I just had this brainwave. I need class-A amplification and also something to warm up my basement during the long drawn Indiana winter and voila, the beginning of a fall project, a 100 to 200 watt class-A amp.
I plan to use 32 Toshiba 2sa1943/2sc5200 devices per channel with 0.22ohm emitter resistors.
Question: How many mA should I see across each resistor to get 200 watts pure class-A?
I am guessing 200mV should do it. (Less than 1amp across each pair) for about 15amps total.
Each channel would have it's own 1kVa toroid and 132000uF of capacitance. The heatsink would be Wakefield type wind tunnels; 2 for stereo). The DC rails would be 85 to 90vdc per rail.
I may try biasing it to 100 watts class-A and let it run upto 400 watt into AB. I hope that power supply would give me 400wpc at 85 to 90v rails @ 8 ohms.
The rectification will be done via discrete hi-speed 80amp Diodes.
The input and VA stages will be basically a butchered Adcom GFA-545 Mk.II power amp whose rails will be increased to 75vdc from the current 55vdc to drive this output stage.
Any ideas and advice will be welcome!
I plan to use 32 Toshiba 2sa1943/2sc5200 devices per channel with 0.22ohm emitter resistors.
Question: How many mA should I see across each resistor to get 200 watts pure class-A?
I am guessing 200mV should do it. (Less than 1amp across each pair) for about 15amps total.
Each channel would have it's own 1kVa toroid and 132000uF of capacitance. The heatsink would be Wakefield type wind tunnels; 2 for stereo). The DC rails would be 85 to 90vdc per rail.
I may try biasing it to 100 watts class-A and let it run upto 400 watt into AB. I hope that power supply would give me 400wpc at 85 to 90v rails @ 8 ohms.
The rectification will be done via discrete hi-speed 80amp Diodes.
The input and VA stages will be basically a butchered Adcom GFA-545 Mk.II power amp whose rails will be increased to 75vdc from the current 55vdc to drive this output stage.
Any ideas and advice will be welcome!
90VDC*2*15A=2700 W(DC).
Think you would need something more in the order of a 5kVA-transformer to let it not heat up as hot as your heatsinks.
In my opinion you should have a look at some Aleph service manuals or the Zen variations and build one of those.
All calculations are correct and done for you.
And get a proper heating system.😉
Cheers
Jens
Think you would need something more in the order of a 5kVA-transformer to let it not heat up as hot as your heatsinks.
In my opinion you should have a look at some Aleph service manuals or the Zen variations and build one of those.
All calculations are correct and done for you.
And get a proper heating system.😉
Cheers
Jens
Hmmmm:
OK, What if I decrease the output to 100watts class-A 🙁 .
90*2*7.5= 1350 watts.
I really want those 400 watt peaks... Just in case.
OK, What if I decrease the output to 100watts class-A 🙁 .
90*2*7.5= 1350 watts.
I really want those 400 watt peaks... Just in case.
That´s still 1350 W(DC) which would call for a 2,5kVA transformer to stay comfortably warm.
You should seriously think about what you wanna drive with it and calculate from there.
Everything else is probably way overkill.
You should seriously think about what you wanna drive with it and calculate from there.
Everything else is probably way overkill.
Interesting...
How about this: Wouldnt the unloaded 90vdc rails sag to something like 68vdc to 75vdc loaded with 675 watts continuous per channel?
In which case 68*2*7.5=1020.
So with that, would a 2.4kVa toroid work? Also to get 2.4kVa I would be stacking three 800VA Toroids on top of each other, would that cause additional heat build up or is that factored into the VA rating? I will probably use Avel.
As for what these will drive... I want to build this to test several speakers... from 2 way mini-monitors to Polks SDA-SRS's.
thanks again!
How about this: Wouldnt the unloaded 90vdc rails sag to something like 68vdc to 75vdc loaded with 675 watts continuous per channel?
In which case 68*2*7.5=1020.
So with that, would a 2.4kVa toroid work? Also to get 2.4kVa I would be stacking three 800VA Toroids on top of each other, would that cause additional heat build up or is that factored into the VA rating? I will probably use Avel.
As for what these will drive... I want to build this to test several speakers... from 2 way mini-monitors to Polks SDA-SRS's.
thanks again!
When using 90V rails you must not use the 2SA1943/2SC5200, because you're running out of the SOAR at very low currents. You're probably not able to drive inductive/capacitive loads. At 25°C (junction temperature) and 90V rails a single 2SC5200 is rated at 0.45 amps only. With some degrees of phase shift the BJT will blow. This is also called second breakdown. 
What devices do you suggest in a TO-3P or TO-3PL package?
The only other device that comes to mind is the MJL21193 and 94 from ON semi.
The only other device that comes to mind is the MJL21193 and 94 from ON semi.
200WPC @ 8 ohm = 5 amps ( 7 amps piek to piek )
7amps class A on push/pull is bias current off 3.5A ???
why you want 15A quisiet current ?
but still 90V*2*3.5A = 630W
7amps class A on push/pull is bias current off 3.5A ???
why you want 15A quisiet current ?
but still 90V*2*3.5A = 630W
Now we are getting closer:
That was part of my original question:
For a 200watt amp, what Quiscent current do I need assuming emitter resistors of 0.22ohm, what should the voltage drop be on each resistor? (Assuming 24 or 32 devices i.e. 12 or 16 complementary pairs.)
I got the 15A from the following assumption (correct or not?)
My yardstick for class-A operation was for every watt of class-A, 5 watts is dissipated as heat. i.e. a 20 watt per channel class-A amplifier will draw 100watts from the AC mains. Or each device will be biased 1.5A (from Self's Trimodial 20 watt class-Amp) and for 200 watts I just extrapolated 1.5*10=15A...😕
Please tell me the error of my assumptions and thanks in advance!!
That was part of my original question:
For a 200watt amp, what Quiscent current do I need assuming emitter resistors of 0.22ohm, what should the voltage drop be on each resistor? (Assuming 24 or 32 devices i.e. 12 or 16 complementary pairs.)
I got the 15A from the following assumption (correct or not?)
My yardstick for class-A operation was for every watt of class-A, 5 watts is dissipated as heat. i.e. a 20 watt per channel class-A amplifier will draw 100watts from the AC mains. Or each device will be biased 1.5A (from Self's Trimodial 20 watt class-Amp) and for 200 watts I just extrapolated 1.5*10=15A...😕
Please tell me the error of my assumptions and thanks in advance!!
feel free to correct me if i am wrong...
depends on the topologie off the amp
-> with Singel ended Ibias = Ioutput
-> with push/pull Ibias = 1/2 Ioutput
so the bias you need ..
200W @ 8E or 100W @ 4E
P=I²*R => I = 5Amps
you need to set your bias to the piek-to-piek value off the current
=> 5 * v2 = 7,07Amps
=> knowing that you are using push/pull this means 3.5Amps
know thats still 315W/side, this is easely done by 6 complementary pairs ( 50W/device, with resistive load )
result
=> 3.6 / 6 = 0.6 Amps
=> 0.22 * 0.6 ( R * I 😛 ) = 0.132V
Still, 400W on 8E is a bit off an overkill imo, you need a transformer of at least 2KVA if you want your amp to be able to drive 4E speakers, and this only for music, if you want the 800W continues you need at least a 3KVA transformer, these don't come cheap you know, and 132000uF of capacitance @ 100V aint cheap at all, this is a megabudget project
Greetz Rudy
depends on the topologie off the amp
-> with Singel ended Ibias = Ioutput
-> with push/pull Ibias = 1/2 Ioutput
so the bias you need ..
200W @ 8E or 100W @ 4E
P=I²*R => I = 5Amps
you need to set your bias to the piek-to-piek value off the current
=> 5 * v2 = 7,07Amps
=> knowing that you are using push/pull this means 3.5Amps
know thats still 315W/side, this is easely done by 6 complementary pairs ( 50W/device, with resistive load )
result
=> 3.6 / 6 = 0.6 Amps
=> 0.22 * 0.6 ( R * I 😛 ) = 0.132V
Still, 400W on 8E is a bit off an overkill imo, you need a transformer of at least 2KVA if you want your amp to be able to drive 4E speakers, and this only for music, if you want the 800W continues you need at least a 3KVA transformer, these don't come cheap you know, and 132000uF of capacitance @ 100V aint cheap at all, this is a megabudget project
Greetz Rudy
Thanks Greetz Rudy!
Yes,
I do plan to use a push-pull design that will go into class-AB beyond 200wpc. So your calculations are good! (I think 😉 ).
• I have a source for Toroids that seems reasonable.
• I have 18 new caps I got from a Surplus auction rated at. 160v and 33000uF each. Cost me peanuts! I salivate just looking at them there in the box all shiny! LOL.
• I have the power section kind of figured out…. I need ideas for an input/VA stage preferably something available in a pre-fab PCB. I was thinking of butchering the boards off an Adcom GFA-545 Mk.II power amp I have. There are few class-AB amps I have heard that sound as natural as this 100 watt amp. (their 555 model is not that nice sounding to me). I will probably tweak it to handle 90 v rails and as apart from the output stage, all preceding stages are class-A anyway, it should work fine.
Question: As a final test, if I measure the AC ampere draw with the amp fully biased, based on your calculations, I should see about 5.5amps all the time (I live in the US, i.e. 120volt mains) correct?
Thanks again!
Yes,
I do plan to use a push-pull design that will go into class-AB beyond 200wpc. So your calculations are good! (I think 😉 ).
• I have a source for Toroids that seems reasonable.
• I have 18 new caps I got from a Surplus auction rated at. 160v and 33000uF each. Cost me peanuts! I salivate just looking at them there in the box all shiny! LOL.
• I have the power section kind of figured out…. I need ideas for an input/VA stage preferably something available in a pre-fab PCB. I was thinking of butchering the boards off an Adcom GFA-545 Mk.II power amp I have. There are few class-AB amps I have heard that sound as natural as this 100 watt amp. (their 555 model is not that nice sounding to me). I will probably tweak it to handle 90 v rails and as apart from the output stage, all preceding stages are class-A anyway, it should work fine.
Question: As a final test, if I measure the AC ampere draw with the amp fully biased, based on your calculations, I should see about 5.5amps all the time (I live in the US, i.e. 120volt mains) correct?
Thanks again!
The MJL21193 seems O.K. There are only very few parts (bjt)available to handle such high voltages. I'd switch to a MOSFET design.
You really want the capacity for 400W/8R and 800W/4R?
A Crown Macrotech 2400 is 850W/4R per channel.
It runs on 114V (its a bridge amplifier), or ±114 in a non-bridge mode.
Here is how you do it:
100W at 8R takes 5A peak bias, or 2.5A in a push-pull design. You run three pair of MJ21193/94 outputs with a total of 2.5A bias on them at ±57V, or 285W at idle. You cascode the three pair of MJ21193/94 with another three pair of MJ21193/94. The second set of three run from ±114V. The ±57V is fed to the collectors of the first set of three through a 6A fast diode. You have a total of twelve TO3 outputs per channel, four TO3P drivers
(use 2SA1302/2SC3281 or similar).
You could wire the whole thing up so it looked like an AKSA with three pair of outputs. I would cascode the diff pair at this kind of voltage. I would also use a Baker clamp.
The amplifier would run like a normal class A push-pull amplifier up to 100W/8R, and cascode itself as it moves into class AB for the balance of its power output. The diode at the ±57V supply is the key to low power dissipation, without it the amplifier would have to get rid of much more heat. Under normal audio use this diode would probably never shut off (switch).
The transformer should be 80-40-0-40-80V at 1.6KVA per channel for 4 ohm loads. You could use two 40-0-40V 800VA units in series too.
I have two 80-40-0-40-80V 2KVA at 25% duty cycle toroids available at USD$100 if you want to try this. I also have about 50 tunnels drilled for twelve TO3 outputs, and 25 tunnels with isolated quadrants (for no insulators on the transistors) for flatpacks (MJL21193/94).
A Crown Macrotech 2400 is 850W/4R per channel.
It runs on 114V (its a bridge amplifier), or ±114 in a non-bridge mode.
Here is how you do it:
100W at 8R takes 5A peak bias, or 2.5A in a push-pull design. You run three pair of MJ21193/94 outputs with a total of 2.5A bias on them at ±57V, or 285W at idle. You cascode the three pair of MJ21193/94 with another three pair of MJ21193/94. The second set of three run from ±114V. The ±57V is fed to the collectors of the first set of three through a 6A fast diode. You have a total of twelve TO3 outputs per channel, four TO3P drivers
(use 2SA1302/2SC3281 or similar).
You could wire the whole thing up so it looked like an AKSA with three pair of outputs. I would cascode the diff pair at this kind of voltage. I would also use a Baker clamp.
The amplifier would run like a normal class A push-pull amplifier up to 100W/8R, and cascode itself as it moves into class AB for the balance of its power output. The diode at the ±57V supply is the key to low power dissipation, without it the amplifier would have to get rid of much more heat. Under normal audio use this diode would probably never shut off (switch).
The transformer should be 80-40-0-40-80V at 1.6KVA per channel for 4 ohm loads. You could use two 40-0-40V 800VA units in series too.
I have two 80-40-0-40-80V 2KVA at 25% duty cycle toroids available at USD$100 if you want to try this. I also have about 50 tunnels drilled for twelve TO3 outputs, and 25 tunnels with isolated quadrants (for no insulators on the transistors) for flatpacks (MJL21193/94).
Bocka,
A guy on the Forum yesterday mentioned new devices from ON semi:
MJL4302 and MJL4281 (along the lines of A1302/C3281)
These are the biggest BJT's I have seen in terms of VCE.
VCE=350 volts!!!!
Watts=230.
Ft=35MHz
hfe=min 50!
I need to find a problem with these devces because on paper they look awesome and dare I say they migh revolutionize the future BJT high power amp designed because till now the max rated VCE device was the MJ15024/25 and MJ21193-6 series at 250v but they were a little sluggish with fT=5Mhz. This guy has awesome specs.
I am not sure about the capacitance on these... that may be an issue because of larger die size...
What do you think?
A guy on the Forum yesterday mentioned new devices from ON semi:
MJL4302 and MJL4281 (along the lines of A1302/C3281)
These are the biggest BJT's I have seen in terms of VCE.
VCE=350 volts!!!!
Watts=230.
Ft=35MHz
hfe=min 50!
I need to find a problem with these devces because on paper they look awesome and dare I say they migh revolutionize the future BJT high power amp designed because till now the max rated VCE device was the MJ15024/25 and MJ21193-6 series at 250v but they were a little sluggish with fT=5Mhz. This guy has awesome specs.
I am not sure about the capacitance on these... that may be an issue because of larger die size...
What do you think?
DJk,
I am interested; but not with a cascode output stage... I would use a straight paralelled OP stage maybe using the MJL4302 pair I mention in the last post or the MJL21193/94.
Cascoding adds a lot of complexity and as I have never etched my own PCB so you can see where this is going.
However if you have PCB's ready for this, I would be more than happy to buy your 1.6kVA Toroids.
Can you send me pictures of the Toroids as well as of the heatsinks?
arifh@prodigy.net
thanks!!
I am interested; but not with a cascode output stage... I would use a straight paralelled OP stage maybe using the MJL4302 pair I mention in the last post or the MJL21193/94.
Cascoding adds a lot of complexity and as I have never etched my own PCB so you can see where this is going.
However if you have PCB's ready for this, I would be more than happy to buy your 1.6kVA Toroids.
Can you send me pictures of the Toroids as well as of the heatsinks?
arifh@prodigy.net
thanks!!
"I am interested; but not with a cascode output stage... I would use a straight paralelled OP stage maybe using the MJL4302 pair I mention in the last post or the MJL21193/94."
The safe area on the MJL4302 pair sucks, only 120W@100V vs the MJL21193/4 are 200W@100V. You will need to use twice as many MJL4302.
Because of the 150*C maximum junction temperature on plastic outputs you will need twice as many as you do if you used 200*C rated metal.
If you don't cascode you will need twice as many outputs, the efficency cuts in half for class A+G vs class A+AB.
My design has 12 pcs X $2 each.
Non-cascode 24 pcs, plastic 48 pcs, MJL4302 96 pcs
"Cascoding adds a lot of complexity "
Uhh, if you say so. In reality it is very simple.
"Can you send me pictures of the Toroids as well as of the heatsinks? "
Get real, a toroid is round and has wires coming out. The heatsinks are from the Hafler DH500.
Don't forget to buy a bunch more heatsink for all the extra devices you will need to mount.
You like the Adcom circuit? Buy the A40 board for $6 from AudioXpress and stuff it the same. Very little difference, Nelson Pass designed both.
The safe area on the MJL4302 pair sucks, only 120W@100V vs the MJL21193/4 are 200W@100V. You will need to use twice as many MJL4302.
Because of the 150*C maximum junction temperature on plastic outputs you will need twice as many as you do if you used 200*C rated metal.
If you don't cascode you will need twice as many outputs, the efficency cuts in half for class A+G vs class A+AB.
My design has 12 pcs X $2 each.
Non-cascode 24 pcs, plastic 48 pcs, MJL4302 96 pcs
"Cascoding adds a lot of complexity "
Uhh, if you say so. In reality it is very simple.
"Can you send me pictures of the Toroids as well as of the heatsinks? "
Get real, a toroid is round and has wires coming out. The heatsinks are from the Hafler DH500.
Don't forget to buy a bunch more heatsink for all the extra devices you will need to mount.
You like the Adcom circuit? Buy the A40 board for $6 from AudioXpress and stuff it the same. Very little difference, Nelson Pass designed both.
can you send me a piccie of the heatsinks too?
i'm making a 600w per side (2 x 1KV toroidals) with the good old MJ21193/94's and need some dangerous heatsinkage, as they're gonna be in a rack case... and how much do you want for them?
(i need 14 devices per amp mounted, 28 overall, i was gonna use 2 heatsinks.)
(mail: matthew@mdtz.com)
i'm making a 600w per side (2 x 1KV toroidals) with the good old MJ21193/94's and need some dangerous heatsinkage, as they're gonna be in a rack case... and how much do you want for them?
(i need 14 devices per amp mounted, 28 overall, i was gonna use 2 heatsinks.)
(mail: matthew@mdtz.com)
DJK,
Cmon pics please!
I know Toroids are round but I want to see:
1) The condition
2) Quality of electrostatic screening/ finishing
3) Potted or not
4) Guage of the feeder cables (approximately) and the length.
5) In metal can or open.
6) What mounting hardware/ accessories.
7) Size dimensions.
I know you are a high IQ guy, but not everyone else is...
I have never seen the DH500, all the more reason to see pics. Also even though in the DIY workd aesthetics are not that paramount, but they don't stand for nada either.
Thanks for the tip on the schematic for the A40 and the source. 🙂
Do you have a schematic of your class-A+G design? I would be interested if so,
Also where do you live, what is the estimated shipping to IN 46038. (Indiana)
thanks again!
Cmon pics please!
I know Toroids are round but I want to see:
1) The condition
2) Quality of electrostatic screening/ finishing
3) Potted or not
4) Guage of the feeder cables (approximately) and the length.
5) In metal can or open.
6) What mounting hardware/ accessories.
7) Size dimensions.
I know you are a high IQ guy, but not everyone else is...
I have never seen the DH500, all the more reason to see pics. Also even though in the DIY workd aesthetics are not that paramount, but they don't stand for nada either.
Thanks for the tip on the schematic for the A40 and the source. 🙂
Do you have a schematic of your class-A+G design? I would be interested if so,
Also where do you live, what is the estimated shipping to IN 46038. (Indiana)
thanks again!
- Status
- Not open for further replies.