How to calculate TL431 feedback bandwidth ?(using it in control loop for SMPS)

the TL431 often use in control loop for SMPS
how to calculate it's feedback bandwidth?

tl431 smps

you need check if you use fast-lane / slow-lane etc.

but welcome my friend , because this is an area of industrial secrecy and all have difficulty in finding out.

Basso book has transfer function for one configuration with tl431, but only one......the one for flyback output where output is CLC.

you should calculate its pole (opto) using bench test

there is probably a book somewhere which details this properly......but as i said to you.....isolated feedback with tl431 and opto is an area of industrial secrecy and you will have the very greatest of difficulty in finding out......as i have already found.................

you understand , if you find out, you are possibly going to make products to compete in the industry....so nobody is wanting you or me or anyone else to find out about it.

so anyway, you need to ask about the opto bandwidth because that is the limiting factor...the tl431 has relatively high bandwidth.

maybe you use powerintegrations software they find it automatically for you............but then you are in the dark....and you must get gain/phase analyser to see what is your bandwidth.

but as i said my friend,......we are deep in the world of industrial secrecy here
.

tl431 schematic

Generally you look at the TL431 just like an opamp and calculate the pole and zero that way. Can you please post your schematic? It's easier for me to explain then.

tl431

please see this picture

smps tl431

my friend you have r(upper) and r(LED) to the SAME node.....i dont think you can do that.....cuzz you are not getting any gain out of the current variation in R(led).

But i tewll you my friend..........nobody will help you.........i have tried many times to find out this......it is a secret of industry..


if you can close the loop......you can compete in industry in SMPS......and that is not wanted......they want to keep you ( and me and others) out.

If you can find a transfer function for that schematic....well.....you wont be able to.

i have already ytried so many times on so many forums.

tl431 circuit

i dont think you can do that.....cuzz you are not getting any gain out of the current variation in R(led

Click to expand...

Sorry, I don't get the point. Of course you can calculate a gain, Vout can be considered as a voltage source independant of the LED current. dILED/dVout is the circuit gain, it's a small signal property that has to be calculated for a specified LED current, the same with the frequency characteristic.

I also don't see general difficulties to determine the overall open loop gain, some parameters as the photo transistor's pole are possibly not exactly specified or show individual variations. But a rough estimation is always possible.

With simple circuits like the present one, many designers skip initial calculations and start with empirical dimensioning of a prototype circuit, I think, using a known working circuit as a starting point.

schematic circuits tl 431

This is a very simple single-pole error amp. The pole is simply at 1/(2Π*Rupper*C). Note that nothing else goes into the equation.
The only thing that is missing here is the secondary of teh optocoupler. Its load resistor is important in caluclating the overall gain of the circuit.

tl431 usage

I apprecaite this VVV, but i believe if our friend wei9321 is going to be able to close his loop, he is going to need the transfer function for this little lot.....including as you say that load resistor on the primary side.

and that , as i say, wei9321, will not find, because it cannot be found.

Even the great text by Basso only includes the transfer function for the version of this circuit that has a CLC filter on the output with the L separating r(upper) and r(led)..........and to be truthful......for a type 2 amplifier it is a hulking great huge big transfer function..........and so too it will be pretty big transfer function in wei9321 's case.

Feedback is the part of SMPS that the industry wants no-one to know.....apart from their own staff.

And put it this way, and engineer who can derive transfer functions for modulators and error amps etc is a rarity.....and a fair genius.

Most will dust off the gain/phase analyzer and go for trial and error......soldering iron burning well!!

empirical dimensioning control

This is a very simple single-pole error amp. The pole is simply at 1/(2Π*Rupper*C). Note that nothing else goes into the equation.

Click to expand...

I think, it's different. Neglecting real TL431 parameters, the V/I transfer characteristic is an ideal integrator. Limited TL431 voltage gain causes a gain bend at low frequencies, finite output resistance results in a feedforward term, a zero that cancels the integrator behaviour at high frequencies, r(LED) matters here.

test tl431

In fact what you say is right FvM, there is much complexity to TL431/OPTO feedback circuits.....

there are two feedback paths in wei9321's schematic......one through Rupper and one through Rled.


And the opto pole must be accounted for and the R(pullup) on the primary side.

Powerintegrations dont trust people to be able to handle it and just calculate it for us.....they go for a very low crossover frequency so that they get under the opto pole without having to account for it.........

tl431 as type 2 amplifier

I see now, that I didn't notice one point: The output current is generated by the difference of Vout and Vk. So the AC transfer function is a PI characteristic, respectively an integrator and a zero. And the corner frequency depends on Rupper and C only (not considering TL431 limited volateg gain), as VVV said. Sorry for causing confusion.

Re: TL431 use in SMPS

eem2am,
I am being puzzled for your insistence on this topology as feedback loop being so secret. It is very likely I am being a complete ignorant on this and be wrong , but then the reason of my reply here in order to understand what I can not so far. Much previous to the start of this thread is this article
www.powersystemsdesign.com/design_tips_june07.pdf
www.powersystemsdesign.com/design_tips_june07.pdf

This article shows exactly the topology discussed here.

Being the SMPS so common and used almost anywhere and made for so many companies why this should be a secret ? . Again my ignorance , but I see no difference with for example saying that a bridge rectifier and the filter is a secret. Also what is so special around the TL431? Any OPAMP could be used to drive an LED and perhaps in a much easier way (in the way to analyse it). Yes , I know that this does not require a separated power but again my ignorance here.

Finally, you can always buy one of those products and reverse engineer it as it is perfectly legal.

Thanks a bunch on any clarification so I can learn.

Jose

TL431 use in SMPS

I agree with your opinion, that the behaviour is basically simple and can be fully understood. A partly undocumented point is the exact frequency characteristic of the optocoupler. Some designs place a dominant pole at the collector node of photo transistor for this reason.