TL431 High Voltage Regulator - How to get it thermally stable?

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Hi,

The idea of this circuit is to design a power regulator that outputs adjustable regulated voltage between 40 and 50 volt. However I seem to unable to get the output thermally stable. The voltage drifts from 42volt to 38 volt in about 10 minutes. Can anybody help me to improve this design? Thanks in advance.

AGJVZslIdxdLAAAAAElFTkSuQmCC
 
There is no source for the Q1 base current. You need to supply
a positive voltage to the TL431 cathode/ 2n5551 base via a resistor from a positive voltage source.


Hi Art,

Thanks a lot for your reply.

Spice simulation shows there is a 7.8uA current through the Q1 base. If I take the 10K R5 out, the current(about 9.2uA) from TL431 is completely fed into Q1 base. Is this enough to drive the Q1 into amplification?

How to calculate the value for the resistor you mentioned?
 
Q1 and R5 are useless and incorrect. Both, Q5 and U1 are inverting amplifying devices. The result will be an oscillator. Connect TL431 cathode directly to the gate of IRFP240 or insert a thermally stable Zener diode between them (level shifte, for higher output voltage).
 
Last edited:
Member
Joined 2008
Paid Member
Looking only at the 431, the minimum current for regulation on the 431 itself is 0.4mA. If you have the capability it's good to give it some margin plus Q1's base current. If you can spare the current then run something like 5 - 10mA through it. In your sim, try a ~7K resistor from the cathode of the 431 to the output voltage rail. Take out the 10K resistor.

Your 33K/1.5K divider on the 431 ref pin is now setting the output V+ at 57.5V which is higher than the input voltage so you'll either have to raise the supply voltage or lower the value of the resistor in the 33K position to get a lower output voltage.
 
Looking only at the 431, the minimum current for regulation on the 431 itself is 0.4mA. If you have the capability it's good to give it some margin plus Q1's base current. If you can spare the current then run something like 5 - 10mA through it. In your sim, try a ~7K resistor from the cathode of the 431 to the output voltage rail. Take out the 10K resistor.

Your 33K/1.5K divider on the 431 ref pin is now setting the output V+ at 57.5V which is higher than the input voltage so you'll either have to raise the supply voltage or lower the value of the resistor in the 33K position to get a lower output voltage.

Do you mean something like this ? TL431+Mosfet.JPG
 
Member
Joined 2008
Paid Member
Sorry, it was late and I locked in on the 431 as the control of a shunt regulator. M1 was then just a problem that needed to be dealt with in the morning.
Here are two circuits from the data sheet that illustrate basic approaches.
 

Attachments

  • Series Pass 431.jpg
    Series Pass 431.jpg
    18.7 KB · Views: 620
  • 431 Shunt Reg.jpg
    431 Shunt Reg.jpg
    15.9 KB · Views: 387
Hi dirkwright,

Here is the LTSpice file. Could you illustrate your circuit?

View attachment 405832

I don't use LTSpice, sorry.

I have a simple regulator, but it's for 300VDC. Basically, the TL431 is in the emitter circuit (as a stable voltage reference, not as a shunt regulator) with the supply for the TL431 coming from the raw DC side of the series pass device. The base of the error BJT is driven by a voltage divider on the output side of the series pass device. The collector of the error amplifier is connected to the base of the series pass device and also, via a resistive device, to the raw DC side of the series pass device. Hope this helps.
 
The correct answer here, and no offense, is stop using the simulator and start reading datasheets and basic literature. You will not learn electronics from a simulator - it is a 'garbage in -> garbage out' device. Most of the components used in it are represented by models which are in most cases (I'd say 99%) valid only for the expected use of the device. If it's not connected correctly, the results will be nonsense, and rarely represent what will happen in a real circuit. This is why you need to understand the basics of a circuit before you even attempt simulating it, otherwise you will simply not know how to interpret the results of the simulation, not what to look for if they are unexpected. This has been said over and over on this forum, and apparently it always needs to be said again.

In your circuit, there are basic errors.
In the original one, the TL431 has no supply. It is a a shunt regulator, which is really a sort of zener diode replacement (albeit adjustable). There is a reason whu it is drawn as a zener diode! So, it needs to have a supply of current through it.
Secondly, there is no need for an added NPN transistor. Although in principle the schepatic could work with it, it can't work woth a TL431 simply because a TL431 can only go as low as 2.5V, and the B-E treshold of the transistor is ~0.7V. So, it will be permanently turned on to some extent, while the TL431 will essentially do nothing. The fact that the lowers voltage the TL431 can regulate is 2.5V (the value of it's internal reference) is well and clearly stated in it's datasheet (with examples of use!) and the datasheet is available from literally hundreds of places on the web. This clearly shows you didn't bother to read it.

Since the TL431 can go up to some 36V, but of course you don't want to use it at it's limit. the simplest way to extend it's voltage handling capability for your case is to add a zener diode in series. It should be for a voltage less than your intended output voltage, and 2.5V plus it's value is the minimum you can get with that circuit, 36V plus it's value is the maximum (again this is for the highest voltage on the TL431). There is no need to do anything special with thermal compensation of the zener - the TL431 ia an active device and it's regulation input is actually a feedback node. It will correct for any thermal drift of the zener.

Again - using a simulator is no shortcut to designing circuits for people who can't design them without one. Especially a simple one like yours - it's design can be done on paper in 5 minutes or less.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.