I would like to create a simple (not very fine tuned) regulated power supply in which with two potentiometers.

I want to adjust the voltage from ~2V-34V, while the output of my 24 V transformer rectified and smoothened with 10000 μF is about 35-36 V. I had the idea to just use approximately 8 LM1084 (5AMP) regulators in parallel and also have 8 TIP122 (5-8A) darlington transistors in parallel, steering their bases to regulate the power. Is this at all feasable/doable?

One thing I realized is that short circuiting the LM regulator breaks them almost instantly.


simulate this circuit – Schematic created using CircuitLab

Obviously I only drew 2 of the 8 planned LM's and transistors and I did not implement the transistors correctly this way. Can anyone please tell me if this could make any sense?

  • 4
    \$\begingroup\$ At 2V 40A you're dropping 34V 40A in the regulators and/or transistors. That's almost 1.4 kW. I've got a fan heater that won't do that. So the first thing you need to do is design the heatsink. More realistically, look at switcher modules. \$\endgroup\$ Jan 14 '20 at 18:42
  • \$\begingroup\$ Switches modules basically turn on/off very fast right? It is not a steady voltage at the output? \$\endgroup\$
    – Niels
    Jan 14 '20 at 19:22
  • \$\begingroup\$ It IS a steady voltage at the output : there is a low-pass filter to smooth out the switching. You'd need a "buck converter" which is based around an inductor which does the smoothing. \$\endgroup\$ Jan 14 '20 at 19:23
  • \$\begingroup\$ And then using these TIP122 in parallel to regulate max current, so after the buck converter... is that a good idea? Let's say combined with something like the BC557 or 2N2222 to take care of all the bases of the TIP222s toghether instead of 1 pot.meter having all the base currents? \$\endgroup\$
    – Niels
    Jan 14 '20 at 19:55

Is it feasible? It really depends on your definition of feasible.

First, let's talk about the heat sink. Any suitable heat sink you might buy would cost much much more than simply buying a commercial, off-the-shelf high current bench power supply, so you'll need to already have a heat sink roughly this size on hand:

my biggest heat sink

If you do, great! You have passed the first feasibility check.

So, you've got your 4 foot heat sink ready to go and are just itching to put it to use.

But are your desired specs feasible? Unfortunately, you'll have to compromise a little bit here. You can have your 40A, but the problem is the voltage differential, the transformer, and the capacitor.

  1. The LM1084 regulators have a maximum input to output voltage difference that they can withstand, which for the adjustable version, is 29V. They need 1.5V of dropout to stay in regulation at maximum output current, so your usable voltage range is going to be 27.5V. You'll either have to be ok with a higher minimum voltage, or lower maximum voltage, as 2-34V is a range of 32V, too high for these regulators to handle.

  2. Unless you measured the output voltage of your transformer while powering a 40A load, then it will not put out anywhere close to 35-36V. It will, in fact, be close to 24V. That is why it is a 24V transformer. Nearly all power transformers, and especially very large ones like this 40A monstrosity you have, put out a substantially higher voltage when unloaded. Their rated voltage (24V in this case) is the voltage they will output when actually under load. That means that the peak voltage generated by the transformer under load will be about 33V. Subtract ~2V due to the bridge rectifier, and you can expect 31V under load. However, with a 10000uF capacitor, at 40A, you will have over 12V of voltage ripple. That means that the voltage feeding your power supply, after the capacitor, will be swinging from 31V to 19V. This is extremely hopeful, in reality it will almost certainly be even worse. The regulators can reject most of that ripple, but with that capacitor, your true maximum output voltage at 40A will likely be somewhere south of 19V, absolute best case.

So you'll need a far, far larger capacitor to achieve higher voltages under load. However, do not expect to get much above 24V, even with the RMS 'bonus' voltage at play. It is a 24V transformer after all. Diodes and ripple will eat up most of that extra peak voltage unless you have a massive capacitor (at least under 40A load).

But lets say you have the necessary capacitor (or more likely, bank of capacitors) on hand and ready to go. Great! Allowing for a more modest 24V at 40A output, you've passed the second feasibility check.

But what about your transistors? They aren't even TO-3 types, they're just regular little TO-220 transistors. Certain component packages have limits on how much power they can dissipate. In the case of the TIP122, it can dissipate 65W if kept at room temperature. Even your behemoth of a heat sink can't do that. You're limited by the thermal interface of the TO-220 case to the heat sink. Let's be generous and assume you have fantastic thermal grease and a perfect thermal interface between a TIP122 and your heat sink, yielding a thermal resistance of 0.3C/W. This is an optimistic value for the contact area available with a TO-220.

At 65W, this wil result in roughly a 20C temperature rise. Unfortunately, we have to derate the maximum power dissipation by 0.52W/C, so the true maximum is a bit less. And you need some thermal head room, so lets just call it 55W. The LM1084 is a bit lower, the datasheet shows that it can only dissipate 30W maximum.

You say you have 8 TIP122 transistors, and 8 LM1084s. Your maximum power dissipation capability here is 8*55 = 440W for the TIP122s, and 8*30 = 240 for the LM1084s. 440W + 240W = 680W. 22V differential at 40A is 880W.

Sorry, but it looks like despite our best efforts, the last feasibility check has failed.

You simply do not have large/powerful enough components to build even a substantially reduced spec (24V @ 40A) power supply, nonetheless a 34V @ 40A behemoth. And in both cases, actually building such a supply would cost you significantly more than simply buying a commercial supply or module capable of your desired output. A supply that would be much safer, more reliable, and vastly superior in every single performance metric.

Even if this was feasible, feasibility doesn't mean its a good idea. It just means it could be done. It doesn't say anything about whether something should be done. And what you are proposing definitely should not be done. When you get into currents and power levels that high, unless you know very well what you're doing, you really ought to buy something rather than build something yourself. For reasons of safety, cost, and performance. And building linear power supply like this is frankly just wasteful. Your transformer is not suitable for use in an adjustable linear power supply. Adjustable supplies use transformers with multiple taps, allowing an input relay (or similar) to switch to different output voltages from the transformer based on the power supply's output voltage. This substantially reduces the power dissipation requirements, the necessary capacitor sizes, the demands on components, cost, everything.

I don't want to be negative, but the power supply you're suggesting is not one that should ever exist. It is possible to do so much better, and there is no reason to build one that pointlessly wasteful and expensive.


Let's see what size Cap you would need for 100Hz.

Ic=CdV/dt for dt=10ms and dV~10% input ripple of 34V ~ 4V
C=Ic*dt/dV = 40A * 10ms/4Vpp = 100 mF = 0.1 farad @50V

I don't think you planned on this.

Above is the discharge equation.

The charge equation depends on the duty cycle which happens to nearly equal the ripple assumed to be near 10~15% of the DC voltage.

This means due to Conservation of Charge Law, if you discharge 40A 90% of the time then you need 10x as much current in 10% of the time to restore the charge or in this case 400 Amps in 1ms every 10ms and hugh dI/dt EMI issues. This has major consequences for I^2*R=Pd and you would need a supercap ESR with the capacity of small lead acid battery. However SuperCaps only come in low voltages and batteries of this voltage have too high ESR for this pulse ripple current. So this is a no-go or a design spec that cannot be reasonably achieved.


You have fairly unreasonable demands for your power supply. As pointed out in the comments, your specs would require a linear supply to dissipate 1.4KW under the worst case condition of output voltage and current. This, as well as the schematic included, indicates that power supply design is not really your area of expertise.

This module appears to meet your requirements, as far as I can tell. It's expensive, but it should work out of the box if you follow the spec. Similar, less expensive modules won't support your wide output voltage range or your 40A current requirement. The details of power supply design are worthy of a couple semesters of study (or experience), and not really practical as an EE.SE post.

  • \$\begingroup\$ That link goes to the first of several search results pages, with a couple hundred different units. Do you have a more specific link? \$\endgroup\$ Feb 7 '20 at 14:05
  • \$\begingroup\$ Odd, for me it goes straight to the page for the module. Search for RBBA3000-50. \$\endgroup\$ Feb 7 '20 at 14:18
  • \$\begingroup\$ Odd, indeed. Here's a link that works for my browser. But there it's out of stock, available mid March. \$\endgroup\$ Feb 7 '20 at 14:21
  • \$\begingroup\$ Octopart octopart.com/search?q=RBBA3000-50&currency=USD&specs=0 shows stock at one vendor, but they seem to like to sell tubes of 12 rather than one unit at a time. I guess you could call them. \$\endgroup\$ Feb 7 '20 at 14:28

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.