TL431/Opto-coupler problem: Bias current of TL431 flowing in opto-diode

[eem2am]

Hello,

I do isolated , offline flyback.

TL431/Opto-coupler feedback.

The problem is, the 1K resistor that's across the opto-diode to provide the TL431 bias current, places a voltage across the opto-diode of 1 Volt....

...this in turn makes it conduct current, which is a big big problem on maximum load, because if 180uA is drawn through the opto-diode then this has the effect of stopping the primary-side PWM control IC from switching.

The control current to stop the PWM controller is just 90uA (actually, 90uA is the lower tolerance level, but it could obviously still occur)

......so

how do you get round the problem of the TL431 drawing some of its bias current through the opto-diode, unwantedly.?

 

[FvM]

I can't imagine this kind of problem. Can you show a schematic to clarify it?

 

[ServoAmp]

A schematic would help with this problem.
Is the circuit simular to the one in this app note
it uses the TL431 with a TCMT1107 opto coupler.

https://focus.ti.com/lit/an/slyt323/slyt323.pdf

 

[eem2am]

Hi,

this is the feedback representation.

https://i53.tinypic.com/300qt8l.jpg

....It is the TL431 bias current (1mA) which flows in the R(bias) resistor (1K) and thus puts a volt drop of one volt across the opto diode.

...this in turn means that th eopto diode will conduct at least 200uA, adn this can be a problem...........

...because it would mean 90uA going through the opto-transistor, which would stop the pwm ic from switching.

....in times of maximum load, we certainly wouldn't want this 90uA flowing out of the controllers feedback pin.


Here is the schematic showing the feedback pin connection.

https://i53.tinypic.com/2rgbypf.jpg


I am wondering if i will have to implement this kind of feedback instead?
https://i54.tinypic.com/295r18x.jpg




[unfortunatley i cannot give the datasheet of the PWM controller IC, as it is a new one, and we are not allowed to divulge it.
It is from Sanken.

I also cannot give you the exact schematics due to security policy

 

[Orson Cart]

Usually this is cured by putting a lower value resistor across the opto -led, e.g. 470 ohm, giving 0.47 volt across the opto led for 1mA of bias in the 431.

alternatively use a different type of 431 (there are some more modern ones that only need 200uA or so)

Regards, Orson Cart.

 

[FvM]

Apparently, the circuit has been designed for typical (0.4 mA) rather than maximum (1 mA) bias current. Below 800 ohm would be required to operate with worst case parameters.