I'm not sure where to post this being that I do not post here frequently, but I'm interested in building a class A design with 30-60w using a single output pair of transistors in a TO-3 package. Here is my post from the SS forum on this site.
"Hello,
I'm looking to build a similar version of "Le Monstre" seen here.
http://diyaudioprojects.com/Solid/Je...s-A-Amplifier/
I scored some really nice looking TO-3 heatsinks, and am looking for a transistor that will be an effective replacement in a TO-3 package. I'd like to get up to the 30-60w range. I was looking at the MJ15003/MJ15004 combination. How do these sound in comparison? Any suggestions for a drop in TO-3 replacement for the original 2SC5200?"
Thanks for any suggestions/assistance!!
P.S. I'm not partial to this design either. If anyone has a nice design that fits my single pair criteria, I am more than accepting to new ideas.
Blair
"Hello,
I'm looking to build a similar version of "Le Monstre" seen here.
http://diyaudioprojects.com/Solid/Je...s-A-Amplifier/
I scored some really nice looking TO-3 heatsinks, and am looking for a transistor that will be an effective replacement in a TO-3 package. I'd like to get up to the 30-60w range. I was looking at the MJ15003/MJ15004 combination. How do these sound in comparison? Any suggestions for a drop in TO-3 replacement for the original 2SC5200?"
Thanks for any suggestions/assistance!!
P.S. I'm not partial to this design either. If anyone has a nice design that fits my single pair criteria, I am more than accepting to new ideas.
Blair
Having in mind your goal (30-60W class A from single pair of TO-3 devices) it might be good to move this thread to "Solid State" section.
A single pair of 250W devices might just meet the 30W of ClassA target for a push pull amplifier.
There is no way you can achieve anywhere near 60W of ClassA reliably from two 250W devices.
eg, Krell use 4 off 250W devices for 50W of ClassA and 8off 250W devices for 100W of ClassA.
The MJ15003/4 is a good basis for a ClassA output stage
There is no way you can achieve anywhere near 60W of ClassA reliably from two 250W devices.
eg, Krell use 4 off 250W devices for 50W of ClassA and 8off 250W devices for 100W of ClassA.
The MJ15003/4 is a good basis for a ClassA output stage
there may be some assumptions not being spelled out above:
push-pull Class A can be 50% efficient at full power into its rated resistive load - output devices dissipate equal power as load
really, really good practical heat sinking can allow up to ~1/2 the Tcase 25 power - it is almost always cheaper to use more devices in parallel on cheaper heatsinks than to buy a big enough heatsink/Cu spreader/fans to do this
with these 2 conditions it appears 120 W Class A into 8 Ohms is possible with "250 W" rated devices in a push-pull pair (250 Wpk ~= 120 Wrms)
loudspeakers can have load impedance dips due to the crossover and driver choices - a "well behaved" loudspeaker shouldn't dip to less than 1/2 its nominal impedance rating
if you add enough bias to keep the same clipping V into 1/2 the load impedance you have to double the current
now we're at ~60 W Class A into "polite" loudspeakers
Krell and other amps designed for more extreme audiophile speakers may double their current again to drive 2 Ohm dips in exotic "8 Ohm" loudspeakers - rather insane to do with Class A as you'd be heating the room with 8X the output rms power rating but the 30 W rating could be managed even in this scenario
if you biAmp then you can expect driver terminal impedance will not have any large dip and not have to allow for such excesses of bias current
another option is to bias the amp for Class A into the nominal load and allow for Class AB operation into load dips
push-pull Class A can be 50% efficient at full power into its rated resistive load - output devices dissipate equal power as load
really, really good practical heat sinking can allow up to ~1/2 the Tcase 25 power - it is almost always cheaper to use more devices in parallel on cheaper heatsinks than to buy a big enough heatsink/Cu spreader/fans to do this
with these 2 conditions it appears 120 W Class A into 8 Ohms is possible with "250 W" rated devices in a push-pull pair (250 Wpk ~= 120 Wrms)
loudspeakers can have load impedance dips due to the crossover and driver choices - a "well behaved" loudspeaker shouldn't dip to less than 1/2 its nominal impedance rating
if you add enough bias to keep the same clipping V into 1/2 the load impedance you have to double the current
now we're at ~60 W Class A into "polite" loudspeakers
Krell and other amps designed for more extreme audiophile speakers may double their current again to drive 2 Ohm dips in exotic "8 Ohm" loudspeakers - rather insane to do with Class A as you'd be heating the room with 8X the output rms power rating but the 30 W rating could be managed even in this scenario
if you biAmp then you can expect driver terminal impedance will not have any large dip and not have to allow for such excesses of bias current
another option is to bias the amp for Class A into the nominal load and allow for Class AB operation into load dips
Last edited:
oops - the bias current math isn't exactly right - the bias current would be increased 1.4X and supply V would be dropped to ~70% to supply the max power into the 1/2 nominal load impedance dip - reducing the "name plate" amplifier power rating into the nominal resistive load by 1/2 as stated
essentially we're designing the bias, supply V for max power Class A at the impedance dip and accepting the (supply V limited) reduced power into the nominal load as the new "name plate " rating - with the power/current reserve to double the Class A power into the 1/2 nominal impedance dip (or 4x power for 1/4 load dip)
essentially we're designing the bias, supply V for max power Class A at the impedance dip and accepting the (supply V limited) reduced power into the nominal load as the new "name plate " rating - with the power/current reserve to double the Class A power into the 1/2 nominal impedance dip (or 4x power for 1/4 load dip)
Last edited:
Hi,
you won't get 50% efficiency for a practical single pair Push/Pull output stage.
I reckon the best you can achieve economically is ~43% Max output power to quiescent dissipation.
50W into 8r0 in ClassA from +-33.5Vdc loaded supply rails is just about achievable. But the devices must run @ Tc<=75degC.
This would require ~0.1C/W/device sink running @ Ts~45degC and delta Ts-a ~12Cdegrees. Ta~30degC
This ClassA output stage could go to ~85W into 4r0 in ClassAB mode.
I would rather go to 2pair, alternatively reduce the target to 40W of maximum ClassA power (this allows Tc<=99degC and much smaller heatsink).
you won't get 50% efficiency for a practical single pair Push/Pull output stage.
I reckon the best you can achieve economically is ~43% Max output power to quiescent dissipation.
50W into 8r0 in ClassA from +-33.5Vdc loaded supply rails is just about achievable. But the devices must run @ Tc<=75degC.
This would require ~0.1C/W/device sink running @ Ts~45degC and delta Ts-a ~12Cdegrees. Ta~30degC
This ClassA output stage could go to ~85W into 4r0 in ClassAB mode.
I would rather go to 2pair, alternatively reduce the target to 40W of maximum ClassA power (this allows Tc<=99degC and much smaller heatsink).
Last edited:
by starting with theoretical limits and then considering how they're modified by real world limitations we can address each one as it affects a particular instance
as you point out efficiency will be reduced - by allowance for supply ripple and device saturation or driver limitations we can modify the theoretical 50% limit for push-pull Class A
and I agree that economic heatsinking leads to more output devices
as you point out efficiency will be reduced - by allowance for supply ripple and device saturation or driver limitations we can modify the theoretical 50% limit for push-pull Class A
and I agree that economic heatsinking leads to more output devices
Build an F5. It doesn't quite meet your criteria; with your desired devices and output, but it's the way to go for sure, and it sounds great!
Thanks everyone for the replies! Someone pointed me towards the JLH higher power design that puts out 28w into 8 ohms and 56w into 4 ohms on 22v rails. It fits the criteria, and is simple enough for my first SS project. It uses 4 MJ15003 transistors which will fit the TO-3 requirement, and I'm probably going to purchase some Asian knockoff boards for the driver circuit. Each transistor will have it's own designated 3.5" x 11" x 1.25" heatsink.
Can anyone reccommend the proper vA transformer for monoblocks? Is 300vA sufficient per channel to eliminate any sag?
Thank you for any assistance!!
Blair
Can anyone reccommend the proper vA transformer for monoblocks? Is 300vA sufficient per channel to eliminate any sag?
Thank you for any assistance!!
Blair
- Status
- Not open for further replies.