5 amp stabilised supply

[Crackle]

It is not difficult to make a 5 amp regulated supply when you have a device that works to spec.

The replacement LM338 regulators arrived today, and were the type seen on the original Ebay listing for my second purchase.
The seller had some still in stock, he has removed the others and says he is going to chase up the issue of the fake ones with his suppliers.
The replacements were supplied free of charge, he had already refunded the original purchase.
A very fair seller, and very fast delivery..
So I am happy.

Here is the proof.
First photo taken with no load.
   
Second photo taken with 5 ohm load
   
Third photo taken with 2.2 ohm load.
   
Last photo taken with the regulator set to 14 volts showing the performance with a 2.2 ohm load.
   
Incidently I am now happy to mention that the actual items I bought were these.  Note the V4 stamped into the body by the adjust pin.
https://www.ebay.co.uk/itm/192135930124

Mike

[Mark Hennessy]

Mike,

As I mentioned earlier, I'd be worried about thermal issues. The TO220 package is not the best at transferring heat to a heat sink, even with the best attention to detail. If you can get away without insulating the device from the heat sink, that helps a fair bit. Alternatively, I did suggest finding a s/h or NOS LM388K in a TO3 package...

But there is another way, and this allows you to make use of your existing "LM338" chips: simply add a PNP power transistor and resistor.

You might already know about this, and you might have already considered and dismissed the idea, but just in case, or for those who haven't come across the notion...

...the power transistor can be anything you like - it just has to be able to handle the power. The value of the resistor decides how the current is shared between the IC and the transistor - for example, if you pick 4.7 ohms, then the IC only sees around 130mA and the transistor handles the rest.

The basic circuit is shown in many books and datasheets - it's well-known, and works well. Here's an example, chosen at random: http://www.bristolwatch.com/ccs/LM317boost.htm

If you like, the regulator IC can be set to carry such a low current that it doesn't even need a heat sink.

You still get the pleasure of DIY rather than buying in a lump, and you can get away with a smaller heat sink that you would with a single LM338 as the junction in a transistor can run hotter than an IC junction. If you pick a power transistor in a TO3 package, the much-improved thermal conductivity between junction and heat sink would allow an even smaller heat sink again.

The only downside is that there is now no short-circuit protection - if that matters in your application, it's simple to include (just another transistor and a resistor or two).



PS: If you only have an NPN power transistor to hand, then you can use it if you include an additional medium-power PNP type to drive the NPN: https://www.eleccircuit.com/best-dc-powe...6v-9v-12v/

[Mark Hennessy]

Posts crossed with seconds of each other!

Good news that it now works. Will be interested to see if more testing reveals any thermal problems. Have you got the means to measure the temperature of the device on the heat sink? I'm a big fan of IR thermometers for that, and they can be found in Lidl/Aldi from time to time for just over a tenner.

[Crackle]

I understand what you are saying Mark and thanks for the warnings and additional information, its useful for others. But I will probably stick with this circuit for this application.
As I said the current demand is probably going to be a maximum of 1.5 amps and that is intermittent, I am sure the LM338T will cope fine bolted on the heatsink I have fitted.

I have a finger which acts as a temperature sensor. That's warm means all OK, Touch for max of 5 seconds means that's working hard. a sizzle sound means it is probably about to blow if it hasn't already.
Unless of course it is a germanium device like an OC26 on a Z12, then it is panic as soon as it gets more than warm.

Mike

[Crackle]

One last note I would add about these regulators is:
Although the regulator worked fine in my test setup, when it was fitted in the unit it was intended for powering, the voltage stabilisation was not very good.
I discovered that the extra length of cable to the ammeter and back was essential to make the regulator work correctly. The regulator seemed to need a resistance of at least .4 ohms in the supply to the load.
The problem was solved by changing the 1uf ceramic capacitor ( I initially used ceramic because of my hatred of Tantalum) on the output pins to ground for a 2.2uF tantalum capacitor.
The notes on the National Semiconductor data sheet say;
"Although the LM138 is stable with no output capacitors, like any feedback circuit, certain values of external capacitance can cause excessive ringing. This occurs with values between 500 pF and 5000 pF. A 1 μF solid tantalum (or 25 μF aluminium electrolytic) on the output swamps this effect and insures stability."

Mike

[Crackle]

The final circuit I used is below.
   

Mike

[Mike Watterson]

1uF ceramic are only good for audio. No use for RF.

[Crackle]

I didn't know that, I thought the larger the value the more effect it had on all frequencies.

Mike

[boater sam]

I didn't know that, I thought the larger the value the more effect it had on all frequencies.

Mike

Its the fact it is ceramic I think, drops off value at RF.

[radioman]

- In reply to Crackle's post - "I thought the larger the value the more effect it had on all frequencies"

In theory maybe, but the inductance of the package plus connecting wires will resonate with the capacitance and this resonant frequency gets lower with increasing capacitance value.
For lots of information about capacitors have a look at this website :-
http://www.iequalscdvdt.com/

Andy