Uc2834 connecting COMP pin to output of op-amp?

Magnethicc

Member level 4
Joined
Feb 7, 2022
Messages
70
Helped
1
Reputation
2
Reaction score
6
Trophy points
8
Visit site
Activity points
914
Hello everyone,
I have a question if anyone have any application notes or schematic which shows it is possible to connect output of op-amp to COMP Pin of PWM Controllers?

I do this because I have a few PWM Controllers which used as slaves so I want to feed them the feedback of the master PWM.

So I buffer the feedback and feed the comp pin of PWM Controllers the output of the buffer op-amp (and the feedback pin of the slave controllers I short to ground)


Thank you.
 

Solution
Since I short FB Pin to ground the COMP pin wants to be high and I force it lower with the buffer op-amp --> I pull current from them.
so by your reply I understand it is a common practice.
Do you have any application notes stating this fact?

Thank you Easy Peasy

as these pins are generally current limited for output ( designed this way ) pulling them down for less power is an accepted design practice,

pushing current into them is an unknown - if the currents are kept low enough that heating does not occur - then the issues become more about other effects inside the chip, an 100 ohm ( or less ) R, in each output would allow you to measure this current - if you are forcing the pin high.

look at the internal schematic for the IC.
 

Since I short FB Pin to ground the COMP pin wants to be high and I force it lower with the buffer op-amp --> I pull current from them.
so by your reply I understand it is a common practice.
Do you have any application notes stating this fact?

Thank you Easy Peasy

 

Solution
You can of course connect opamp to COMP pin when doing "external control" as such....just use a diode....anode to COMP pin.

You can demonstrate this on the LTspice sim using the LT1243 part, which is pin4pin with the uc3843 etc.

In fact, you can do this with almost all pwm controllers......you will find that the "compensation pin range of control" is usually "0.7V to 2V", or "1.4V to 5V" in the case of UC2843.......so the COMP pin range of control is literally lifted up so that the said diode can be put there and not mess it up.
 

I'm assuming the OP is referring to UC2843, not UC2834 (which doesn't have a COMP pin)

Generally I only overdrive if the error amplifier is an OTA with a high output impedance. That's not the case for the UC2843. The datasheet does say the COMP pin has limited source and sink current, and mentions pulling COMP low to force PWM to shut off.


But it doesn't mention overriding COMP for any other purpose. It might work, but should be tested carefully.
 

It does show an offline application which connects an opto-coupler from COMP to GND in-order to force the COMP voltage by sinking the sourcing current limit from the COMP pin.
Is it not the same as overriding the COMP voltage using an op-amp?
--- Updated ---

Why diode is needed?is it to disconnect the buffer Incase output of buffer is higher than the maximum output voltage of the internal UC2842 error amplifier?
 
Last edited:

Is it not the same as overriding the COMP voltage using an op-amp?
...Yes diode is deffo needed...else you could damage the internal opamp.....its not really much of a hassle/restriction anyway.......so just using the diode is fine
 

UC2843 comp pin sink current is specified with 6 mA typically, there's IMHO no guarantee that it keeps absolute maximum of 10 mA if low comp output is externally driven high.

The obvious solution is to connect Vfb to GND to force comp high.
 

Hi FvM,
Yes, indeed the connection is FB to GND and buffer output to COMP pin.
I suspect that the diode cupoftea recommends disconnect the buffer to block any sinking current of the internal error amplifier
UC2843 comp pin sink current is specified with 6 mA typically, there's IMHO no guarantee that it keeps absolute maximum of 10 mA if low comp output is externally driven high.

The obvious solution is to connect Vfb to GND to force comp high.
 

Check the COMP pin I-V and find two things - the weaker side of the pin current (leading to best setup for FB) and how much, for selecting an op amp capable of doing what must be done.

My POL DC-DC designs were made with all key tap points looped via pads to make this kind of thing easy / no worries.

If you are especially lucky the slave FB path might be jammable to a place that delivers no current and then COMP:COMP could work. On mine the fully pinned out error amp os slaves was used as a local unity gain buffer driving ISET, easy as pie.
 

Yes, just like any opamp, you wouldnt connect another opamp output directly to it, so the diode is needed.
You also woudlnt connect say a voltage source directly to an opamp output...its the same kind of thing.
 

Thanks and yes, then you have to make sure your opamp isnt supplied by higher than the Vcc (5V) of the UCC28C43, because then it would be overvoltaging it, unless you had a blocking diode. This is why its good routine just to put the diode in.....another point is what if your external opamp 's vcc comes up before the vcc of the ucc28c43...then again you are going to be overvoltageing the internal opamp of the ucc28c43....unless you have the diode
 

@cupoftea - you can also put a simple clamp on the driving opamp output - there are many ways of doing this - including a zener to one of the inputs,

however a simple schottky on each IC EA o/p is not such a bad idea, assuming the volt drop across the schottky is the same for all controlled units - i.e. each EA in each chip must send out the same current ( and we know from typ data sheets this is not quite the case ) . . .

So assuming the OA can pull down ALL the connected IC OA outputs - but wait this is unlikely ( to many mA ) - so even better to drive an NPN in a reduced gain ckt to pull all the connected IC EA o/p's low for less power . . . no diode now needed.
 

Also, you could do a simple buffer opamp but with the diode as before (anode to comp) , and the inv pin of external opamp connected to COMP pin......

There is a microchip controller which better facilitates this than uc28c43...its the MCP1631(?) or something...its the microchip MCP16XX...i cant remember last two...its like the ucc28c43 but doesnt have those 2 series diodes in it which mess things up a bit.
 

Similar threads

Cookies are required to use this site. You must accept them to continue using the site. Learn more…