High current variable linear supply - Page 3 - UK Vintage Radio Repair and Restoration Discussion Forum

Radio Wrangler · 30th Jul 2015, 6:59 am

One of the most sensitive instruments around, the Agilent Noise Figure Analyser, measures noise contributions down to the theoretical thermal noise floor of temperatures getting down to below liquid nitrogen. It does this with switch-mode power supplies in the box.

Switchers are useful, but you need toplan out your screening and filtering.

With linear supplies you wind up with a reservoir voltage set by your chosen max output voltage, and a current set by your chosen current capability and this sets the inevitable heat dissipation. Full stop.

You can have smaller sinks, but the higher temperature means lower reliability. Power supplies were always the cinderella - given to a student or a new grad to design, yet one whoopsie and they could destroy the rest of whatever they powered. Reliability is just as important as the sexier bits of a product. You can have smaller and smaller sinks at the expense of noisier and noisier fans.

The big HP supplies with heavy power output uses thyristor switch pre-regulators to get out of the heat trap.

It's difficult to get good bandwidth and good cleanliness out of a pure switcher (difficult, not quite impossible) but switching pre regulators with a final linear regulator to remove noise can be a good compromise.

You have to think how your load current might fluctuate, and how many millivolts you're prepared to let the output swing. Output capacitors help, but for high currents and small voltage swings, they soon become immense. The art is in haveing some capacitance and having a regulator bandwidth that smoothly takes over the job as the capacitor rolls off at lower frequencies.

If it wasn't for the shipping costs to NZ... Europe is awash with surplus test equipment after most of the electronics industry ran away to join the chinese circus.

David

trh01uk · 30th Jul 2015, 8:39 am

I am in the middle of building a linear power unit with an output of about 7V up to 13 amps. Not as powerful as yours, but it has similar problems. Its not difficult to get it all sorted when everything is working fine - but things like a short on the output are tricky to handle because the output transistor dissipation suddenly goes through the roof, if you have simple current limiting. The entire output voltage is across the output transistor and the current is at absolute maximum.

There are various strategies for dealing with this problem. Current limiting is fine, if you have a big enough heatsink, but at the power levels we are talking about here, with 30V at 20A, that's at least 600W - a small bar fire.....Monitoring the temp and switching on a fan is a possibility. Current foldback is another good solution. Blowing a fuse is another, or perhaps a mechanical current trip, which is more convenient.

Another rather nice solution is to cycle the output if an overload is detected. The regulator IC switches on and off the output in overload conditions - so that the output is only on for a short time. The average dissipation in the output transistor is then multiplied by Ton/Toff, which if its 3% or so, becomes a fairly small number again: 600W gets reduced to 18W.

I am currently playing with a nice chip, which does this for you, the Texas Instruments UCC3837. Details at http://www.ti.com/product/ucc3837. I managed to buy some of these - I forget where - possibly Digikey. I could only find them in SO8 package, so a bit of a pain - but I used a SO8 to DIP8 converter board (very cheap on Ebay) which sorted that. I have it running on a veroboard layout for the moment.

Richard

Bazz4CQJ · 30th Jul 2015, 10:07 am

Can anyone offer any advice re component selection wrt the reservoir cap in a linear PSU? Am I right in thinking that capacitance and voltage are not the only parameters which ought to be specified here?

B

Radio Wrangler · 30th Jul 2015, 10:20 am

Ripple current and ESR

Design is best done by checking with a simulator. LT spice is worth learning and is free (Windows/Linux/OSX versions available)

David

trh01uk · 31st Jul 2015, 8:03 am

Quote:

Originally Posted by Bazz4CQJ

Can anyone offer any advice re component selection wrt the reservoir cap in a linear PSU? Am I right in thinking that capacitance and voltage are not the only parameters which ought to be specified here?

Ripple current is pretty important if the supply is to have long life. If you going to make the psu in a small box with no cooling, then internally the temperatures may go quite high - so you will want to check on the temp rating of the capacitors too. If you are bothered about cost (who isn't?), then I have generally found its better to use several caps in parallel, connected by hefty wires. The design I mentioned uses 3 x 22000uF caps in parallel.

I have found the best way to understand what is going on with a design is to use a simulation program. I can recommend a couple. One I use a lot is PSU Designer II available from http://www.duncanamps.com/psud2/. With this program you can get a graphical plot of just about any parameter you want, including the current flowing in the reservoir caps. You can play with the value of the capacitor's ESR and see what effect it has.

An alternative that runs in Excel is "fwrecta-1.xls". I've just tried to find a link to it, but can't. I can send you my copy if you want it - just PM me.

Richard

turretslug · 31st Jul 2015, 10:33 am

Several smaller capacitors also offers a surface area advantage for dissipating that ripple current heating. Theory says that, by time they're clamped, they'll occupy a greater volume but in practice they may end up being easier to accomodate than a single large cylinder of food-tin size or larger. Also, having several terminals and bus-bar linking means that the likely and often hefty several wires to/from rectifier, array of pass transistors, bleeder, control amp etc. is easier and neater to accomodate than bunched in one spot. Don't let the "common" reference point deviate too far from ideal, though.

I recall seeing a Philips application note for their PSU-intention capacitors where it stated that removing the plastic insulating sleeve (in this case, the characteristic translucent blue) would improve the ripple rating by typically 20% by improving cooling (though it could be argued that if you need to do this, the design is getting a bit marginal anyway!). It would also reduce the effectiveness of the isolation capability between capacitor plates and chassis- though I would never rely on sleeve alone if there was a need for serious "floating", e.g. voltage doubling, as the sleeve is all too easily damaged by metal clamps. If you need to do this, there are nylon clamps with bottoming flanges.

QQVO6/40 · 1st Aug 2015, 2:47 pm

Hello all.

An interesting thread.
I have just completed setting up a 13V power supply at 150Amps. Quite easy.

I had an inspiration. I used a second hand battery charger for an electric forklift.
If you dig in the right places you can get them for the price of the scrap copper.
It comes with 240V mains input. Phase control for voltage control. Fixed voltage for this one but with a little research I could make it variable. Crowbar over voltage protection. 2 large chokes and filter caps.
I have put it under load with a cro connected and there is little switching noise. A couple more caps might take care of it.
And no chance of overloading the thing!

Hardest thing I had to do was clean and panelbeat the case then spray paint it.

My next project is to work on another forklift battery charger for 28Volt at 100Amps. I have a GRC-106 radio transceiver that draws 47Amps at 28Volts.

Welding cables to connect this stuff!

Love it!!

Just my 2d worth.

Cheers,
Robert.

Diabolical Artificer · 2nd Aug 2015, 7:15 am

Regarding cooling of heatsinks, would a Peltier device be useful, perhaps with a fan as well ? I've not seen them used in commercial devices.

From what I've read so far the way to go seems to be a transformer with various taps, a triac pre regulator, then use of big R's to dissipate heat on the final stage.

I mentioned building a DC load earlier that uses op amp's and logic level mosfet's.
Would the same principle be useful for this application?

Andy.

30th Jul 2015, 6:59 am	#41
Radio Wrangler Moderator Join Date: Mar 2012 Location: Fife, Scotland, UK. Posts: 22,902	Re: High current variable linear supply One of the most sensitive instruments around, the Agilent Noise Figure Analyser, measures noise contributions down to the theoretical thermal noise floor of temperatures getting down to below liquid nitrogen. It does this with switch-mode power supplies in the box. Switchers are useful, but you need toplan out your screening and filtering. With linear supplies you wind up with a reservoir voltage set by your chosen max output voltage, and a current set by your chosen current capability and this sets the inevitable heat dissipation. Full stop. You can have smaller sinks, but the higher temperature means lower reliability. Power supplies were always the cinderella - given to a student or a new grad to design, yet one whoopsie and they could destroy the rest of whatever they powered. Reliability is just as important as the sexier bits of a product. You can have smaller and smaller sinks at the expense of noisier and noisier fans. The big HP supplies with heavy power output uses thyristor switch pre-regulators to get out of the heat trap. It's difficult to get good bandwidth and good cleanliness out of a pure switcher (difficult, not quite impossible) but switching pre regulators with a final linear regulator to remove noise can be a good compromise. You have to think how your load current might fluctuate, and how many millivolts you're prepared to let the output swing. Output capacitors help, but for high currents and small voltage swings, they soon become immense. The art is in haveing some capacitance and having a regulator bandwidth that smoothly takes over the job as the capacitor rolls off at lower frequencies. If it wasn't for the shipping costs to NZ... Europe is awash with surplus test equipment after most of the electronics industry ran away to join the chinese circus. David __________________ Can't afford the volcanic island yet, but the plans for my monorail and the goons' uniforms are done

30th Jul 2015, 8:39 am	#42
trh01uk Octode Join Date: Nov 2010 Location: Ross-on-Wye, Herefordshire, UK. Posts: 1,654	Re: High current variable linear supply I am in the middle of building a linear power unit with an output of about 7V up to 13 amps. Not as powerful as yours, but it has similar problems. Its not difficult to get it all sorted when everything is working fine - but things like a short on the output are tricky to handle because the output transistor dissipation suddenly goes through the roof, if you have simple current limiting. The entire output voltage is across the output transistor and the current is at absolute maximum. There are various strategies for dealing with this problem. Current limiting is fine, if you have a big enough heatsink, but at the power levels we are talking about here, with 30V at 20A, that's at least 600W - a small bar fire.....Monitoring the temp and switching on a fan is a possibility. Current foldback is another good solution. Blowing a fuse is another, or perhaps a mechanical current trip, which is more convenient. Another rather nice solution is to cycle the output if an overload is detected. The regulator IC switches on and off the output in overload conditions - so that the output is only on for a short time. The average dissipation in the output transistor is then multiplied by Ton/Toff, which if its 3% or so, becomes a fairly small number again: 600W gets reduced to 18W. I am currently playing with a nice chip, which does this for you, the Texas Instruments UCC3837. Details at http://www.ti.com/product/ucc3837. I managed to buy some of these - I forget where - possibly Digikey. I could only find them in SO8 package, so a bit of a pain - but I used a SO8 to DIP8 converter board (very cheap on Ebay) which sorted that. I have it running on a veroboard layout for the moment. Richard

30th Jul 2015, 10:07 am	#43
Bazz4CQJ Dekatron Join Date: Apr 2009 Location: Oxfordshire, UK. Posts: 4,935	Re: High current variable linear supply Can anyone offer any advice re component selection wrt the reservoir cap in a linear PSU? Am I right in thinking that capacitance and voltage are not the only parameters which ought to be specified here? B

30th Jul 2015, 10:20 am	#44
Radio Wrangler Moderator Join Date: Mar 2012 Location: Fife, Scotland, UK. Posts: 22,902	Re: High current variable linear supply Ripple current and ESR Design is best done by checking with a simulator. LT spice is worth learning and is free (Windows/Linux/OSX versions available) David __________________ Can't afford the volcanic island yet, but the plans for my monorail and the goons' uniforms are done

31st Jul 2015, 10:33 am	#46
turretslug Dekatron Join Date: Nov 2011 Location: Surrey, UK. Posts: 4,400	Re: High current variable linear supply Several smaller capacitors also offers a surface area advantage for dissipating that ripple current heating. Theory says that, by time they're clamped, they'll occupy a greater volume but in practice they may end up being easier to accomodate than a single large cylinder of food-tin size or larger. Also, having several terminals and bus-bar linking means that the likely and often hefty several wires to/from rectifier, array of pass transistors, bleeder, control amp etc. is easier and neater to accomodate than bunched in one spot. Don't let the "common" reference point deviate too far from ideal, though. I recall seeing a Philips application note for their PSU-intention capacitors where it stated that removing the plastic insulating sleeve (in this case, the characteristic translucent blue) would improve the ripple rating by typically 20% by improving cooling (though it could be argued that if you need to do this, the design is getting a bit marginal anyway!). It would also reduce the effectiveness of the isolation capability between capacitor plates and chassis- though I would never rely on sleeve alone if there was a need for serious "floating", e.g. voltage doubling, as the sleeve is all too easily damaged by metal clamps. If you need to do this, there are nylon clamps with bottoming flanges.

1st Aug 2015, 2:47 pm	#47
QQVO6/40 Hexode Join Date: May 2009 Location: Newcastle, Australia Posts: 321	Re: High current variable linear supply Hello all. An interesting thread. I have just completed setting up a 13V power supply at 150Amps. Quite easy. I had an inspiration. I used a second hand battery charger for an electric forklift. If you dig in the right places you can get them for the price of the scrap copper. It comes with 240V mains input. Phase control for voltage control. Fixed voltage for this one but with a little research I could make it variable. Crowbar over voltage protection. 2 large chokes and filter caps. I have put it under load with a cro connected and there is little switching noise. A couple more caps might take care of it. And no chance of overloading the thing! Hardest thing I had to do was clean and panelbeat the case then spray paint it. My next project is to work on another forklift battery charger for 28Volt at 100Amps. I have a GRC-106 radio transceiver that draws 47Amps at 28Volts. Welding cables to connect this stuff! Love it!! Just my 2d worth. Cheers, Robert.

2nd Aug 2015, 7:15 am	#48
Diabolical Artificer Dekatron Join Date: Jul 2009 Location: Sleaford, Lincs. UK. Posts: 7,670	Re: High current variable linear supply Regarding cooling of heatsinks, would a Peltier device be useful, perhaps with a fan as well ? I've not seen them used in commercial devices. From what I've read so far the way to go seems to be a transformer with various taps, a triac pre regulator, then use of big R's to dissipate heat on the final stage. I mentioned building a DC load earlier that uses op amp's and logic level mosfet's. Would the same principle be useful for this application? Andy. __________________ Curiosity hasn't killed this cat...so far.