multisim11 optocoupler error

[bwana1]

Hello, apparently I have stumbled upon an error on Multisim 11 optocouplers.
HCNR201, a dual photodiode-single LED precision optocoupler is shown as it
were made by a LED lighting two phototransistors, together with its SPICE model.
The worst thing, in my opinion, is that original photodiode cathode and anode are
now connected to collector and emitter of their transistor counterparts, resulting
in opposite polarization of the real devices if mounted on a pcb, not to mention the
bunch of doubts related to the effectiveness of a simulation of a circuit empoying
such device.

 

[FvM]

Accuracy of a SPICE model isn't related to how it connects in a schematic symbol.
You need to check the SPICE model itself, e.g. if it's supplied by the manufacturer, if it includes dynamical parameters etc.

 

[bwana1]

You need to check the SPICE model itself

Yes, you need it.
I already did it before opening this thread.
If you go and open that model, you'll see two transistor models instead of two photodiode models, as reported in the external schematic.
But I haven't yet heard that a transistor model equals that of a photodiode.
And the corresponding datasheet with no doubt indicates two photodiodes.
Repeating my argumentation, if I have to polarize a photodiode, I polarize it inversely, if instead I have to polarize a transistor, I put positive voltage on collector and negative on emitter, right the opposite.
While simulating, the simulation can run even if there are (photo)transistors instead of diodes, though nobody can bet on an identical and correct behaviour, so this kind of simulation in my opinion is quite useless and misleading.
Worse things happen if, by chance, you go on and design your circuit's PCB relying on the current pinout.
The best surprise will happen when you then mount the **real** component, because you end with two photodiodes directly polarized.
Best regards

 

[FvM]

Did you check the SPICE model given in the HCNR201 data sheet? It's in fact using transistors to model the LED/photodiode coupling. Obviously this is more a kind of behavioral than a physical model. But this are still photo diodes rather than photo transistors.

I guess MultiSim is using the same model.

If so, you should address your doubts about correct model behavior to Avagotech. Generally, it's quite common that a SPICE model doesn't represent all properties of a device. At best, the restrictions are clearly told.

 

[bwana1]

Avago datasheet states the following:
pin 1 : LED cathode
pin 2 : LED anode
pin 3 : Photodiode 1 cathode
pin 4 : Photodiode 1 anode
pin 5 : Photodiode 2 anode
pin 6 : Photodiode 2 cathode

Multisim schematic states the following:
pin 1 : LED cathode
pin 2 : LED anode
pin 3 : Phototransistor 1 emitter
pin 4 : Phototransistor 1 collector
pin 5 : Phototransistor 2 collector
pin 6 : Phototransistor 2 emitter

So, whatever the model, if I correctly connect for multisim pinout, pins 4 and 5 will go to a higher voltage than pins 3 and 6.
Following the datasheet statement, the real component will also have pins 4 and 5 higher than 3 and 6, leading to a direct polarization of the claimed photodiodes.
The opposite (and wrong) polarization of every application circuit.
Quod erat demonstrandum. Fullstop.

 

[FvM]

If it's the MultiSim model is different, use the model from Avagotech.

 

[bwana1]

I don't know if they have one, please note that my first effort was to warn people agaist a possible bug in Multisim, that our discussion seems to confirm. If you have a link to an Avago model, that could be the cure.
That specially because, as you can see the bug is somewhat involved and possible to discover only at the end of a whole process.

***Some time later***
I followed your suggestion and found Avago's Spice models.
That's the link:
SPICE Models
Now I'll see if this can solve the problem.
Best regards.

 

[FvM]

please note that my first effort was to warn people agaist a possible bug in Multisim

That's good indeed.

As I said, there's a model printed in the data sheet (on pg. 14). **broken link removed**

The model can be also downloaded SPICE Models

 

[bwana1]

What the hell!
I was really blind! I watched that datasheet hundreds of times and did never mind of it!
I owe you a lot of excuses.
The most you search a thing, the most it's under your eyes!
Best regards