Flyback converter with poorly controlled peak current.

[treez]

Hello,

We are doing a FAN6300 based offline flyback converter which operates open loop and repeatedly charges a capacitor (to 300V). The flyback is simply disabled when the capacitor is charged, then re-enabled so as to re-charge the capacitor after a discharge. (The capacitor is 600uF)

We are setting the flyback’s peak current by simply connecting a 10k resistor to the FAN6300’s ‘FB’ pin. The problem with this method is that the internal pull-up resistor is poorly toleranced (3k to 7k). Also, the internal rail to which the internal pull-up resistor is connected is not actually defined in the datasheet (ie, neither its voltage , nor the tolerance on this voltage, is described in the datasheet).

Can you help us to find out what value of internal pull-up resistor we are likely to end up with in any particular batch of FAN6300 chips? (ie statistical probability of receiving any particular value of pullup resistance within that tolerance range)? Also, do you know what is the value of the internal rail to which the internal pull-up resistor connects?

Not only does the poor tolerance of the internal pull up resistor mean that the flyback’s peak current is then not well defined, but it also means that the ‘FB’ pin could end up reaching the 3.9V overload level which would result in the flyback wrongly shutting down.
Another point is that the two series diodes that connect to the FB pin will also have their own Vf tolerance, as well as variation with temperature. Again, the datasheet does not give Vf values for these diodes.


FAN6300 datasheet:
https://www.fairchildsemi.com/datasheets/FA/FAN6300.pdf

- - - Updated - - -

(By the way the external resistor connected to the FB pin goes down to ground)

 

[dick_freebird]

5K +/- 40% is a p!ss-poor tolerance indeed. Some
of that might be temperature? I've never seen a
process (and I've seen many) with worse than
about 20% PCM limits. If it's temperature and the
system has a narrower range, maybe there's some
relief there.

Only Fairchild would know the production stats for
that attribute. Maybe (but I doubt) a FAE could get
you a better range or an idea of what's normal. But
if you have to deliver a lot of these over the long
haul, it could be exposing you to "lot bomb" risk.

Seems like the regulation scheme is quasi hysteretic,
maybe you want an even simpler chip that lets you
pick better components for feedback? If it's just got
to squeal until the bucket's full, and then quit, that
could be real cheap and simple I'd think.

 

[treez]

5K +/- 40% is a p!ss-poor tolerance indeed. Some
of that might be temperature?

yes i agree its very very poor. The thing is, as you know, this FB pin is "supposed" to have a feedback optocoupler connected to it as part of a negative feedback loop, and therefore, for that particular case, the wide tolerance of the pullup resistor is not quite so important.
But as you know, our problem is that we are not connecting it up like that, and we are simply putting in a 10k resistor from the FB pin to ground…and so for our case this poor tolerance is a real problem. I believe you recognise this as you have described exactly how our actual “hysteretic style” feedback loop works.

The other big problem is that the internal rail to which the internal pullup resistor connects to is not detailed….that is, its voltage is not stated. Do you know what it is?
Not knowing the internal voltage rail means we do not know how far away from the 3.9V Overload shutdown level that we will be when we get FAN6300A’s with the 3k internal pullup resistor.
Also, do you know the values (and tolerances) of the 2R/R resistors which are also involved with the FB pin?

 

[FvM]

Poorly controlled peak current? FB pin simply not designed to provide an exact current limiting. I have used a voltage clamp for FB in an application of a current mode controller with similar internal circuit and it did at least "work". Getting specifications for tolerances etc. of this unsupported function is expecting too much.

 

[treez]

Thanks, but surely fairchildsemi would be able to tell us what is the voltage of the internal rail in the FAN6300A which the internal pullup resistor pulls up to? (and its tolerance).

And also, they would surely be able to tell us the value of the 2R/R resistors that are also involved in the FB pin? (or at least the range of values that they may end up being)
............................................................................
The very interesting thing about this, is that this power supply was actually designed for us by a Chinese company. On the schematic, they show the external 10k resistor that’s connected to the FB pin as a capacitor(!!)….but have it 'correctly' designated on the schematic as a resistor (R30), and have it correctly stated as a resistor in the Bill Of Materials. This seems ominous to me, I mean, was the resistor drawn in the schematic as a capacitor deliberately, so as to distract us away from the fact that it is actually a resistor?…with all the problems of having a resistor there “obscured”.
As you know, the FB pin should “normally” have the collector of a feedback opto connected to the FB pin, and a capacitor there as well is normal, as part of the feedback loop compensation. –It’s the resistor to ground from the FB pin that’s not normal, and so we are wondering why it got drawn as a capacitor……because somehow the schematic package has given the capacitor an “R” based designator. (R30).