Maker Pro
Maker Pro

2N3055's and heat

John Potter

Nov 20, 2016
4
Joined
Nov 20, 2016
Messages
4
I have a 52V transformer that after rect. produces 3.3A at 52V DC - 4.44A at 36V AC
I wish to use either a LM317 or a LM338 with 4 x 2N3055 power transistors.

If you run 2N3055's in parallel you are supposed to have a resistor like a 0.1Ohm 10W from each emitter so that thermal runaway doesn't take place. Surely the resistor should be at least the output in watts ?

There are circuits on the Internet with this resistor on the base or collector but I think the emitter is correct ?

Could you possibly tell me how to calculate the value - Ohms and watts of this resistor. ? I understand this can be critical.

Honestly the Internet can be a nest of worms, there are many circuits that don't use a resistor at all.


On her page is a link to the circuit that states she removed the 0.1 Ohm 10W resistors as it 'limited her current output'. This 20 odd Amp supply works fine.

Please advise.

I was considering using 4 x 2N3055's each at about 0.75A at 52V (39W each). I thought 3A from the transformer a safe limit. 1 heat sink per 2N3055 in an x pattern under a 125mm 0.5A fan blowing from outside the case down the heat sink fins with a 30mm gap at the bottom. The rear of the case is all vents. I may fit a 150mm 0.4A fan if the 125mm doesn't cover the heat sinks.

Mounting 2N3055's and heat dissapation.

There are 3 videos on this subject. Quite a down to earth approach but unfortunately nothing with metal to metal with heat compound - AND fan. Metal to metal with compound was by far the best. I can insulate the heat sinks. Extrapolating his results I should be able to draw 39W per 2N3055 at about 40C case temperature or less. The thermal resistance junction to case is 1.5 C/W. The max junction tempis 200C. I should be running around 60 C - or not?


Any advice would be greatly appreciated.
 

Herschel Peeler

Feb 21, 2016
401
Joined
Feb 21, 2016
Messages
401
I have a 52V transformer that after rect. produces 3.3A at 52V DC - 4.44A at 36V AC
I wish to use either a LM317 or a LM338 with 4 x 2N3055 power transistors.

If you run 2N3055's in parallel you are supposed to have a resistor like a 0.1Ohm 10W from each emitter so that thermal runaway doesn't take place. Surely the resistor should be at least the output in watts ?

There are circuits on the Internet with this resistor on the base or collector but I think the emitter is correct ?

Could you possibly tell me how to calculate the value - Ohms and watts of this resistor. ? I understand this can be critical.

Honestly the Internet can be a nest of worms, there are many circuits that don't use a resistor at all.


On her page is a link to the circuit that states she removed the 0.1 Ohm 10W resistors as it 'limited her current output'. This 20 odd Amp supply works fine.

Please advise.

I was considering using 4 x 2N3055's each at about 0.75A at 52V (39W each). I thought 3A from the transformer a safe limit. 1 heat sink per 2N3055 in an x pattern under a 125mm 0.5A fan blowing from outside the case down the heat sink fins with a 30mm gap at the bottom. The rear of the case is all vents. I may fit a 150mm 0.4A fan if the 125mm doesn't cover the heat sinks.

Mounting 2N3055's and heat dissapation.

There are 3 videos on this subject. Quite a down to earth approach but unfortunately nothing with metal to metal with heat compound - AND fan. Metal to metal with compound was by far the best. I can insulate the heat sinks. Extrapolating his results I should be able to draw 39W per 2N3055 at about 40C case temperature or less. The thermal resistance junction to case is 1.5 C/W. The max junction tempis 200C. I should be running around 60 C - or not?


Any advice would be greatly appreciated.

I have never seen an LM317or LM338 working at over 35 V in.
 

John Potter

Nov 20, 2016
4
Joined
Nov 20, 2016
Messages
4
Your half right and I'm half wrong. The LM338 won't handle 52V but the LM317AHVT will handle 60V. I was stumped until I found it.
 

AnalogKid

Jun 10, 2015
2,884
Joined
Jun 10, 2015
Messages
2,884
I have a 52V transformer that after rect. produces 3.3A at 52V DC - 4.44A at 36V AC
I wish to use either a LM317 or a LM338 with 4 x 2N3055 power transistors.
To do what? A little clarity would be nice.
If you run 2N3055's in parallel you are supposed to have a resistor like a 0.1Ohm 10W from each emitter so that thermal runaway doesn't take place.
Thermal runaway, as in what happens in a bipolar transistor audio amplifier without output transistor compensation, cannot happen in a power supply.
Surely the resistor should be at least the output in watts ?
Surely not. Not ever. Use Ohm's law.
Nothing is stated explicitly, so let's assume you are talking about building a power supply. If the supply can supply 4 A at 40 V, that's 160 W. If there are 4 output transistors that are perfectly (magically) balanced, then each one is passing 1 A. If each emitter has a 1 ohm resistor in the emitter, P = I^2 x R, or 1 W. Bump that up to a 2 W resistor to lower its operating temperature.
There are circuits on the Internet with this resistor on the base or collector but I think the emitter is correct?
Depends on the circuit. Schematic?
Could you possibly tell me how to calculate the value - Ohms and watts of this resistor. ? I understand this can be critical.
The purpose of the emitter resistors is to lessen the effect of variations in Vbe of the paralleled transistors. As the Vbe of one transistor decreases below that of the others, its base current increases so share of the total output current increases, increasing its power dissipation. Adding a small resistance to each emitter circuit adds a voltage drop that is relatively stable with temperature. So if the difference between two unballasted transistors was 0.2 V out of 1.0 V, enough for one transistor to completely starve the other of any base current, adding resistors can turn this into 0,2 V out of 2.0 V, and guarantees that the base currents (and emitter output currents) are within 10% of each other. If you need better matching, increase the resistance.
On her page is a link to the circuit that states she removed the 0.1 Ohm 10W resistors as it 'limited her current output'.
Emitter ballast resistors limit the available output voltage range, not the current.

ak
 

Ratch

Mar 10, 2013
1,099
Joined
Mar 10, 2013
Messages
1,099
I have a 52V transformer that after rect. produces 3.3A at 52V DC - 4.44A at 36V AC
I wish to use either a LM317 or a LM338 with 4 x 2N3055 power transistors.

If you run 2N3055's in parallel you are supposed to have a resistor like a 0.1Ohm 10W from each emitter so that thermal runaway doesn't take place. Surely the resistor should be at least the output in watts ?

There are circuits on the Internet with this resistor on the base or collector but I think the emitter is correct ?

Could you possibly tell me how to calculate the value - Ohms and watts of this resistor. ? I understand this can be critical.

Honestly the Internet can be a nest of worms, there are many circuits that don't use a resistor at all.


On her page is a link to the circuit that states she removed the 0.1 Ohm 10W resistors as it 'limited her current output'. This 20 odd Amp supply works fine.

Please advise.

I was considering using 4 x 2N3055's each at about 0.75A at 52V (39W each). I thought 3A from the transformer a safe limit. 1 heat sink per 2N3055 in an x pattern under a 125mm 0.5A fan blowing from outside the case down the heat sink fins with a 30mm gap at the bottom. The rear of the case is all vents. I may fit a 150mm 0.4A fan if the 125mm doesn't cover the heat sinks.

Mounting 2N3055's and heat dissapation.

There are 3 videos on this subject. Quite a down to earth approach but unfortunately nothing with metal to metal with heat compound - AND fan. Metal to metal with compound was by far the best. I can insulate the heat sinks. Extrapolating his results I should be able to draw 39W per 2N3055 at about 40C case temperature or less. The thermal resistance junction to case is 1.5 C/W. The max junction tempis 200C. I should be running around 60 C - or not?


Any advice would be greatly appreciated.

You might want to study post #2 of this link. http://www.electro-tech-online.com/threads/bullet-proof.146490/#post-1240701 .

Ratch
 

(*steve*)

¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥd
Moderator
Jan 21, 2010
25,510
Joined
Jan 21, 2010
Messages
25,510
Thermal runaway, as in what happens in a bipolar transistor audio amplifier without output transistor compensation, cannot happen in a power supply.

In this case he I'd taking about current sharing, and the same mechanism which causes thermal runaway causes instability in the amount of current shared leading to one (or more) transistor handling more power than they would if the current was balanced.

The emitter resistor adds negative feedback which tends to force balance between the transistors.

So while it may not be thermal runaway in a classic sense, it's caused by the sane thing and results in the same end point.
 

JohnP

Oct 22, 2016
31
Joined
Oct 22, 2016
Messages
31
Thank you Herchel Peeler - sorry I forgot.
Thank you Analog

I am starting back into electronics as a hobby after a 50 year break. I was given a 36V AC 4.44A toroidal transformer for my 70th birthday (strange present) and decided to build a general purpose power supply. I have a 400W 60V buck V & A regulator from China somewhere on a bullock cart on it's way. However I thought doing my own linear supply might at least get me back into newer technology. Google 'Mullard 20W amplifier' and you'll see what I mean.
I did actually make a Bailey OC35 transistor amplifier that worked. On polystyrene ceiling tiles and stuffed under the bench seat of my Ford Consul. Good old bus.

I thought I was making a 36V DC power supply. It was a bit of a shock when I metered the DC. I forgot 1.4141. 50 years is a long time. I don't mind loosing a few volts using these emitter resistors.
I have a very old Fluke but it checked out OK with a newer one. I also have some 50 year old Multicore solder, seems OK.

I made 2 mistakes. The LM338 won't take 60v and the LM317 should read LM317AHVT which does.
I was stuffed until I found that.

If I knew what a bipolar transistor was I'm sure I would agree with you.

Thank you for the clarity on the emitter resistor. It was like a breath of fresh air - seriously.
I have seen the resistor at 0.47Ohms and wondered why. One less thing to worry about.
Would it be safer to use a 0.47Ohm resistor anyway ?

Ratch thanks.

I will study that link but a brief look lost me in quite a few places. I'll do my best though. That's quite a friendly Forum.

My main worry now is heat dissipation. I hope my assumptions from what I have learned so far are reasonable. I have considered a thermal cut out on each of the 2N3055's and a 4A fuse. I doubt I'll ever use more than 100W but while I'm 'at it' I might as well do the best I am able to - which isn't a lot.

Thank all for the help so far.
 

AnalogKid

Jun 10, 2015
2,884
Joined
Jun 10, 2015
Messages
2,884
Three-terminal voltage regulators that have a floating adjust pin, as opposed to a fixed ground pin, are different in the way they function and the way they are specified. For example, there is no input voltage spec for an LM317. What is specified is the max. input-to-output voltage drop across the device. So, if you are certain that your input DC voltage never will be greater than 1200 V, you can use a standard LM317 to regulate that down to 1170 V with no problems, because the differential voltage is only 30 V, safely below the part's specification.

ak
 

(*steve*)

¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥd
Moderator
Jan 21, 2010
25,510
Joined
Jan 21, 2010
Messages
25,510
On that subject, there are some circuits on the net which are tracking preregulators for lm317s.

The employ a transistor to drop the input voltage to (say) 10V above the output.

This allows you to short the output (or for the device to go into current limiting or thermal shutdown) without exceeding the Vin/Vout max voltage.

Its in the ti application note 583

Http://www.ti.com/lit/an/snva583/snva583.pdf
 
Last edited:

JohnP

Oct 22, 2016
31
Joined
Oct 22, 2016
Messages
31
Hi AnalogKid. Thanks

According to the data below the max input voltage for a LM317 is 40V. LM317AHVT is 60V

http://cpc.farnell.com/stmicroelect...hLookAhead&searchView=table&iscrfnonsku=false

LM317AHVT can regulate between 1.2V and 57V

http://cpc.farnell.com/fairchild-se...Categories&searchView=table&iscrfnonsku=false

If I have got anything wrong please advise - and Thanks again.

Hi Steve
This circuit would certainly be a very good idea but it's getting rather close to my comfort zone. In other words some of the terminology is in the 'what does that mean' area. However it does make the power supply far more robust and I do have a habit of - OOPS! I have a feeling I may well be getting back to you on this one for some advice on Q1 & Q2 and what resistor values etc. to use.

This could well end up as part of 'Phase II' along with current regulation which I'm hoping can be added onto the end of the V regulation. Current Regulation is a real nest of worms (for me). The circuit below might be OK for low current applications, I don't really know.


Please remember Steve I don't know what the hell I'm doing and it's only because of kind people like yourself that I can slowly progress. A month ago or so ago I didn't know the case of a TO-3 2N3055 was the collector and I had forgotten 1.4141 x AC = DC.

Thanks for the help Steve.
 

AnalogKid

Jun 10, 2015
2,884
Joined
Jun 10, 2015
Messages
2,884
In both link's datasheets, the Absolute Maximum Ratings sections state the input-to-output max differential voltage. Nowhere in that section or in any of the standard parameters tables does it say anything about a max input voltage. National Semiconductor made a big deal about this when the part came out.

ak
 

JohnP

Oct 22, 2016
31
Joined
Oct 22, 2016
Messages
31
Hello Analog - I guess CPC Farnell must be wrong - thanks.





The LM317AHVT at 57V covers my needs I believe. They got that wrong too at a Max input of 60V input and variable between 1,2V and 57V. I wonder what's going on, am I missing something ?

How strange their PDF Data sheet says the same :- Output Adjustable Between 1.2V and 57V
Top left first page.

I think you would agree anybody reading that whose a novice would assume it's adjustable between 1.2V and 57V.
.
LM317T - From the CPC advert :-

Output Type:
Adjustable
Input Voltage Min:
4.2V

XXXXXXXXXXXXXXXXXXXXXXXXXXXXX
Input Voltage Max:
40V
XXXXXXXXXXXXXXXXXXXXXXXXXXXXX

Fixed Output Voltage Nom.:
-
Adjustable Output Voltage Min:
1.2V
Adjustable Output Voltage Max:
37V
Output Current:
2.2A
Linear Regulator Case Style:
TO-220
No. of Pins:
3Pins
Operating Temperature Min:
0°C
ETC.
Operating Temperature Max:
125°C
Packaging:
Each
Product Range:
 

cjdelphi

Oct 26, 2011
1,166
Joined
Oct 26, 2011
Messages
1,166
Why in earth would you use a linear regulator in the first place? Total waste of power, a switching regulator is much much more preferable in this instance
 

Ratch

Mar 10, 2013
1,099
Joined
Mar 10, 2013
Messages
1,099
Why in earth would you use a linear regulator in the first place? Total waste of power, a switching regulator is much much more preferable in this instance

What instance? Did the OP say what his needs were? Perhaps a switching regulator is too noisy for his application.

Ratch
 

JohnP

Oct 22, 2016
31
Joined
Oct 22, 2016
Messages
31
It's 50 years since I was making valve amps as a hobby. A lot has changed. I picked a linear regulator as the circuit is simple and I thought building one would help me catch up with modern components and how to use them.

I need a power supply so that's where I started.

I was advised by a friend that a linear regulator would be far easier than a buck regulator. Maybe not at 160W it isn't.

If you like it's a learning exercise that won't be wasted. What I didn't know was that the Volts and Amps I am using was going to push building a linear regulator so difficult. Maybe that is not a bad thing - I am learning. I would be quite pleased if I could hang a 2V 20mA LED on the end of it. I joke. It would be more like a few thousand.

On it's way from Banggood in China somewhere stuck on a bullock cart is a 400W 60V buck regulator with V & A regulation, memory for settings, etc. and display.

http://www.banggood.com/DKP6008-400...age-Current-Meter-p-1082317.html?rmmds=search

I have received 3 x 5A 1.2v - 37V buck V regulators to 'play' with.

I am also building a 4A 20V linear regulator as a warm up. LN317 and 2 x LM338's.

I can honestly say it's the steepest learning curve I have ever been on.

It's people like yourselves that are making it possible - even if you think I'm mad. Your not the first.

I do thank you for your help. I am finding it rather difficult to comprehend at times but I have all the time in the world and needed something to 'get stuck into'.

After this I want to build a transistor amplifier - no chips. EL34's are a bit expensive now. Pity.
 

cjdelphi

Oct 26, 2011
1,166
Joined
Oct 26, 2011
Messages
1,166
What instance? Did the OP say what his needs were? Perhaps a switching regulator is too noisy for his application.

Ratch

OK fair enough there could be too much noise for his needs... however a nice capacitor / inductor combo to filter out the noise should suffice, but you could then use an lm317 and pass transistor(s)

My only concern was the huge power waste going on, you'll need a hell of a heatsink lol
 

cjdelphi

Oct 26, 2011
1,166
Joined
Oct 26, 2011
Messages
1,166
John it's all fun and games until you forget you left it on then you'll whinge a little when you burn your hand on it lol (been there done that with a heatsink to dissipate say 1 or 2 watts, but i was actually losing about 20 watts across it lol that burn took ages to heal lol

Basically, linears should only be required when dropping a couple of volts say 12 to 9v or 30v to 5v for a small load say 20ma

For example...

The lm317 will cost you 15w of power to charge your phone at 5v 1amp from a 15v battery

Your phone consumes 5w of power to charge, so where's that 10watt going? Heat it's shunted you might as well warm your coffee up on it on it with small heatsink

With a switching regulator, 12v down to 5v can be up to 98% efficient, meaning instead of wasting 10watt you'll waste less than a watt, heatsink can be smaller, runs far cooler, but yes noise can be an issue.

You can turn an lm317 into a switching regulator and say a 555 to drive it, but as you found out buying one is far easier lol, instead of a constant supply of dc power, it gets chopped up and depending on current draw it can use pwm to reduce or increase voltage depending what it sees in it's feedback pin

If it's learning and not a power supply that will be in use daily, great :)

youtube for how linear works vs switching
 

shrtrnd

Jan 15, 2010
3,876
Joined
Jan 15, 2010
Messages
3,876
Everybody else here presented good advice with well thought-out analysis.
You asked for 'any advice', so here's mine.
Aluminum cased TO-3 packages don't hold-up thermally as well as steel TO-3 packages.
When I build something that I want to last, I use the steel TO-3 packages in my devices.
I had a problem with 2N3055's about 20 years ago that I didn't expect (steel TO-3 packages).
My application was serious enough, that I started talking to engineers at some of the component manufacturers.
I was told that the 2N3055's made today, are designed with different dies than the original early 2N3055's, and
are much less robust than the early versions. People use them because they've been around for a long time,
and because of that they're cheap. (And less heat effiecient).
The component manufacturer engineers suggested I use MJ15003's as they are more robust than the 2N3055's
made today. They cost more, and I don't know your application.
Some other specs you might want to check-out for your application would be the MJ15015G or MJ21194G.
... you see what you get when you ask for 'any advice'?
Good luck with your experimentations.
 
Last edited:

Ratch

Mar 10, 2013
1,099
Joined
Mar 10, 2013
Messages
1,099
OK fair enough there could be too much noise for his needs... however a nice capacitor / inductor combo to filter out the noise should suffice, but you could then use an lm317 and pass transistor(s)

My only concern was the huge power waste going on, you'll need a hell of a heatsink lol

If you peruse post #5 in this thread, you will see a circuit that uses a power resistor to dissipate most of the heat, thereby lowering the heat sink requirement.

Ratch
 

JohnP

Oct 22, 2016
31
Joined
Oct 22, 2016
Messages
31
Whilst in the loo contemplating the meaning of life this morning I had a revelation. It didn't hurt much.

I have seen 'centre tap' circuits and had a quick look but I'm afraid it was a case of 'duh'.

However, this transformer I have is 18VAC x 4.44A + 18VAC x 4.44A therefore I think centre tapped ?
I had it wired for 18VAC at 8.88A. I did check I could with CPC Farnell first. I then decided on 36VAC at 4.44A.

Working on the theory 'you can only ask'.
Is it possible to switch between 26V DC x 3.3A and 52 DC x 3.3A. I don't think I see a problem with the switching arrangement at the output from the transformer. Would everything that follows on work ?
If you halve the difference in Volts between Input and load then you would halve the Watts on the 2N3055's - and heat ?
It would be nice if I was right otherwise I shall have to go back on my medication.
 
Top