0-30 Vdc Stabilized Power Supply

[redwire]

U2 Drives the output transistors.    With a minimum gain of 1,000 for the MJ11016, a 3A output will require a  3/1000 = 3 milliamp current from the BD139.  The BD139 has a minimum gain of about 40, so 3milliamps /40=  nothing.
The C-E saturation voltage for the MJ11016  is 200mA at 20Amps,  so for max output the BD139 would have to provide 200mv.  With a gain of 40, U2 would only  need to provide 5 milliamps.  I think the TLE2141ID can provide at least 20ma.

 

[liquibyte]

Redwire, care to share pics of your rig that you built with the microwave transformer?  I'm curious about how you modified that transformer to output the 50+V @ 7A.  Did you replace the primary or the secondary side?  I've seen an article about doing this but the guy was calling the secondary side the primary.  He replaced the primary and turned the secondary into the primary and wound insulated stranded wire to produce a new secondary.  I'm curious if that's what you did.

 

[redwire]

liquibyte ,  I'm searching for some pictures of the final build but I haven't found any yet.  I took a MOT from a large microwave (1100w) that I found at the dump. I removed only the secondary windings because the primary seemed large enough to handle the load.  I purchased a thick copper tape that was approximately the width of  transformer opening with conductive adhesive on one side and carefully doubled the thickness of the tape.  I wasted a bit of tape until I got the hang of sticking both sides together.  I wound the transformer using mylar tape ? between the layers of copper tape until it was full.  I ordered the mylar tape slightly larger than the copper and may have had to trim the mylar a bit to fit the opening with.      I can get  8 A at 12V but not at 50V.  I don't recall the thickness(copper wt) of the tape or length but my intent was to have the cross sectional area equivalent to a 16 - 14g wire when doubled up. 

 

[liquibyte]

I seem to remember reading that you used copper tape.

I also have an 1100W transformer but all the tuts I've seen have the author removing the thinner wire windings.  Correct me if I'm wrong on this, but isn't the thicker wire the secondary and the thinner wire the primary or are they reversed in microwave transformers?  The one I took apart yesterday had the thinner stuff hooked up to the ac and the thick stuff hooked up to the circuitry.  I can count 8x11 windings on the bottom thicker wire which appears to be 12 or 13awg.  I'll probably use standard wire to see what I can get but I don't want to remove the wrong side and have to rewind both.  Considering that this thing has 4 welds on each side, I don't see this as feasable.

I'd like to do the higher current version as a one off supply but am also planning on the 3 amp version x 2 for a positive and negative serial circuit.  I've found some transformers on Digikey that I think will work wired in serial at 28V @ 4.6A.  The 28V @ 4.2A thing just doesn't seem to exist as a common off the shelf option.  These were actually fairly cheap as a laminated option.  I'd love to do these with toroids but, price aside, I think they'd need soft start circuits to be stable.

I plan on doing a schematic in Eagle and then a tutorial as I build later on after I accumulate all the parts.  Looks like a fun project and I may end up with several supplies.  I appreciate the work that you and audioguru have put into this in correcting the original fiasco.

 

[redwire]

On a MOT the secondary windings are very small wires with many windings  because it steps up the voltage  to thousands of volts.  Remove these very small wires and rewind these.    This should be consistent with many of the tutorials.  The problem with using standard wire is that the insulation is thick and you will not get as many windings resulting in a lower voltage.  Wires for  transformer typically have a thin lacquer or enamel coating to allow tight, compact placement.

 

[liquibyte]

This is the tutorial I was looking at.  According to this guy, he got 18V out of this technique: http://www.users.on.net/~endsodds/psrewind.htm

According to the table at: http://ludens.cl/Electron/trafos/WireTable.txt

AWG  Diam    DiaTot  AreaCu  AreaSQ    I        Ohm/m    m/kg
13     1.828    1.88    2.62       3.53      7.40   .00670    43.2

I should be able to get 7A out of this.

I've also been looking at http://ludens.cl/Electron/Magnet.html in regards to just making my own from scratch.  I just need to find a supplier of the EI cores and bulk wire locally.  I think this would solve quite a few problems when a custom winding is needed.  I'm not sure I'd enjoy winding toroids but might try in the future.

I was looking to purchase two of these and use them for two 3A units: http://www.digikey.com/product-detail/en/VPS28-4600/237-1281-ND/666167?itemSeq=141856235&uq=635238377888372307 .  These are the cheapest I could find that won't break the bank.

 

[redwire]

audioguru,      Finally got everything tied together for a quick test of the smd board.  I ran a 3156 taillight bulb (about 2 A) for around 4 minutes.  No heat sink on the bridge diode or BD139.    The heat sink on the MJ11016 output transistor was quite warm,  the bridge rectifier diode was very warm and the BD139 was cool.  The hottest part seemed to be the 1.5K, 1/2 W transistor for the power-on led I added.  Got a few more tests to run but it looks good so far.  U2 was warm.

View attachment 41988

 

[redwire]

One more picture of the rig operating

View attachment 41989

 

[liquibyte]

That's pretty cool.  How's it performing?  Did you outsource the board or do it yourself?

I also wanted to ask about the TO-3's and heatsink performance.  I have a really big case that I want to put two of these in and was wondering if a 7-9in x 3-4in would be enough for four devices at ambient or would active cooling be a better idea if say you were at 5V and 3A?

 

[audioguru2]

I hope that the surface-mount TL2141 does not exceed its maximum allowed temperature when the ambient temperature is high, the output current is high, the voltage is set low or a shorted output, the quiescent current of the TLE2141 is high and the hFE of the driver and output transistors are low.

 

[redwire]

liquibyte,  I had the boards made by a commercial pcb shop.  On my other PS,  I have (2) 2N5886G's on a aluminum heatsink about 4" x 7" x 1" with a small computer fan blowing across it.  It is amazing how effective a small amount of air blowing over the heat sink is.  I would not recommend containing the output transistors inside of the case without a fan.  Others have mounted the transistors to the outside back of the unit.

More tests tomorrow with low output, high current.

 

[liquibyte]

The case I have is 100% aluminum and rather large, being about 5x10x11 and I was thinking of hanging the heatsink off the back like you see in the older linear supplies.  I think maybe a temp controlled fan circuit may be in order so I don't have to listen to one all the time.  I had planned on using a third supply in there as well to power a couple of displays for voltage and current.  I just need to work out a decent circut using the MAX7219 drivers and a decent voltage ref.

I thought maybe you had the boards fabbed because of the silkscreen but you never know today.  I've seen some really professional looking boards done at home.  Either way, nice looking work there.

 

[redwire]

I hope that the surface-mount TL2141 does not exceed its maximum allowed temperature when the ambient temperature is high, the output current is high, the voltage is set low or a shorted output, the quiescent current of the TLE2141 is high and the hFE of the driver and output transistors are low.

I ran a test with  a 0.47 10W resistor at about 1.2 V.  1.2/0.47 = 2.6 A .  I only have it attached to a 100VA, 28V transformer so I guess it was about pegged out because the current control led did not come on.  After a while, the heat sink on the one output transistor was very hot to the touch.  U2 was warm. The BD139 was cool.  U1 and U3 were warm.  I then turned the voltage to 0.5 and didn't notice any significant  change from the previous test.

 

[liquibyte]

Would it be possible to replace the BD139 with a 2N5886 or equivalent TO-3 package?  I understand why the venerable 2N3055 was used but not a heavier version of the BD139 in the darlington.

 

[audioguru2]

The BD139 is a fast driver transistor. The 2N5886 that you want to use is a slow power transistor.
The driver transistor needs to be fast to avoid overshoots or oscillation when the output current changes abruptly.

 

[liquibyte]

Is there an equivalent in TO-3 that has these characteristics that you know of?  It's not that big a deal, I just thought that it would be nice to mount all of them on an external heatsink and the TO-3 package seems to be the best way to go as far as thermals are concerned.

 

[redwire]

For my smd version I was looking for a smd package replacement for the BD139 but it was difficult to determine how fast the BD139 switches from the data sheets to find a suitable substitute.   What are your thoughts on the STN851  transistor?    STN851_transistor.pdf

 

[Kevin Weddle]

I have this other 1-25volt variable power supply that may compare to this design. This one is great by design because the change in output voltage is inverted at the opamp input. The 1-25volt supply has higher gain.

 

[audioguru2]

The transition frequency for the STN851 transistor is 130MHz, it is 190MHz for the BD139 and it is 250MHz for the original 2N2219 transistor. The STN851 will work fine.

 

[redwire]

I have this other 1-25volt variable power supply that may compare to this design. This one is great by design because the change in output voltage is inverted at the opamp input. The 1-25volt supply has higher gain.

Was your goal to provide a easy to build, basic PS or will this be part of a more comprehensive design that may include current control?  I would have a concern about having the output shorted.