Find output of optocoupler for variable input

[polashd]

In a smps the mcu (at secondary side) for measuring various things (including voltage and current at primary side as well as secondary output). I’m using EL817 optocoupler to transfer info of primary to secondary. But the output of the optocoupler is not linearly dependent on its input. How do I calculate the input at primary from output (of optocoupler) at secondary? Is there any equation. The primary side is a rectified dc (from ac sin wave 50hz, 0v to 10vdc) so the input varies much. The CTR of the optocoupler is not fixed for all input range. How can I measure the input at primary side in real time?

 

[Diver300]

The transfer ratio of an optocoupler is the ratio of output current ( through the phototransistor) to the input current (through the LED). The transfer ratio varies enormously and it depends on the voltage across the photo transistor and the temperature. There is a big piece-to-piece variation between different optocouplers and the transfer function isn't very linear.

Much like with the gain of a bipolar transistor, it's not possible to predict the voltage and current at all accurately in the linear region.

Optocouplers are almost always used as switches. In an SMPS that uses an optocoupler, there is usually a voltage detector connected to the output voltage (often a TL431 or similar) which powers the optocoupler's LED when the output voltage gets to the desired voltage. That causes the optocoupler to turn on, which stops the input circuit from working and so stops the output voltage from getting larger.

The TL431 contains an op-amp so it has huge gain, so it does not matter if the optocoupler needs a bit more current in some conditions than in others.

A lot of SMPS circuit will run in burst modes, where they run for a bit and then stop running for a bit. When they are running, and the optocoupler is off, the voltage across the LED in the optocoupler could be any voltage up to the forward voltage of the LED, it depends on the design, so the average voltage could vary over a wide range.

Similarly on the input circuit, that could have a large or small voltage across the phototransistor of the optocoupler. Also the voltage may only be present some of the time.

If you want to measure the voltages using an oscilloscope, you need an isolated input probe, something like this:-

https://www.picotech.com/accessorie...-high-voltage/25-mhz-700-v-differential-probe

 

[polashd]

If you want to measure the voltages using an oscilloscope, you need an isolated input probe, something like this:-

https://www.picotech.com/accessorie...-high-voltage/25-mhz-700-v-differential-probe

Thanks for ur help.
My goal is to measure and display (on an LCD display) the input, output (voltage, current, watt) and efficiency of the smps in realtime (& continuously) using MCU. Using an oscilloscope is not an option in this case. Can u pls suggest any other way for my case?

 

[Nigel Goodwin]

Have a second micro-controller, on the primary side, which measures the voltage and current on the primary, then use an opto-isolator to send serial data to the main micro-controller on the secondary side.

 

[Diver300]

Have a second micro-controller, on the primary side, which measures the voltage and current on the primary, then use an opto-isolator to send serial data to the main micro-controller on the secondary side.

If you want to know the efficiency, you should consider the input current waveform. The power factor of small power supplies is quite poor and the current waveform is far from sinusoidal. You may want to measure the current and voltage much more often than once per cycle so that the power factor can be measured.

The basic idea is that you measure the instantaneous current and voltage at 1 kHz or faster, and multiply them together to get the power. You would take an average of the power to display. That may give a very different reading than averaging the current and voltage and displaying the value of those multiplied.

 

[danadak]

Analog devices has an extensive line of analog and digital isolators that
could handle the galvanic isolation needs in SMPS.

This is an older ap note, they have a much broader portfolio of isolation
solutions than this portrays.

https://www.analog.com/media/en/training-seminars/tutorials/MT-071.pdf

Regards, Dana.

 

[crutschow]

I’m using EL817 optocoupler to transfer info of primary to secondary. But the output of the optocoupler is not linearly dependent on its input. How do I calculate the input at primary from output (of optocoupler) at secondary?

If the feedback is an analog voltage-error signal to the PWM controller, as typical for many designs, then you use the CTR value for the optocoupler, which typically has a wide tolerance.
Normally the drive to the opto is a current, such as supplied from a TL431 reference, so you include the voltage-to-current gain of that into the equation.
This total gain is, of course, part of the feedback path gain, so would be included with the PWM controller error-voltage to PWM duty-cycle gain to give the total loop gain.

Here's an SMPS paper that may be of interest.