Mosfets are no different from BJTs when it comes to dissipation.
At 50% of the SOA a 150 watt device is only good for 75 watts.
The output devices of an output stage capable of 150 watts in 4 Ohm should therefore have a total power figure of 300 watts minimum.
20V/6A= 3.33 Ohms. Continuous power in 8 Ohms should be close to 75 watts, two 150 watt devices would be sufficient.
But in the words of Nelson Pass: heatsinks are the most important parts of amplifiers.
At 50% of the SOA a 150 watt device is only good for 75 watts.
The output devices of an output stage capable of 150 watts in 4 Ohm should therefore have a total power figure of 300 watts minimum.
20V/6A= 3.33 Ohms. Continuous power in 8 Ohms should be close to 75 watts, two 150 watt devices would be sufficient.
But in the words of Nelson Pass: heatsinks are the most important parts of amplifiers.
It's not as simple as giving some percentage value. You need to do a proper SOA analysis to make an informed judgement.
As an example, here's the quick and dirty thermal analysis I did for my amplifier. The power supply is +/-56V, output maximum power is 100W, and it uses two Hitachi 2SK1058 and 2SJ162 pairs, with thermal transfer pads to a Conrad 300mmx75mmx48mm heatsink:
Rth (transistor) = 1.25 deg/watt
mounting pad = 0.2 deg/watt
Rth (heatsink) = 0.37 deg/watt
Max Temp (transistor) = 150 deg
Dissipation at 50W out = 76.8W (from simulation - 1/2 power point is generally worst case, but it's a good idea to do a few simulations to find the worst case)
Ambient = 25 deg
Temp (transistor) = Ambient + dissipation / thermal resistance
Temp (transistor) = 25 + 76.8 / (((1.25+0.2)/4)+0.37)
Temp (transistor) = 130 degrees.
My design works reasonably comfortably with two pairs of output transistors and a fairly hefty heatsink.
Of course, you should also check that you won't exceed current and voltage ratings for your devices, as well.
Hope that's been of some help.
Cheers,
Suzy
As an example, here's the quick and dirty thermal analysis I did for my amplifier. The power supply is +/-56V, output maximum power is 100W, and it uses two Hitachi 2SK1058 and 2SJ162 pairs, with thermal transfer pads to a Conrad 300mmx75mmx48mm heatsink:
Rth (transistor) = 1.25 deg/watt
mounting pad = 0.2 deg/watt
Rth (heatsink) = 0.37 deg/watt
Max Temp (transistor) = 150 deg
Dissipation at 50W out = 76.8W (from simulation - 1/2 power point is generally worst case, but it's a good idea to do a few simulations to find the worst case)
Ambient = 25 deg
Temp (transistor) = Ambient + dissipation / thermal resistance
Temp (transistor) = 25 + 76.8 / (((1.25+0.2)/4)+0.37)
Temp (transistor) = 130 degrees.
My design works reasonably comfortably with two pairs of output transistors and a fairly hefty heatsink.
Of course, you should also check that you won't exceed current and voltage ratings for your devices, as well.
Hope that's been of some help.
Cheers,
Suzy
Miss Jackson,
your quick calculation is a bit off.
At 76.8 watts dissipation divided over 2 pairs of your Hitachis, the die temperature is 81 C. (with the assumed data you posted)
At a transistor temperature of 105 C above ambient dissipation is 143 watts.
Max dissipation will be a lot higher if you hook your 100 watt in 8 to a 4 Ohm loudspeaker.
your quick calculation is a bit off.
At 76.8 watts dissipation divided over 2 pairs of your Hitachis, the die temperature is 81 C. (with the assumed data you posted)
At a transistor temperature of 105 C above ambient dissipation is 143 watts.
Max dissipation will be a lot higher if you hook your 100 watt in 8 to a 4 Ohm loudspeaker.
Hi,
you are asking a lot from 1pair of 150W FETs to drive a 3r3 load to 120W. To then ask the same amp to cope with the reactive impedance of a real speaker is going too far, in my opinion.
Compare the SKA GB150, it uses 2pair of 150W devices in an efficient output stage configuration to achieve 150W into 4ohm. Ampguru recommends 0.9C/W sink.
If you keep the FETs cool with a BIG sink (0.5C/W) it may survive on gentle speaker loads. No guarantees, but then my software seems to be pessimistic.
you are asking a lot from 1pair of 150W FETs to drive a 3r3 load to 120W. To then ask the same amp to cope with the reactive impedance of a real speaker is going too far, in my opinion.
Compare the SKA GB150, it uses 2pair of 150W devices in an efficient output stage configuration to achieve 150W into 4ohm. Ampguru recommends 0.9C/W sink.
If you keep the FETs cool with a BIG sink (0.5C/W) it may survive on gentle speaker loads. No guarantees, but then my software seems to be pessimistic.
You're quite right - Greg's GB150D appears a lot more capable with 4 x 150W devices and the modern TO3P package allowing easy direct connection to the conrad, for better thermal transfer.
Then there's the lower dissipation common source MOSFETs. You really don't see the real value of a great design until it's finer points are compared with others.
I note Suzy uses a transfer heatsink not included in her thermal calculations -that's gotta be another 1 deg/watt and another 19 degC in die temp.
Also there's probably another 5deg C in the fact that the conrads 0.37 deg/W is for an 80 deg rise not 56 , but the transfer heatsink probably corrects that.
Good thing music isn't static power.
Then there's the lower dissipation common source MOSFETs. You really don't see the real value of a great design until it's finer points are compared with others.
I note Suzy uses a transfer heatsink not included in her thermal calculations -that's gotta be another 1 deg/watt and another 19 degC in die temp.
Also there's probably another 5deg C in the fact that the conrads 0.37 deg/W is for an 80 deg rise not 56 , but the transfer heatsink probably corrects that.
Good thing music isn't static power.
Sorry mate,
but Mr Ball's GB150 is 150 watts in 8 Ohms, not in 4 !
If the GB150 does much less than 240 watts in 4 Ohms it wouldn't be very impressive. The GB150 is likely to get in reach of 275 watts in a 3.3 Ohms load.
Long time ago i've constructed a 100 watt/8 design with 4 Mosfets in the output stage, the TO220 types as used in the Pass/Thagard A75 class A amplifier. Those have only half the max power of the TO247 IRFPs.
The mono amplifiers had no trouble driving QUAD ESL63s whatsoever, 90 degree phaseshift reactive enough ?(plenty heatsinking and higher voltage front end rails up to the drivers, admitted)
With doubled number of output devices they run for years as a 50 watt class A duo. (15 watts continuous dissipation per device)
Total power capability to continuous output power in 8 Ohms ratio of the GB150 is 4 to 1. That used to be the classic ratio for all the Hitachi output mosfets class AB designs ranging from Perreauxs to the Sphinx amplifiers by the Silltech cable guys overhere.
Miss jackson is right to mention that a solid thermal calculation is appropiate, just the power figure without a heatsink doesn't add up to much. Max dissipation occurs at 1/2 output voltage though, not half output power. I'd like some of those TO247 thermal pads that can do 0.2 C/W.
but Mr Ball's GB150 is 150 watts in 8 Ohms, not in 4 !
If the GB150 does much less than 240 watts in 4 Ohms it wouldn't be very impressive. The GB150 is likely to get in reach of 275 watts in a 3.3 Ohms load.
Long time ago i've constructed a 100 watt/8 design with 4 Mosfets in the output stage, the TO220 types as used in the Pass/Thagard A75 class A amplifier. Those have only half the max power of the TO247 IRFPs.
The mono amplifiers had no trouble driving QUAD ESL63s whatsoever, 90 degree phaseshift reactive enough ?(plenty heatsinking and higher voltage front end rails up to the drivers, admitted)
With doubled number of output devices they run for years as a 50 watt class A duo. (15 watts continuous dissipation per device)
Total power capability to continuous output power in 8 Ohms ratio of the GB150 is 4 to 1. That used to be the classic ratio for all the Hitachi output mosfets class AB designs ranging from Perreauxs to the Sphinx amplifiers by the Silltech cable guys overhere.
Miss jackson is right to mention that a solid thermal calculation is appropiate, just the power figure without a heatsink doesn't add up to much. Max dissipation occurs at 1/2 output voltage though, not half output power. I'd like some of those TO247 thermal pads that can do 0.2 C/W.
Hi Jacco,
that is not what Ampguru says.
He recommends reducing the transformer voltage significantly for driving 4ohm speakers.
He is effectively claiming 150W into 8ohms speakers OR 150W into 4ohm speakers. Although there is mention of 200W ability for 4ohms but no confirmation of resistive or reactive loading.
that is not what Ampguru says.
He recommends reducing the transformer voltage significantly for driving 4ohm speakers.
He is effectively claiming 150W into 8ohms speakers OR 150W into 4ohm speakers. Although there is mention of 200W ability for 4ohms but no confirmation of resistive or reactive loading.
I believe Mr Ball makes a distinction between 4 ohm use and heavy 4 ohm use. The GB150 design and the kind of output devices aren't fit for very low impedance use to begin with.
Now, take for instance my personal all-time No1 horrible amplifier, the 1982 Crescendo. That design did 140 watts continuous in 8 Ohms and 250 watts peak output in 4 Ohms on four 100-watt Hitachi output devices only. Prescribed heatsink was a 6" SK53 profile.
The Creshitty ran on 70-75 Vdc rails, but if one could keep it from oscillating it was virtually impossible to kill, not even with rotten loads. Wish it did, sounded horrible despite its large bandwidth. (imo).
(Mosfets and vertical ones in particular are only good for Class A output stages)
Now, take for instance my personal all-time No1 horrible amplifier, the 1982 Crescendo. That design did 140 watts continuous in 8 Ohms and 250 watts peak output in 4 Ohms on four 100-watt Hitachi output devices only. Prescribed heatsink was a 6" SK53 profile.
The Creshitty ran on 70-75 Vdc rails, but if one could keep it from oscillating it was virtually impossible to kill, not even with rotten loads. Wish it did, sounded horrible despite its large bandwidth. (imo).
(Mosfets and vertical ones in particular are only good for Class A output stages)
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.