Maker Pro
Maker Pro

Question regarding pass transistors

Resqueline

Jul 31, 2009
2,848
Joined
Jul 31, 2009
Messages
2,848
The circuit with the NPN after the regulator is (like I said earlier) a less than ideal regulator (poor load regulation).
The b-e junction starts to conduct slightly over 0.5V, conducting progressively more as the voltage increases.
As with any other diodes it may end up well above one volt at high currents.
You may see up to 1V drop in output voltage (no-load to full-load) with that circuit.
Most datasheets have a max base current spec for power transistors.
You need to specify how you need the PSU to perform. Do you need current limiting for instance?
 

jackorocko

Apr 4, 2010
1,284
Joined
Apr 4, 2010
Messages
1,284
My whole goal was to build something that would be useful and hopefully let me move from theory to reality so that I can start to understand the concepts behind finished designs and their limitations. That is why my first post was simply, "What are the advantages and disadvantages to the two designs."
Nothing here is mission critical. The goal was to build a linear power supply that had variable output voltages with as high as possible current that was as stable as real life would allow it to be. I am not shooting for the sky. Just to build a robust linear supply that I can count on, whatever the finished product is, it's what I will have to live with. I hope that helps explain my thinking. So far I have learned tons about component limitations and that was exactly my purpose when I started this.

Here is my last effort at a schematic, does anyone see anything wrong with this? My only question here, if all else is correct, will the TIP2955 handle the 95W it is expected to dissipate under worse case being voltage of 3.3V
attachment.php
 

Attachments

  • psu_selectable_lm317t_2.PNG
    psu_selectable_lm317t_2.PNG
    24.9 KB · Views: 4,128
Last edited:

davelectronic

Dec 13, 2010
1,087
Joined
Dec 13, 2010
Messages
1,087
pass transistor

Hi jackorocko.
I think it will work ok, you will need both LM317 and the TIP2955 on the supply's heat sink, the LM317 is a TO220 washer and bush and TIP2955 is a TO-3P washer and bush, you need to use thermal compound on both, or a heat conductive silicone washers, to achieve 95 watts you will need a large heat sink, and or air cooling on demand or continuous, demand is better as unloaded there is no need for cooling. as i said a thermal switch can take care of cooling, or you can build a second fan control circuit thats variable in its action.

A ventilated case is a must, i expect you will want panel meters, optional fine and course current control is another option, and a known switchable internal load for testing circuits is another good thing to have, up to you.
The full current range across all voltage is unlikely, but 95 watts is big power, i doubt your need that in electronics projects, IE not much at 3.3 volts or 24 volts will ever need 95 watts.

My first builds a while ago know look easy on paper, until you get to the lay out of the circuit, but it comes together in the end, pcb, or copper strip.

If you follow the data sheets you cant go far wrong, the reference voltage resistor commonly 120R should be tied to the regulators pins as close as possible to avoid voltage drift. But most of this is in the applications data sheet. Dave. :)
 
Last edited:

Resqueline

Jul 31, 2009
2,848
Joined
Jul 31, 2009
Messages
2,848
Take a look in the LM317 datasheet. It has many examples of just what you'd need, even simple switchmode varieties that'll dispense with most of the dissipation.
Or; here's my version of your version:
 

Attachments

  • psu_selectable_lm317t_2.PNG
    psu_selectable_lm317t_2.PNG
    27.3 KB · Views: 2,359

davelectronic

Dec 13, 2010
1,087
Joined
Dec 13, 2010
Messages
1,087
pass transistor

Hi Resqueline.
The extra transistor will help in terms of power and heat dissipation, will the make before break be a smooth transition ? ive never used a make before break, is there not a split second that one voltage will impose on the other ? Dave. :)
 

jackorocko

Apr 4, 2010
1,284
Joined
Apr 4, 2010
Messages
1,284
Hi Resqueline.
The extra transistor will help in terms of power and heat dissipation, will the make before break be a smooth transition ? ive never used a make before break, is there not a split second that one voltage will impose on the other ? Dave. :)

Yes, but as resqueline stated before, it is better then not having a resistance there or the output voltage will be close to the input voltage.
 

jackorocko

Apr 4, 2010
1,284
Joined
Apr 4, 2010
Messages
1,284
Take a look in the LM317 datasheet. It has many examples of just what you'd need, even simple switchmode varieties that'll dispense with most of the dissipation.
Or; here's my version of your version:

Ok resqueline, I think I might have found an issue(possible in mine too since I drop 1.4V across R7, leaving me 26.5V to the LM317 input.) In your schematic R7 is a 22 Ohm resistor. I did the math and I see you are dropping ~2V across R7 with a current of 100mA. If that is the case then wouldn't the 28V input be to small to fully provide 24V output at 4.5A. Which I am gonna assume is the near the full potential of this circuit.

How did I do with my mathematical analysis?

Take a look in the LM317 datasheet. It has many examples of just what you'd need, even simple switchmode varieties that'll dispense with most of the dissipation

http://www.sparkfun.com/datasheets/Components/LM317.pdf
You mean figure 13? Must be all datahseets are not equal because I could have sworn I seen better examples in another datasheet? Maybe that was the LM338 I was looking at.

BTW, I like how you wired the rotary switch, neat design implementation.
 
Last edited:

davelectronic

Dec 13, 2010
1,087
Joined
Dec 13, 2010
Messages
1,087
pass transistor

Hi all.
I dont know if its me, its late, tired, but the more i look at Resqueline circuit the resistors are all in series from R2 all the way up, but open circuit until switched in line, so all the resistors become a successive series resistance progressively from R2 3.3 volts right up to 24 volts, as only R2 is connected to ground alone, R3 R4 ....... and so on are all in series, or am i missing something , not sure, something dont look right, its probably me. Dave. :) PS it seems all from and including the last resistor have to travel through the previous resistances for the reference pin to meet GND.
Maybe thats the schematic for make before break, but only R2 goes to ground above the 200R resistor the others are in series, so as the voltage goes up the reference pin has to travel through successive series resistances ?
 
Last edited:

jackorocko

Apr 4, 2010
1,284
Joined
Apr 4, 2010
Messages
1,284
Hi all.
I dont know if its me, its late, tired, but the more i look at Resqueline circuit the resistors are all in series from R2 all the way up, but open circuit until switched in line, so all the resistors become a successive series resistance progressively from R2 3.3 volts right up to 24 volts, as only R2 is connected to ground alone, R3 R4 ....... and so on are all in series, or am i missing something , not sure, something dont look right, its probably me. Dave. :) PS it seems all from and including the last resistor have to travel through the previous resistances for the reference pin to meet GND.
Maybe thats the schematic for make before break, but only R2 goes to ground above the 200R resistor the others are in series, so as the voltage goes up the reference pin has to travel through successive series resistances ?

No, I think you got it. As the switch transitions from 3.3 to 24V the resistors will be added in series and be additive to the total resistance. No resistor will be in series to ground until the switch is rotated to that position. Take for instance R6, the only way to get to ground is by turning the rotary switch to the 24V position which then sends current through R6 R5 R4 R3 R2, which will equal 2k Ohms resistance.

The switch itself is a make then break design, AFAIK.
 

davelectronic

Dec 13, 2010
1,087
Joined
Dec 13, 2010
Messages
1,087
pass transistor

You say it has to pass current from R6 down through to and incuding R2, then thats the total resistance in series, if power has to pass through all the resistance from the 24 volts position down to 3.3 volts. you just said they pass through the resitance total, if thats not in series what is it, its not individual switched in line, and it certainly not a parallel cuicuit of resistance. Look at the diagram from reference R2 upwards to R6, unless iam missing something. Dave.
 

jackorocko

Apr 4, 2010
1,284
Joined
Apr 4, 2010
Messages
1,284
You say it has to pass current from R6 down through to and incuding R2, then thats the total resistance in series, if power has to pass through all the resistance from the 24 volts position down to 3.3 volts. you just said they pass through the resitance total, if thats not in series what is it, its not individual switched in line, and it certainly not a parallel cuicuit of resistance. Look at the diagram from reference R2 upwards to R6, unless iam missing something. Dave.


correct, the higher the resistance the more voltage output you will get, so each time the switch is rotated from 3.3v to 24 volt another resistance is added in series. As we know resistors in series are additive.

If you look resqueline changes the values(black text) of the resistors. These will add up to equal the total resistance you need for each voltage setting.

example: 9V setting needs 750Ohms of TOTAL resistance. R2 200 Ohms + R3 160 Ohms + R4 360 Ohms = Rt 750 Ohms
 

davelectronic

Dec 13, 2010
1,087
Joined
Dec 13, 2010
Messages
1,087
pass transistor

Ok point taken, iam aware of resistance configarations. So your schematic with the individual resistances to GND make the same resistances for a given selected voltage ? i cant vue the diagrams yet, iam using this dodgy phone for the net. What i mean is, so your switched selections are the same as Resqueline's schematic on the rotary switch ? for the same given switch posítion equal in both diagrams. As your shematic uses single switced resistances for each given voltage. Dave.
 

Resqueline

Jul 31, 2009
2,848
Joined
Jul 31, 2009
Messages
2,848
Ok resqueline, I think I might have found an issue (possible in mine too since I drop 1.4V across R7, leaving me 26.5V to the LM317 input.) In your schematic R7 is a 22 Ohm resistor. I did the math and I see you are dropping ~2V across R7 with a current of 100mA. If that is the case then wouldn't the 28V input be to small to fully provide 24V output at 4.5A. Which I am gonna assume is the near the full potential of this circuit.
How did I do with my mathematical analysis?

I think it's better to take it from the output & back, because the transistors will amplify & deliver the current that the LM can't.
UR8 = 4.5A / 2 * 0.1ohm = 0.225V
UR7 = 0.225V + 0.775V = 1V (approx)
IR7 = 1V / 22ohm = 45mA
Ib = 4.5A / 15 = 0.3A (or less)
I(LM317) = 0.345A (or less)
That adds up to 4.845A and so the transistors & the LM will strike a balance will a little less current from both to arrive at 4.5A.
Note however that this supply is not current limited in a way that I'd deem sufficient (and I'd also like to have a variable current limit, for my needs).
Once upon a time I made a 0-30V 0-12A supply using an LM723 and four 2N3055's (+ one driver). It was a bit more complex though.

http://www.sparkfun.com/datasheets/Components/LM317.pdf
You mean figure 13? Must be all datahseets are not equal because I could have sworn I seen better examples in another datasheet? Maybe that was the LM338 I was looking at.
No, that datasheet is from TI. I believe I was referring to a National Semiconductor datasheet. Please click on my link (blue letters).

BTW, I like how you wired the rotary switch, neat design implementation.

The difference will be found only in the switchover instants.
Without a make-before-break switch you'll get a full-voltage pulse out when switching.
With a make-before-break switch & separate resistors to ground you'll get a dip in voltage upon switching since two resistors will then be paralelled.
With a make-before-break switch & series resistors there will be no disturbances (other than the wanted transition).
 

davelectronic

Dec 13, 2010
1,087
Joined
Dec 13, 2010
Messages
1,087
pass transistor

Thanks thats explained the voltage selection for me and the resistive voltage make before break transition, crystal now. Dave.
 

jackorocko

Apr 4, 2010
1,284
Joined
Apr 4, 2010
Messages
1,284
I think it's better to take it from the output & back, because the transistors will amplify & deliver the current that the LM can't.
Yep that makes perfect sense, I should have known that. I still find it hard to analyse circuits because I am not always sure where I should start. I am gonna assume by a statement below that the 4.5A you use in the calculations below is what we want and is only an arbitrary number?
UR8 = 4.5A / 2 * 0.1ohm = 0.225V
UR7 = 0.225V + 0.775V = 1V (approx)
I don't see how you are figuring the voltage drop across R7, where are you getting the 0.775V from?
IR7 = 1V / 22ohm = 45mA
Ib = 4.5A / 15 = 0.3A (or less)
I(LM317) = 0.345A (or less)
15 must be the gain of the transistor you are using?
That adds up to 4.845A and so the transistors & the LM will strike a balance will a little less current from both to arrive at 4.5A.
Note however that this supply is not current limited in a way that I'd deem sufficient (and I'd also like to have a variable current limit, for my needs).
Would you use a sense resistor and a transistor from Vout to ground to dump the excessive current with an inline fuse. Or is there a better way( of course there is a better way)
Once upon a time I made a 0-30V 0-12A supply using an LM723 and four 2N3055's (+ one driver). It was a bit more complex though.
Your knowledge is immense in this area and I am glad you have taken the time to help me. Maybe when I advance more I can finally make complex circuits. Right now I am like a minnow in an ocean.
 

Resqueline

Jul 31, 2009
2,848
Joined
Jul 31, 2009
Messages
2,848
Yep that makes perfect sense, I should have known that. I still find it hard to analyse circuits because I am not always sure where I should start. I am gonna assume by a statement below that the 4.5A you use in the calculations below is what we want and is only an arbitrary number?
Yes, that's always the challenge in any profession. Yes, the 4.5A is a figure you brought up as the max output current.

I don't see how you are figuring the voltage drop across R7, where are you getting the 0.775V from?
It's the sum of the drop across R8 plus the b-e drop. 0.775V is just an arbitrary but plausible Vbe that miracously results in a nice round 1V total drop.

15 must be the gain of the transistor you are using?
Yep. It really can get that small for power transistors under unfavorable conditions.

Would you use a sense resistor and a transistor from Vout to ground to dump the excessive current with an inline fuse. Or is there a better way( of course there is a better way).
A sense resistor in series with Vout yes, but not to ground. Check out the datasheet I linked to, it has a circuit/suggestion.

Your knowledge is immense in this area and I am glad you have taken the time to help me. Maybe when I advance more I can finally make complex circuits. Right now I am like a minnow in an ocean.
Only glad when I can help and ppl listen & understand. I was where you are now, and I would not have been able to design that 723 circuit all by myself back then. I found it in a magazine as a 0-3A supply and just "spritzed" it up a little. You'll get there.
 

jackorocko

Apr 4, 2010
1,284
Joined
Apr 4, 2010
Messages
1,284
A sense resistor in series with Vout yes, but not to ground. Check out the datasheet I linked to, it has a circuit/suggestion.

If I understand you, if you limit the current on Vout of the voltage regulator then you also control the current coming into Vin and also the base of the power transistors. But my only concern here is if I add in a resistor to limit current on the output then doesn't this also change my VR resistors R2-R6? which is no big deal, I am just making sure I understand.
 

Resqueline

Jul 31, 2009
2,848
Joined
Jul 31, 2009
Messages
2,848
There are more & less ingenious & complex ways of doing it with more or less ideal results. See page 18 in my link. Page 19 has the switchmode versions.
"Of course" Iin = Iout. One can make use of the existing 1.24V reference, & usually pulls down the Adj pin to lower the output voltage & thus restrict the current.
 

jackorocko

Apr 4, 2010
1,284
Joined
Apr 4, 2010
Messages
1,284
There are more & less ingenious & complex ways of doing it with more or less ideal results. See page 18 in my link. Page 19 has the switchmode versions.
"Of course" Iin = Iout. One can make use of the existing 1.24V reference, & usually pulls down the Adj pin to lower the output voltage & thus restrict the current.

I just don't get what you are trying to tell me. I looked at page 18 and I do not see a schematic on that page that that uses a resistor in series with the output. As far as the switchmode schematics go, I am a little scared of those. I have never used, discussed, even experimented with inductors so I don't even know where to begin. :(

Edit: So in typical "me" fashion I was annoyed that I could not understand what you where talking about so I broke out google and found a PDF from national that described how they do current limiting inside a voltage regulator. The first schematic on page 18 now makes sense to me, but I am completely dumb founded how to mathematically analyse the circuit, since my knowledge of opamps is very limited. But I do get the jist of it, the op-amp changes the voltage applied to the ADJ. It is essentially running in differential mode, and will adjust it's output to try and equalize the two inputs I believe? How close am I? The only thing that bothers me about this design is the need for a negative voltage. I guess I can easily get that with a center tapped transformer and another -VR. But I guess the question remains, is there an easier way of doing it?

I also thought about using another Lm338 in series with the output and run it in constant current mode, but at 24V and 5A of current that resistor would need to drop 120W of power. I could cook an egg on that sucker!!

I wish I had you on speed dial, kills me waiting till tomorrow to see your replay ;)
 
Last edited:

davelectronic

Dec 13, 2010
1,087
Joined
Dec 13, 2010
Messages
1,087
pass transistor

Hi jackorocko.
Inductors are just enamel copper wire wound on a set diameter core and length, the core is normally ferrite but for small stuff its often axial lead components, look like resistors but chunkier, the schematic tells the inductance, wire / turns / AWG / SWG first is the US standard the other is UK, and Europe i suppose.

Why not have a go at a linear continuous variable voltage supply first, just use the LM317, ok dont complicate things with current limiting yet, build the basic supply a 2 amp transformer is ok for a start, build it on matrix or copper strip board, or even point to point wiring, get a feel for the lay out, i know its a basic no frills start, but a simple variable supply would give you the foot up on the practical side, not to put you off but the schematic you want to start with would have swamped me in my first few builds.

It does get mind testing on copper strip, you got to keep every thing clear and focused, when you apply power is nerve racking the first few times, its all so easy to create an unwanted solder bridge, or put a link in a track containing other component leads, and yes its happened to me, despair best sums it up, as Resqueline said he was where you and me are, not to far apart, the first successful build will give you the boost, some might say to simple, but better a good first ending than disaster, i hated my failures, once in a while i make a mistake now and again, ie, no power out, sit there going over every track link until i find my mistake, making less now as iam getting more confident.

Visualizing the components lay out is not as easy as you might think, kits can be ok, but there not rewarding like building up a circuit from scratch.

You talk about the theory which is good, although my maths is not brilliant, i did it in tandem, as i read about a circuit i attempted it, and that way of the theory and practical running in parallel worked for me, still is.

The basic supply you will use, and you can still case up your first psu, going with linear voltage travel, and voltage and current analogue panel meters, ok so you can only drop to the reference voltage, and you have small constraints with current out put, but theses are ok for a first psu build, your still power plenty of other projects with such a first power supply.

Build the board up, rectifier use individual diodes, not a package, even a full wave rectifier can be a challenge for a first circuit, but a goal once built, then your filter capacitor, regulator, and so on, your be surprised what looks basic in lines on paper is way more time consuming on the board, and of course fused and grounded case if steel or alloy.

All the above is not to put a damper on your circuit, i think its ok, the above is not taking huge steps to quick in the build process, any other member thats started on copper strip or similar will know what i mean. Your look back in a few months, and have had little or no failures. Please dont be offended by my suggestions, as thats all they are, the bottom line starts with you and your choices. Dave. :)
 
Last edited:
Top