High power transistor

[CJay]

Honestly, I would skip experimenting with transistors and 78xx regulators, just go straight to the LM723 and experiment with the transistors there, it offers so much more and is RF quiet/immune.

Grab a copy of the PW Marchwood PSU articles, there are many versions kicking around the 'net and I think Warrington Amateur Radio CLub hosts a known good version on their website.

[davelectronic]

Cheers Cjay ! I will certainly have a look, I've  been over it once, but that was a while back. I came across a blinding enclosure that would be ace for such a project.

[David Hess]

Honestly, I would skip experimenting with transistors and 78xx regulators, just go straight to the LM723 and experiment with the transistors there, it offers so much more and is RF quiet/immune.

I might not simply because the 723 is more complex to use.

When I hacked together a quick 10+ amp power supply for a mobile VHF/UHF transceiver, I used an LM337, 2N3055, and an operational amplifier to control the variable speed cooling fan. 

[davelectronic]

12 Volt 20 Amp Power Supply http://www.zen22142.zen.co.uk/Circuits/Power/1220.htm   

I'm  not done with the 3 terminal regulator yet, repairing a 3 to 5 Amp Altai ebay buy with the LM317 and a MJ2955 first.

But what do you make of the circuit in the link above ? Its been a good source before now. Thoughts anyone !

[Seekonk]

If your transformer is a little shy, people often supply the driver with a little booster supply. That is preferable to extra heat in the output transistors.  A 5V wall wart sitting on top would work well.

[2N3055]

Well that is one of those 723 designs with paralleled transistors we all spoke about...
Seems OK on a first look, also there a many more like this, with some better protection circuits..
I would definitely try to find one with foldback current limiter, when so much watts are available...
Also, some people find it nice to have overvoltage (crowbar) circuit, to protect equipment if PSU blows transistors and apply full rectified voltage to the output...

http://www.americanradiohistory.com/Archive-Practical/Wireless/80s/PW%201983%2006.pdf
http://www.americanradiohistory.com/Archive-Practical/Wireless/80s/PW%201983%2007.pdf

Take a peek here, it is a very good article that goes into practical design, cooling and other...

[David Hess]

The Astron 723 designs which I linked earlier are built that way.  They include a bias supply for the 723 so it can drive the output transistors with a minimum of dropout, an output crowbar circuit to protect the load in case of failure, and foldback current limiting.

I accidentally had a grounded coaxial cable touch the exposed case of the drive transistor on my big Astron which melted one lead of the bias supply winding where it came out of the transformer.  To fix it, I added a small power transformer to supply the bias to the 723.

[davelectronic]

I'm  not sure why you'd want to use a plug pack for power, you get any additional voltage from the source. The schematic i posted reference to was a concept for the main circuit, any additional protection is easy to add on. Such as current limiting and over voltage protection. I'd  opt to put these on a separate board. So i guess the circuit in the link i posted as a concept is a viable one ? Unfortunately i never download pdf files I've not sourced myself, for security reasons.

[2N3055]

Okeyyyy.......

http://www.americanradiohistory.com/Practical_Wireless_Magazine.htm

June and July of 83, Marchwood power supply

[davelectronic]

Thanks, but not sure where it is on that page. Thanks for posting anyway, I probably could find the Marchwood project if i hunted it down. I often look all over the net for psu and other circuits. Its a win to find working examples.

[David Hess]

Thanks, but not sure where it is on that page. Thanks for posting anyway, I probably could find the Marchwood project if i hunted it down. I often look all over the net for psu and other circuits. Its a win to find working examples.

The design shows how complicated a 723 can get although it includes several safety features.

[davelectronic]

Thanks, but i never download unknown pdf files, sorry, as a Unix /Linux user you learn a lot.

Back on track, I'm  not phased by complexity  by the Marchwood project. I can visualise lay out and schematics very well, complex math formula can be my weak spot, although i usually get there in the end.

Cost is a factor, and some physical ability problems as well at the moment )-: more the hardware work involved, although ABS can make life easier in some areas. Just thinking lately, after nearly 12 years as a hobbyist I've never bread boarded a project, maybe its time i did.

[davelectronic]

Build A 10 Amp 13.8 Volt Power Supply http://www.rason.org/Projects/powsupply/powsupply.htm(Share from CM Browser)

This one and its associated protection circuit looks good, a variation on similar circuits. Any negatives to this one ? I know current is a little lower, but this looks like not to much of a monster to breadboard for proof of concept, thoughts anyone ?

[davelectronic]

Link don't  work.

[Kleinstein]

The circuit has a few points to improve:
1) The compensation cap looks quite small, so C5 might need to be larger (e.g. 500 pF-1nF) to make the circuit stable.

2) There is no need to use an 2N3055 for Q1, a smaller (e.g. TO220) transistor should also do it.

3) For lower noise a capacitor from pin5 to GND can be added.

4) 5 A for each of the 3055 is already quite a lot of current, if no foldback current limiting is used.
    So either add a 3rd power transistor or fold back current limiting (resistor from pin 2 to GND).

5) The drop out voltage is relatively high, similar to the 7812+PNP+NPN. I might be worth using separate diodes and filter cap for the regulator supply. This could save same 2 V on the transformer or about 1/3 the filter caps, as the regulator supply would have less ripple and thus higher min voltage.

6) The output capacitor likely does not need to be that large. 100 µF electrolytic + 100 nF ceramic should be OK

7) There should be a fuse at the transformer, at least one.
 

[davelectronic]

Thank you for the tips, yes they all make sense to me. I did think the driver for the 2x 2N3055's was a bit hefty. And any saving on voltage drops is really welcomed. Thanks again for looking the circuit over.

[davelectronic]

Its been a few days now... I finally built the circuit for restoring this old Alti 3 - 5 Amp 13.8 Volt linear psu. It was an ebay buy that someone had been happy with the side cutters and more.

Using an LM317K and an MJ4502 power transistor, it running again. I've set it up at 12.70 Volts with fixed value resistors. Its where i like to power radio gear. A test unmounted outside the case (for tesing) I'm getting a shade under 12.70 Volts.

Tested with 20 watt load, then 40 watt load, the transformer is only rated at the above maximum current. With a 40 watt load the regulator was steady at 45°C and the power transistor was steady at 52°C. I measured the voltage across the base resistor ( 10R 0.5 watt ) and the ballast resistor ( 7 watts 1.0R ) These where 180mV for the base resistor, and 0.95 Volts for the ballast resistor.

It run with the 40 watt load for a couple of hours very steady, so just have to mount it in the case now. I sould have taken some more measurements, base emitter voltage, and base collector etc. I'll do that when its wired in the case. One positive was the MJ4502 ran much cooler than the same circuit with an MJ2955. Might try a few other high power transitors in other circuits. I will be happy with a 20 Amp psu built from scratch using the LM317K, then i will move on to the LM723. With the above my voltage dropped 0.3 Volts with that 40 watt load.

[2N3055]

...............One positive was the MJ4502 ran much cooler than the same circuit with an MJ2955........

Well everything else equal, it should... MJ4502 has junction to case thermal resistance of 0,875 °K/W , while MJ2955 has about 1.5 °K/W... so at 40W that should be approx 20-25 °C less inside transistor...

[Cliff Matthews]

Dumb question: I thought a "ballast resistor" is only named as such if >2 parallel pass elements use low value emitter resistors to "balance" away from single element thermal runaway and possible cascade failure. Is this not the reason we employ a nautical term?

[davelectronic]

Yes I'm  sure it is, in more than one its a ballast resistor, as a single i presume it prevents thermal runaway of the transistor. So not strictly ballast, as a single resistor.

[David Hess]

...............One positive was the MJ4502 ran much cooler than the same circuit with an MJ2955........

Well everything else equal, it should... MJ4502 has junction to case thermal resistance of 0,875 °K/W , while MJ2955 has about 1.5 °K/W... so at 40W that should be approx 20-25 °C less inside transistor...

What temperature was davelectronic measuring?  The larger MJ4502 will have lower junction temperature but the external temperatures should have been the same.

[davelectronic]

I was measuring to side of the TO3 package, as close to the junction as i could get. This is what i was after from the start of this thread. Whilst i knew the same power would be dissipated, i wanted a higher power device to run cooler, so a larger overhead margine in power terms.

[davelectronic]

I've got a square bridge rectifier bolted to the same heat sink as the regulator, not perfectly ideal. The regulator would be much cooler on its own. But the trade off from heat there is not bad really. This supply won't see more than 3 Amps from my intended use anyway. As long as it can handle that continuous, and say a bit over at 50% duty cycle, then I'm happy with that.

[David Hess]

I was measuring to side of the TO3 package, as close to the junction as i could get. This is what i was after from the start of this thread. Whilst i knew the same power would be dissipated, i wanted a higher power device to run cooler, so a larger overhead margine in power terms.

That is only going return an unspecified Tcase which will vary depending on both the temperature measurement and the thermal resistance between the case and heat sink.  The transistor with the larger die will have a lower Tjunction for a given Tcase but that is not what you measured.

Typically Tjunction is measured via the change in Vbe at a given current and this works very well but the since Tjunction has a time constant of milliseconds, the measurement needs to be made while power is applied or immediately after power is removed.

That is an awful lot of work for a single project so another way I have done it in the past is to measure Tcase as specified by the manufacturer which usually involves drilling a small hole in the heat sink and then estimate Tjunction from the power level and specified junction to case thermal resistance which is sort of what you did.

[davelectronic]

That's a step to far for me, i will take my chances on what i measured, i know its far from perfect, but it will do for me. If i have over specified the components for a project (like this one) I'm less likley to run into trouble. Thanks for the info. Things in my world move slow....

I've still got to get it cased up, and put in service.