No.130 - Thurlby 30V-2A Bench Power Supply Repair 

2N3055 was designed to be a very robust part. When properly heatsunk, the SOA is large. It’s sort of a no-nonsense trusted building block. Old reliable. Proper heat sinking is key though.

I have power supply with the same spec's, 0-30V 2A, but it uses only one 2N3055. That transistor can handle 115W 10A, if properly cooled. Here we see FOUR of them.

@@erikdenhouterMakes it robust. A 2N3055 will dump a 100W, but not at full rated C-E voltage IIRC.

@@absurdengineering I think they included some headroom indeed, it is the high transformer voltage that makes a large C-E voltage to begin with. Now I think of it, the power supply I was talking about uses a relais that kicks in when the unit is regulated over 15 volt. The reason for that is to go from one to two secondair windings, so below 15 Volt only one winding is used, hence half the needed C-E voltage is there :)

Solid kit for the tech colleges back in the day I seem to remember?

@@PhatTony-km3fl Deffo. We had some in daily use at an FE college I worked at in Liverpool only a few years ago.

My college had them too, late 80's

No doubt the amount thermal expansions and contractions this unit has had it its life much be pretty high!

So had I until I saw the tell tale blobs of solder on the topside of the Pcb. Thinking back, I must have inserted 1000's of track pins onto bare Pcb's back in the day. I always remember some disasters when the soldering iron slipped!

@@IanScottJohnstonWhen I was a wee lad, I went fancy with resistor leg off-cuts. But I did the off-cuts before even installing the resistor. Resistors make for nice little wire holders. So first I’d solder a resistor forest - a “tree” in half the vias. Then I’d cut them off, and repeat with the other not-yet-cut leg in the rest of the vias. Cut those off to free the resistors again. Then the resistors were soldered onto the board. Worked fine for hobby projects. I dread fixing the PCBs that had little rivets for the vias. Those would always get unreliable and were such a pain to get solder to stick to. Atmosphere - good for us, bad for the connections 😅

@@absurdengineering Never much had a problem with track pins unless they had been sitting around for a good few months, then they had oxidized a little and were hard to solder on the underside. But, the biggest pain with them was teaching the junior techs how to install them......need quite a bit of force with the soldering iron to get them down into the hole properly......and mucho scope for skating the soldering tip across the board!

@@IanScottJohnstonMaybe yours were better made. The usual old US-stuff was all riveted on with a press, straight onto tinned traces. The rivets were tin plated too. No soldering whatsoever. Maybe that was the problem.

I’ve had one for 35+ years and it’s still my main go-to PSU. Only thing that ever failed on it was the mains switch power LED!

The original cartoon character was a graphic I found on the web many years ago......he was holding a flag or something, I contacted the author and paid a couple pounds for the royalty, then I modified it so he holds test gear/tools instead (with the authors blessing). Eventually, I went to fiverr and got a guy to create small video clips of him doing various things. I have a wee collection now for various purposes which I have integrated into Davinci Resolve.

Reason: It's too easy in KiCad to drop a ground pour I think. I know I'm old.......but I really like working on this older gear and working with the technologies and implementations we used to do daily in the 80's.......sniff!

@@IanScottJohnston these were the power supplies i grew up with. Kingshill power supplies for the higher voltage or currents. and Levell TM3B's for the audio measurements. Still have an HP distortion analyser under the bench, still works!

I've since bought a longer 300mm screwdriver so next time I don't have to remove the heatsinks...:-)

I don’t always wick off old solder, it’s a case of resoldering the first and seeing how it accepts new solder and more importantly new flux. Balance this up with desoldering the entire IC which can subject the IC to a bit more stress.

I have a small valve amplifier kit that I never assembled somewhere.....hmmm, I should dig it out.

@@IanScottJohnston Thats sounds like a good idea to me, it would also introduce some jeopardy and fear into the post with the high HT rails present Ian, ouch!?🤣

OK their input AC voltage is reasonably high when it gives 55V DC. So it gives 110W with max load at lowest output voltage, so I can agree to use two 2N3055.

Like almost all figures in the glossy sales brochures, the 15A, 115W figures are theoretically possible, but only in *very* specific but usually unrealistic conditions. In this case there are three different operating regions to consider, which are shown in the 'safe operating area' chart, or SOA. (Bottom of page 2 for the ON-semi 2N3055-D): Region 1) Voltage < 7.5V. Here the 2N3055 is specified at 15A max; that is determined by the current carrying capability of the wire bonds connecting the silicon to the case connections. But note that all the other specification graphs for the 3055 only go up to 10A meaning useing above 10A is at your own risk and performance is unspecified - it just shouldn't destroy itself below 15A. Region 2) Voltage between 7.5V and 40V. Here the maximum current is determined by the 115W power disspation limit of the transistor. Eg. at 25V, the maximum collector current is 115/25 = 4.6A. Not too shabby eh? *Except* this is only true when the case is kept at 25C (or less). This arises from the maximum junction temperature limit of 200C and the junction to case thermal resistance specified at 1.52 C/W. Keeping the case at 25C when the transistor is dissipating 115W is virtually impossible without liquid Nitrogen or the like. A very good heatsink (without forced air) would be 0.5 C/W, but the TO3 case to heatsink insulator (mica or whatever) might add another 1 C/W. Bench PSUs are typically specified to operate in an ambient temperature of up to 40C. So (200C - 40C)/(1.52 + 1 + .5) = 53W maximum dissipation. So now we need at least two devices for a 32V/2A supply. Region 3) Voltage > 40V. This region is subject to secondary breakdown of the transistor where the reverse biased collector-base junction's breakdown voltage decreases with increasing collector current so the maximum allowable current is much less than that calculated in region (2). In the video, Ian measured the voltage on the main capacitors as 55V. When the supply is shorted, current is limited to 2A max but the total dissipation of the 2N3055s is now approx 110W. In reality the 55V is the unloaded transformer voltage so at 2A this will drop by perhaps 10%. *But* the PSU designer has to allow for the mains voltage to vary by +/- 10% from nominal so the 55V might be the worst case (but we don't know what the mains voltage was in Ian's lab at the time). The SOA graph shows the max current is approx 1A at 55V, (ie. 55W) - again with the case temperature at 25C. Taking all factors into account it is hard to (safely) get a 2N3055 to dissipate more than 20 to 25W, especially when operating at high voltages. Don't forget to add a large margin for long term reliable operation and you can see why they use so many devices in these linear supplies.

Tony did the math. Those infinite heatsinks are rare beasts indeed 😂

@@tonyh6309 Many thanks for the thorough and instructive answer. You are correct, with the high COLLECTOR−EMITTER VOLTAGE (VOLTS) I can see that it is necessary to have at least four 2N3055 at 2 Amp. Due to (THERMALLY LIMITED @ TC = 25°C) according to ON Semiconductor data sheet. Thanks for the reminder.

The more transistors in parallel the better as it makes it easier to control and keep lower temps, especially making sure the temps are equal across them all. Any difference in temp between each of them can mean unequal collector currents and thus more likely thermal runaway can occur. Also, each transistor has an emitter resistor to help balance each transistor load, so ohms law comes into it now and the heat generated on each resistor.

Thank you for the explanation, I just thought that 4 would be overkill for the amount of current being drawn, and, would have thought that 2 would've sufficed. But, I suppose, that the over engeneering is why they are still around. Thank you again Ian... Have a good day...

[Part 1] Like almost all figures in the glossy sales brochures, the 15A, 115W figures are theoretically possible, but only in *very* specific but usually unrealistic conditions. In this case there are three different operating regions to consider, which are shown in the 'safe operating area' chart, or SOA. (Bottom of page 2 for the ON-semi 2N3055-D): Region 1) Voltage < 7.5V. Here the 2N3055 is specified at 15A max; that is determined by the current carrying capability of the wire bonds connecting the silicon to the case connections. But note that all the other specification graphs for the 3055 only go up to 10A meaning useing above 10A is at your own risk and performance is unspecified - it just shouldn't destroy itself below 15A. Region 2) Voltage between 7.5V and 40V. Here the maximum current is determined by the 115W power disspation limit of the transistor. Eg. at 25V, the maximum collector current is 115/25 = 4.6A. Not too shabby eh? *Except* this is only true when the case is kept at 25C (or less). This arises from the maximum junction temperature limit of 200C and the junction to case thermal resistance specified at 1.52 C/W. Keeping the case at 25C when the transistor is dissipating 115W is virtually impossible without liquid Nitrogen or the like. A very good heatsink (without forced air) would be 0.5 C/W, but the TO3 case to heatsink insulator (mica or whatever) might add another 1 C/W. Bench PSUs are typically specified to operate in an ambient temperature of up to 40C. So (200C - 40C)/(1.52 + 1 + .5) = 53W maximum dissipation. So now we need at least two devices for a 32V/2A supply.

[Part 2] Region 3) Voltage > 40V. This region is subject to secondary breakdown of the transistor where the reverse biased collector-base junction's breakdown voltage decreases with increasing collector current so the maximum allowable current is much less than that calculated in region (2). In the video, Ian measured the voltage on the main capacitors as 55V. When the supply is shorted, current is limited to 2A max but the total dissipation of the 2N3055s is now approx 110W. In reality the 55V is the unloaded transformer voltage so at 2A this will drop by perhaps 10%. *But* the PSU designer has to allow for the mains voltage to vary by +/- 10% from nominal so the 55V might be the worst case (but we don't know what the mains voltage was in Ian's lab at the time). The SOA graph shows the max current is approx 1A at 55V, (ie. 55W) - again with the case temperature at 25C. Taking all factors into account it is hard to (safely) get a 2N3055 to dissipate more than 20 to 25W, especially when operating at high voltages. Don't forget to add a large margin for long term reliable operation and you can see why they use so many devices in these linear supplies.

@@tonyh6309 Yes, that is more or less the assumption I came to when I saw the 55V across the caps, that-that would be the contributing factor as to why there were 4 of them as the 2A seemed like child's play..... You have clarified that for me..... Thank you kindly for the VERY detailed writeup. It is not often that one still gets responses like yours to individuals like myself who genuinely ask a valid question in order to be able to gain some good, sound insight into why things are done in the manner they are, hence why I said that this is why this power supply is still around and working...