Panel Mount Voltage Regulator / Knob

[Robin Hewitt]

I'd still go with the LM338T, it might run a tadge warm but it's so cheap and easy. Heck, just PM me a mailing address and I'll send you a freebie

[irving2008]

I'd still go with the LM338T, it might run a tadge warm but it's so cheap and easy. Heck, just PM me a mailing address and I'll send you a freebie

Robin, it'll run rather more than warm. On a decent 4degC/W heatsink the dissipation limit is 12.5W (at 25degC ambient) so at 20v input and 5V output the allowable output current before it goes into thermal shutdown is less than 800mA, rising to 2.5A at 15v output. Maximum output would be 18.8v at 5A.

Also there's no current limit, only short circuit protection, again not meeting expectations...

I'm not saying that you couldn't use this, and its cheap and simple, but Lee will have to downgrade his expectations and thats not making best use of his 160VA transformer when a 50VA would do.

regards,
Irving...

[Jeremy]

re: using an LM338.

I guess it depends as to whether or not current limiting is really needed and whether or not there is a need to supply a constant high current.

The power supply in that pic I posted is a good example. It uses a 5A linear voltage regulator IC rather similar to the LM 338, but only needs to deliver full current for short periods when charging up the 1.445 F, 20V capacitor bank. The same might be the case when driving steppers with PWM chopper control, for example, where the average current might actually be pretty low much of the time.

Jeremy

[irving2008]

I agree Jeremy. The point is that Lee wants a bench supply with a specific capability. If I was making a fixed supply then the LM338 is an ideal candidate. IF you want full output current to a given spec over a wide voltage range then linear regulators arent going to cut it. All depends on whats required. If Lee wants to reduce his expectations to 0 - 20v at .5A max guaranteed then the LM338 will do the job without overheating across the range. If he wants >1A output guaranteed then the LM338 isnt the way to go.... as the heatsink will be more expensive than the rest of it all together....

There is a secondary issue - is the goal just to produce something that'll 'just about do' or also to learn more about how to design a proper supply?

Irving...

[Robin Hewitt]

Robin, it'll run rather more than warm. On a decent 4degC/W heatsink the dissipation limit is 12.5W

OTOH if he's bolting it to an aluminium enclosure he's going to get heaps better than 4degC/W.

I'm not saying there isn't anything "better", but does this particular pudding really require that many eggs?

[irving2008]

OTOH if he's bolting it to an aluminium enclosure he's going to get heaps better than 4degC/W.

I'm not saying there isn't anything "better", but does this particular pudding really require that many eggs?

4degC/W is 300 sq cm of 1.5mm ali sheet held vertically and supported at corners only - I'd be surprised if the enclosure gave anything close to around 10degC/W - aluminium boxes are incredibly poor heatsinks (the type of alloy used makes a huge difference as well).

[Eddy]

A small fan will do wonders for heat dissipation. You can pick up small fans that are 5v and 12v at several surplus shops.

I placed a 50mm fan over a dual 15A motor controller with no heatsink and it kept it sufficiently cool even at the full 15A pull. The most current I could pull with a heatsink and no fan was about 8-9A.

Eddy
www.wrighthobbies.net

[irving2008]

A small fan will do wonders for heat dissipation. You can pick up small fans that are 5v and 12v at several surplus shops.

I placed a 50mm fan over a dual 15A motor controller with no heatsink and it kept it sufficiently cool even at the full 15A pull. The most current I could pull with a heatsink and no fan was about 8-9A.

Eddy
www.wrighthobbies.net

You are indeed correct, however there is a world of difference between a pulse width modulated motor controller with a high peak current but low RMS value and a linear regulator with a constant high dissipation.

[Eddy]

You are indeed correct, however there is a world of difference between a pulse width modulated motor controller with a high peak current but low RMS value and a linear regulator with a constant high dissipation.

You missed my point. The motor controller has to dissipate the heat caused by the voltage loss of the high and low side Vsat. A loss of 2V at 15A is 30 watts of heat to dissipate per IC (there's 2 on the board), a total of 60 watts with no heatsink. The movement of the air across the surface of the IC and the PCB is able to dissipate the 60 watts.

Eddy

[irving2008]

You missed my point. The motor controller has to dissipate the heat caused by the voltage loss of the high and low side Vsat. A loss of 2V at 15A is 30 watts of heat to dissipate per IC (there's 2 on the board), a total of 60 watts with no heatsink. The movement of the air across the surface of the IC and the PCB is able to dissipate the 60 watts.

Eddy

Eddy, I didnt miss the point - if its a motor controller is not dissipating 60w continuously, as its almost certainly a PWM arrangement. Its only dissipating that wattage during the 'on' cycle at a high frequency. take for example a common controller chip from ST. Its capable of 30A without a heatsink... but thats because the switching frequency is 10khz. for a single pulse at that frequency the thermal characteristic is <0.1degC/W so a heatsink of only 16sq cm of copper on the PCB will suffice to keep the mean junction temperature below 125degC. You could never run a linear circuit at that sort of load. The actual static chip thermal resistance is typical at 15decC/W... at 60W the jiunction temp would be 15*60deg over ambient = 900deg... it would have long since melted!

For the record I have designed systems for both miltary (-55 -> +125degC) and space applications (-100 -> +250degC)!