Optical Isolation w/ Temperature Compensation

[Nick Naylor]

Hey all,

Trying to transfer a control signal across an opto-isolator. Control
signal goes anywhere from 0-5Vdc. Current is minimal, as it is driven
with an op-amp (AD822). My opto is currently a CNY17-1. I tried
simply hooking the op-amp directly to the opto and putting a load on
the output of the opto. It drifts 20mV just by itself, but if I heat
it up, it can drift all way up to 200mV! That's a lot. This circuit
may be near other things that will get warm. Are there any ways to
overcome the temperature drift? Any ideas are welcome.


Thanks Everyone,
Nick

 

[Spehro Pefhany]

Hey all,

Trying to transfer a control signal across an opto-isolator. Control
signal goes anywhere from 0-5Vdc. Current is minimal, as it is driven
with an op-amp (AD822). My opto is currently a CNY17-1. I tried
simply hooking the op-amp directly to the opto and putting a load on
the output of the opto. It drifts 20mV just by itself, but if I heat
it up, it can drift all way up to 200mV! That's a lot. This circuit
may be near other things that will get warm. Are there any ways to
overcome the temperature drift? Any ideas are welcome.

Have a look at the Infineon (maybe now Vishay) IL300 (used in a closed
loop), or drop in an isolation amplifier, or use some digital means.
Optoisolators age as well as drift, BTW.

Best regards,
Spehro Pefhany

 

[John Popelish]

Hey all,

Trying to transfer a control signal across an opto-isolator. Control
signal goes anywhere from 0-5Vdc. Current is minimal, as it is driven
with an op-amp (AD822). My opto is currently a CNY17-1. I tried
simply hooking the op-amp directly to the opto and putting a load on
the output of the opto. It drifts 20mV just by itself, but if I heat
it up, it can drift all way up to 200mV! That's a lot. This circuit
may be near other things that will get warm. Are there any ways to
overcome the temperature drift? Any ideas are welcome.

Thanks Everyone,
Nick

You might change to a dual channel isolator (with the LEDs driven in
series), using the second output as part of the feedback for the
opamp. Then the drift will consist mainly of the difference between
the two isolators so most thermal effects will track.

 

[Joerg]

Hi Nick,

One way might be clamping if you had some clock on the other side and
can spare a "clamp moment" every now and then. That still leaves the gain
drift which could also be clamped.

If clamping isn't an option I usually modulate this stuff and then it
transfers across nearly anything without a worry in the sky about drift.
One way to perform modulation is to use a V/F converter such as the LM331.

On the other side you convert back. This adds a few Dollars in cost but
now you only transfer digital pulses and none of the drift matters much.

Regards, Joerg

 

[John Larkin]

Hey all,

Trying to transfer a control signal across an opto-isolator. Control
signal goes anywhere from 0-5Vdc. Current is minimal, as it is driven
with an op-amp (AD822). My opto is currently a CNY17-1. I tried
simply hooking the op-amp directly to the opto and putting a load on
the output of the opto. It drifts 20mV just by itself, but if I heat
it up, it can drift all way up to 200mV! That's a lot. This circuit
may be near other things that will get warm. Are there any ways to
overcome the temperature drift? Any ideas are welcome.


Thanks Everyone,
Nick

Several people (Burr-Brown, Analog Devices at least) make official
isolation amplifiers, and they're not too expensive. Unless you need
large quantities, that's probably the way to go.

John

 

[Ban]

Hey all,

Trying to transfer a control signal across an opto-isolator. Control
signal goes anywhere from 0-5Vdc. Current is minimal, as it is driven
with an op-amp (AD822). My opto is currently a CNY17-1. I tried
simply hooking the op-amp directly to the opto and putting a load on
the output of the opto. It drifts 20mV just by itself, but if I heat
it up, it can drift all way up to 200mV! That's a lot. This circuit
may be near other things that will get warm. Are there any ways to
overcome the temperature drift? Any ideas are welcome.

Nick,
you can make a pulse width modulator with another AD822 and recover the
signal with a simple lowpass secondary. This thing will oscillate at about
5kHz, maybe you can even simplify your circuit.


5V
|
| |\
o---)---------|-\ ___
1.7...3.3V | | >-|___|--+
.-. +-----|+/ 470 |
100k| | | |/ ,---.
| | | AD822 opto| X |
'-' | '---'
| | |
| | ___ ===
+---)----|___|-+ GND
| | 100k |
| | |\ |
+---)-----|+\ |triangle
| | | >-+generator
| +-----|-/ |
| | |/ |
| | ___ |
| +----|___|-+
.-. | 100k
100k| | ---
| | ---
'-' |10n
| |
=== ===
GND GND
created by Andy´s ASCII-Circuit v1.24.140803 Beta www.tech-chat.de

 

[John Fields]

Hey all,

Trying to transfer a control signal across an opto-isolator. Control
signal goes anywhere from 0-5Vdc. Current is minimal, as it is driven
with an op-amp (AD822). My opto is currently a CNY17-1. I tried
simply hooking the op-amp directly to the opto and putting a load on
the output of the opto. It drifts 20mV just by itself, but if I heat
it up, it can drift all way up to 200mV! That's a lot. This circuit
may be near other things that will get warm. Are there any ways to
overcome the temperature drift? Any ideas are welcome.

 

[Frank Miles]

Nick,
you can make a pulse width modulator with another AD822 and recover the
signal with a simple lowpass secondary. This thing will oscillate at about
5kHz, maybe you can even simplify your circuit.

[snip nice drawing]

In fact if your accuracy requirements are sufficiently modest you can
even do the measurement side with a 7555 timer, driving the control port.
Attenuate the input signal to avoid pathological timer responses. The
pulse width varies but the duty cycle is fairly linear over 20-30% of
the power supply.

-frank
--