Need help understanding TL431

[hailteflon]

I said this in a former post. If anyone was interested I wanted to make sure I didn't give any eroneous info.

JAN zeners are extremely quiet. If you frequent surplus houses picking up JAN diodes is adviseable. They have a J, JX, JTX, or JTV in the part number. The milspec is M10950/159. There is often a code like CCCN or such. This is an obsolete manufacturing code for the military.

I am not sure about the lower noise. I have not presently been able to substantiate the lower noise in military diodes via technical documents on the web. I did test some on an oscilloscope years ago and they were much quieter, but I cannot say this is the case for all military diodes.

The military specification is MIL-PRF-19500, not 10950. There are over 700 specification sheets for semiconductors in this specification. The /159 is the 159th and it is a reference zener.

Type 19500 in the Document Number box. http://assist.daps.dla.mil/quicksearch/[/URL]

 

[AndrewT]

36V to 57V regulator

Hi,
schematic of the cascaded 431 +CCS shunt regulator as tested.

edit.
max output is about 40mA with the CCS set to 46mA.

All the caps can be safely increased in value.
Try upto 220uF

 

[hailteflon]

Andrew-thanks again for confirming the dual 431 setup.


Perhaps a small ceramic in parallel with the 1uf would eliminate a bit of noise.

There is a minimum voltage (or is it current) called "punch through" if I recall correctly. Below this threshold the wiper does not properly conduct to the resistive element.

Manufacturers rarely publish such info in potentiometer specs.

If this pot is passing a very low current then a small ceramic may eliminate wiper noise. Mark

 

[Archwn]

Re: shunt regulators

First, go to datasheetarchive.com and download the motorola datasheet. Type in TL431 and scroll down near the bottom of the page to get the motorola sheet.

From the way you spoke it sounds like you are confusing a current reg with a voltage reg. The 431 keeps a constant current if used as a CCS (constant current source). It is actually a programmable zener.

Get the motorola datsheet and then post any problems that you have. We will then have common data.

Mark

Hi Mark, Thanks for your advise and help. I really appreciate indeed.

And yes, I'm so confuse with the TL431 implimentation really. I did download TL431 from many manufacturers and they didn't answer all my noob question I have.(and I will tell you that I even didn't know how to properly use normal zener as a fixed voltage source - I mean the concept)

Could you also please advise how can I easily get hold those JAN Zener? 😀

The big advantage of a shunt regulator is that the series resistor isolates the load from the main supply. With a larger difference between supply and shunt that resistor will have a larger value so isolation factor increases.

In some of our circuits (DAC and Phono preamp) we use an LM317/337 bringing the supply voltage down from 45 Volt to 30 Volt. Then a shunt regulator follows bringing the Voltage down to 18 - 24 Volt.

The TL431 is a nice thing but it has the disadvantage that at frequencies over 100 kHz the regulation is less effective. So it's a good thing always to have a capacitor there with good HF specifications. In our DAC we use an Os-Con 10 µF and a polyester 10 NF in parallel for the digital circuits.

When designing audio circuits, discrete or with op amps, a better option is to use a discrete shunt. Two transistors and a zener diode do the trick far better then the TL431.

In tube circuits a tube can be used for the shunt regulation.

Hi johnrtd, thanks for you advice and info.

If the series R increase, will it affect the transient current response? I can see that we might get away with that with the cap at down stream like you say 10u+0.01u. you reckon it's enough for 100mA drain?

I will also have a look into two transistors and a zener configuration. would be good for the comparision.

Your last paragraph is quite interesting and I defenately will have a look into that. Thanks

Hi,
Now calculate the power dissipation in the 431 when (a) the load is drawing maximum current, (b) when the load is drawing nominal current, (c) when the load is drawing minimum current.

Is the TL431 overloaded during any of these conditions?

I have 32mA +-8mA and I opted for the PMA version (which is mistakenly shown in the datasheet with an NPN pass transistor instead of PNP).
Vout=18V and Iout=30mA CCS=52mA, the BD140 gets quite warm. A To92 will overheat.

Hi Andrew, thanks again.

So now I can see it's ok as long as the case is not overheated. 🙂

A shunt regulator is NOT in series with the load. Therefore the 431 cannot have a Vdrop like a conventional series regulator. That explains why you can find no reference to Vdrop.
BUT, the R or CCS feeding current to the load + shunt MUST drop voltage. The lm317 I am using in CCS mode drops a minimum of between 2.2V and 3.1V over a fair range of CCS currents. The 317 runs from warm to hot depending on Vin.
I intend fitting a 20C/W to the lm317 and 14C/W to the bd140.

What I try to understand is that if I have 5.2v from the usb v+ and I need 5 or 5.1v with TL431 without LM317 will the small value of series R work? ??? or it is useless in a designer point of view. (then the real question would be do I have to regulate the usb +v with TL431 as it seems to be better? ??? )

no.
The dropping medium is required to drop a voltage from Vin to Vout.
The load pulls a nominal current.
The shunt pulls an extra current to ensure that Vdrop=Vin-Vout.
The CCS is an alternative to the R. It matters not that you are choosing alternative currents. The R is a cheap and easy way if the Vin does not vary much.
If Vin from typical mains transformers varies as much as usual with both loading and with mains voltage then the CCS works better provided Vdrop does not exceed the maximum that the 317 (or it's replacement) can withstand.

Thanks for your timely input. what you say no with a profund explanation is what I read as a yes. 😀 It took me a while to understand but it is much clearer now. 😎

 

[Eva]

Andrew according to some of the facts that you have explained I think that your TL431 are sometimes oscillating depending on load current, load capacitance, parasitistic capacitances of the test fixture, beta of the buffer transistor, etc... TL431 has already a very high open loop gain on its own and adding a buffer transistor without proper frequency compensation is likely to cause trouble.

 

[AndrewT]

Hi Eva,
thanks for looking in and giving comment.

Can you direct me to which numbers/description indicates the possibility of oscillation? Which version, PMA's original, my first single 431, my second single + Zener or the final twin 431?

 

[theChris]

probably would want to lower source inductance by adding some local decoupling capacitance on the input.

the TL431 datasheet lists some stability regions that show stability as a function of output capacitance and shunt current.

 

[hailteflon]

Archwn:

Could you also please advise how can I easily get hold those JAN Zener?

Actually, the 431 is more practical. JAN (military type) are usually several dollars and are probably special order items if the distributor will even order them.

Buying them surplus is probably the only option and the selection will be sparse.

The 431 has little variation with load. A zener has a large variation.

In case you haven't gotten the basic principle figured out here:

The R between the CCS and the load determines the output voltage. The 431 (programmable reference--a zener diode and a reference diode are the same thing) varies the current through R such as to keep a constant voltage at the load. V=IR is the basic formula. That is, the voltage is determined by R times the current going through it. Mark

 

[Eva]

Hi Eva,
thanks for looking in and giving comment.

Can you direct me to which numbers/description indicates the possibility of oscillation? Which version, PMA's original, my first single 431, my second single + Zener or the final twin 431?

The traces of 1.2Mhz RF and the abrupt changes in noise floor depending on load current should have been enough to make you think of HF instability (applies to all versions).

 

[AndrewT]

Hi Eva,

As I turn Vin down further Vout drops in sync, since regulation has been lost and the 20MHz stays constant @ 1.2mVpp.

this condition only arises when the circuit is NOT operating as a regulator.
Vdrop across the CCS exceeds the difference between Vin-Vout.
The result is that the shunt transistor is near enough switched off and there is insufficient current left to operate the two flow paths into the TL431.

That does not sound like a sign of an unstable regulator.

But if you can offer any advice on how to stop even this then please make some suggestions. I ask because during the turn on phase and during the turn off phase these oscillations may be present if the cause was not an interaction between the lab supply and the CCS. As one poster has already noted, the normal smoothing/and/or stability cap before the regulator has been omitted in the test circuit.

Yes, there is a change in the noise floor in some of the arrangements. But the noise does not have any predominant frequency that I could identify nor lock on to.

 

[triode_al]

TL431 and a series pass regulator

I am rebuilding a power supply, and raising it from 18V to 23 V, for classical J50/K135 mosfet output.

The schema that is provided by Philips and others for a series pass regulator, based on the TL431 have a capacitor Crk between the cathode (top) and the reference. It is drawn as an electrolytic, so this must be several microfarads.

Question:
- what is the role of Crk (soft start? )
- What size is recommended.
- why is there no capacitor from output to reference, that is what my intuition tells me.
- some say "Consider a MOSFET power amp. - you can use a regulator for everything but the output stage.". Isn't this exagerated?

albert

 

[hailteflon]

As I recall this is done to smooth the regulation. It isn't necessary unless there is a lot of fluctuation of the input voltage, or possibly a stability problem at the load.

The value is up to you on an experimental basis. This is often done across a zener in a zener regulator.

The problem is that is slows the response of the regulator.

It depends on the application. I would try the circuit without it and add it if the output is not acceptable.......................as I recall.

The T.I. 431 datasheet shows it to be .01uf, in a similar series regulator circuit.

I recall that it usually made things worse.

Look around on the web for the Motorola Linear/Switchmode Voltage Regulator Handbook. I recall (or maybe I was just looking for it) seeing it on the ON Semiconductor (formerly Motorola) website about a year ago. Mark

 

[triode_al]

As I recall this is done to smooth the regulation. It isn't necessary unless there is a lot of fluctuation of the input voltage, or possibly a stability problem at the load.

The value is up to you on an experimental basis. This is often done across a zener in a zener regulator.

The problem is that is slows the response of the regulator.

It depends on the application. I would try the circuit without it and add it if the output is not acceptable.......................as I recall.

The T.I. 431 datasheet shows it to be .01uf, in a similar series regulator circuit.

I recall that it usually made things worse.

Look around on the web for the Motorola Linear/Switchmode Voltage Regulator Handbook. I recall (or maybe I was just looking for it) seeing it on the ON Semiconductor (formerly Motorola) website about a year ago. Mark

Thanks, [making things worse] this is indeed the type of warning needed.
In the past I made a powersupply for a Hiraga 'Le Monstre' - 8 watt, some 120 000 muF, using a IRF250 Mosfet - and wow - it blew the fuse on my powerline. It started to kick. This was the PS model with a zener and a differential transistor (source to base) so I reverted back to another transformer and all passive R/C regulation. That 8-watter is doing fine with my ESL57s but at times a tad too small.
I do have a 300B PP too, (with 4700 mF at 400V very well behaved) but just need to keep working a bit...

 

[hailteflon]

120,000 uf is going to have a HUGE inrush current that will blow fuses and/or windings. An inrush current limiter helps.

datasheetarchive.com has the Motorola Voltage Regulator Handbook

Type HB206 into the search box. ON Semiconductor doesn't list it anymore.
Good luck with your projects, Mark