Opto-isolation and Voltage stepdown Question

[ericmar]

pwm opto dac

Hi everybody,

I hv some questions abt the opto-isolation and the voltage stepdown. Let's say i were to convert a signal from a MCU to a medical device and i use a DAC to convert the signal passed from the MCU. Since the device is expected to receive only uV signal and thus, i will need to stepdown the voltage with a factor of 124 to get the desired signal. If i use an opto-coupler in between of the DAC and the interface connected to the medical device, should i do the voltage stepdown b4 or after the opto-coupler?

In fact, i hv a prototype which does the voltage stepdown right after the signal comes out from the DAC and then the signal is passed to an op-amp b4 sending to the medical device. With this configuration, i can get pretty accurate results on the medical device but the isolation was not considered during the prototype so need to integration the isolation in the new design with an opto-coupler.

Another modified prototype was done by passing the signal comes out from the DAC to the op-amp, and then through an opto-coupler. The voltage stepdown is done after the signal comes out from the opto-coupler b4 sending to the medical device. But with this configuration, i always get inaccurate results.

I am not sure where went wrong in the circuit. Could anyone pls help me with this?

Thanks a lot!

Best regards,
Eric Mar

 

[IanP]

liniar optoisolation

If option with MCU-->DAC-->Buffer-->VoltageDivider-->MedicalDivice works,
maybe you should try MCU-->DAC-->OptoCoupler-->Buffer-->VoltageDivider-->MedicalDevice ..
I haven't come accross good linear optocouplers which would work correctly in the µV range; in fact, they have even problems with low mVs .. so you have to consider generating higher voltages from DAC and bigger division ratio before the Medical Device ..
Regards,
IanP

 

[cyberblak]

analog signal optoisolation

I don't think that an optocoupler could transfer a signal in the uV range... You probably should scale it after the optocoupler.

good luck,

cyberblak

 

[VVV]

analog signla optoisolation

Can you power the DAC from the medical device side? With a serial DAC, you would only need 3 opto's.
That way you can do MCU-OPTOs-DAC-Divider-Medical device.

All your headaches would be gone.

 

[ericmar]

pulse amplifier with optoisolation

Can you power the DAC from the medical device side? With a serial DAC, you would only need 3 opto's.
That way you can do MCU-OPTOs-DAC-Divider-Medical device.

All your headaches would be gone.

Hi VVV,

Why do I need 3 opto's? And why do I need the medical device to power up my DAC?
In fact, I m not drawing any power from the medical device to support any of my components on the board!

Actually, I m not really sure if the isolation circuit is designed correctly or not since it is not my design. I would appreciate if anyone could kindly show me some examples of the isolation circuit.

Thanks million!

Regards,
Eric Mar

 

[cyberblak]

The benefit of this configuration (MCU-OPTO-DAC-MED) is that you no longer need to use a linear optocoupler since the optos will transfer digital signals. Then you can use low-cost standard optocouplers.

In order to be properly isolated, you must isolate the power supply and the signals that cross the isolation barrier. With the optos you take care of the signals, so that leaves the power supply to isolate.

What is the type of power supply that you intend to use?

cyberblak

 

[VVV]

What I meant was if there is power available on the medical device side, you can power the DAC from it and so you only transmit digital signals over the optos. That solves the problem of accuracy.

You need 3 optos, since the A/D will communicate serially with the MCU. You can do it with only two optos, if you are careful. One will be the clock, one the data. But some DACs require a chip select or similar signal, so that will be the third opto.

 

[dpsman]

You can also implement VVV's idea with a small DC-DC converter or even better, you can power the whole thing with the DC-DC and keep bidirectional communication with uC with only two optos and forget about thermal stability of opto devices!

regards,

dpsm

 

[bicho]

Why opto-isolation? It would be a good solution if you could put it (and DAC) on medical device side like vvv wrote.
If you need isolate an analog signal, maybe an amplifier isolator like ISO124 would be better...

Regards,
bicho

 

[ericmar]

You can also implement VVV's idea with a small DC-DC converter or even better, you can power the whole thing with the DC-DC and keep bidirectional communication with uC with only two optos and forget about thermal stability of opto devices!

regards,

dpsm

The smallest DC-DC converter I m aware of now is the DCR010503U from Texas Instruments. But it is not very small and a bit expensive!

 

[artem]

you can use optocouplers photodiode as source voltage for microvolts . But load must be high resistive otherwise nonlinearity will be there . IL300 can help you achieve stability . Or use transformator modulated with PWM (say DAC) - that gives much stronger power irrespective to load .

 

[ericmar]

you can use optocouplers photodiode as source voltage for microvolts . But load must be high resistive otherwise nonlinearity will be there . IL300 can help you achieve stability . Or use transformator modulated with PWM (say DAC) - that gives much stronger power irrespective to load .

I m using IL300 but it cannot achieve the stability. I m trying to make it provides a unity gain but the output is always lower than the input which causes some losses on the results.

 

[artem]

The question is to design low voltage bandgap you put to the photodiode of IL300 . Then modulate optocoupler with PWM to achieve needed voltage . Bandgap will ensure that your output for pulse will have fixed voltage . Regulating PWM due will give you needed voltage level assuming passive LPF will be after . Try to measure photodiode generated voltage for maximum current supplied on LED . let say it will be 0.5 v . Then put bandgap directly to photodiode output . This allow you to have fixed voltage pulse necessary for PWM handling .

Even better result can be achieved if two photodiodes will be connected serially - open circuit voltage for photodiode is 500 mv . 2 of them will provide 1 V should be enough to supply Si bandgap. Again , this requires that oiutput voltage to be regulated with PWM instead of conventional DAC.

Max pass frequency for IL300 is about 200 kHz . What is your requirements for output signal in terms of frequency?

 

[artem]

just an example . Bandgap made on diode is very poor , but you have 1 V on IL300 output, that allows you to put temperature stable bandgap.

PWM acts as DAC . If output impedance does not need to be low you can try it

Play with C for PWM freq filtering.