5 amp stabilised supply

[Crackle]

Following on from my thread on Fake LM338T regulators, which is just about exhausted now. (unless someone else wants to use the thread to notify of other fake devices on the market, which I feel would be a most useful thread.)

I wondered if the experts here could comment on my proposed new layout for when I do finally manage to buy a LM338T that actually works as it is supposed to.
I may yet be reduced to paying the RS price of £9.87 (for 5, min quantity)

My layout is on Veroboard and looks as below.
   
it will be supported by the LM338T and an additional bracket.

thanks
Mike

[Refugee]

I would interchange the capacitors and resistor/pot to reduce the risk of oscillation.

[Crackle]

do you mean move all the capacitors to as close to the IC and the 2 resistors away from the IC.
I thought that R1 was best close to the out pin.
Mike

[Nick]

The decoupling capacitors need to be as close to the IC pins as is possible.

Note that Veroboard has quite thin tracks: The track of a stripboard is 2mm (78 mil) wide. The holes drilled into them are 0.8mm (31 mil) in diameter, leaving a total (0.6mm [23 mil] on each side of the hole) of 1.2mm (47 mil) for a maximum usable current path. Thickness of Veroboard/Stripboard at 1 ounce per square foot is 35 microns (1.37 mil) thick.

Assuming that you have an ambient of 25 degrees C (77F) and can tolerate of up to 50 degrees C (90F) more, then fed into a PCB track width calculator it spits out:
5.4 Ampere

So, at 5A you are near the recommended limit for the track size. I'd bulk up the tracks with either plain solder or a bit of tinned copper wire buried in solder on the IN and OUT tracks.

Edit: ISTR that Veroboard has a recommended maximum current capacity of 2A, but I can no longer find that - the product is now owned by Vero Technologies Ltd whose website is useless.

[ppppenguin]

Mike, I think you've discovereed that an apparently simple regulator isn't as easy as it might seem.

Is it worth asking the question about what you want the PSU for? Your proposed design is going to need a very hefty heatsink as well as all the other issues that you've had with 338 regulators. If I wanted a 5A PSU I'd be using a switchmode unless there was some very good reason not to.

You can buy good quality mains input switchmode PSUs. Both chassis mount boards and "lump in cord". Don't know if you need variable voltage which makes thing s a bit harder. For 5V at 5A there are low cost regulator modules such as: http://www.hobbytronics.co.uk/d24v50f5-5...-regulator haven't looked for other output voltages.

Incidentally it's probably not worth building your own switchmode regulator. I've designed with the original National "simple switcher" chips but would now always look to a packaged solution if possible. If I want several amps at low voltage and low disspation it's much easier to buy than make.

If you must have a linear reg, the also consider the LM1084, a more modern part that may be less expensive than a good 338.

PS: The LM1084 is udner £3 from RS: https://uk.rs-online.com/web/p/low-dropo...s/5339331/

[Crackle]

The decoupling capacitors need to be as close to the IC pins as is possible.

I'd bulk up the tracks with either plain solder or a bit of tinned copper wire buried in solder on the IN and OUT tracks.

Thanks Nick
I always bulk up the track on a device like this with a run of solder and could always add a copper wire to it as well.
Like this then.
   
Jeffrey
I want 14-15 volts out, so the 5v one is no good for me.
The current will be an average of 1.5 to 2 amps, Which my LM317's and the fake ones just cant supply without dropping the out volts.

Mike

[ppppenguin]

No idea about quality but here's 15V 2A for £7.99: https://www.amazon.co.uk/DC-1520-supply-...B00ECWZKDE

Or various options from CPC, so presumably reputable: https://cpc.farnell.com/c/electrical-lig...r-supplies

I'm sure there's a UK supplier for this Meanwell unit: https://www.rhydolabz.com/miscellaneous-...-2401.html Meanwell have a good reputation.

I'm struggling to find just regulator modules. Sure they exist but the search results are swamped by PSUs.

As I said before, unless I truly had to have a linear unit I'd just buy off the shelf. Even if a switcher is a bit noisy you can bung it in a metal box with a mains inlet filter and some decent output filtering.

[ppppenguin]

If you insist on DIY then loads of low cost 15V 2A linear reg chips. Example: https://www.rapidonline.com/st-l78s15cv-...or-47-3306

It's still a big heatsink though if you want full output and a reasonable input margin of input voltage.

[Mike Watterson]

I've already suggested that at more than 1.5A it's better to buy a ready made PSU. Especially at 5A, absolutely at 10A or more.
Also there are decent very low RF noise 14V 5A, 10A, 20A SMPSUs for two-way radios. The Amateur market ones are better than the ones for illegal CB. Any legal CB will run off 1.5A or even 1A.
Some people use a car or truck battery and a smart charger.

I've used a big transistor on giant heatsink driven by a regulator for 10A for battery replacement. The regulation doesn't need to be good. 14.5V on receive and 13.5V on transmit is fine. Many decent amateur rigs are OK down to 11V (reduced output) and up to 15V or even 16V. Most important aspects are lack of mains hum and lack of RFI, not absolute regulation like logic boards.

[BusyBee]

I am not sure about noise induced in the other circuitry in one of my homemade multi output units (although it's not been an issue) but I designed the unit to provide somewhat greater than one amp output on 6.3V 5V and 1.25V with fairly small heatsinks. I just sat them on top a fan providing constant cooling. It has seemed to have worked Ok as the unit does what I ask. Not quite as far as you are going, but seems to readily provide well over one amp at 6.3V indefinatly and copes with short lived surges well. The biggest problem I get with linear regulators seems to be the input / output differential putting up dissipation. I have used tandem circuits to reduce this but it starts to get hard work and is obviously inefficient. This is seldom too much of an issue though. When I mount the components close they seem to work without problem. Just overdesign for current flow. If trying to go as far as you I would probably use a heatsink which works well with fan cooling, although my tendency these days is to go with switch mode units I have to hand and apply extra smoothing etc if required. I have also implemented smaller local regulators for individual stages to split the problem a bit. I remember I do have a LM338 circuit for power to higher current valve heaters (usually powered from a good 12V swith mode unit). It does run 2.5A Ok if I remember although is still heatsink limited if left running for any thing over a few miniutes (external heatsink so I gauge limits by hand).

Tracy