The Karan KA-600 is an interesting example because the amount of capacitance is given in the specs (132000 uF per channel). Assuming minimum 100V rails to get 600W into 8 ohms, that's 660 joules per channel. Or 11 joules per watt, counting both rails. Give or take a few.
In contrast, Hiraga's 300B as sold in kit form had about 3.7 joules per watt, single ended.
In contrast, Hiraga's 300B as sold in kit form had about 3.7 joules per watt, single ended.
I find that it is virtually impossible to give a definitive statement about how much power supply capacitance and transformer size, because the 'target' is just too nebulous.
What do you want into 8 ohms? 50W, 100W, 200W, 400W. The amount of designed in output swing will change as the square root of the power, so: 100W 22,000 may be OK, but 400W, 44,000 would be necessary, EVEN WITH A RESISTIVE LOAD.
Now, what about loading? For example, the loudspeaker that I use, drops to .33 ohm at 2KHz! Now what?
Then again, my big amp, the JC-1, often drives large full range electrostatics, you know, 8uF in series with perhaps 1 ohm.
It isn't JUST the ripple, but the SPEAKER GROUND RETURN IMPEDANCE, that might seriously affect the low frequency damping factor. This implies, at least 10,000 uF, and more is better.
Then there is the condition, in the real world, of a speaker needing short term peak current that is twice as much as its worse case DC resistance would imply. (ref. Otala, Peak current requirements for loudspeakers) That is why many serious designers put lots of peak current (for 10ms or so) and you don't want to collapse the power supply completely when pushing out 50A or so for a short time.
OK, how do we put a NUMBER on all these conditions?
What do you want into 8 ohms? 50W, 100W, 200W, 400W. The amount of designed in output swing will change as the square root of the power, so: 100W 22,000 may be OK, but 400W, 44,000 would be necessary, EVEN WITH A RESISTIVE LOAD.
Now, what about loading? For example, the loudspeaker that I use, drops to .33 ohm at 2KHz! Now what?
Then again, my big amp, the JC-1, often drives large full range electrostatics, you know, 8uF in series with perhaps 1 ohm.
It isn't JUST the ripple, but the SPEAKER GROUND RETURN IMPEDANCE, that might seriously affect the low frequency damping factor. This implies, at least 10,000 uF, and more is better.
Then there is the condition, in the real world, of a speaker needing short term peak current that is twice as much as its worse case DC resistance would imply. (ref. Otala, Peak current requirements for loudspeakers) That is why many serious designers put lots of peak current (for 10ms or so) and you don't want to collapse the power supply completely when pushing out 50A or so for a short time.
OK, how do we put a NUMBER on all these conditions?
Last edited:
Last time I saw him, he demonstrated just that model driving his own, one-off loudspeakers based on JBL professional units, a 15" bass, a 10" midbass, and a few horn loaded drivers upstairs, with an efficiency of no less that 95 dB/2.83/1m, and probably more. If memory serves, each box weighs in at around 90 kilos (app. 200 lbs).
Some dynamics, huh?
I think it's not that simple.
Not disputing your maths, but if memory serves, all Milan's amps are fully balanced in topology. I never asked, he never told.
What I do know is that he currently uses custom capacitors, made for him by Germany's Fisher & Tausche.
It all comes back to what I said some pages ago - there can be no definite answer until you put your requirements on the table.
100 WPC/8 Ohms is one requirement, but adding 200 WPC for the same into 4 Ohms is quite another; add nominally 400 WPC/2 Ohms, and you have a whole new ball game.
So what we might percieve at first sight as an overkill might in fact be just enough to meet the designer's data.
100 WPC/8 Ohms is one requirement, but adding 200 WPC for the same into 4 Ohms is quite another; add nominally 400 WPC/2 Ohms, and you have a whole new ball game.
So what we might percieve at first sight as an overkill might in fact be just enough to meet the designer's data.
Is this why full regulation on the output stage becomes an issue ..?
Hiraga ! Explains why you don't exceed 25 watts ...
Which is a different ball game, because it's not just twice the capacitance, divided in half per amp side, but also ripple suppression at half the impedance.
Joule per watt, or C per A(V), in both cases it all depends on which load impedance is targetted.
Similarity from the same batch => http://www.diyaudio.com/forums/atta...3476548-authentic-toshibas-faitesvosjeux2.jpg
(peanut money is a different story too)
Last edited:
100% agree. It's another example of people wanting The Answer without having to do the work of figuring out what they need for a given design target.
Cause I don't need 'em! 5W peak (6 volts) is all I ever need at home.Hiraga ! Explains why you don't exceed 25 watts ...
Now if I'm trying to fill a hotel ballroom or stadium, that's another matter all together.
Really? So design rules and practices aren't part of figuring out what you need?
If I know how much power (or current) my amp is likely to supply, then trying to figure out the size of the power supply in joules per watt or farads per ampere or whatever is lazy? It's wanting The Answer handed to me?
I don't get it.
- Status
- Not open for further replies.