UC3842/3 Current Sense Pin

[hesam_m]

I want to design a constant current supply based on the UC3842/3.

I see the chip provides the current sense input, can I use this pin as the current feedback to limit the current?
The mentioned threshhold voltage is 1V. I'm not sure exceeding 1V will turn-off the PWM (short circuit use case) or reduce PWM duty cycle as voltage feedback error amplifier does (2.5V reference).

 

[gophert]

It looks like the current sense pin connects to a comparitor. If the current is too high, the comparitor toggles a latch (flip flop) to disable the PWM output. The latch can be reset with each PWM cycle but is immediately retriggered if there is an over current situation.

note the comparitor reference input is 1V so the sense resistor must be at 1V to activate the sense pin to interrupt the PWM. Example, a 2 ohm resistor will allow up to 0.5A but disable over 0.5 A.

 

[rjenkinsgb]

It seems to use a combination of voltage and current feedback.

The "current sense" comparator is what controls the PWM pulse duty cycle, comparing the voltage from the sense resistor to the output of the voltage feedback amp.

eg. If the voltage is correct, the threshold is around 1V, but that will increase if the output is undervoltage or decrease if the output is overvoltage.
[It relies on the current through an inductive load ramping up over time, to give a variable duty cycle].

If I'm following the operation correctly, you could configure the voltage feedback amp as a unity gain inverter with a couple of high value resistors for in and feedback, which should allow you to use a pot fed from Vref & 0V to adjust the amp output and so the current sense threshold, giving direct current control.

 

[hesam_m]

I think the block diagram won't handle a good picture. I personally think the current sense pin can NOT be used to build CC and it is just a threshold to detect short circuit and spikes and it turns-off the PWM if it detects such thing. So I should use the voltage feedback pin, because it is an error amplifier not just a comparator

 

[gophert]

I think the block diagram won't handle a good picture. I personally think the current sense pin can NOT be used to build CC and it is just a threshold to detect short circuit and spikes and it turns-off the PWM if it detects such thing. So I should use the voltage feedback pin, because it is an error amplifier not just a comparator

Yes and no.
The current sense pin only senses over current and limits current to that point. However, the load on a constant current supply cannot be so high in resistance that it Reduces current flow below the constant current target.

for example, if you have a 10v max and you put a 100 ohm load, there is no way to maintain a 1 amp constant current. But, any load below 10 ohms will keep the current sense pin "activated" and limit the current to 1 amp.

 

[hesam_m]

Yes and no.
The current sense pin only senses over current and limits current to that point. However, the load on a constant current supply cannot be so high in resistance that it Reduces current flow below the constant current target.

for example, if you have a 10v max and you put a 100 ohm load, there is no way to maintain a 1 amp constant current. But, any load below 10 ohms will keep the current sense pin "activated" and limit the current to 1 amp.

Yes, but the idea behind the CC (constant current) is to compensate from the output voltage and keep the current constant all the time. so if 1A is the target and the load tries to consume more, the current remains constant and the voltage reduces. that's why I think I should use the FB pin, not the current sense pin. Current sense should be connected to ground in this case. Do you agree?

 

[gophert]

Yes, but the idea behind the CC (constant current) is to compensate from the output voltage and keep the current constant all the time. so if 1A is the target and the load tries to consume more, the current remains constant and the voltage reduces. that's why I think I should use the FB pin, not the current sense pin. Current sense should be connected to ground in this case. Do you agree?

it will work as you described for constant current - I was just saying that your constant current is limited by the voltage head space, you cannot keep a 0.5 amp current constant through a 1000 ohm load because you would need 500v to push 0.5A, ohms law still applies to all constant current supplies. All a constant current supply can do is limit the current (as this device does) - lowering voltage by skipping PWM pulses to maintain constant current.

 

[hesam_m]

This is still confusing, does this pin provides only a threshhold for over current and turns off the PWM, or behaves the same as the Feedback (error amplifier) pin and reduces the duty cycle till the current draw falls below the reference threshold

 

[gophert]

This is still confusing, does this pin provides only a threshhold for over current and turns off the PWM, or behaves the same as the Feedback (error amplifier) pin and reduces the duty cycle till the current draw falls below the reference threshold

The current sense pin shuts down the pulse if over current is sensed which effectively lowers the voltage of the smps and keeps the current in compliance.

 

[rjenkinsgb]

The device does not use conventional PWM with an internal ramp and comparator; it uses the load current ramp, caused by the load inductor, as the PWM ramp.

The output is switched on at the start of each cycle, then turned off when the current sense voltage exceeds the output of the voltage feedback circuit.
That's why it is called a current mode controller, presumably.

 

[hesam_m]

I want to get a feedback from a low side shunt resistor that is amplifierd by an opamp, it won't be connected to the source of the mosfet, still applicable to use the current sense pin?

 

[rjenkinsgb]

It should be OK as long as the opamp can follow the current waveform accurately.

I think you may just be able to link the voltage feedback and compensation pins, so the voltage error amp tracks the 2.5V ref.
That should make the current sense threshold somewhere around 0.4 to 0.5V, looking at the device diagram.

 

[hesam_m]

It should be OK as long as the opamp can follow the current waveform accurately.

I think you may just be able to link the voltage feedback and compensation pins, so the voltage error amp tracks the 2.5V ref.
That should make the current sense threshold somewhere around 0.4 to 0.5V, looking at the device diagram.

The opamp output would be a simple flat voltage, as it senses the "output shunt" that naturally passed through some heavy filtering.

I want to sense the current by the output shunt resistor and opamp, however, I have a doubt whether this pin only used to turn-off the chip (PWM) by over-current, or I could use it to keep the current constant and compensate it by reducing the voltage (constant current). if that's not applicable, then I'm gonna tie it to the ground and use the feedback pin for sensing both voltage and current like many other chips.

Gophert says it will work.

 

[Nigel Goodwin]

You might find this useful?

Adding a Current Limit Feature to a Buck/Boost Converter

Adding a Current Limit Feature to a Buck/Boost Converter: In this project we will have a closer look at a common buck/boost converter and create a small, additional circuit that adds a current limit feature to it. With it, the buck/boost converter can be used just like a variable lab bench power...

www.instructables.com

While it says "current limit", there's no real difference if you set the output voltage to a sufficiently high one - which applies to any constant current source anyway.

It's essentially the same technique as used on the Chinese CV/CC charger boards, using an opamp to monitor the current and over-ride the voltage feedback.

There's a couple of their schematics on this page, the boards with three pots:

Lm2596 Dc Dc Step Down Schematics

Pcb design download and schemas

pcbdesignsdl.blogspot.com

 

[rjenkinsgb]

It will not work at all with the current sense pin connected to ground.
Anything other than an instantaneous current measurement point will mess up the operation; as I said above, the inductor current forms part of the overall PWM signal loop.

If you want to use a filtered current sense, use a voltage mode controller than generates an internal PWM signal, rather than a current mode one.

Consider a SG3525 / UC3525 instead; that produced the PWM signal totally internally, so you have absolute control over it.

https://www.ti.com/lit/ds/symlink/uc3525a.pdf?ts=1603207589658&ref_url=https%253A%252F%252Fwww.ti.com%252Fproduct%252FUC3525A

https://docs.rs-online.com/962b/0900766b80dad829.pdf

 

[gophert]

Gophert says it will work.

I didn't say adding an external op amp would work, I only described how the chip you posted does work. I don't understand the need for external devices when all the functionality seems to be accessible on various pins on this chip.

 

[ronsimpson]

Here is a buck supply that is almost CC. Here I have a capacitor across the LED but when cost is important I do not use the cap. L must be large. Note the Schottky diode. Size Rs so the current turns off at 1A. Size the inductor so ripple is small and does not go to zero like in the picture. The average LED current is the dotted line. Disable the error amp.

This works but there is some errors in measuring the current. (you are regulating the peak current which is related to the average current)There are small 5 pin IC that work like this. No error amp. This is not what the IC was designed to do.
I see I missed a line. VCC need to go to + supply.

 

[ronsimpson]

Here is a common way to do CC in a led power supply. The 10 ohm resistor measures the average current in the LEDs. Feed back is from the current in the LEDs. In this example the supply is a "boost" type, where the input voltage is less than the LED voltage. With CC power supplies the "FB" voltage is usually around 0.1 to 0.2 volts depending on what IC. It would never be 2.5V or 5V like in voltage feed back power.

 

[ChrisP58]

The uc384x series of controllers are classic current mode devices for switch mode power supplies.

When running properly, the signal at the current sense must hit it's limit once per each switching cycle. But that limit is adjusted by the voltage feedback error amplifier.

If the voltage at the FB pin is lower than the reference, then the output of the error amp rises, allowing more current per pulse. If the FB voltage is higher than the reference, then the current trip level is reduces, and less current is allowed per pulse.

So the output voltage is actually controlled by adjusting the current.

It's actually very similar to how we control the speed of a car. We don't directly control the engine RPMs. We adjust the amount of fuel into the engine to control the speed.