UK Vintage Radio Repair and Restoration Powered By Google Custom Search Vintage Radio and TV Service Data

Go Back   UK Vintage Radio Repair and Restoration Discussion Forum > Other Discussions > Homebrew Equipment

Notices

Homebrew Equipment A place to show, design and discuss the weird and wonderful electronic creations from the hands of individual members.

Closed Thread
 
Thread Tools
Old 25th Jul 2015, 1:59 am   #1
Otari5050
Pentode
 
Otari5050's Avatar
 
Join Date: Jun 2015
Location: Wellington - New Zealand
Posts: 112
Default High current variable linear supply

I'm gearing up to build a high current variable linear power supply and I'm curious to know if anyone has a tried and tested design they'd like to share.

My plan is for 0 - 30 volts at 20 amps max.

I've got some nice 2SD797 pass transistors rated at 30 amps each, so 6 of those in parallel (with associated resistors) should handle the current without too much stress.

The design I am looking at uses an LM723 regulator driving a darlington pair then the pass transistors.

If you have a unit you've built that does the biz and is reliable, I'd love to see it.
Otari5050 is offline  
Old 25th Jul 2015, 5:09 am   #2
Radio Wrangler
Moderator
 
Radio Wrangler's Avatar
 
Join Date: Mar 2012
Location: Fife, Scotland, UK.
Posts: 22,800
Default Re: High current variable linear supply

The LM723 is the classic part for the job and it works very well.

BUT

723's are very sensitive to any RF in their environment. RF gets self-rectified, causes large DC offsets and seriously affects operation. If you're going to use the supply to power or anywhere near a radio transmitter, take great care of filtering and decoupling.

Most power supplies in that sort of range use things like thyristor pre-regulators to reduce heat dissipation when sourcing high current when set to low output voltages.

Manuals and circuits or a lot of HP stuff are available on BAMA and on the Agilent website (now renamed again as Keysight ! )

David
__________________
Can't afford the volcanic island yet, but the plans for my monorail and the goons' uniforms are done
Radio Wrangler is offline  
Old 25th Jul 2015, 6:49 am   #3
Diabolical Artificer
Dekatron
 
Diabolical Artificer's Avatar
 
Join Date: Jul 2009
Location: Sleaford, Lincs. UK.
Posts: 7,637
Default Re: High current variable linear supply

Here's a schematic for a PSU that I've built several of. It uses an LM317 with TIP3055's. It's very stable and quiet with no ripple. I used a scaled down version for a 24v PSU for a mixing desk, no hum at all. Obviously you'll need to tweek it a bit especially as it won't go down to 0v because of the 1.25 reference voltage, but there's a way to connect the 317 with negative V I think to get your 0v. Others here can help there.

The HV diodes like the BYV26E can be left out as can the HV caps, as the original design was intended for use with HV testing.Hope this helps,

Andy.
Attached Thumbnails
Click image for larger version

Name:	LM317-30V-variable-high-power-supply-6-20A-max.jpg
Views:	3046
Size:	69.3 KB
ID:	111011  
__________________
Curiosity hasn't killed this cat...so far.
Diabolical Artificer is offline  
Old 25th Jul 2015, 11:36 am   #4
'LIVEWIRE?'
Rest in Peace
 
'LIVEWIRE?''s Avatar
 
Join Date: Mar 2006
Location: N.W. Oxfordshire(Chipping Norton)
Posts: 7,306
Default Re: High current variable linear supply

That circuit is very similar to an ex-demo stand PSU which I've used for many years now to power things needing up to 15-16v. @ up to 10A. It uses 3 x 2N3055 (TO3 versions) in parallell. TBH I forget what the regulator circuit is (have the diagram somewhere!) but mine will go down to 0v.
Regarding Otari 5050's post, I'm just wondering why 6 x 2SD797s in parallel would be needed for a 20Amp PSU if eaxh is rated at 30Amps Max.?. Surely two in parallell would be sufficient?
'LIVEWIRE?' is offline  
Old 25th Jul 2015, 12:27 pm   #5
turretslug
Dekatron
 
turretslug's Avatar
 
Join Date: Nov 2011
Location: Surrey, UK.
Posts: 4,385
Default Re: High current variable linear supply

Generally, getting bucket-loads of amperage rating for pass transistors isn't the head-ache- it's when you add up all the thermal hindrance points between individual device junctions and free air that the bottle-neck occurs and it means that a surprisingly large array of transistors ends up being needed to handle the dissipation- especially, as RW points out, when there is a high input voltage and a requirement for the output to go low.
turretslug is offline  
Old 25th Jul 2015, 1:00 pm   #6
mhennessy
Dekatron
 
Join Date: Jul 2007
Location: Evesham, Worcestershire, UK.
Posts: 4,241
Default Re: High current variable linear supply

For such a large range of output voltage, I'd consider using a centre-tapped secondary winding, and use a relay (or SCRs if you're feeling clever) to switch between tappings. For example, to get 30V DC, you'll need a minimum of 30V AC from the transformer (and that's cutting it a bit fine). I'd start with a transformer with dual 18V secondaries put in series.

The control circuit is simply a comparator looking at the output voltage, and will switch from 18 to 36V AC when the output voltage reaches a certain threshold (say, 14V DC). Or, instead, you could look at the differential voltage across the regulator (Vin-Vout). You would need to build in some hysteresis, and also a time delay to avoid problems with dynamic situations.

Such a scheme is employed in many commercial bench power supplies - often with many more tappings to give more options. A lot of complexity, perhaps, but heat sinks are very expensive!

Alternatively, some used a pre-regulator scheme, based on SCRs. These work a bit like a lamp dimmer, chopping up the AC from the mains transformer so that the DC voltage going in to the regulator was no higher than it needed to be. Somewhere I'll have some schematics from Wier - IIRC, those units went from 0 to 60V.

The more modern equivalient is to use a switched-mode DC-DC converter between the smoothed DC and the linear regulator. The latter is able to remove most of the switching noise, so such a unit is "cleaner" than a totally switched-mode power supply (and has better transient response). But that gets complicated for DIY, especially at those currents.

Whatever way you look at it, this is going to be big, heavy and hot! A decent chassis is a must. You might need inrush protection to stop the mains fuses blowing (a thermal breaker rather than a fuse might be worth doing - it can double as the mains switch). I also recommend simple bi-metallic switches on the heatsink(s), and if you use fans, you'll want to include thermal control (a thermistor and 2 transistors). The more transistors you can afford to put in, the better - they lead a hard life in a power supply. The thermal resistance from junction to heat sink will be paralleled up, which reduces the size of the heat sink. One of the hardest things to resolve at that sort of current is what exactly is earth? In a power supply, the output terminals are the point of reference, so expect to have to think carefully about where to return every component that appears to need a ground connection.

Good metering is nice. Be aware that some of the very cheap modules that you can get on eBay are very good indeed - others are not.

Good luck,

Mark
mhennessy is offline  
Old 25th Jul 2015, 1:59 pm   #7
KeithsTV
Nonode
 
KeithsTV's Avatar
 
Join Date: Aug 2007
Location: South Bradford, West Yorkshire, UK.
Posts: 2,571
Default Re: High current variable linear supply

I'll echo what Mark has said and would suggest adding remote sensing to ensure the load receives the set voltage.

The more pass transistors the better. Each transistor should have a low value resistor in its emitter which helps to equalise the current distribution between the transistors. These resistors can be used to sense the current for any current limit circuit and also for an output current meter.

A switched secondary is also a good idea to minimise the dissipation in the transistors, at a low voltage output and high current without this switching you could be dissipating around 600W. The secondary could be switched by a relay of SCR but what ever is used will have to be rated at 30A or more.

Keith
KeithsTV is online now  
Old 25th Jul 2015, 2:27 pm   #8
mhennessy
Dekatron
 
Join Date: Jul 2007
Location: Evesham, Worcestershire, UK.
Posts: 4,241
Default Re: High current variable linear supply

I've just looked up SOA graph on the datasheet for the 2SD797.

If you didn't switch the secondary, you'd have somewhere in the region of 40 to 50 volts going in to the regulator. Worst-case - 20 amps into a short circuit - these transistors are only good for 2.5 to 3.5 amps each. And that's with good heat sinking!

With half that input voltage, then they'll do 7.5 to 10 amps each. Much, much safer...

It has to be said that the SOA is usually a problem with big BJTs. MOS-FETs are usually much better in this regard.

Interestingly, Thurlby-Thandar use four 2N3055s in their PL320 design. That's 0.5 amps each!

With 50V in, 20A out, into a short-circuit, that's a power dissipation of 1000 watts. That's 167 watts per device. From the data sheet, it appears that the thermal resistance from junction to case is 0.8 degrees C per watt. That implies a temperature difference of 133 degrees across just the thermal resistance of the device itself. Before considering the thermal resistance of the case-to-sink interface, and the thermal resistance of the heat sink itself. Ouch

Switching the secondaries suddenly seems more attractive. But as Keith says, the switch will need to be chunky. With a full-wave rectifier, peak charging currents are usually around 3 to 4 times higher than the average DC current.
mhennessy is offline  
Old 25th Jul 2015, 3:41 pm   #9
Herald1360
Dekatron
 
Herald1360's Avatar
 
Join Date: Feb 2007
Location: Leominster, Herefordshire, UK.
Posts: 16,527
Default Re: High current variable linear supply

As a rule of thumb, you can dissipate about 40W in a "250W" TO3 on a decent fan blown heatsink, or up to perhaps 100W if you go for a watercooled copper plate arrangement. RCS258, 2N6259, MJ15024 were all used.

I had some fun years back with linear magnet power supplies with either motorised variac or SCR preregulators servoed from the passbank Vce. They were on the primary side of the three phase mains transformers.
__________________
....__________
....|____||__|__\_____
.=.| _---\__|__|_---_|.
.........O..Chris....O
Herald1360 is offline  
Old 25th Jul 2015, 4:59 pm   #10
mole42uk
Nonode
 
mole42uk's Avatar
 
Join Date: Aug 2010
Location: Resolfen, Wales; and Bristol, England
Posts: 2,588
Default Re: High current variable linear supply

My DIY version uses 4 2N3055s, works from about 5 - 25V and will pass 20A all day if it's needed. It has very large heatsinks so each transistor running at 100W keeps sufficiently cool. There is a variable current-limit control so it can operate in constant-current or constsnt-voltage mode. Two analogue meters - volts and amps and a brightly-lit 'ON' switch to remind me to switch it off.
It was cobbled together about 15 years ago and has run faultlessly ever since whenever I need a low-voltage supply.
I'm sure I could dig out a circuit from some file somewhere.....
__________________
Richard

Index:
recursive loop: see recursive loop
mole42uk is offline  
Old 25th Jul 2015, 6:44 pm   #11
Herald1360
Dekatron
 
Herald1360's Avatar
 
Join Date: Feb 2007
Location: Leominster, Herefordshire, UK.
Posts: 16,527
Default Re: High current variable linear supply

Sorry, there's no way a 2N3055 can run 100W unless its case is held at 25C. That's not physically possible on an aircooled sink given even the best possible 0.5C/W interface between the device and the heatsink unless you blast the sink with some arctic air.

100W between the four is feasible, though.
__________________
....__________
....|____||__|__\_____
.=.| _---\__|__|_---_|.
.........O..Chris....O
Herald1360 is offline  
Old 25th Jul 2015, 6:47 pm   #12
mhennessy
Dekatron
 
Join Date: Jul 2007
Location: Evesham, Worcestershire, UK.
Posts: 4,241
Default Re: High current variable linear supply

The 2N3055 has a small-ish die, meaning that the thermal resistance between junction-to-case is quite high at 1.52C/W. Asking one to dissipate 100W means a temperature difference of 152C between case and junction - so although the case might seem to be at a reasonable temperature, the junction inside might not be. The maximum junction temperature is 200C.

The On-Semi datasheet shows a helpful graph on the front page, which says that to dissipate 100W, the case must be kept at around 50C or less, which fits in with what I've said above.

Of course, you need a decent thermal path between the case and the heat sink in order to keep the case down at 50C. A mica washer and properly applied grease might be around 0.3 to 0.5 C/W, so that's a temperature difference of 30 to 50 degrees C at 100 watts. Which suggests that the heat sink itself needs to be kept in the range of 0 to 20 degrees to keep the die at 200C when dissipating 100W. That could be tricky

Upgrading from 2N3055 to something like the MJ15003 might be worthwhile as they have a thermal resistance between junction and case of 0.7C/W - less than half of the 2N3055. But they're about 3 times the price.

As a principle, it's good to over-engineer anything that runs hot. Reducing temperatures will increase the life-span of not just the semiconductors, but nearby capacitors, for example. Also, thermal drift is minimised. I "compulsively collect" heat sinks because I enjoy power engineering

Another thought - if you're thinking of using the ballast resistors to double as the current sense resistors (for limiting and the current meter), then use resistors to take a "mix" of the voltage across all of the ballast resistors. If you just sample 1, it's likely to be inaccurate, and the error will change with operating current and die temperature. I tend to use a separate sense resistor, but at 20A, you'd probably be looking to minimise resistive losses so might prefer to not do that. Something like 0.1 ohms would be OK; it would drop 0.3 volts (20A/6) - more than enough to swap the differences in Vbe - and only wast around 1W of heat.

However you sense the current, choose resistor(s) that have a low temperature coefficient to maintain accuracy. The metal-clad aluminium types are very good for their price. Use the 4-wire technique when hooking them up.

EDIT: Post crossed with Chris...
mhennessy is offline  
Old 25th Jul 2015, 7:02 pm   #13
KeithsTV
Nonode
 
KeithsTV's Avatar
 
Join Date: Aug 2007
Location: South Bradford, West Yorkshire, UK.
Posts: 2,571
Default Re: High current variable linear supply

Another suggestion is to add foldback current limiting. In case of a short on the output the current could be limited to 1-2A limiting the transistor dissipation to 10s rather than hundreds of watts.

I've attached the circuit of the series pass transistors for a 13.8V 10A power supply I build back in 1983. This uses summed voltages across the emitter resistors for current limiting and driving a meter.

There is also an SCR across the output to crowbar the output in the event of the output overvolting.

Keith
Attached Thumbnails
Click image for larger version

Name:	PSU Transistors.jpg
Views:	979
Size:	41.1 KB
ID:	111024  
KeithsTV is online now  
Old 25th Jul 2015, 7:37 pm   #14
mhennessy
Dekatron
 
Join Date: Jul 2007
Location: Evesham, Worcestershire, UK.
Posts: 4,241
Default Re: High current variable linear supply

Looks good

Also, it highlights that the driver transistor might need to be a surprisingly "meaty" device. Some of the TTi supplies I've see use a 2N3055 as the driver!

Foldback is relatively rare in a general-purpose supply, but it might be worth adding it as a switch on the front panel. Of course, the thermal cutouts I mentioned in my first post are a useful fall-back for when the heat sinking isn't up to the job for continuous worst-case operation.

Some supplies have the crowbar with a continuously variable trigger voltage. It's an extra bunch of circuitry, but definitely worth considering
mhennessy is offline  
Old 25th Jul 2015, 8:34 pm   #15
Diabolical Artificer
Dekatron
 
Diabolical Artificer's Avatar
 
Join Date: Jul 2009
Location: Sleaford, Lincs. UK.
Posts: 7,637
Default Re: High current variable linear supply

Fascinating thread. I love building linear PSU's, being the first circuit's I,ve "designed" - pinched a few idea's)from working out component value's to making a PCB etc.Still lot's to understand and improve on.

I too collect heatsink's compulsively,usually off scrap PCB's. PC CPU heatsink's make good HS's as theyre chunky and have a fan fitted. The problem is you can't really calculate heat transfer as pulled HS's have no data sheet;heatsink's don't have ID numbers.

A few questions. Why would mosfet's be better in this application? By a switched secondary do you mean as in a switch mode PSU employing an IC, PWM or similar. I'm familiar with foldback on a mixing desk but not in PSU; what does it do in this case?

I'd like to learn more on the topic of dissipating heat for semiconductors; any good articles you'd recommend? I'm looking at making a simple DC load for testing PSU's. It uses an dual op amp as voltage followers to switch a pair of mosfet's (FQA24N60) The problem of heat dissipation is shared by both PSU's and DC load's.

Going to have to learn about crowbar protection, it seems to be a basic building block of a lot of circuits.

Andy.
__________________
Curiosity hasn't killed this cat...so far.
Diabolical Artificer is offline  
Old 25th Jul 2015, 9:27 pm   #16
Radio Wrangler
Moderator
 
Radio Wrangler's Avatar
 
Join Date: Mar 2012
Location: Fife, Scotland, UK.
Posts: 22,800
Default Re: High current variable linear supply

A point made a few posts back by mhennessy is important:

If using the emitter 'current sharing' resistors for your current meter shunts, use a resistor mixer to sum them together instead of relying on just one.

Take this a bit further. You're building a supply with current limiting and a constant current mode. I've seen designs where the current sensing is on one transistor only. Power transistors run hard do sometimes fail, and if that is the one that fails, suddenly the PSU loses its current limiting function, and will supply current to the point of destruction. If working into a heavy load, the remaining transistors share the full current taken and get even hotter, so the next one fails quite soon. Then it gets even worse and the next one pops in an accelerating 'zipper' failure.

A design challenge is to control the current of each transistor so that if one fails, the current limit scales down proportionately.

Another worry is that '2N3055' is the dumping ground for any sexier power transistor wafers that don't meet spec. The 2N3055 spec are open ended and rather easy to meet. It allows scrapped wafers to be flogged and still make money.

Consequently, that 2N3055 bought almost anywhere could have an awful lot more Ft than the 800kHz minimum spec might imply.

I've de-lidded a number and tried to identify just what's inside.

Of course, some of the firms buying dud wafers and packaging them don't even care whether it really meets or exceeds the 2N3055 minima. The no-name things in steel cans are the worst lottery.

It's a bit like the business of fake capacitors with little ones mounted inside larger cans, only the shonky 2N3055 got there years earlier.

David
__________________
Can't afford the volcanic island yet, but the plans for my monorail and the goons' uniforms are done
Radio Wrangler is offline  
Old 25th Jul 2015, 9:35 pm   #17
GMB
Dekatron
 
GMB's Avatar
 
Join Date: Aug 2003
Location: near Reading (and sometimes Torquay)
Posts: 3,086
Default Re: High current variable linear supply

I always add a simple thyrister "self destruct" circuit to the output so that if the regulation fails then the thyrister blows the fuse.
GMB is offline  
Old 25th Jul 2015, 9:38 pm   #18
KeithsTV
Nonode
 
KeithsTV's Avatar
 
Join Date: Aug 2007
Location: South Bradford, West Yorkshire, UK.
Posts: 2,571
Default Re: High current variable linear supply

With a lower power PSU with a current limit circuit if you short circuit the output the current is limited to that set by the limit circuit. E.g. if the PSU has a maximum current of 1A, the current limit is set to 1A and when the output is short circuited the maximum current that can flow is 1A. This does mean that the series pass transistor(s) are passing 1A. If the output of the rectifier is 30V this means there will be 30V across the transistors and they will dissipate 30W. In the case of the proposed 30V PSU with a current limit of 20A under the output short circuit condition the transistors would be dissipating 600W and would probably eventually fail.

With a foldback current limit the current limit falls as the output voltage falls once the maximum current has been reached. The PSU I mentioned in post 13 has foldback current limiting. If the load tries to draw more than 10A the output voltage will fall and the current will also fall until under a short circuit condition the current is limited to 1A as shown in the diagram. The output transistors will be dissipating about 15W under a short circuit condition compared to about 150W with a conventional current limit.

Designing a foldback limit can be difficult as it's effectively a negative resistance device and can be prone to go unstable. When I was originally building the PSU at one point it did oscillate with a 10A square wave passing through a large wirewound rheostat which was actually audible.

HTH

Keith

PS take a look at the notes I posted for the 2014 Golborne talk on Power supplies. This may help you with the thermal calculations for the heatsinks.

https://www.vintage-radio.net/forum/...=110286&page=3
Attached Thumbnails
Click image for larger version

Name:	current limit.jpg
Views:	236
Size:	10.6 KB
ID:	111030  

Last edited by KeithsTV; 25th Jul 2015 at 9:44 pm. Reason: added link
KeithsTV is online now  
Old 25th Jul 2015, 10:06 pm   #19
Radio Wrangler
Moderator
 
Radio Wrangler's Avatar
 
Join Date: Mar 2012
Location: Fife, Scotland, UK.
Posts: 22,800
Default Re: High current variable linear supply

I was in the right place at the right time and picked up a large Farnell PSU for £20 it does 0-35v at 10A and 0-70v at 5A and only needed a bit of a clean-up.

Big monster power supplies aren't in as much demand as handier sizes, and market prices reflect this. It may be worth looking around before you start to build.

David
__________________
Can't afford the volcanic island yet, but the plans for my monorail and the goons' uniforms are done
Radio Wrangler is offline  
Old 26th Jul 2015, 4:49 am   #20
mole42uk
Nonode
 
mole42uk's Avatar
 
Join Date: Aug 2010
Location: Resolfen, Wales; and Bristol, England
Posts: 2,588
Default Re: High current variable linear supply

Quote:
Originally Posted by Herald1360 View Post
Sorry, there's no way a 2N3055 can run 100W unless its case is held at 25C. That's not physically possible on an aircooled sink given even the best possible 0.5C/W interface between the device and the heatsink unless you blast the sink with some arctic air.

100W between the four is feasible, though.
I'm going to have to dig it out of the shelf now and check that my memory is still functioning! It may be that I've used something else for the power transistors after all.
__________________
Richard

Index:
recursive loop: see recursive loop
mole42uk is offline  
Closed Thread

Thread Tools



All times are GMT +1. The time now is 9:55 pm.


All information and advice on this forum is subject to the WARNING AND DISCLAIMER located at https://www.vintage-radio.net/rules.html.
Failure to heed this warning may result in death or serious injury to yourself and/or others.


Powered by vBulletin®
Copyright ©2000 - 2024, vBulletin Solutions, Inc.
Copyright ©2002 - 2023, Paul Stenning.