Author Topic: TL431 linear power supply  (Read 9119 times)

0 Members and 2 Guests are viewing this topic.

Offline mike_mike

  • Frequent Contributor
  • **
  • Posts: 522
  • Country: ro
Re: TL431 linear power supply
« Reply #75 on: December 17, 2018, 04:23:37 pm »
There should be a NPN transistor where it writes "vdd" (with the base on vdd) ?
 

Offline spec

  • Frequent Contributor
  • **
  • Posts: 833
  • Country: england
  • MALE
Re: TL431 linear power supply
« Reply #76 on: December 18, 2018, 02:19:56 am »
some notes:
- The common emitter stage inverts the phase so you have to swap inverting and non-inverting inputs.
Thanks- standard error ;D

- I've not made any calculations yet, but I think it would be safer to replace some transistors with BD139/40.
Good observation. I did look at the BC546/BC556 dissipation which will be 300mW worst case. I haven't done a thermal budget, but the BC546/BC556 limit on the datasheet is 500mW. I am not too keen on the BD139/140 for those circuit functions for various reasons. I was intending to recommend that mike_mike checks how hot the transistor cases are (by the finger test), and if necessary, add small heatsinks- no big deal.

- Probably it would be hard to compensate.
Not sure why you think this.

I'll see if I can simulate it.
Be great if you did do a bode plot of the circuit. :)
« Last Edit: December 18, 2018, 02:27:49 am by spec »
 
The following users thanked this post: AngraMelo

Offline spec

  • Frequent Contributor
  • **
  • Posts: 833
  • Country: england
  • MALE
Re: TL431 linear power supply
« Reply #77 on: December 18, 2018, 02:50:41 am »
mike-mike

I have been away from your PSU thread for a few days, but have now caught up with the latest posts.

Attached is a revised schematic for the 25V, 5A Version 3 PSU with the opamp phase error corrected and other modifications as follows.

C2, C17 and C15 are 10uF non polarized capacitors ?
Yes, they are X7R ceramic types for good high frequency performance.

Does they really need to have such a big value and they really need to be non polarized ?
  Yes, as above, but I have changed the schematic in view of your comments so that all solid capacitors are 100nF except one 1uF.

R11, R16, R6, R13 could have a more usual value ?
R11 was 330k, revised to 220k (330k is a normal value though),
R16 was 24k and is now 240k and must be this value.
R6 is 56R and must be this value (56R is a standard value)
R14 is 14k and must be this value. But you will notice that it is annotated AOT (Adjust On Test). The actual value should be adjusted to give exactly 25V output voltage with the output voltage setting potentiometer set to maximum, but this is optional and up to you. I could have put another zener diode in the circuit to eliminate the need for AOT and then R14 could be a standard value. If you would like this option just shout and I will modify the schematic accordingly.

About odd/unavailable values, you do know that you can make any value capacitor and resistor by connecting them in parallel/series.

« Last Edit: December 18, 2018, 02:57:32 am by spec »
 
The following users thanked this post: mike_mike, AngraMelo

Offline mike_mike

  • Frequent Contributor
  • **
  • Posts: 522
  • Country: ro
Re: TL431 linear power supply
« Reply #78 on: December 18, 2018, 04:34:23 am »
@spec Thank you.
D5 should be a simple zener diode or it is better to use a zener diode and a npn transistor ? Or there can be used an LM317 ?
 

Offline Kleinstein

  • Super Contributor
  • ***
  • Posts: 6222
  • Country: de
Re: TL431 linear power supply
« Reply #79 on: December 18, 2018, 06:49:31 am »
mike-mike

I have been away from your PSU thread for a few days, but have now caught up with the latest posts.

Attached is a revised schematic for the 25V, 5A Version 3 PSU with the opamp phase error corrected and other modifications as follows.

...
The regulator shown looks like it is rather slow with 1 µF at the OP and 100 µF after the transistor.  The more normal values are about a factor of 1000 smaller.  Is the stability (e.g. loop gain) check in a simulation ? With 2 low pass filtering stages and no phase boost there might be a stability issue, at least with a capacitive load.

Even if the BC546/BC556 are specified at 500 mW, a TO92 case still gets rather hot at 300 mW, if there is no fan.
So for the 10 mA current sink at the output the BD139 is a good idea. For the current source for the regulator one should get a way with a TO92 case transistor if the current is a little lower (e.g. 5 mA). This should be sufficient even for 5 A output.
 

Offline spec

  • Frequent Contributor
  • **
  • Posts: 833
  • Country: england
  • MALE
Re: TL431 linear power supply
« Reply #80 on: December 18, 2018, 09:22:44 am »
@spec Thank you.
No probs- apologies for not correcting the obvious opamp phase error sooner. :)

D5 should be a simple zener diode or it is better to use a zener diode and a npn transistor ? Or there can be used an LM317 ?
You could do any of those things: it is not necessary though. But, if for example, you would rather use an LM317, just say and I will post a modified circuit for you.

D5 in the isue03 version is a simple zener diode and has two functions:
[1] To provide 15V Vcc for the opamp
[2] To provide the reference voltage for the PSU voltage stabilization loop.

Because of the architecture of the issue 3 PSU, where there is a large voltage difference between the raw supply voltage and 15V rail, a single zener will be quite adequate for your stated requirements. Bear in mind that this PSU has been designed specifically with your stated requirements that in mind. Given a free choice the PSU architecture would be different, as would a few other areas, including the opamp and driver transistor which would probably be a depletion NMOSFET. The 2N3055 transistors would also change to a transistor with a much lower thermal resistance, junction to case.

The other thing is that the LM358 current drawn from the 15V supply rail will be more or less constant at a maximum of 400uA,  as the LM358 is not sourcing any current, only sinking current. This is one of the handy characteristics of the LM358: a constant current drain, as opposed to a varying current drain, is beneficial for the voltage stability of the 15V rail.

But, I will post another PSU circuit to show how I would like to do the voltage reference using this architecture and if I have time, I will also post an LM317 reference version.

By the way, this PSU gives an output voltage of 0V to 25V at 0A to 5A and will current limit around 6A. But the OV needs to be qualified: exactly 0V output is impossible to achieve without extra circuitry. This problem holds for any linear PSU, not just this one. On balance, and in the interests of simplicity and your requirements, I omitted the extra circuitry. But if you would like true 0V just say.

The 0V ambiguity is due to leakage currents in the driver and output transistors, and there are effectively six transistors in parallel. Also the output transistors and driver will have a high junction temperature, especially with high currents and a low output voltage so the 0V potential will vary depending on usage. Temperature increases leakage exponentially.

You will have, no doubt read the posts about the compensation being too heavy. Don't worry about this, it is intentional, as I have previously stated. There is little point in fine tuning the compensation until the PSU has been prototyped and tested. Once that is done we can optimize the stabilization, if necessary, by a few simple modifications, mainly capacitor changes.

In common with many PSU designs, the approach is that the opamp voltage servo loop defines the absolute DC output voltage and also provides the very low frequency currents, the big electrolytic capacitor provides the medium frequency currents, and the solid capacitor (ceramic) provides the high frequency components. You can get a more in depth coverage of this area from the manufacturers application notes and from individual component data sheets.

« Last Edit: December 18, 2018, 10:35:59 am by spec »
 
The following users thanked this post: AngraMelo

Offline spec

  • Frequent Contributor
  • **
  • Posts: 833
  • Country: england
  • MALE
Re: TL431 linear power supply
« Reply #81 on: December 18, 2018, 09:37:42 am »
mike-mike

I have been away from your PSU thread for a few days, but have now caught up with the latest posts.

Attached is a revised schematic for the 25V, 5A Version 3 PSU with the opamp phase error corrected and other modifications as follows.

...
The regulator shown looks like it is rather slow with 1 µF at the OP and 100 µF after the transistor.  The more normal values are about a factor of 1000 smaller.  Is the stability (e.g. loop gain) check in a simulation ? With 2 low pass filtering stages and no phase boost there might be a stability issue, at least with a capacitive load.

Even if the BC546/BC556 are specified at 500 mW, a TO92 case still gets rather hot at 300 mW, if there is no fan.
So for the 10 mA current sink at the output the BD139 is a good idea. For the current source for the regulator one should get a way with a TO92 case transistor if the current is a little lower (e.g. 5 mA). This should be sufficient even for 5 A output.
Both of these areas have already been discussed.

5mA Ik will not give sufficient current, worst case, to drive the output transistors.
 
The following users thanked this post: AngraMelo

Offline mike_mike

  • Frequent Contributor
  • **
  • Posts: 522
  • Country: ro
Re: TL431 linear power supply
« Reply #82 on: December 18, 2018, 06:56:14 pm »
@spec
I need a final version of schematic, which uses LM317 instead of Zener diode and I will start drawing the layout.
 

Offline Kleinstein

  • Super Contributor
  • ***
  • Posts: 6222
  • Country: de
Re: TL431 linear power supply
« Reply #83 on: December 18, 2018, 07:56:14 pm »
With the extra charge pump stage as shown the voltage might get too high for an LM317.
So if at all one should use the charge pump to only give something like 2 times the normal voltage (diode to ground and not the normal positive supply).

I don't see a real need for the charge pump. Already just extra diodes and Filter capacitor would give some extra headroom for the current source, as with an extra capacitor one can have less ripple than the main supply. For the low current needed by the OP, the zener might be sufficient - the LM317 would give some extra loss in voltage, but could still work.
 

Offline spec

  • Frequent Contributor
  • **
  • Posts: 833
  • Country: england
  • MALE
Re: TL431 linear power supply
« Reply #84 on: December 19, 2018, 04:13:59 am »
@spec
I need a final version of schematic, which uses LM317 instead of Zener diode and I will start drawing the layout.
OK will do :)
 
The following users thanked this post: AngraMelo

Offline spec

  • Frequent Contributor
  • **
  • Posts: 833
  • Country: england
  • MALE
Re: TL431 linear power supply
« Reply #85 on: December 19, 2018, 04:20:51 am »
With the extra charge pump stage as shown the voltage might get too high for an LM317.
So if at all one should use the charge pump to only give something like 2 times the normal voltage (diode to ground and not the normal positive supply).

I don't see a real need for the charge pump. Already just extra diodes and Filter capacitor would give some extra headroom for the current source, as with an extra capacitor one can have less ripple than the main supply. For the low current needed by the OP, the zener might be sufficient - the LM317 would give some extra loss in voltage, but could still work.
:-//
 
The following users thanked this post: AngraMelo

Offline not1xor1

  • Frequent Contributor
  • **
  • Posts: 376
  • Country: it
Re: TL431 linear power supply
« Reply #86 on: December 19, 2018, 08:13:00 am »
There should be a NPN transistor where it writes "vdd" (with the base on vdd) ?

the current regulated by the zener is just few mAmperes so there is no need to transform it in a series regulator (if you meant that)
 

Offline not1xor1

  • Frequent Contributor
  • **
  • Posts: 376
  • Country: it
Re: TL431 linear power supply
« Reply #87 on: December 19, 2018, 08:16:02 am »
[...]
I'll see if I can simulate it.
Be great if you did do a bode plot of the circuit. :)

I simulated just AC. The bodeplot doesn't look so good. It needs some more work on compensation.

 

Offline not1xor1

  • Frequent Contributor
  • **
  • Posts: 376
  • Country: it
Re: TL431 linear power supply
« Reply #88 on: December 19, 2018, 08:28:56 am »
With the extra charge pump stage as shown the voltage might get too high for an LM317.
So if at all one should use the charge pump to only give something like 2 times the normal voltage (diode to ground and not the normal positive supply).

I don't see a real need for the charge pump. Already just extra diodes and Filter capacitor would give some extra headroom for the current source, as with an extra capacitor one can have less ripple than the main supply. For the low current needed by the OP, the zener might be sufficient - the LM317 would give some extra loss in voltage, but could still work.

The LM317 request concerned spec's schematic, not mine.
Since the purpose of that circuit of mine was to get an output of 25V from a LM358 without any voltage boost stage, the charge pump ensures a proper and stable supply voltage for the LM358 even when the transformer voltage gets too low.
Of course the circuit may work even without charge pump as showed in the other schematic I posted
 

Offline spec

  • Frequent Contributor
  • **
  • Posts: 833
  • Country: england
  • MALE
Re: TL431 linear power supply
« Reply #89 on: December 19, 2018, 08:57:08 am »
UPDATE #2 of 2018_12_21 (R4 changed from 1k to 56R)
UPDATE #1 of 2018_12_19 (add resistor to pot wiper)

mike_mike

Schematic for 25V 5A PSU version 06, using an LM317 regulator, attached below. There are a few other minor changes here and there, including replacing all BC547/BC557 transistors with BC337/BC227 types. This will give more power dissipation margin for Q3 and Q8 and will make the constant current handover a touch sharper. But you can still use the original BC547/BC557 types if you like. The 1N400x diode in Q3 collector has been removed as it was unnecessary.

I have done a thermal budget for BC547/BC557 in Q3 and Q8 positions and the maximum junction temperature, with 300mW dissipation and an assumed ambient temperature of 70 degC, is 125 degC, which gives a 25 deg C margin which is safe, especially as it is unlikely that the ambient temperature will be as high as 70 degC. All the same, with either transistor family, it would be wise to attach small heatsinks to the cases of Q3 and Q8.

All resistors have been normalized and the need for an AOT resistor has been eliminated.

A 1N400x diode has also been added at the output.

All solid capacitors are X7R ceramic, through hole, and all polarized capacitors are aluminum electrolytic.
« Last Edit: December 21, 2018, 06:21:57 am by spec »
 

Offline mike_mike

  • Frequent Contributor
  • **
  • Posts: 522
  • Country: ro
Re: TL431 linear power supply
« Reply #90 on: December 19, 2018, 09:04:42 am »
@spec Thank you.
Can you please replace the transistors that will generate more heat (Q3, Q8) with some transistors that will support more power in order to not heat so much ?
 

Offline spec

  • Frequent Contributor
  • **
  • Posts: 833
  • Country: england
  • MALE
Re: TL431 linear power supply
« Reply #91 on: December 19, 2018, 09:11:17 am »
@spec Thank you.
Can you please replace the transistors that will generate more heat (Q3, Q8) with some transistors that will support more power in order to not heat so much ?
No sweat mike_mike.

I would rather not fit larger transistors because their other characteristics are not so good.  This whole dissipation business for Q3 and Q8 has been blown up out of all proportion (like a lot of things).  Don't even give it a thought. Either transistor type will be fine. :)
« Last Edit: December 19, 2018, 09:13:44 am by spec »
 

Offline imo

  • Super Contributor
  • ***
  • Posts: 1757
  • Country: 00
Re: TL431 linear power supply
« Reply #92 on: December 19, 2018, 09:23:51 am »
I would add 4x4n7/100V (4n7-10n) capacitors in parallel with the bridge's diodes, and an 100k (100k-470k) resistor from the RV3's wiper to gnd (in case the wiper looses its contact).
« Last Edit: December 19, 2018, 09:28:37 am by imo »
 

Offline spec

  • Frequent Contributor
  • **
  • Posts: 833
  • Country: england
  • MALE
Re: TL431 linear power supply
« Reply #93 on: December 19, 2018, 09:32:29 am »
I would add 4x4n7/100V (4n7-10n) capacitors in parallel with bridge diodes, and an 100k (100k-470k) resistor from the RV3's wiper to gnd (in case the wiper looses its contact).
Thanks. I did consider commutation capacitors across the diodes, but on balance decided to pass, besides which they would mess up my drawing. :D

Good suggestion about the resistor on the wiper. I will embody that. :-+
 

Offline blackdog

  • Frequent Contributor
  • **
  • Posts: 580
  • Country: nl
  • Please stop pushing bullshit...
Re: TL431 linear power supply
« Reply #94 on: December 19, 2018, 09:48:44 am »
Hi,

Every time I have to laugh a lot about the standard errors that are made at Linear power supply circuits here on the forum...

Why let's take the last schemes, C8 100uF, and R13 100 Ohm, you must be jokink...

If your reaction is to make the circuit slower because it generates, then you're in a wrong way to solve the problem, YOU NEED SPEED!!!

Every component within de opamp loop creats delay, this delay is your main problem, and also placing extra delay components (C8 and R13) wil make your problems worse.

Extra components in de control loop of a opamp may not create more delay so that the fase margen of de whole circuit comes below 45 degrees.
If you design a goed linear power supply schematic then there wil be 60 a 70 degrees phase margin.

Small tip, place a capacitor over R16 a 240K say 0.1uF.

If you ignore this, you will always start crying.  :box:  :-DD

Kind regards,
Blackdog

“Two things are infinite, the universe and human stupidity, and I am not yet completely sure about the universe.”
 

Offline spec

  • Frequent Contributor
  • **
  • Posts: 833
  • Country: england
  • MALE
Re: TL431 linear power supply
« Reply #95 on: December 19, 2018, 10:14:01 am »
Here we go again. Another smart arse pops out of the woodwork.

As I have said before to others, lets see your complete circuit for this application. :-//
 

Offline spec

  • Frequent Contributor
  • **
  • Posts: 833
  • Country: england
  • MALE
Re: TL431 linear power supply
« Reply #96 on: December 19, 2018, 10:19:54 am »
mike-mike

I have updated the schematic for version 6 of the PSU attached to reply #89: nothing major just added a 100k resistor to the pot wiper.

Keep an eye on reply #89 because I will update that as necessary.
« Last Edit: December 19, 2018, 10:30:57 am by spec »
 

Offline imo

  • Super Contributor
  • ***
  • Posts: 1757
  • Country: 00
Re: TL431 linear power supply
« Reply #97 on: December 19, 2018, 10:23:48 am »
Also add a 10uF/16V from LM317's ADJ pin to GND (noise reduction).
« Last Edit: December 19, 2018, 10:28:36 am by imo »
 

Offline spec

  • Frequent Contributor
  • **
  • Posts: 833
  • Country: england
  • MALE
Re: TL431 linear power supply
« Reply #98 on: December 19, 2018, 10:28:54 am »
Also add a 10uF/16V from LM317's ADJ pin to GND (noise reduction).
I also considered that but decided there was no need. Take a look at the many posts on this thread to get a feel for the background to this power supply project. Still, at least you haven't started  banging on about how slugged the opamp is at this stage.  :)
 

Offline imo

  • Super Contributor
  • ***
  • Posts: 1757
  • Country: 00
Re: TL431 linear power supply
« Reply #99 on: December 19, 2018, 10:43:33 am »
You have done a nice LTSpice simulation above - you may try with say 15V output, switching from 10mA -> 5A -> 10mA output current (with, say, 200ms period and 10us edges) and adjust the loop when required..
« Last Edit: December 19, 2018, 10:45:35 am by imo »
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf