So many memebers here are aware that I make PCBs and one aspect of making PCBs is the electroplating process. Now my tank was finished about a month ago and is working great on the old constant current source but this could only throw out 800mA at the most. For those who are not aware the required amp per square foot is around 10A and with my larger PCB sizes this means that I need 9.2A more (the boards are approx 1 ft square).

Two design options where available (and still are):

• Linear constant current source
• SMPS based design

The linear constant current source is the easiest option because all I need is a big mosfet and some simple control circuitry. However as this is a linear regulator and the voltage across the tank is around 2V ~ 3V the power dissipation will be around (12V - 3V) * 10A = 90W. That is ALOT of power wasted!

The switch mode power supply (SMPS), however will dissipate a fraction of that energy. So this was the obvious choice and thus I designed and built a SMPS using LTSpice, KiCad and some copper clad. The device worked....but not as far as I had hoped. The fet (also tried a PNP) in the SMPS would heat up insainly despite operating in the fully saturated region. After some research it turns out that even if a TO-220 dissipates 1W that can still lead to a huge amount of heat. So until my new order arrives (low on resistance low gate voltage power mosfets), I have gone the linear route (pictures soon!).

So the mosfet chosen for the linear constant current source was an IRF450 (I have many of those!). The linear regulator was built onto stripboard and the schematic was similar to the one shown below.

However there are some slight differences:

• The reference is not fed from a zener but a linear 100K potentiometer
• The sense resistor is a chunky 5W 0.1 ohm. I also attached a crude heat sink to help keep it cool
• The op-amp used was an LM358. The second amp was used as a voltage amplifier across the sense resistor. This was then fed to a cheap chinese 3 pin LED volt meter. That way, I can see how much current is flowing through. A trim pot is used in the amplifier to adjust the output reading so it is someone accurate
• A fan is attached (pictures soon), to the project box that holds the controller to keep things cool!
• The TO-3 MOSFET (IRF450), has a minature CPU fan on top while also screwed into a very large piece of copper clad (and some thermal paste).

So far the device can chuck out 8A and be happy (the TO-3 needs more cooling).
Like I said, pictures will arrive soon (dont want people to be left in the dark)!

I also forgot to mention that the power supply used is an old computer PSU (cant really beat that for its low voltage high current outputs).

 

Why do you start with 12 volt rail if your tank voltage is only 2 to 3 volts? Your heat issues will be greatly minimized if you use the 12 volt rail only to power your op amp and reference but use the 5 volt rail as your current source through your transistor. That saves you from dissipating 7v x 10 amps = 70 watts if you continue down this linear path.

If you have the right parts, almost eliminate major heat issues if you start with the 3.3 volt rail. A 500 W atx power supply should handle 25 amp on the 3.3v rail.

 

That is not a bad idea at all! I will try to change that now. But before I do I am sure there was a reason why I could not do this but I will find out soon.

 

Just tried it and I get up to 6.3 amps with the 5V rail and over 10 with the 12V. I know the policy this forum (no schematic/no help), and thus will not ask for help. But I do hope to gets some pictures soon!

 

The power dissipation problem manifests itself as heat in the regulator transistor, beit an NPN (as you show) or an NFET (as I prefer). To calculate your heatsinking requirement for the regulator, multiply the Collector-to-Emitter (Drain -to-Source) Voltage by the current into your tank. If your tank voltage is 5V, but you start with 12V, then the voltage drop across the regulator is 12-5 = 7V. If your cell current is 8A, then you are dissipating 7*8 = 56W, which requires forced air and/or a huge heatsink.

Now imagine that the input voltage was reduced to 6V. 6-5 = 1V. 1V*8A = 8Watts, hugely easier to deal with....

 

I have an 450W ATX PS that I picked up at Goodwill for $5 that outputs 3.3V @ 32 Amps.

 

Thanks for the replies. I understand that the smaller the voltage the lower the drop across the CE on the transistor (I actually used an NMOS instead so that would be drain source voltage). But what amazes me is that for some reason using a lower voltage rail on the PSU results in less current flowing through the tank which I find odd considering it is a constant current source where the reference current is independent of the supply voltage (using a zener to produce the voltage which is then divided and set via a potentiometer).

My only guess is that the supply is not capable of such currents on the lower voltage wires so I will check those soon!

My heat sink is crude at most but it works very well. A large piece (about A5 size), of FR4 double sided with a CPU fan and the metal case (TO-3) bolted down. It is also thermally bonded to help transfer the heat. I then made the heat sink more efficient by coating with a black acrylic paint near the package which has cooled it down significantly.

 

I have an 450W ATX PS that I picked up at Goodwill for $5 that outputs 3.3V @ 32 Amps.

5 Bucks for a PSU!!!

Would love to find a deal like that!

 

Just tried it and I get up to 6.3 amps with the 5V rail and over 10 with the 12V. I know the policy this forum (no schematic/no help), and thus will not ask for help. But I do hope to gets some pictures soon!

Try again with 5V rail but only connect the 5 volt rail to the NPN collector (or nFET Drain) to +5V. Keep the op amp and the reference connected to 12V rail. That way, the FET has plenty of Gate voltage (or drive current, for NPN).

 

@GopherT

THAT sounds like the problem. I will try later as I am currently plating but will let you know of the results.

 

@GopherT

THAT sounds like the problem. I will try later as I am currently plating but will let you know of the results.

For even less heat, you could try your 3.3V rail to the power transistor.

 

I think I know the reason now.

The resistance reading of the electrolyte is around 1 ohm. With a 5V rail the max voltage across it would be 5V which would give 5A. This means that I need to decrease the pH to lower the resistance so that a smaller voltage will give a larger current.

Thanks for the help though!