My F-5 has fairly large heat sinks. 12.25 by 8 by 2.3. They do a great job, the rest of the aluminum chassis helps as well. It is easy to hold on to these as long as you want.
I know that stressing the outputs can be detrimental to their reliability. Is the point at which damage occurs determined by going over the recommennded .59, or is it from the heat of the heat sink going to high? I realize one is connected to the other, but I'm wondering if it is ok to up the bias, as long as the heatsink can be touched for 5 seconds, or is it more to it than whether the heat sink is sufficient to keep itself with in the 5 second rule?
How high is too high, a particular number on the multimeter, or the actual heat of the sink? Where do most of you let yours run and how hot does your sink get?
Thanks,
Russellc
I know that stressing the outputs can be detrimental to their reliability. Is the point at which damage occurs determined by going over the recommennded .59, or is it from the heat of the heat sink going to high? I realize one is connected to the other, but I'm wondering if it is ok to up the bias, as long as the heatsink can be touched for 5 seconds, or is it more to it than whether the heat sink is sufficient to keep itself with in the 5 second rule?
How high is too high, a particular number on the multimeter, or the actual heat of the sink? Where do most of you let yours run and how hot does your sink get?
Thanks,
Russellc
The manufacturer tells us that the maximum Tj for long term reliability is somewhere in the range 125degC to 200degC.
Most plastic packaged output devices are either Tjmax=150degC or Tjmax=175degC
The SOA curves for Tc=25degC show how much current and voltage the device can take and for how long to just bring Tj = Tjmax.
The hotter you run Tc then the lower is the SOA (temperature de-rated SOAR).
If your sink at the contact face is 50degC then Tc~55degC to 65degC.
Draw your temp de-rated SOAR to find the allowable currents/voltages/durations.
raise the device dissipation (increase Ib) and Ts, Tc both rise. New set of de-rated SOAR for the new Tc.
Eventually as SOAR moves down, the output devices pass current for the duration that the speaker demands and you eat into the reliability reserve. Not instant failure, but likely to be a shortened lifetime.
Most plastic packaged output devices are either Tjmax=150degC or Tjmax=175degC
The SOA curves for Tc=25degC show how much current and voltage the device can take and for how long to just bring Tj = Tjmax.
The hotter you run Tc then the lower is the SOA (temperature de-rated SOAR).
If your sink at the contact face is 50degC then Tc~55degC to 65degC.
Draw your temp de-rated SOAR to find the allowable currents/voltages/durations.
raise the device dissipation (increase Ib) and Ts, Tc both rise. New set of de-rated SOAR for the new Tc.
Eventually as SOAR moves down, the output devices pass current for the duration that the speaker demands and you eat into the reliability reserve. Not instant failure, but likely to be a shortened lifetime.
Russellc, it doesn't matter how hot or cold your heat sinks, it does matter how high the Tj is. You can have big enough heat sinks and crappy thermal interface between them and transistor - heat sinks will be cold and transistors are hot.
I usually measure the temperature on transistor case (Tc) with IR thermometer then, using formula Tj = Tc + Pt * K, define junction temperature.
Pt - applied power
K - junction to case heat transfer - I'm using 1.0 for transistors like IRFP240
Acceptable junction temperature is less than 100C.
So, for example, if Tc = 60C and Pt=25W, junction temperature Tj will be 85C which are OK.
PS. Mica insulator with 0.05mm thickness will give you 8-10C difference in temperature between heat sink and transistor case.
I usually measure the temperature on transistor case (Tc) with IR thermometer then, using formula Tj = Tc + Pt * K, define junction temperature.
Pt - applied power
K - junction to case heat transfer - I'm using 1.0 for transistors like IRFP240
Acceptable junction temperature is less than 100C.
So, for example, if Tc = 60C and Pt=25W, junction temperature Tj will be 85C which are OK.
PS. Mica insulator with 0.05mm thickness will give you 8-10C difference in temperature between heat sink and transistor case.
Last edited:
Simple answer is NO, it is not OK. I understand why Mr. Pass introduced this rule of the thumb for touching heat sinks for DIYers but it is not applied for any heat sinking. For each situation you are better to calculate junction temperature to make sure it's acceptable.
Jeez, if I remember at what power the fets in the old Alephs were cooking in comparison.
Personally, I would not mind running the heatsinks at even 60 degrees or a little more.
In fact many commercial amps do that (and higher) if the heatsinks are internal and there's no risk of the owner to get burned.
Have fun, Hannes
PS: wouldn't hurt if you make sure first that the thermal pads are of high quality for good heat conduction of fet to sink.
Personally, I would not mind running the heatsinks at even 60 degrees or a little more.
In fact many commercial amps do that (and higher) if the heatsinks are internal and there's no risk of the owner to get burned.
Have fun, Hannes
PS: wouldn't hurt if you make sure first that the thermal pads are of high quality for good heat conduction of fet to sink.
I hadnt really thought about the juncture not sending the heat to the sink.
The surface was really smooth and the fit is tight. I am using silicone silpads now, and am running the IRF devices.
Sounds like the best bet is to get a thermometer and measure the actual output device temp. Then is seems from what everyone says that it would be safe to up it until the heat of the output device is maximal.
Thanks,
Russellc
The surface was really smooth and the fit is tight. I am using silicone silpads now, and am running the IRF devices.
Sounds like the best bet is to get a thermometer and measure the actual output device temp. Then is seems from what everyone says that it would be safe to up it until the heat of the output device is maximal.
Thanks,
Russellc
Thanks for the response, but it wasnt Mr pass saying this to DIYers that I am referring to, I believe I saw that in the F-5 owners manual, Speaking to his own commercial F-5s. Doesnt mean the rule holds true for my DIY effort based on what everyone's saying here. If my ouputs crap out, I have spares and several sets of Toshiba outputs. Next time I'll use them and the mica pads with thermal greas instead of silpads.
Westend, the member who cut the aluminum for my case and also used the same heat sinks in his effort, polished the surface to a almost mirror finish
and based on his description to my queries, runs no hotter or cooler than my effort. I need to go buy a suitable thermometer!
thanks again,
Russellc
Now that I've thought about it a minute, I still have the same question based on a different set of circumstances. If I am measuring the heat of the output device ITSELF, is it OK to up the bias as long as the heat of the device remains within recommendation?
Where is everyone running their F-5 in terms of bias?
Russellc
Where is everyone running their F-5 in terms of bias?
Russellc
I would avoid using them for high power applications. The best transfer rate you can get with them for TO-247 is about 1C/W. If you are running 25W through a transistor you will get 25C difference between transistor case and heat sink.
I don't think I would exceed 50W with even a cold sink.
Thank you for the response, much appreciated!
Russellc
How high the bias (current) depends on how much power you want to put into the device, and your rail voltage.
Current = Power / Voltage.
I guess we all agree up to this point.
How much power you want to put into one device depends very largely on its Rthjc, or thermal resistance junction to case. For a well dimensioned heatsink plus whatever insulator you might choose to use, the case temperature at the device case (i.e. the metal plate making up the drain for most MOSFETs) is not likely to exceed 70 degC. If you believe in what Bob Cordell says in his book, and I do tend to agree with him based on my semicon knowledge, then you can use the device up to 150 degC junction temperature continuously, even in industrial applications. If you want to be safe, then use 120 degC.
A good TO247 / TO3P device has a Rthjc of somewhere between 0.83 to 1 K/W. That convertes to 50-60W per device, assuming junction temperature of 120. The number from Nelson I consider to be very much on the safe side.
I have already submitted an article to Jan Didden's Linear Audio on Class A amplifier case design, especially on various aspects of thermal design, including comparison of insulators, ...., etc. I was told that unfortunately it would probably not make the second issue. So it will be a long wait I am afraid.
PS I am of course not implying that you should go and lobby Jan !!
Patrick
.
Current = Power / Voltage.
I guess we all agree up to this point.
How much power you want to put into one device depends very largely on its Rthjc, or thermal resistance junction to case. For a well dimensioned heatsink plus whatever insulator you might choose to use, the case temperature at the device case (i.e. the metal plate making up the drain for most MOSFETs) is not likely to exceed 70 degC. If you believe in what Bob Cordell says in his book, and I do tend to agree with him based on my semicon knowledge, then you can use the device up to 150 degC junction temperature continuously, even in industrial applications. If you want to be safe, then use 120 degC.
A good TO247 / TO3P device has a Rthjc of somewhere between 0.83 to 1 K/W. That convertes to 50-60W per device, assuming junction temperature of 120. The number from Nelson I consider to be very much on the safe side.
I have already submitted an article to Jan Didden's Linear Audio on Class A amplifier case design, especially on various aspects of thermal design, including comparison of insulators, ...., etc. I was told that unfortunately it would probably not make the second issue. So it will be a long wait I am afraid.
PS I am of course not implying that you should go and lobby Jan !!
Patrick
.
Last edited:
How high the bias (current) depends on how much power you want to put into the device, and your rail voltage.
Current = Power / Voltage.
I guess we all agree up to this point.
How much power you want to put into one device depends very largely on its Rthjc, or thermal resistance junction to case. For a well dimensioned heatsink plus whatever insulator you might choose to use, the case temperature at the device case (i.e. the metal plate making up the drain for most MOSFETs) is not likely to exceed 70 degC. If you believe in what Bob Cordell says in his book, and I do tend to agree with him based on my semicon knowledge, then you can use the device up to 150 degC junction temperature continuously, even in industrial applications. If you want to be safe, then use 120 degC.
A good TO247 / TO3P device has a Rthjc of somewhere between 0.83 to 1 K/W. That convertes to 50-60W per device, assuming junction temperature of 120. The number from Nelson I consider to be very much on the safe side.
I have already submitted an article to Jan Didden's Linear Audio on Class A amplifier case design, especially on various aspects of thermal design, including comparison of insulators, ...., etc. I was told that unfortunately it would probably not make the second issue. So it will be a long wait I am afraid.
PS I am of course not implying that you should go and lobby Jan !!
Patrick
.
Thank you for your response Patrick. Some of the responses contain excellent info, but yours pulled all of them together for me. Seems to me buying a good thermometer is a good idea.
Over on the AudioKarma site, one of the builders (Westend, who is also a member here,) informed me of temp measurements he made on his effort, which used the same sinks and case s mine. I'll see if I still have those messages.
Russellc
There was a very old discussion about mica & grease, silpad, Aluminium oxide, Kapton, & Kerafol.
http://www.diyaudio.com/forums/pass-labs/37262-mica-goop-2.html#post1032023
These different types will be analysed in detail in my article in Linear Audio. So I cannot publish here.
But the simple conclusion is that Kerafol lowers the temperature difference by a factor of 3 compared to mica, followed by aluminium oxide. These were dealt with in that thread with some simple mathematics.
Patrick
.
http://www.diyaudio.com/forums/pass-labs/37262-mica-goop-2.html#post1032023
These different types will be analysed in detail in my article in Linear Audio. So I cannot publish here.
But the simple conclusion is that Kerafol lowers the temperature difference by a factor of 3 compared to mica, followed by aluminium oxide. These were dealt with in that thread with some simple mathematics.
Patrick
.
Last edited:
Thanks again for the info on the high performance pads. i'll read up on you thread provided.
Much appreciated,
Russellc
Yeah the Kerafol-stuff is about the best thing one can get from Conrad here in Europe
I find it a bit surprising junction temperature is currently of so large interest here; given that most fets run so far away from their limits
Thread concerns "turning up the bias"......this brings them closer to and past their limits....thats why junction temp is being discussed.
Russellc
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.