I like your idea of using two optoisolators acting as a current mirror.

if that Dim Control current varies, then both my R1 and R2 need to change.

The circuit can be modified to overcome any change in the 10uA current, by reducing resistor values drastically, thus swamping the 10uA current so that it becomes insignificant, as below :-



D-ctrl voltage (referenced to Gnd2) now follows the input PWM-derived voltage closely. A maximum D-ctrl voltage of 2.5V corresponds to a 75% duty cycle of your 3.3V PWM input signal.
The optos might need to be a thermally-coupled matched pair.
An advantage of the higher LED current is that the opto transistor minimum Vce is reduced, giving a lower D-ctrl voltage for better dimming.

 

What is your thoughts on this;

You are depending upon the two optos having identical characteristics, which the simulated models do and real devices will not.

 

Very true. I suspect any arrangement using a current mirror is going to require careful matching/tweaking of components. A small imbalance can result in quite marked deviation from a linear response.

 

Hello,

The IL300 is a so-called linear optocoupler.
The datasheet shows circuits how to use it.

Bertus

 

Here's a simplified circuit which, like Crutschow's one, uses the opto as an on/off switch. The output response, albeit not so linear as before, is substantially independent of opto linearity, temperature and the Dim-ctrl constant current value.



The response curve shown is for an input PWM duty-cycle range of 0% to 100%.
D-ctrl voltage now goes down to ~33mV.

 

Hello,

The IL300 is a so-called linear optocoupler.
The datasheet shows circuits how to use it.

Bertus

Thanks, I've seen this one. Bit pricey for what I need it for - if I went down this path, I'd probably use a dual optoisolator and use the second one as feedback - similar to how the IL300 is supposed to be used. I'm hoping that being a dual one in the same package that they might be reasonably closely matched?

 

You are depending upon the two optos having identical characteristics, which the simulated models do and real devices will not.

What if it was a dual opto - same package?

 

Here's a simplified circuit which, like Crutschow's one, uses the opto as an on/off switch. The output response, albeit not so linear as before, is substantially independent of opto linearity, temperature and the Dim-ctrl constant current value.

View attachment 208205

The response is for an input PWM duty-cycle range of 0% to 100%.
D-ctrl voltage now goes down to ~33mV.

I like that. But I lose a lot of granular control at lower duty cycle


Setting R4 lower fixes this, but crashes the FL5150.
Perhaps a 'Digital Output Optocoupler' (with schmitt trigger for PWM shaping) will alleviate this and allow me to use higher value resistors without suffering this issue?

 

Hey @crutschow @Alec_t
I picked up these H11L1M digital (schmitt trigger) optoisolators to help with the wave shaping on the output.
Not quite sure how to wire it up to
a) give me 0-2.5V output,
b) have a NON-inverted output (if possible), and
c) where to put the load resistor (remembering that the FL5150 cuts out if there is too much current draw. I had RLoad at 4.7k but it crashes. Works when it is 10k)

With the above, I get inverted operation.
And strangely - when I turn the light off for longer than a brief moment, it won't turn back on!? I have the low power mode transistor connected as per the example schematics. Perhaps in enters low power mode and isn't coming out of it...?

 

Try this :


The PWM is used as a low-side drive for the opto, to avoid inversion.
Edit: If you combine R2 and R3 into a 5k trimpot you'd have some tweakability on the Dim-ctrl voltage limit.
Edit 2: S1 should be omitted from the H11L1M model.

 

I've just converted the contents of the dotted rectangle in post #30 into LTspice models for the H11L1M, H11L2M and H11L3M.
If anyone's interested here they are :

Edit: Don't use these, they may be buggy. I'm looking into it.

 

I've just converted the contents of the dotted rectangle in post #30 into LTspice models for the H11L1M, H11L2M and H11L3M.
If anyone's interested here they are :

Wow, thanks!
I've wired this;

But I am only getting 0V->0.054V at the dim pin.

If I put the load resistor in;

I get 1.195V->0.087V (inverted).
However, as soon as I hit that 0.087V (fully off), the light will no longer turn back on unless I cycle the power to the FL5150 chip. Not sure why... If I use a normal optoisolator instead of the H11L1M, that issue doesn't occur...

EDIT:
When I simply short the DIM Control pin to GROUND, the light turns off and when I remove it, it turns back on. No problem.
I put a 22k load resistor on (R4), and it works OK then. So again, it's an issue with using that Vdd pin... However, now with a 22k R4 resistor and 4.7k for R2, I have a voltage divider occurring... Odd - because that doesn't show up in the simulations. Is it because it is an 'open-collector' for Vout? Seems to exhibit the same voltage response as if it was a normal optoisolator.

With the following setup (using a H11L1M instead of the 817C shown here), I get the same voltage response/output at DIM Control;

So that H11L1M hasn't helped me at all.....
At 0%, it still has enough voltage to turn the light on ~0.180V

So.
a) Can we avoid using that load resistor? Or is this a function of how the H11L1M works (and it being an 'open-collector' output)?
b) if we HAVE to have that 22k load resistor, how can I get 2.5V at the DIM Control without having too much affect from that 10uA?

 

I tried this;

When I use R4 = 10k, I get a strange issue as I increase the dimming, it gets to about 70% and then actually starts DIMMING again. When R4 = 22k, the issue doesn't present. Again; an issue using Vdd?
Unfortunately with R4 = 22k, the wave shaping of the schmitt trigger doesn't help enough at lower duty (as shown).

 

take step back :
Your dimming an AC load of what voltage / power / frequency ?
what dimming range do you want to cover ?
what input voltage range do you want to achieve that control ?

 

take step back :
Your dimming an AC load of what voltage / power / frequency ?
what dimming range do you want to cover ?
what input voltage range do you want to achieve that control ?

240V, 300W, 50Hz. Using an FL5150 to do the dimming 'smarts'.
Dimming range...? 0-100%...
I just need to provide it a 0-2.5V signal. However, each of the gangs (one FL5150 each) reference GND are at different voltages at any one time. Which means my control signal GND cannot be connect to it - so I require isolation.

 

So that H11L1M hasn't helped me at all....

Sorry. I'll check the model for a bug.

 

Sorry. I'll check the model for a bug.

Be great to have a good model for it, but so far my practical tests aren't giving me promising results. I need to have that 22k resistor otherwise the FL5150 shuts down. And it appears to give me the same voltages on the DIM control pin as a standard 817C optoisolator. So the H11L1M hasn't helped with linear output...

 

240V, 300W, 50Hz. Using an FL5150 to do the dimming 'smarts'.
Dimming range...? 0-100%...
I just need to provide it a 0-2.5V signal. However, each of the gangs (one FL5150 each) reference GND are at different voltages at any one time. Which means my control signal GND cannot be connect to it - so I require isolation.

Your hooked on the FL5150 ?

its always much more difficult to generate an isolated linear voltage than it is to isolate a "digital" PWM signal,

 

Your hooked on the FL5150 ?

its always much more difficult to generate an isolated linear voltage than it is to isolate a "digital" PWM signal,

I’m generating an isolated PWM and then filtering it. My issue is the FL5150 isn’t happy when I draw current from the 5V Vdd pin. It wasn’t designed to be a current source. So I have to put a large resistor as the Rload. Esteem Vdd and the optoisolator collector pin. When I do that, the PWM shape gets smoothed/distorted so much, that I get poor linearity. So, I’ve been trying a Schmitt trigger to help with wave shaping. Unfortunately it hasn’t been successful; a 22k load resistor simply causes too much distortion.
If you have any other ideas to try, paste a schematic.

 

I’m generating an isolated PWM and then filtering it. My issue is the FL5150 isn’t happy when I draw current from the 5V Vdd pin. It wasn’t designed to be a current source. So I have to put a large resistor as the Rload. Esteem Vdd and the optoisolator collector pin. When I do that, the PWM shape gets smoothed/distorted so much, that I get poor linearity. So, I’ve been trying a Schmitt trigger to help with wave shaping. Unfortunately it hasn’t been successful; a 22k load resistor simply causes too much distortion.
If you have any other ideas to try, paste a schematic.

Can I summarise , see if I have this right.

The FL5150 is expecting a 0 to X linear analog control voltage, and the opto-isolators are as you have seen not linear and not repeatable across temperature and different devices. ( diode forward current /light is none linear , transistor is none linear )

I see you have tried PWM , but that as you have seen , this needs power.