Author Topic: TL431 linear power supply  (Read 38038 times)

0 Members and 1 Guest are viewing this topic.

Offline xavier60

  • Super Contributor
  • ***
  • Posts: 2783
  • Country: au
Re: TL431 linear power supply
« Reply #25 on: December 14, 2018, 06:50:30 am »

And I want to be very stable and I want to NOT oscillate... this is the hardest thing for me.

Most power supply designs will oscillate without proper compensation, especially those that use op-amps.
The power supply circuit in your first post will work well  enough with compensation.
HP 54645A dso, Fluke 87V dmm,  Agilent U8002A psu,  FY6600 function gen,  Brymen BM857S, HAKKO FM-204, New! HAKKO FX-971.
 

Offline spec

  • Frequent Contributor
  • **
  • Posts: 833
  • Country: england
  • MALE
Re: TL431 linear power supply
« Reply #26 on: December 14, 2018, 07:11:32 am »
Hi mike-mike,

Attached below, is a schematic for your PSU as promised. Apologies for the delay, but it took longer to sort than expected, and other things got in the way too. >:(

Bear in mind that this is a paper design and I do not have a full specification for your mains transformer so I have made some assumptions. Also, right now, I do not have access to a lab so I cannot build and test a prototype for you. But I have analyzed the circuit and used a conservative approach whenever possible.

You may find the circuit odd, but it is not an outline circuit like you normally see, and shows the required wiring including star points (SPs), padding, and decoupling.

The two opamp options shown on the schematic are unique, and not any old opamp will do, although you could use other opamps with reduced performance. The OPA192 is the better of the two, but costs a bit more. But, in terms of your PSU there would be little difference in the performance. Mainly the accuracy of the 25V setting would be affected.

The frequency compensation is very heavy, but that is intentional. If necessary the frequency compensation can be optimized at a later date, by a few simple component changes. The important thing is to get the basic circuit working at this initial stage.

You will need very good heat-sinking for the output transistors and fan cooling would be advisable, but we can talk you through all that.

Finally, I cant emphasis how important it is to do a good layout with the connections as shown in the attached schematic. The thick lines on the schematic indicate where thick wires/PCB traces are required. It is also essential to keep wires and PCB traces short and compact. The 100nF capacitor across the supply pins of the opamp, must have short leads and be physically connected to the opamp pins, or as close as possible to the pins. Likewise the padding resistors and other decoupling capacitors must be directly on, or close to the opamp pins.

There is a physical layout that I would recommend, which will simplify construction a lot and optimize performance. Once you have had a look at the schematic, we can discuss the layout if you like.

Unless otherwise stated, all solid capacitors are ceramic X7R through hole types, the physically bigger the better (not surface mount).

The specification for the PSU is 0V to 25V (exactly), 0A to 5A output current.
« Last Edit: December 14, 2018, 08:22:59 am by spec »
 

Offline spec

  • Frequent Contributor
  • **
  • Posts: 833
  • Country: england
  • MALE
Re: TL431 linear power supply
« Reply #27 on: December 14, 2018, 08:08:36 am »
The only reason I have not used the TL431 that way is why would I bother when better performing and easier to use operational amplifiers are plentiful?  It is routine in isolated switching power supplies however.

The problem is a little like using the 723 or most switching regulator controllers where stabilizing feedback to the inverting input comes from the transconductance output instead of the low impedance buffered output.  The TL431 has a transconductance output, despite what the datasheet specifications show, so gain varies considerably from device to device and with load.  If you build a series regulator with it, then the variable biasing conditions are altering the transconductance unless steps are taken to prevent it so the frequency compensation becomes difficult to control.



A better question might be *why* the TL431 operates so benignly when used as a simple shunt regulator.  It will normally have a relatively large load capacitance which rolls off the gain of its transconductance output for dominant pole compensation which is exactly how a transconductance amplifier is compensated.  But in a circuit where it controls a series regulator, that cannot be done if fast response is desired and instead feedback from the squirrely output to the input implements the frequency compensation because people are used to working with voltage feedback amplifiers.  It can work as the 723 shows but can also be hit or miss.

This brings up an alternate compensation scheme; place a series RC network from the TL431 output to ground.

I take it then that you yourself or anyone that you know has built a successful TL431 PSU.
 

Offline not1xor1

  • Frequent Contributor
  • **
  • Posts: 716
  • Country: it
Re: TL431 linear power supply
« Reply #28 on: December 14, 2018, 08:11:59 am »
No probs :)

I will have a look for a suitable PSU, but I am quite busy at the moment. A number of members are experienced in PSUs, perhaps they could recommend some suitable designs as well.

But I can recommend a complete lab PSU that will do your job for around £47UK, including shipping (the transformer for a DIY PSU would cost more than that). The lab PSU provides 0V to 30V and 0A to 10A, with digital readouts for both voltage and current. There are controls for coarse and fine voltage output and the same for current. I have three of these and use them for electronics, battery charging, you name it.

https://www.ebay.co.uk/p/Adjustable-DC-Power-Supply-30v-10a-Variable-Precision-Digital-Lab-PSU-Clip-Cable/23025566378

You can probably get this generic lab PSU delivered to Romania.

From the pictures (if you look through the case aeration slots it looks empty inside) I guess it is a switching PSU.
The linear ones are better as they can be easily improved by replacing the electrolytic caps (they are low quality and too low voltage) and adding a proper TO3 heatsink (as minimal cure).
It is a linear PSU, with a very low level noise output of 3mV, why do any guessing. :-//

There are lots of chinese PSUs looking the same and with small or large differences inside (some with a huge heatsink on the back some with just a piece of aluminium inside).
I bought a similar one (30V 5A) for about 45€ including shipping.
It has a huge transformer.
The one you linked is 30V 10A. So it should have a 500VA transformer, but since they are coarsely overrated a 300VA one would be more likely.
But if you zoom the pictures and look at the last one, you can clearly see that the case is empty.
So unless it uses a low profile toroid it can be just a switching PSU.
OK ...I found a youtube video... it is definitely a switching PSU.
https://youtu.be/1_ZK68_c3Ok?t=145
 

Offline not1xor1

  • Frequent Contributor
  • **
  • Posts: 716
  • Country: it
Re: TL431 linear power supply
« Reply #29 on: December 14, 2018, 08:22:22 am »
Hi mike-mike,

Attached below, is a schematic for your PSU as promised. Apologies for the delay, but it took longer to sort than expected, and other things got in the way too. >:(


you must be kidding  ;D
a 20V/µs opamp with a huge 1µF compensation capacitor...
and what would be the advantage versus a proven and foolproof design using the traditional LM723 ?
 

Offline spec

  • Frequent Contributor
  • **
  • Posts: 833
  • Country: england
  • MALE
Re: TL431 linear power supply
« Reply #30 on: December 14, 2018, 08:30:25 am »

And I want to be very stable and I want to NOT oscillate... this is the hardest thing for me.

The power supply circuit in your first post will work well  enough with compensation.
Can you show a circuit for this compensation using a TL431, especially one that you have built and tested successfully. Or perhaps know of such a PSU, that someone else has built and tested.

I am genuinely interested as stated before.
 

Offline spec

  • Frequent Contributor
  • **
  • Posts: 833
  • Country: england
  • MALE
Re: TL431 linear power supply
« Reply #31 on: December 14, 2018, 08:57:24 am »
No probs :)

I will have a look for a suitable PSU, but I am quite busy at the moment. A number of members are experienced in PSUs, perhaps they could recommend some suitable designs as well.

But I can recommend a complete lab PSU that will do your job for around £47UK, including shipping (the transformer for a DIY PSU would cost more than that). The lab PSU provides 0V to 30V and 0A to 10A, with digital readouts for both voltage and current. There are controls for coarse and fine voltage output and the same for current. I have three of these and use them for electronics, battery charging, you name it.

https://www.ebay.co.uk/p/Adjustable-DC-Power-Supply-30v-10a-Variable-Precision-Digital-Lab-PSU-Clip-Cable/23025566378

You can probably get this generic lab PSU delivered to Romania.

From the pictures (if you look through the case aeration slots it looks empty inside) I guess it is a switching PSU.
The linear ones are better as they can be easily improved by replacing the electrolytic caps (they are low quality and too low voltage) and adding a proper TO3 heatsink (as minimal cure).
It is a linear PSU, with a very low level noise output of 3mV, why do any guessing. :-//

There are lots of chinese PSUs looking the same and with small or large differences inside (some with a huge heatsink on the back some with just a piece of aluminium inside).
I bought a similar one (30V 5A) for about 45€ including shipping.
It has a huge transformer.
The one you linked is 30V 10A. So it should have a 500VA transformer, but since they are coarsely overrated a 300VA one would be more likely.
But if you zoom the pictures and look at the last one, you can clearly see that the case is empty.
So unless it uses a low profile toroid it can be just a switching PSU.
OK ...I found a youtube video... it is definitely a switching PSU.
https://youtu.be/1_ZK68_c3Ok?t=145
Hell, you are right. I specifically ordered the linear type. My apologies.
 

Offline spec

  • Frequent Contributor
  • **
  • Posts: 833
  • Country: england
  • MALE
Re: TL431 linear power supply
« Reply #32 on: December 14, 2018, 09:01:14 am »
Hi mike-mike,

Attached below, is a schematic for your PSU as promised. Apologies for the delay, but it took longer to sort than expected, and other things got in the way too. >:(


you must be kidding  ;D
a 20V/µs opamp with a huge 1µF compensation capacitor...
and what would be the advantage versus a proven and foolproof design using the traditional LM723 ?
I said it was slugged. :)
I made no claim about any advantage over anything. Why didn't you post a suitable 723 circuit? You could have saved me a load of bother. >:D
 

Offline not1xor1

  • Frequent Contributor
  • **
  • Posts: 716
  • Country: it
Re: TL431 linear power supply
« Reply #33 on: December 14, 2018, 09:11:26 am »
45Vdc means the voltage after rectifier and filtering.
I would still like to remain to LM358 since my electronic components vendor does not have this part number (the one with LT).
I just need something like a error amplifier controlling one or more transistors (npns), a ref voltage, a current limit (not variable). Also I want to have a output current of about max 5A and a voltage of about 3-25Vdc. A overload protection led should be very useful, but only if it is simple to implement.
I can use a 30vca 10A transformer or an 24vca, 6A transformer.
And I want to be very stable and I want to NOT oscillate... this is the hardest thing for me.

Could you please help me with such a schematic ?

With a 30VAC 10A transformer or even with the smaller 24VAC 6A transformer you are building more an expensive electric heater than a PSU  :)

You must first take into account that transformer are rated for resistive loads. If you put a rectifier bridge and a capacitor, you get narrow pulses of high value currents which increase power dissipation through the transformer winding, so according to the value of the capacitor you have to properly derate the transformer. For average values (around 2000-3000µF per ADC) 60% would work.
That means that a 30VAC 10A transformer is suitable for a 30VDC 6A PSU (with multiple electrolic capacitors in parallel for a total value of 12000-20000µF).
The 24VAC 6A transformer is suitable for a 30VDC 3.6A PSU (7000-10000µF).
And even with derating, in the worst cases you would need to dissipate approximately 200-220W or 100-110W. That is a lot of power.

It would be much better to just use a rectifier bridge and capacitors to get a DC voltage and then use one of those programmable switching modules to get a regulated output. That would save lot of wasted power and expensive heatsinks.

If you still want to build yourself such electronic heaters, given that you do not want current regulation, then the easiest solution is to just use the old LM723, even one of those datasheet circuits.
 

Offline not1xor1

  • Frequent Contributor
  • **
  • Posts: 716
  • Country: it
Re: TL431 linear power supply
« Reply #34 on: December 14, 2018, 09:19:59 am »
Hi mike-mike,

Attached below, is a schematic for your PSU as promised. Apologies for the delay, but it took longer to sort than expected, and other things got in the way too. >:(


you must be kidding  ;D
a 20V/µs opamp with a huge 1µF compensation capacitor...
and what would be the advantage versus a proven and foolproof design using the traditional LM723 ?
I said it was slugged. :)
I made no claim about any advantage over anything. Why didn't you post a suitable 723 circuit? You could have saved me a load of bother. >:D

there are loads of 723 circuit diagrams/projects on the net, and lots of them even in the datasheets (ST, NSC, etc). So what would be advantage of yet another (in this case untested) design ?  :)
 

Offline xavier60

  • Super Contributor
  • ***
  • Posts: 2783
  • Country: au
Re: TL431 linear power supply
« Reply #35 on: December 14, 2018, 09:29:48 am »

And I want to be very stable and I want to NOT oscillate... this is the hardest thing for me.

The power supply circuit in your first post will work well  enough with compensation.
Can you show a circuit for this compensation using a TL431, especially one that you have built and tested successfully. Or perhaps know of such a PSU, that someone else has built and tested.

I am genuinely interested as stated before.
The only times that I have used a TL431 as an active part of the regulation loop rather than as the voltage reference, are as low voltage control rail supplies.  I used the same topology as the circuit in the first post, but with a medium power MOSFET or BJT acting as follower for the K voltage. Very easy to compensate with a capacitor between K and R.
Just like with the circuit  you posted, even if the feedback capacitor is oversize, the op-amp or TL431 becomes a very slow Miller integrator.
Transient response remains reasonable because all that the op-amp or TL431 needs do is compensate for small changes in B-E voltage as load current changes.
Response can optimized by reducing the size of the capacitor  and adding some series resistance.
HP 54645A dso, Fluke 87V dmm,  Agilent U8002A psu,  FY6600 function gen,  Brymen BM857S, HAKKO FM-204, New! HAKKO FX-971.
 

Offline spec

  • Frequent Contributor
  • **
  • Posts: 833
  • Country: england
  • MALE
Re: TL431 linear power supply
« Reply #36 on: December 14, 2018, 09:44:42 am »

And I want to be very stable and I want to NOT oscillate... this is the hardest thing for me.

The power supply circuit in your first post will work well  enough with compensation.
Can you show a circuit for this compensation using a TL431, especially one that you have built and tested successfully. Or perhaps know of such a PSU, that someone else has built and tested.

I am genuinely interested as stated before.
The only times that I have used a TL431 as an active part of the regulation loop rather than as the voltage reference, are as low voltage control rail supplies.  I used the same topology as the circuit in the first post, but with a medium power MOSFET or BJT acting as follower for the K voltage. Very easy to compensate with a capacitor between K and R.
Just like with the circuit  you posted, even if the feedback capacitor is oversize, the op-amp or TL431 becomes a very slow Miller integrator.
Transient response remains reasonable because all that the op-amp or TL431 needs do is compensate for small changes in B-E voltage as load current changes.
Response can optimized by reducing the size of the capacitor  and adding some series resistance.
Thanks- very informative.
 

Offline spec

  • Frequent Contributor
  • **
  • Posts: 833
  • Country: england
  • MALE
Re: TL431 linear power supply
« Reply #37 on: December 14, 2018, 09:47:57 am »
Hi mike-mike,

Attached below, is a schematic for your PSU as promised. Apologies for the delay, but it took longer to sort than expected, and other things got in the way too. >:(


you must be kidding  ;D
a 20V/µs opamp with a huge 1µF compensation capacitor...
and what would be the advantage versus a proven and foolproof design using the traditional LM723 ?
I said it was slugged. :)
I made no claim about any advantage over anything. Why didn't you post a suitable 723 circuit? You could have saved me a load of bother. >:D

there are loads of 723 circuit diagrams/projects on the net, and lots of them even in the datasheets (ST, NSC, etc). So what would be advantage of yet another (in this case untested) design ?  :)
As you say, there are stacks of 723 circuits around, some better than others, so I would suggest that linking to a suitable circuit would help the OP a lot.
 

Offline spec

  • Frequent Contributor
  • **
  • Posts: 833
  • Country: england
  • MALE
Re: TL431 linear power supply
« Reply #38 on: December 14, 2018, 10:02:24 am »
you must be kidding  ;D  a 20V/µs opamp with a huge 1µF compensation capacitor...
I took this as a joke, but just in case, I chose those two opamps because of their low input ofset voltage, very low input bias current, high output current, RRO, RRI(100mV beyond) and the ability to drive a relatively large capacitor. The cost is reasonable too. The slew rate is not relevant in this application.

But, in general, there are endless situations, in fact most, where components are used way under their performance: 150mHz transistors used as audio amps, 20ns logic gates used to turn a LED on and off. A 10MIPS microprocessor used to alter the speed of a motor ...  :)
« Last Edit: December 14, 2018, 10:09:41 am by spec »
 

Offline spec

  • Frequent Contributor
  • **
  • Posts: 833
  • Country: england
  • MALE
Re: TL431 linear power supply
« Reply #39 on: December 14, 2018, 10:33:23 am »
mike_mike,

Attached is a 723 PSU schematic which I found drifting around my laptop. This, is the kind of thing that not1xor1 and I were discussing.
 

Offline mike_mikeTopic starter

  • Frequent Contributor
  • **
  • Posts: 885
  • Country: ro
Re: TL431 linear power supply
« Reply #40 on: December 14, 2018, 10:55:44 am »
 Thank you for your help.
What about the attached schematic ?
 I already asked on the forum but I still not get a response for some of the questions.
Will it (the power supply) oscillate ?
« Last Edit: December 14, 2018, 10:59:01 am by mike_mike »
 

Offline IanMacdonald

  • Frequent Contributor
  • **
  • Posts: 943
  • Country: gb
    • IWR Consultancy
Re: TL431 linear power supply
« Reply #41 on: December 14, 2018, 12:38:02 pm »
The main issue with amplified IC regulator arrangements is high volts drop. The LM317 has a rather high burden voltage in itself, and adding a further transistor and series resistor considerably worsens that. Although, it is a workable arrangement if top efficiency is not important.

The gotcha with your TL431 arrangement is that if the volts setting pot develops a bad spot on its track, the load is going to have its magic smoke forcibly removed, because the output will go straight to max. Putting the pot on the low side of the potential divider eliminates this risk, at the expense of the control being rather nonlinear. If it's a one-time preset that doesn't matter too much though.

To achieve proportional control of voltage whilst being 'o/c wiper safe' generally requires a noninverting opamp arrangement fed from the pot's wiper, as in spec's reply.
« Last Edit: December 14, 2018, 12:41:18 pm by IanMacdonald »
 

Offline David Hess

  • Super Contributor
  • ***
  • Posts: 16545
  • Country: us
  • DavidH
Re: TL431 linear power supply
« Reply #42 on: December 14, 2018, 12:51:06 pm »
you must be kidding  ;D
a 20V/µs opamp with a huge 1µF compensation capacitor...
and what would be the advantage versus a proven and foolproof design using the traditional LM723 ?

The compensation capacitor is way too large but a fast operational amplifier like that would be selected for lower noise.  I have done the same thing many times using an LT1007 or OP27.
 

Offline T3sl4co1l

  • Super Contributor
  • ***
  • Posts: 21606
  • Country: us
  • Expert, Analog Electronics, PCB Layout, EMC
    • Seven Transistor Labs
Re: TL431 linear power supply
« Reply #43 on: December 14, 2018, 05:56:17 pm »
I take it then that you yourself or anyone that you know has built a successful TL431 PSU.

Good evening spec,

Have you? :)

Tim
Seven Transistor Labs, LLC
Electronic design, from concept to prototype.
Bringing a project to life?  Send me a message!
 

Offline mike_mikeTopic starter

  • Frequent Contributor
  • **
  • Posts: 885
  • Country: ro
Re: TL431 linear power supply
« Reply #44 on: December 14, 2018, 06:11:32 pm »
mike_mike,

Attached is a 723 PSU schematic which I found drifting around my laptop. This, is the kind of thing that not1xor1 and I were discussing.
I built this schematic and it works as expected.
The problem is that LM723 will not be available more time from now and I need something that is always available from more than 2 manufacturers.
So this is why I want to use LM317 and pass transistors or TL431.
« Last Edit: December 14, 2018, 06:14:29 pm by mike_mike »
 

Offline spec

  • Frequent Contributor
  • **
  • Posts: 833
  • Country: england
  • MALE
Re: TL431 linear power supply
« Reply #45 on: December 14, 2018, 07:07:56 pm »
UPDATE #1   2018_12_18 The attached schematic has the opamp input connections reversed- see reply #77 for revised schematic.

Hi mike_mike

I got the message that you wanted to use parts from your spares box, so I did the attached schematic for you:
(and from the above post I see you need components second sourced in Romania)
« Last Edit: December 18, 2018, 03:09:29 am by spec »
 

Offline mike_mikeTopic starter

  • Frequent Contributor
  • **
  • Posts: 885
  • Country: ro
Re: TL431 linear power supply
« Reply #46 on: December 14, 2018, 07:11:36 pm »
Thank you.
But I cant see were is the current limit or the short circuit protection...
 

Offline spec

  • Frequent Contributor
  • **
  • Posts: 833
  • Country: england
  • MALE
Re: TL431 linear power supply
« Reply #47 on: December 14, 2018, 07:17:55 pm »
Thank you.
But I cant see were is the current limit or the short circuit protection...
No sweat.  :)
There isn't any current limit- didn't you say that you didn't need current limiting in one of your posts?
« Last Edit: December 14, 2018, 07:19:36 pm by spec »
 

Offline spec

  • Frequent Contributor
  • **
  • Posts: 833
  • Country: england
  • MALE
Re: TL431 linear power supply
« Reply #48 on: December 14, 2018, 07:21:14 pm »
you must be kidding  ;D
a 20V/µs opamp with a huge 1µF compensation capacitor...
and what would be the advantage versus a proven and foolproof design using the traditional LM723 ?

The compensation capacitor is way too large but a fast operational amplifier like that would be selected for lower noise.  I have done the same thing many times using an LT1007 or OP27.
I don't mess about with compensation. :-DD
 

Offline mike_mikeTopic starter

  • Frequent Contributor
  • **
  • Posts: 885
  • Country: ro
Re: TL431 linear power supply
« Reply #49 on: December 14, 2018, 07:21:48 pm »
I need current limiting. I dont need current limiting led.
I need something like a transistor and a power resistor for current limiting. And I dont need that the current limit to be variable.
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf