input/drive circuit PS filtering, cascodes, referencing compensation to AC gnd instead of a supply are all good
but I did enjoy Self's argument on power amp output Q supply regulation - when asked for it in a new design assignment he replied by asking what was the logical improvement vs putting hard clipping on the input
but I did enjoy Self's argument on power amp output Q supply regulation - when asked for it in a new design assignment he replied by asking what was the logical improvement vs putting hard clipping on the input
There are techniques for paralleling any number of reasonably low cost pass transistors, or equivalents, as necessary to meet the peak current demand, while minimising the dissipated energy, so this should not be used as an excuse. Again, intelligent engineering can come to the rescue ...
Frank
Frank
Although I've never worked with any power chip, I agree with you, Frank. I do so because of two things.
One of the most sensible and meaningful ways of upgrading almost any equipment, and especially the budget sector products, is to upgrade their filter capacitors. Since 1980, I have not heard a single product which was not improved by better qualiy, bigger capacitors; some more, some less, but all sounded better. Should you do a bit more than just the filter caps, such as upgrading the rectification with some fast diodes, and perhaps substituing the usually tight power transformer with something a bit more relaxed, even lowly budget sector amps start to sound serious.
Another reason is that again, I have never yet encountered a power amplifier which did not sound better when its VAS was fed off its own, separate regulated power supplies. Again, some more, some less, but all tend to do from better to much better.
I am now slowly wrapping up my power amp project, and have come to the last, but by no means the least section, the power supply.
I'd like to hear your comments on what I have in mind.
The first thing I am set on doing is slightly spreading arund the filter caps. So, instead of putting all power amp filters on one board, I intend to put 2 10,000 uF caps on their own board, but also a 4,700 uF cap (same model, same manufacturer) on the actual power amp board, followed by 100 uF and 0.1 uF caps. It seems to me it's practical to have it as near the output stage as possible. Views? Opinions? Experience?
The voltage gain stage will be fully regulated, as I have been doing since the early 80ies. The questions which now arise are:
1. Should these regulated lines also feed the predriver only, or predriver AND driver, leaving only the output section to run off capacior fed lines?
2. Or should I go for full voltage/current regulation of the power section as well, separately from what precedes it?
Remember, I am very flexible. It can be a single case stereo amp, or two separate monoblocks, meaning PCB real estate is very flexible.
I am not bound by any price constraints, this is for me and my pleasure, so cost is relatively unimportant.
I have a relative abundance of serious power devices (all by Motorola/ON Semi, MJ 21195/21196 TO-3 250W devices, MJL 3281/1302 plastic pack 200W devices), so there's no problem in using 2, 3 or 4 pairs of power devices, on a separate hefty heat sink.
So, I plan to use 3 or 4 pairs of output devices in the audio section, and it's not a problem to use exactly the same number and type of devices in the regulator, because remember, I do want it to be capable of some rather serious current peaks.
All ideas and views invited.
I'd like to hear your comments on what I have in mind.
The first thing I am set on doing is slightly spreading arund the filter caps. So, instead of putting all power amp filters on one board, I intend to put 2 10,000 uF caps on their own board, but also a 4,700 uF cap (same model, same manufacturer) on the actual power amp board, followed by 100 uF and 0.1 uF caps. It seems to me it's practical to have it as near the output stage as possible. Views? Opinions? Experience?
The voltage gain stage will be fully regulated, as I have been doing since the early 80ies. The questions which now arise are:
1. Should these regulated lines also feed the predriver only, or predriver AND driver, leaving only the output section to run off capacior fed lines?
2. Or should I go for full voltage/current regulation of the power section as well, separately from what precedes it?
Remember, I am very flexible. It can be a single case stereo amp, or two separate monoblocks, meaning PCB real estate is very flexible.
I am not bound by any price constraints, this is for me and my pleasure, so cost is relatively unimportant.
I have a relative abundance of serious power devices (all by Motorola/ON Semi, MJ 21195/21196 TO-3 250W devices, MJL 3281/1302 plastic pack 200W devices), so there's no problem in using 2, 3 or 4 pairs of power devices, on a separate hefty heat sink.
So, I plan to use 3 or 4 pairs of output devices in the audio section, and it's not a problem to use exactly the same number and type of devices in the regulator, because remember, I do want it to be capable of some rather serious current peaks.
All ideas and views invited.
Where to start ...?!
In summary, a Yes to all your questions! I'm of a mind now where the power supply is the most important part of the amplifier, the amplifying topology is almost a by-the-way! In order of sensitivity to power supply fluctuations, typically the output stage is the least affected, the VAS the most sensitive, which as you say you've addressed before. So add regulation with those priorities in mind; in my experience every effort you make to improve the power supply, in every area, will be rewarded many times over. And be very particular in the routing of power lines or traces so that current pulses do not impact where they shouldn't.
I would read up every tip and thread about optimising power supplies, the important things have been said many times over, and the gains from doing it right can be tremendous ...
Frank
Where to start ...?!
In summary, a Yes to all your questions! I'm of a mind now where the power supply is the most important part of the amplifier, the amplifying topology is almost a by-the-way! In order of sensitivity to power supply fluctuations, typically the output stage is the least affected, the VAS the most sensitive, which as you say you've addressed before. So add regulation with those priorities in mind; in my experience every effort you make to improve the power supply, in every area, will be rewarded many times over. And be very particular in the routing of power lines or traces so that current pulses do not impact where they shouldn't.
I would read up every tip and thread about optimising power supplies, the important things have been said many times over, and the gains from doing it right can be tremendous ...
Frank
Thank you, Frank, we seem to agree across the board.
I also think it's very difficult to overdo it with power supplies.
A few weeks ago, I took out some 20+ year old 6,800 uF caps out of a Toshiba integrated amp and soldered a pair of 10,000 uF brand new caps, nicely formed. Bearing in mind that the old gave place to the new, in itself a leap forward, the extra 47% of capacitance did wonders for the unit. I would have gone even further, but there was no more space left for still bigger caps.
And that's strictly a low level, budget amp from 1982, delivering a nominal 50WRMS into 8 Ohms, using 80W SEPP output stages. I should add that I was VERY surprised at the way it was constructed, I would never have thought it possible to use such circuitry in such a budget model, belying its price grade and far belying its perceived level of technology. Then again, Toshiba does make all those components.
I understood it as an object lesson - serious people will be serious at any price level.
And it made me promise myself to throw away my wallet when it comes to power supplies.
On top of what you plan to use , I would use one 4700 uf per output device and place them 20 -25mm away from the ouputs, CRC WITH full regulation for vas, obligatory soft start , leave out all sonic degrading protection BS limiters , use gold plated high grade fuses ...
Regards
Last edited:
So would I - if I had a room to dedicate to just this, Wayne.
Work it out yourself - that takes a lot of space, add a large power transformer, the basic capacitor block (4 10,000 uF caps), the not at all insignificant heat sinking I need to use and so forth, and it becomes overwhelming all too soon.
I agree with all of that ( fas42 + a w ) . It is also an economical way to do it ( a . w ) . Stability should be enhanced .
My own experiments also show to try under voltage to driver stage . You can have the option of optimum voltage also ( power out ) . Call it soft clip if you like . Do it on the scope , you will see what I mean . At clipping you should find less harmonics if driver stage clips first ( 1 V under will do ) . That way you get great book figures for power . Bet you anything the customers choose soft clip . I found it by accident . I was using my ears one day and the oscilloscope said my ears were right . I took so much trouble to build my clean high voltage supply , only to find I ran it as before or lower . Worth a thought ?
My own experiments also show to try under voltage to driver stage . You can have the option of optimum voltage also ( power out ) . Call it soft clip if you like . Do it on the scope , you will see what I mean . At clipping you should find less harmonics if driver stage clips first ( 1 V under will do ) . That way you get great book figures for power . Bet you anything the customers choose soft clip . I found it by accident . I was using my ears one day and the oscilloscope said my ears were right . I took so much trouble to build my clean high voltage supply , only to find I ran it as before or lower . Worth a thought ?
Err ok , so when you said this ....!!!
"Remember, I am very flexible. It can be a single case stereo amp, or two separate monoblocks, meaning PCB real estate is very flexible. I am not bound by any price constraints, this is for me and my pleasure, so cost is relatively unimportant."
What you really meant was .........
Err ok , so when you said this ....!!!
"Remember, I am very flexible. It can be a single case stereo amp, or two separate monoblocks, meaning PCB real estate is very flexible. I am not bound by any price constraints, this is for me and my pleasure, so cost is relatively unimportant."
What you really meant was .........
... in terms of cost, not space. My room has only 12 m2 of area (about 109 sq. ft).
My fault, I wasn't specific enough, sorry.
Who's complaining?
John, the key difference between you and me in this case is that you have manufacturing capabilities, both via the Parasound company and as a private citizen, that I can only dream of.
Just to illustrate the point, it is not possible for me to obtain locally manufactured anodized aluminium plates. I have to import them from Slovenia, which is a hassle and a half, believe me. And, because of taxes and levies, does become quite expensive in the end. The best I can hope for is painted aluminium, and even that is of mediocre quality only. I am not particularly consoled that the paint is an ecological, water based imported Italian paint.
By pure blind luck, a friend of mine does aluminium shaping using his own CNC machine, and he spent 20 years making gyroscopes, so a deviation of one micrometer is enormous to him (in the aiming gear of a T84 tank, that deviation is the difference between hitting and missing the target).
Don't worry, I'll figure out a way with a little help from my friends, but it will never look like an industrial product because of lack of support of local manufacturing facilities.
Sorry, I forgot.
As for the output stage, I plan on using 4 pairs of Motorola/ON Semi MJL 3281/1302 200W devices per channel. According to Data Sheet specs, at 85 deg. Centigrade, these four should still be capable of delivering 950 Watts of power, which I reckon is just about right for a nominally 100 Watts into 8 Ohms power amp, even with doubling down to 2 Ohms.
I plan on a total of 69,400 uF per channel, as you could have seen from previous posts.
That's about 44% of what (the transistors) you use in JC-1. I don't know its power rating, so I cannot compare that. A link to it would be appreciated.
Last edited:
- Status
- Not open for further replies.