My Variable DC Power Supply (30V/1A)

[birrbert]

Hi Folks!

Just a quick update on this project which has come to a longer halt, unfortunately. I had to travel abroad for the past two weeks and as I returned I had to start some renovation in my apartment so that kept and still keeps me busy. Plus I've got a kid on the line, so in about a month I'll be a father.

The good news is that I have all the necessary parts. In addition I bought some nice TO-3 adapters, pin headers, connecting terminalsl all I need now is one or two free afternoons to put everything together.

[runem]

edit:
What kind of heat-sink compound do you recommend? I found too many types (with silicone, without silicone, silver based, etc).

edit2:
I went for some very cheap silicone based compound.

I think you made a good choice there. I have bad experience with the silver stuff and mica, because the first and last time I tried, the silver particles somehow found their way through the mica insulation sheet and made a strange metallic fume i couldn't understand where was coming from in the beginning.
Of course I wouldn't hesitate using silver paste if the component can be mounted directly on the cooling element.
Good luck on your project!

[birrbert]

Hi! I'm back!

I am in the need of some clarification: on blankfield's complementary schematic is Q4's Emitter connected to Q1, Q2 and Q3's Base? I mean, Q4's Emitter has to be connected with the other three transistor's Base? And then Q1's Emitter connected to Q2's and Q3's Base simultaneously, plus in the same time Q2's Emitter connected to Q3's Base?

Or I'm misunderstanding the schematic? I see some dots there; they represent junctions and where there is no dot then those lines go to their own separate ways?

That's in mind so confusing as I wrote it down... Please clarify! I have everything laid out on the table and wiring up the power transistors and the resistors would be the final step.


edit:
I think I figured it out by trying it live this way: a) connected Q4's Emitter to Q1's, Q2's and Q3's Base by wiring three bases together b) the Emitter of Q1, Q2 and Q3 connected one by one to a resistor c) wired Q1's, Q2's and Q3's Collector together to power them from the positive rail. Then I hooked up the three resistors to the big one and this big one got wired as it would have normally been wired on the original PCB.

But I'm disappointed. The system is working, but everything is running very very hot. No change whatsoever compared to previous attempts, except for the fact that it can deliver higher currents. As a load I used a 24V/100W light bulb. With the current limit set to 2 Amps, after just 4 or 5 minutes the radiator was burning. The KBPC1510W also. I put a cheap thermometer's probe between the wings and the temp rose to 60 degrees Celsius, but I'm sure it was hotter than that. I don't know how to continue. I worked hard in the last three days to wire most of the components nicely and make it work, but I expected better results. Or I shouldn't have?

I made it as modular as possible, so I can change components without having to redo everything. I might try a fan tomorrow, though the whole thing is big already; I wanted it to run passively, so I don't know if I want to make it even bigger by adding fans.

Photos attached (the last one is high res). Every constructive thought is highly appreciated! I'm about to give up, but I spent too much time and money to stop... I guess. Thank you!

[birrbert]

So, the thing is that I can't get away without active cooling. I tried that today. I put a 12 V DC fan just a few centimeters from two power transistors that are closest to each other and everything looked fine. Bot transistors got hot while powering the light bulb, but the temperature seemed to be nicely controlled by the added airflow. I'm glad that at least it works. I went up to 4.2-4.5 Amps, but at that point I realized how hot the bridge rectifier became. Now it's placed just behind the LCD so it gets no extra cooling and it got so hot that it was almost melting the plastic around it. By the way, I swapped the 15 A bridge with a 25 A one, but I'm afraid I'll have to mount that too on the radiator. Or maybe find a more efficient type that runs cooler.

Next week I have some other stuff to care of, but I'm planning to finalize the power supply by the end of the week. In the meantime, every additional thought from you - regarding this project - is highly appreciated! Two more high res photos attached. Thank you!

[SeanB]

Those bridges are meant to be mounted to a heatsink in normal use, preferably a big one if you are drawing full load. You need a sheet of aluminium there to mount it to, probably best to bring it around and mount to the heatsink as well. If you use Shottky diodes you will get less heat, though you will need 4 devices, or 4 PC power supplies to rob the output rectifiers out of if they have a high enough voltage, generally they have a 45V device on the 12V rail that will be in a TO220 package, typically rated at 30A. Use both halves in parallel and make a bridge out of them. Otherwise you can buy them from RS or other suppliers, at 2A they will not need a heatsink, though you can use one if you want.

[mariush]

birrbert, hello from a fellow Romanian

If you think it would help, I could mail you a VSIB620 (200v 6A bridge rectifier) : http://www.vishay.com/docs/84656/vsib620.pdf

I could possibly also add some anodized aluminum heatsink, the one you see in the picture, which would work for this style of bridge rectifer... it's a 7.6c/w heatsink so with a bit of thermal paste, it should still keep the rectifier at about 70-90c when you drive it at 4-5 Amps - it would be hot, but safe to run.

Otherwise, i have some big chunky UF3007 ultra fast around and I could spare 4... but they're only 800v 3A diodes

[Strada916]

Remember what goes in must come out. So. You apply 30 something volts at the input. Or on the collectors of the transistors. Vin - Vout x I is what is dissipated into the heat sink. for example. Vout is 5 and Iout is 2amps the you are dissipating 25 x 2 50 watts in heat. Does that make sense?

[runem]

The downside of linear power supplies is that you simply cannot skimp on the cooling. You will need some big chunky heat sinks to keep it cool. It seems to me, from the size of it, that yours is generally too small in size, has not the sufficient surface and so on.
I would try to get hold on a heat sink from a junked 100W (or something like that) audio amplifier, with lots of thin fins on it that gives plenty of surface. Some scavenged CPU coolers with fans should also be worth a try. I believe you can by them pretty cheaply on e-bay. Or, you could also try a small liquid radiator and let the thermal convection do the "pump job", and then have a silent system. I have such a radiator (Thermaltake CPU cooler) here and you can have it for free if you pay the shipping cost, but unfortunately I don't have the cooling block anymore.

[Strada916]

you might want to solder the transistor connections as just wrapping bare wire over the pin is not a good idea as it can fall off and there would be some volt drop there too.



Reason for edit
suggested to use modern transistors, turns out you are.

[Strada916]

MJ15003 might be a better power transistor, it has half the junction to case thermal resistance than the 2n3055, and better electrical characteristics, although the mj is a little more expensive around $5US plus shipping also you would not need three as they can handle more current too. 

[birrbert]

Hey guys! I just checked in quickly and am very glad to see so many replies! I can't read them in detail, because my daughter was born today. Our very first child, her name is Sára (Sarah in English).

This means that I have to pause the project again for two weeks or so, but when I come back I'll resume and promise to finish the power supply.

Hello mariush! Thanks for the offer. I might test that b.r. I will send you a message later.

Strada916, thanks for the explanation. The dissipated heat make total sense. Thanks for the tip about the transistor too.

runem, the reason I got this radiator is because it had the TO-3 holes. I'll see how I can mount two 80 mm fans on it and if it's efficient and it looks cool then I'll stick with this. Of course, I'm always looking for improvements so I would appreciate if you could send me photos/description about the Thermaltake CPU cooler and the cooling block (just to know what to look for).

Only the best and hope to be back soon!

edit:
Speaking of bridge rectifiers, this question got stuck in my head: what influences how hot or cool a bridge rectifier will run? Is it the current rating? Is it the type/package? Are there know types that keep their cool more than others under stress?

[ttyz]

Congratulations birrbert! 

[dr_p]

Congrats for Sara !

Speaking of bridge rectifiers, this question got stuck in my head: what influences how hot or cool a bridge rectifier will run? Is it the current rating? Is it the type/package? Are there know types that keep their cool more than others under stress?

It's the forward instantaneous voltage drop and the forward instantaneous current. Let's take for example, the part mariush mentioned ( http://www.vishay.com/docs/84656/vsib620.pdf )

On page 3 you have Figure.3:



You are loading your power supply with a continuous 5A of DC current. So the AC side is loaded with 5A RMS (AC) as well, being it's a linear supply (almost all the input current goes to the output). The AC current is jumping up'n'down according to the 50Hz sine, the voltage drop on each diode varies, etc. If we look at the graph and assume a current of 5A DC (constant), we get a voltage drop of 0.62V (at 125C).

But I want to consider the worst case scenario:



Even though the test scenario is at 3A, I don't think I'm making a mistake by considering a constant 5A at 0.95V drop on each diode. That should be more than needed.

At each one time current passes through 2 diodes of the bridge, so the total power dissipation in the bridge is: 2 x 0.95V x 5A = 9.5W

So what heatsink to choose? Well, Dave made a video about that:

http://www.youtube.com/watch?v=8ruFVmxf0zs

Long story short, having a look in the datasheet also reveals this:



Add the Thermal Resistance from Junction to Case and the Thermal Resistance from Case to Ambient and you get R(JA) - from junction to ambient, expressed in degrees C per W. In this scenario 3.4 + 7.6 = 11C/W

So you dissipated 9.5W earlier so the junction temperature will rise by 9.5 x 11 = 104C above ambient, so probably around 130C.

Now we look at the datasheet again:



and we know that at 130C the bridge can pass 5A maximum. So it's fine to use and we expect a lower junction temperature than 130C, since we over rated the calculations. We then expect a much lower heatsink temperature, because of the thermal resistance between junction and case (also heatsink). Still, probably too hot to touch (70-80C). Incidentally, it's just like mariush said

[runem]

Of course, I'm always looking for improvements so I would appreciate if you could send me photos/description about the Thermaltake CPU cooler and the cooling block (just to know what to look for).

Congrats with a new family member birrbert!
You were asking for some pics and here they are! If you mount the radiator horizontally the hot water will always rise. Sorry, I don't have the cooling block, but it shouldn't be to difficult to make something yourself, or have someone with a mill to machine one out of an aluminum block for you.

[birrbert]

Thank you very much everybody! Both the mother and child are well.

What do you think about installing the cooling fans?
A) Should I bother with temperature based cooling?
B) Should I consider to let the system vary fan speeds as temperatures go higher or lower?
C) Or should I simply buy some nice silent fans, wire them directly and let them be on/blow/spin non-stop?

Maybe option C is the most realistic at the moment. Although option B would be professional, I would need to set up a complex system with some micro controller and stuff, right?