Any CFP output stage combined with an appropriate VAS can achieve rail output levels - minus the usual RE drop for protection circuits and/or current-sharing of parallel outputs.
Because music is dynamic and it is best to avoid hitting the rails, it is inherently best to have excess voltage for headroom. I used to try to count every potential watt from the PSU but eventually realised that is a silly parameter to put before low-THD and other things that are much more critical to a pleasant listening experience.
Rail-to-rail opamps are designed for low-voltage environments, particularly battery-sourced products. It is typical in such products that the signals within are of known and limited values and the efficient use of available voltage IS critical.
Is there a voltage shortage where you live?
Even though my power needs for my listening space do not exceed about 4W total, and my amps are therefore quite low-power (still comparatively over-rated at maybe 20W per channel), I do enjoy the exercise of designing very high power amps with low-THD. In all of the "super amps", there is a hit on how completely the rail voltage is used, yet one can achieve kilowatt levels at good efficiency and ppm-20k THD using a hybrid class-G approach.
Regarding the rail-to-rail output: Has anyone mentioned bootstrapping the driver or VAS? Old and trusty and simple.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.