Author Topic: current sharing on psu pass transistors  (Read 2481 times)

0 Members and 1 Guest are viewing this topic.

Offline m3vuvTopic starter

  • Super Contributor
  • ***
  • !
  • Posts: 1738
  • Country: gb
current sharing on psu pass transistors
« on: September 28, 2019, 02:37:52 pm »
hi all im building a psu that should be good for 0-30v 0-20a,i have 6x tip35 pass transistors on a common heatsink insulated,im using 5 watt 0.2 ohm emitter resistors ,it works ok but one of the pass transistors gets  way hot,the others are luke warm,its running 60 watts of bulbs at 12v,any ideas?
 

Offline amspire

  • Super Contributor
  • ***
  • Posts: 3802
  • Country: au
Re: current sharing on psu pass transistors
« Reply #1 on: September 28, 2019, 02:47:56 pm »
0.2 ohms should share the current well enough, so something seems wrong.

It just seems that the hot transistor just does not have a 0.2 ohm emitter resistor. Have you measured the voltages across each 0.2 ohm resistor?
 

Offline drussell

  • Super Contributor
  • ***
  • Posts: 1855
  • Country: ca
  • Hardcore Geek
Re: current sharing on psu pass transistors
« Reply #2 on: September 28, 2019, 02:48:28 pm »
Measure the current flowing through each by measuring the voltage drop across each of the emitter resistors. 

Are they all actually conducting the same current?  If so, then you must have an issue with the thermal connection of the one transistor to the heatsink.  (It could be an internally defective transistor, I suppose.)  ??

Edit:  Seems amspire beat me to the same suggestion by 30 seconds.  Kudos!  :)

If they're not conducting the same current, double check the actual value of your emitter resistors and also let us know how you are driving the bases.  Also double check that there is not some unexpected resistance in the collector connections to the source of the power.
« Last Edit: September 28, 2019, 03:01:28 pm by drussell »
 

Offline Audioguru again

  • Regular Contributor
  • *
  • Posts: 115
  • Country: ca
Re: current sharing on psu pass transistors
« Reply #3 on: September 28, 2019, 03:39:35 pm »
Measure the current flowing through each by measuring the voltage drop across each of the emitter resistors. 

Are they all actually conducting the same current?  If so, then you must have an issue with the thermal connection of the one transistor to the heatsink.  (It could be an internally defective transistor, I suppose.)  ??
Many cheap fake Chinese transistors are like that.
 

Offline drussell

  • Super Contributor
  • ***
  • Posts: 1855
  • Country: ca
  • Hardcore Geek
Re: current sharing on psu pass transistors
« Reply #4 on: September 28, 2019, 03:52:26 pm »
... (It could be an internally defective transistor, I suppose.)
Many cheap fake Chinese transistors are like that.

Indeed!  That was why I pointed that out in the edit. 

If everything else really does seem to be correct but one transistor behaves poorly, it certainly can be a horribly made defective transistor, especially these days, especially with commonly cloned parts like TIP- series or 3055s or whatever....  This was much less of a problem "back in the day" but sadly is very high on the suspect list of late.
 

Offline Kleinstein

  • Super Contributor
  • ***
  • Posts: 14067
  • Country: de
Re: current sharing on psu pass transistors
« Reply #5 on: September 28, 2019, 04:04:06 pm »
A little over 3 A give quite a lot of power per transistor, especially if at worst case voltage (e.g. 35 V). The 0.2 Ohms are about sufficient for current sharing, but there are still tolerances so that one of the transistors may see some 4 A.

It may still be acceptable if good transformer tap switching is used. Otherwise I would suggest reducing the current per transistor to some 2 A.

Compared to 0.2 Ohms the wiring can also add some resistance and thus effect the current sharing if in a poor way.
 

Offline duak

  • Super Contributor
  • ***
  • Posts: 1038
  • Country: ca
Re: current sharing on psu pass transistors
« Reply #6 on: September 28, 2019, 04:32:46 pm »
Heatsinks tend to have a temperature gradient from the bottom to the top where the temperature is highest.  Are the devices arranged horizontally or vertically?

Are you using a nylon shoulder washer to insulate the mounting screw from the collector tab?  If so, it should be replaced with a glass-filled nylon shoulder washer or better.  Simple Nylon is not up to the task because it creeps at temperature and you lose clamping force leading to thermal runaway.  A better way to clamp the devices is with a springy bracket pressing down on the case over the die.

Please provide a picture of the heatsink.  If you can't, then a description will help.

If all else fails, It should be possible to reduce the temperature of an individiual device by putting some resistance in series with its base connection.  This is easier than fiddling with the value of its emitter resistor.
 

Offline m3vuvTopic starter

  • Super Contributor
  • ***
  • !
  • Posts: 1738
  • Country: gb
Re: current sharing on psu pass transistors
« Reply #7 on: September 28, 2019, 11:47:42 pm »
well the transistor that was getting hot went sc e/c,ive removed it,the voltage  accross the balast resisters are readings are  tr1/173mv tr2/175.9mv tr3/178.mv tr4/158mv tr5/180.6mv,the heatsink is black anodised aluminium and measures 150mm x 150mm x 30 deep,all transistors are mounted on silpads with dow corning 340 thermal compound and bur free etc.i assume it was a faulty transistor.
 

Offline dom0

  • Super Contributor
  • ***
  • Posts: 1483
  • Country: 00
Re: current sharing on psu pass transistors
« Reply #8 on: September 29, 2019, 12:23:34 pm »
Compared to 0.2 Ohms the wiring can also add some resistance and thus effect the current sharing if in a poor way.

That's why for many transistors in parallel the C and E connections should be mirrored, i.e.

,
 

Offline Kleinstein

  • Super Contributor
  • ***
  • Posts: 14067
  • Country: de
Re: current sharing on psu pass transistors
« Reply #9 on: September 29, 2019, 02:13:37 pm »
Compared to 0.2 Ohms the wiring can also add some resistance and thus effect the current sharing if in a poor way.

That's why for many transistors in parallel the C and E connections should be mirrored, i.e.

(Attachment Link)
That type of connection is good for switching, but not for analog control. In analog control the collector resistance has very little effect, but emitter side wiring drop would add up.  It should be more like the emitter resistors meeting at one point.
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf