Push-Pull / Bridge-H / Feedback Isolation / Control

[unavezmasysale]

Isolated Feedback for Push Pull SMPS

Dear,

I have already design a Push-Pull SMPS but I do not know how to design the isolated voltage/current feedback. Can you help me with some kind of amplifier (Rshunt and a instrumental amplifier, for example)?
I have to feedback, the current one in the primary side and the voltage in the output of the SMPS.

Thanks.

 

[crutschow]

Re: Isolated Feedback for Push Pull SMPS

One way to provide galvanic isolation is to use an opto coupler in a linear mode. an example for the voltage feedback. A normal instrumentation amplifier won't provide such isolation.

If the current feedback is from the primary side, I see no reason for it to be isolated. :?:

 

[andre_luis]

Re: Isolated Feedback for Push Pull SMPS

He´s probably concerning to take control of output current limit. I presume it´s not too usual, once the value is damped for output capacitor. Furthermore, an instantaneous current limitation is already provided by shunt resistor at primary side of transformer.

 

[KlausST]

Re: Isolated Feedback for Push Pull SMPS

Hi,

there are several ways even with optocouplers.

The simple and good working is like crutschow showed. Look for optocouplers designed for linear/analog operation.

The disadvantage is that optocouplers drift with time and temperature. This drift you may see 1:1 in tranfer rate.

To avoid this you can generate on the LED side a sawtooth or triangle waveform and use a comparator to compare it with your voltage (Vx). Feed the comparator´s output to the optocouplers LED.
Now the optocoupler works as a digital transmitter without the disadvantage of it´s drift causing erros. On the ouput side of the optocoupler is a PWM with duty cycle proportional to voltage (Vx).

Also there are optocouplers with an extra monitoring receiver to the LED side. With the monitoring current you can cancel a lot of the errors caused by drift.

The most precise solution i can think of is to use an ADC feed the digital information with an optocoupler to a DAC.

Klaus

 

[FvM]

Re: Isolated Feedback for Push Pull SMPS

You'll notice that most industry standard SMPS have the reference and error amplifier for voltage or current feedback on the secondary side and are transmitting the error signal with an opto coupler. This way, the accuracy and linearity requirements for the analog isolator are much lower and can be usually achieved with standard optocouplers.

Recent linear isolators with high accuracy are mostly using A/D and D/A conversion with digital isolation, as already mentioned.

You should decide if your application actually requires analog isolation, and if so, what's the required accuracy.

 

[crutschow]

Re: Isolated Feedback for Push Pull SMPS

As FvM stated the opto coupler is normally in the feedback loop so any effects of gain non-linearity or drift are significantly reduced by the loop gain and would have no significant effect on the power supply performance for a properly designed loop.

 

[unavezmasysale]

Hi,
I created a similar thread but I think that i was not so specific as I need so I decided to create a new one that surely will help to the community.

The system that I am designing works likes an inverter. A push-pull smps step-up a voltage from 12 to 400Vdc, then a H-bridge (working to a constant cycle) generate a semi-sinusoidal wave (Vpk = 400V - Imax = 1A - frec = 100kHZ - Max duty cicle = 45%).
Well, I need to control the duty cicle of the smps to achieve specifics current levels ( 10mA, 100mA, ..1A) at the output of the h-bridge. I made a differencial circuit to reduce the signals levels and now I need to isolated this voltage to connects it to an ADC (DSP uC).
The idea is to use an optocoupler working in the lineal zone but I do not know how to implement and I do not really know if it is possible or if I have to think another solution.
I attached the image of the current and the circuit:





Any idea?

Thanks.

 

[KlausST]

Hi,

for good isolated feedback you may use optocouplers with integrated monitoring diode. Like HCNR200 and HCNR201.

One additional hint: Avoid saturation, don´t ever regualte to the power rails or to 0mA. Neither with OPAMPs nor with the optocoupler input or output.

Klaus

 

[unavezmasysale]

Hi KlausST,

Like you said use a high linearity analog optocouplers may be the best solution, but in my country is not available and another similars components are very expensive and difficult to find like the IL300 or ISO130. So I have to deal with other circuits, more easy to get in the local market.
Now I am trying to do this with a TL431 and a CNY17-2 before de differential amplifier that I attached in the previous post.
Do you ever work with them?
This is the circuit that I am simulating now:



Thanks.

 

[chuckey]

It could be an idea to use a pair of opto couplers, one monitors the output current, the other takes the "set" current, if the transistors are wired as a long tailed pair their collector voltages should represent the two currents so the difference could be used to control the output current. Of course the balance depends on the match of the opto isolators, but at least you have tried!
Frank

 

[unavezmasysale]

Hi Chuckey,

Do you have the schematics of the circuit that you propose?

Thanks.

 

[unavezmasysale]

Hi,

What do you thing about this design?





Thanks.

 

[KlausST]

Hi,
i don´t see any benifit in the first differential opamp stage ( besides of compensating for gnd bounce)
A simple 2 resistor 1 capacitor solution may work also.

If the result is good for you then it´s OK.


Filters:
You have two RC filters. one with 1600Hz, the other with 15600Hz.
If you design both to about 2kHz it may decrease ripple, while phase shift is about the same.

Klaus

 

[unavezmasysale]

KlausST,
Do you mean to use a resistor divider with a capacitor to replace the differential AO stage?

- - - Updated - - -

I used a resistor divider instead of the differential AO and this was the response:


I had changed the filters, thanks.

 

[KlausST]

Hi,

Do you mean to use a resistor divider with a capacitor to replace the differential AO stage?

Yes

The values are 907K, 33k, but use 2n2.
(as second stage you can use 100k and 750p)

Klaus

 

[unavezmasysale]

KlausST,

Well I can not solve my problem with this circuit, once I changed the Vin everything go wrong.

My problem is that after the voltage follower I need to have 4V when the inputs is 10mV and 10V when the input is 1000mA.

I can not really understand how to do it...

 

[FvM]

I see the problem that the operation principle of the TL431 circuit is contradicts the specification in post 1 (linear signal transmission). TL431 is intended for feedback controllers with a fixed setpoint, given by the internal reference voltagee and the Ra/Rb ratio.

 

[KlausST]

Hi,

Well I can not solve my problem with this circuit, once I changed the Vin everything go wrong.

What Vin do you change?

Klaus

 

[unavezmasysale]

Hi LausST and FvM,

Finally I have changed the way in wich the circuit works and therefore the solution has changed too.

I need to keep 5V constant at R14, the incoming signal is a PWM pulse with a 400Vpeak. R13 can change all the time so R14 will automatically change too, but this is not important now.

What do you thing about this circuit? do you know the real funcion of C1, C2 and R3 ?

Thanks again.

 

[FvM]

The circuit doesn't keep the maximum ratings of 4N27, at least the load resistor must be respectively corrected. R5 should be e.g. > 400 ohm.

C1 and C2 are modfying the frequency characteristic of th ecircuit, e.g. to tune a feedback loop response. C3 is presently only working as supply bypass capacitor. The small value of 100 pF is mostly useless.