Announcement - Double adjustable power supplies

Thread Starter

bloguetronica

Joined Apr 27, 2007
1,541
I'm here to announce that these projects have errors, unnoticed until now. By the time, I considered the junction to case thermal resistance (TO-220) to be 0.3ºC/W instead of 5ºC/W. This means that, even considering an infinite heatsink, the temperature of the junction will rise above 125ºC, and the heatsink temperature will not rise above 70ºC, seeming to work fine (I normally consider that Tcase < 70ºC and Tjunction < 110ºC for an ambient temperature of 40ºC. This also give me margin to work at Tamb = 50ºC, since with an increase of 10ºC, the temperatures will still be acceptable.).

All other projects requiring heatsinks are not affected by this error, due to the nature the following considerations that I normally use:
Heatsinks are calculated for Tcase < 70ºC as long as Tjunction doesn't go higher than 110ºC. The formula is:
Rs-a= (Tc - Ta) / q - Rc-s

If Tjunction does go higher than 110ºC, then:
1- I consider Tjunction to be the limiting factor, so it should be < 110ºC (this happens with higher wattages, and in those cases, Tcase might be well below 70ºC). The formula is:
Rs-a= (Tj - Ta) / q - Rj-c - Rc-s;
2- I reduce the dissipated power when step 1 is impossible.

So this means that for lower wattages, any similar error around junction to case resistance does not affect the calculations. That's why all the other projects were unaffected. The solution here is to reduce the dissipated power of the regulators, with the implied redesign.
 

Dave

Joined Nov 17, 2003
6,969
cumesoftware,

Thanks for the clarification.

Would you like me to copy this into the "Double adjustable power supplies" thread as a record of this information?

Dave
 

Thread Starter

bloguetronica

Joined Apr 27, 2007
1,541
cumesoftware,

Thanks for the clarification.

Would you like me to copy this into the "Double adjustable power supplies" thread as a record of this information?

Dave
Yes. This is a really severe error and the thread should be copied.
 
Top