30v 10A bench supply schematic. Is it any good???

[FotatoPotato]

Hello everyone,

I have been interested in building a bench power supply for a while now and just recently I got a 24v 10A transformer, and I want to put it to use. Because I am new to building power supplies I decided to just look up schematics for power supplies that would meet my requirements. I found this schematic and it looks pretty good. Even though it's only rated at 4A I’m sure that if I just beef up the transistors and power traces/ components then it would be fine for delivering 10A. Let me know if this is a good schematic and if I should build it.

Thanks

EDIT: I'm looking for 12v @10A for the High current and 30v @ 5A

[schmitt trigger]

Question yourself: 300 watts output from a 240 VA transformer?
Because this is a linear power supply, expect no more than about 170 watts output.

[FotatoPotato]

Sorry, I should have been more specific. Im looking for 12v @ 10A and 30v @ about 5A

[Audioguru]

The schematic is a partially modified Greek kit from at least 13 years ago and Chinese copies are available today. I fixed some of its problems by beefing up many parts and using two output transistors (with emitter resistors) and a 28V/4.3A (120VA) power transformer.
When it had a 24V transformer its output voltage could not reach 30V when loaded to 3A, maybe 25VDC.

MC34071 opamps in a DIL package are not available anymore but a TLE2141 is available in a DIL package,has the same pin numbers and has the same spec's.

For an output of 10A then the BD139 driver transistor might have a heart attack and the opamp driving it might not be powerful enough.

[FotatoPotato]

Would replacing the Drive and Power transistors with a high power Mosfet and adding a mosfet gate driver Ic help with achieving a 10A output?

[Audioguru]

A Mosfet will produce exactly the same heat as a power transistor. But its gate-source voltage is high which will reduce the output voltage of the supply.

[FotatoPotato]

Ah, Ok so pretty much no difference to efficiency and power output between a Mosfet and a power trannie. Would It make more sense to build a Buck-boost converter and feed it 15v @20A so that I can get a higher efficiency? I get the 15v @ 20A from the transformer because it can be configured to output 12v @ 20A or 24v @10A, the 15v is what I get after it goes through a full bridge rectifier and filter capacitor.

[Cliff Matthews]

...MC34071 opamps in a DIL package are not available anymore..

There's always these cheap helpers (SOIC-8 to DIP-8) https://www.sparkfun.com/products/13655


Perhaps the OP hasn't seen this one (using 2 vintage DIL's to control TIP142's)
http://www.ve2ums.ca/chasse/Serge/Atelier/Projets/Membres/VE2EMM/alimentation_ang.htm

[FotatoPotato]

The PSU you linked is a great help and I really appreciate it, I might actually make it because it meets my needs and the components aren't too expensive. Do you think it would be possible to replace the TIP142 NPN Darlington BJT with a N-channel Mosfet for better efficiency?

Thanks!   

[Cliff Matthews]

The PSU you linked is a great help and I really appreciate it, I might actually make it because it meets my needs and the components aren't too expensive. Do you think it would be possible to replace the TIP142 NPN Darlington BJT with a N-channel Mosfet for better efficiency?

Thanks!   

http://home.earthlink.net/~schultdw/psupply/

[FotatoPotato]

Again, another SUPER useful resource, Thanks so much 

[dmills]

Your 12V @ 10A is actually by far the nastier requirement for a simple linear supply capable of 30V at any current at all.

The reason is that for ANY simple linear supply the power dissipated in the pass transistor is (input voltage - output voltage) * current, essentially independent of the pass device technology.
For a simple minded linear like your are trying for the input voltage must be greater then 30V (So that you can get 30V at the output), which means that at 12V @ 10A you are dissipating at least 30 - 12 = 18V * 10A = 180W, and it only gets worse as the output voltage drops.
The '3055 lacks the safe operating area to do this at temperature, quite apart from being the '741 of the power transistor world (Old and crap, and beloved of a certain generation of hobbiest), I would be looking at a dozen of the things to handle this, and would almost certainly use something else in practise.
If you wanted such a supply to supply say 1V @ 10A, you would have at least 290W being dissipated in the pass bank.

Yea there is a reason everyone gets clever to a greater or lesser extent when building high current variable output power supplies.

One nice trick for example is to use a transformer having a couple of secondary windings and have a 'range' switch which changes the connections from parallel to series as the voltage is turned up.
If for example you had such a thing putting out say 18V in parallel mode and 36V in series mode (quite reasonable) then @ 12V, 10A your pass device is now only dropping (18 - 12) * 10A = 60W, a third of the case for the simple minded design.

The ultimate expression of this idea is a switched mode preregulator setup to maintain the input to the linear regulator at say a volt above its output, which is what most designers these days seek to do, it trades complexity for a MASSIVE reduction in heat, and components are cheaper then heatsinks. 

Regards, Dan.
 

[FotatoPotato]

I like the points you bring up and I never thought about the fact that I would be dissipating almost 300W of heat stepping down 30V to 12v @10A. I do however think the design will be fine because I can think of only an EXTREELMY small amount of projects where I would actually need 12v @ 10A. I just want that power as a possibility so that if I need to supply something for maybe 30 seconds to a minute with such a large amount of current then I Could. I have played around with building a SMPS but I never got anywhere and it eventually proved to be too complicated for my beginner skills (or lack there of). For now I’ll probably just stick to the more simple linear design.

[dmills]

There is not actually all that much in hobby electronics that needs anything like that much current, personally I would build a supply capable of say 20- 25V @ 1 amp and see if you ever need more (20 or so V chosen to allow a 12-0-12 transformer secondary).
This will suffice for most of the little micro sort of projects, and if you were to build a dual output one would also cover most audio projects (+-15V is common there).

This is a decidedly easier project in all sorts of ways, and in all probability by the time you really need more then that you will have the skills to cope with building something clever.

I have **MUCH** bigger power supplies in the lab, I nearly never use them.

regards, Dan.

[FotatoPotato]

That is ture, I have only had a handfull of times when I ever needed more than 5A, mostly with 100W LED's and peltier modules, I'll still build this supply so that it is "capable" of delivering more than 5A but I'll probably never use more than 3A. And like you said, by the time I need more than 10A, I'll have the know-how to build a proper switchmode supply that can handle that kind of power. 
 

[ogden]

I have only had a handfull of times when I ever needed more than 5A, mostly with 100W LED's and peltier modules

Exactly. When it's about higher power than my small lab supply can do, I just bring out my old 400W desktop PC power supply (and some fuses too). There's quite a lot of information around about using such for loads that are not PC. You can even slightly adjust voltage up or down with simple tweak: http://electronics.pl7.de/power-supplies/converting-computer-power-supplies-psu-to-stabilized-13-8-v-dc-20-a/

When you need higher voltage @ high power - just buy it. Today for ~30$ you can get quite decent 24V 20A "brick". If you need variable voltage and/or current regulation - add DPS5015 (Dave recently blew such up) and you are quite covered. Rather than having 300W linear lab supply "beast" on the table, I put my projects there.

[FotatoPotato]

I have wanted to learn how to build a completer SMPS, not just a large step down transformer combined with a Buck-boost converter, but a true SMPS that converts the 110v AC directly to my desired output. The only thing keeping me away is the complexity and fact that I would be dealing with 110V AC and somewhere around 230V DC, so if I make one mistake then there goes my face, my room and my circuit breakers. And I'm not too keen on having something like that happening. Maybe one day I'll make a schematic and post it here and have people tell me if I made a bomb or a functional power supply.

[ogden]

I have wanted to learn how to build a completer SMPS, not just a large step down transformer combined with a Buck-boost converter, but a true SMPS that converts the 110v AC directly to my desired output. The only thing keeping me away is the complexity and fact that I would be dealing with 110V AC and somewhere around 230V DC

You don't build soldering iron and multimeter, right? Or did you? - Then why do you need to build supply? SMPS design is very complex stuff, only knowledge you can learn while designing SPMS is that you shall not design SMPS, especially while you are beginner. Better build something that is really worth your time and resources.

[FotatoPotato]

Definitely, trying to design one now would be a waste of time, there are many other more useful things that I could work on that would teach me some really useful stuff.

[Audioguru]

A TIP142 darlington has a maximum base-emitter voltage loss of 3.5V at 10A. Many Mosfets have a 10V gate-source voltage loss at 10A.
So if you use Mosfets then the maximum output voltage is 6.5V less, unless you are lucky enough to find very sensitive Mosfets that have a loss of only 4.5V.

[dmills]

Yea but producing an isolated gate drive voltage 12V or so above the source terminal is not exactly rocket science....

Point still stands that a linear passbank will be horrible efficiency wise a lot of the time, and not just when it is up against the input rail (Which is actually best case).

Regards, Dan.

[rob77]

A TIP142 darlington has a maximum base-emitter voltage loss of 3.5V at 10A. Many Mosfets have a 10V gate-source voltage loss at 10A.
So if you use Mosfets then the maximum output voltage is 6.5V less, unless you are lucky enough to find very sensitive Mosfets that have a loss of only 4.5V.

there is no such a thing as  "gate-source voltage loss" , so please don't spread BS.

you need a voltage at least Vgs above source's potential to controll the N-chan mosfet. but that has nothing to do with the output voltge drop on the mosfet you can still have a voltage drop Vds of few millivolts while Vgs at 12V! the gate is isolated and you can think of a N mosfet as a variable resistor which needs to be controlled by a voltage higher than the voltage on the source pin. the source pin is usually the output of a PSU so you need higher gate voltage than the output. you can easily solve this by a small dc-dc converter (e.g. MC34063) providing the needed higher voltage for the control circuit of the mosfet and clamping the Vgs with a zener.

[ogden]

there is no such a thing as  "gate-source voltage loss" , so please don't spread BS.

mosfet you can still have a voltage drop Vds of few millivolts while Vgs at 12V!

Now explain please - supply voltage coming into N-FET linear regulator is 30V and it is only voltage rail of the supply. To open gate, 12V voltage above source needed which would be at 29.99V considering few mV drop as you say. But hey - we have single rail which is 30V. How do you open N-mos? IMHO in this case gate shall receive 29.99+12V above ground Or am I missing something here?

[edit] Ok. Didnt come to hang yourself. In short: introducing switching crap into regulator chain like MC34063 and clamping(!) Vgs of N-MOS linear regulator with zener is worst crime which can be done to linear lab supply.

[rob77]

there is no such a thing as  "gate-source voltage loss" , so please don't spread BS.

mosfet you can still have a voltage drop Vds of few millivolts while Vgs at 12V!

Now explain please - supply voltage coming into N-FET linear regulator is 30V and it is only voltage rail of the supply. To open gate, 12V voltage above source needed which would be at 29.99V considering few mV drop as you say. But hey - we have single rail which is 30V. How do you open N-mos? IMHO in this case gate shall receive 29.99+12V above ground Or am I missing something here?

[edit] Ok. Didnt come to hang yourself. In short: introducing switching crap into regulator chain like MC34063 and clamping(!) Vgs of N-MOS linear regulator with zener is worst crime which can be done to linear lab supply.

i don't see how a zener used to protect the gate of as mosfet  is a crime    and a low current switcher can be easily filtered down to virtually no noise , so i can't see a crime there either.

but after your last post i'm pretty sure you have a major gap in understanding of a mosfet as a series pass element... the voltage at the gate is NOT DIRECTLY RELATED to the output voltage , it's not like a NPN emiter follower. with a mosfet you control the resistance of the drain-source path by changing the gate voltage . if the threshold voltage of the mosfet is for example 5V and saturation is at 10V Vgs then you control the whole output range (for example 0 to 30V) of the PSU with that 5V swing (5 to 10V)  at the gate.

[dmills]

The low noise, simple way to get that extra rail is usually to just do a voltage doubler off the main transformer secondary, current is pretty negligible, so the caps can be small.
Small isolated switchers can work, but why bother when you can get something usable with a couple of diodes and caps?

It has to be said, that I have personally never really seen the virtues of mosfets in this application, the limiting factor if usually more Safe Operating Area then saturated switch efficiency, and jellybean BJTs usually have much better SOA then modern jellybean power mosfets for the simple reason that power mosfets are usually designed as switching devices (There are exceptions mainly aimed at audio, but they are less common then they were).

BJTs also usually parallel more easily then mosfets in practise because they do not have that annoying variation in threshold voltage to the same extent so simple emitter resistors are sufficient to get reasonable current sharing out of the things.

Regards, Dan.

[Audioguru]

The original supply was linear, not switching and used emitter-follower 2N3055 output transistors that needed a base voltage of 31.5V max to produce an output of 30.0V at 4A. The question was if a Mosfet would increase efficiency. I replied that the 10V gate-source voltage loss (of a source-follower) would require a minimum gate voltage of 40V then the opamp driving it would need a very high supply voltage that might destroy it. Even a logic level Mosfet would produce a gate-source voltage loss of 4.5V and would push the opamp to its maximum allowed supply voltage limits and maybe destroy the opamp when the load current is low then the transformer voltage is a little higher than its spec.

[Kleinstein]

Using MOSFETs can reduce the voltage lost in the high current path, but as a source follower it needs an extra higher supply voltage. MOSFETs also have difficulties having them in parallel. There are cases where MOSFETs are a good idea, but there are also cases where they are not.

The source follower has another problem: with BJTs the current gain tends to go down at high current. This tends to limit the peak current on a short. With MOSFETs the trans-conductance usually is going up with high current. So peak currents can be rather high. Related to this MOSFET speed also depends a lot on the current: at higher currents they a rather fast, but at low currents they are slow. This can make the design of the regulator more difficult. So often one needs a relatively high bias current to avoid the slow range. 

The circuit with the power stage as an emitter follower is usually chosen to have a low output impedance and this way allow for a smaller output capacitor and simpler regulator. A source follower (especially with those FETs with a good FBSOA) tends to have a higher output impedance than the emitter follower and this way might not work well in the similar simple circuit. It might take a rather high bias current to make it acceptable.

[ogden]

i don't see how a zener used to protect the gate of as mosfet  is a crime 

Protection is not a crime but assumption that mosfet gate voltage can be exceeded in linear supply is sign that you do not understand how linear supplies/regulators work. Basically you accused Audioguru of spreading BS w/o proper understanding what actually he is talking about.

low current switcher can be easily filtered down to virtually no noise , so i can't see a crime there either.

Well well. Whole essence of linear supplies and big, heavy mains AC transformers used is to avoid switching noise, bring regulation into low, AC mains 50/60hz frequency domain. If noise can be easily filtered in the regulation circuit, then it can be easily filtered in the power circuit as well - using much smaller inductors than that big heavy transformer. Think about it. Do you even know what is inductor parasitic capacitance and how it affects ability of filter to do actual filtering? Of course it is crime to introduce switching regulator into pure linear supply unless it is not a part of output regulation.

but after your last post i'm pretty sure you have a major gap in understanding of a mosfet as a series pass element...

No a lot of understanding is actually required to comprehend that MOSFET in linear supply is not used as a switch. Power transistors in linear regulators are never widely/fully open because there always  shall be some voltage headroom for regulation - in case mains (input) voltage drop. So your "voltage drop Vds of few millivolts" when we talk about linear regulator is BS instead.

[FotatoPotato]

So I just finished building the 0-20v 0-10A linear power supply. Everything works flawlessly and it is almost perfect. If you look at the schematic that I linked you can see that the main transistor is a TIP142 Darlington BJT. The original schematic is only rated for 1A but by adding 4 of them in parallel you can obtain a 10A output. So I did that, the only problem is that only 2 of the BJT's work at any given time. Even weirder is that its random. If you look at a pic of the PSU you can see that they are in groups of 2 (2 on the left and 2 on the right) and randomly only one of those pairs of 2 will work. They are all linked together identically and all their bases are connected as well.

So I have no idea why this is happening, maybe you do. If you have any advice on changes to make for all 4 BJT's to work please let me know! 

[ogden]

the only problem is that only 2 of the BJT's work at any given time. Even weirder is that its random.
So I have no idea why this is happening, maybe you do. If you have any advice on changes to make for all 4 BJT's to work please let me know! 

For 2 or more transistors to work in parallel - they shall be identical, but usually they are not. To have better chances of equalizing current between transistors, you shall add small value (0.2 to 0.5 ohm resistor) to emitter of each transistor as shown in article below. Make sure you calculate worst case dissipation and size those resistors accordingly. IMO each transistor shall have it's own base resistor as well (or maybe not). Try just emitter/ballast resistors first:



Article:
https://www.allaboutcircuits.com/textbook/semiconductors/chpt-4/bjt-quirks/

[Twoflower]

Make sure you don't exceed the 25mA the V_Z pin of the 723 can provide. The TIP142 states for the ON-Characteristics I_B = 10mA or even 50mA depending on the I_C. That's per transistor not for the the four you have.

[FotatoPotato]

Would something like this work to balance the current between all four transistors?  And I do have a 220 resistor on the output signal to limit the current, is that ennough or should it be lower?

The 5th transistor is a 2N3904 if you couldnt read it (sorry the schematic immage is a bit blurry)

[ogden]

What exactly 2n3904 transistor is doing there?! Please explain your intention.

This 4-transistor power stage operates as emitter follower, no any additional trickery/transisors needed - connect bases of power transistors together, add individual ballast resistor to emitter of each transistor.

220Ohm is fine. For starters. The higher value resistors - the better balancing of the power, but higher dissipation/losses. You can use even smaller ballast values if transistors are well matched, but only measurements will tell. Precisely measure resistance and voltage drop on each ballast to estimate current flowing, share your results

[FotatoPotato]

OK, so I added a 0.22 resistor to the emitter of each transistor to the supply rail. This seems like it has worked to an extent because when I put a large load on the supply, all 4 transistors begin to get hot which to me indicates that all 4 of them are working. This is all fine and well but there is one problem. On larger loads over 1A one of the pairs of 2 transistors still gets WAY too hot. They are hitting 180c at 4A of current draw while the others are sitting at about 65c. I know they should get hot but not that hot! Is this just a case of crappy thermal pads or is something else wrong here?

Also it seems like I’m well within the current limit of the 723's Vz pin because I have pulled 8A so far with all 4 transistors and it hasn’t popped. Let’s hope it stays that way!

[glarsson]

Measure the voltage drop over each 0.22 ohm resistor. If the transistors share the load equally the voltages should be close. If the voltages are close (i.e. same current, same voltage drop over the transistor, same power dissipation in the transistors) then check if there are differences in cooling, e.g. thermal pads, paste, etc.

[FotatoPotato]

Huh, this doesn’t seem right.... when I pull 500mA at 20V there is a different drop across each resistor. On the first transistor there is a drop of 60mv with a resistance of 0.226 , on the second transistor there is a drop of 7mv with a resistance of 0.226 again, on the third transistor there is a drop of 3mv with a resistance of 0.227 , again, and on the final transistor there is a drop of 27mv with a resistance of 0.229 ..... I don’t think that is what is supposed to happen. Any advice?

[ogden]

Huh, this doesn’t seem right.... when I pull 500mA at 20V there is a different drop across each resistor. On the first transistor there is a drop of 60mv with a resistance of 0.226 , on the second transistor there is a drop of 7mv with a resistance of 0.226 again, on the third transistor there is a drop of 3mv with a resistance of 0.227 , again, and on the final transistor there is a drop of 27mv with a resistance of 0.229 ..... I don’t think that is what is supposed to happen. Any advice?

Perhaps some or all power transistors are already damaged because you said that transistors (their cases?) approached 180oC temp, but at 150oC junction temp manufacturer says they are basically dead. I would never let transistors to approach more than 90oC or so.

I would derate/downgrade my power supply power plans, leave two most similar transistors and debug circuit with just two, obviously without overcooking them.

[FotatoPotato]

After some more testing I have concluded that it was shitty thermal pads I took them off and replaced it with some nice thermal paste, now all 4 transistors sit comfortably at 50c while pulling 6.5A at 13v. Welp, problem solved!

It also seems like the 25W 0.22 resistos that I'm using are a bit overkill as they dont even get mildly warm under a heavy load.

[ogden]

It also seems like the 25W 0.22 resistos that I'm using are a bit overkill

It is. You really shall use Ohms law *before* you pick resistors.

Power_dissipated=Current*Current*Resistance