Electronics > Projects, Designs, and Technical Stuff
Any tips to improve my filter?
Psi:
Hi, I'm using a H11F1 photo fet optocoupler for isolated resistance control (the H11F1 does 100R to 300meg)
and i'm driving the opto LED through a filter from a 5V MCU PWM based DAC (30kHz).
The filter is two RC filters in a row. Both using 100R and 100uF. (The 100R also function as the LED current limit).
If i go any lower than 100uF for the capacitors i can start to detect some of the PWM DAC noise getting through the opto and changing its output resistance.
The resistors are somewhat fixed due to needing the current to run the opto LED.
The product works fine as is, but I would really like to use less than 100uF capacitors, it seems rather excessive.
Does anyone have any better ideas to filter this sort of thing. Maybe a few LC filters or a RC then a LC would work better? what about LC etc..
I'm just not very knowledgeable about the type of passive filters and what works better in what situation.
I could do an active filter but i would prefer not to add any more ICs to the BOM. (Unless you tell me it will work MUCH better in this situation)
Any ideas?
Thanks.
brabus:
The first problem here is C16. At powerup the capacitor is quickly charged by the poor LED, with no current limitation.
I would drive the LED using a bipolar transistor, linearizing the current with a resistor on the emitter. I would then drive the transistor base via an RC filter with way smaller C and way higher R values. Full flexibility regarding NPN/PNP choice, inverted logic, etc. We can easily switch to an LC if we want a steeper filter response.
Could this be viable in your project?
If you give more details regarding the current range through the LED, we can define some values.
Psi:
--- Quote from: brabus on May 20, 2020, 11:15:57 am ---The first problem here is C16. At powerup the capacitor is quickly charged by the poor LED, with no current limitation.
--- End quote ---
That is a very good point. I will fix that in next version. Thanks
--- Quote from: brabus on May 20, 2020, 11:15:57 am ---
I would drive the LED using a bipolar transistor, linearizing the current with a resistor on the emitter. I would then drive the transistor base via an RC filter with way smaller C and way higher R values. Full flexibility regarding NPN/PNP choice, inverted logic, etc. We can easily switch to an LC if we want a steeper filter response.
Could this be viable in your project?
If you give more details regarding the current range through the LED, we can define some values.
--- End quote ---
Anything is viable for the project, but simple and low part count is better.
Currently the IR LEDs are being driving at 18mA. but 15-20mA should be fine.
Ideally it would be nice if i could drive the LED in way that was more linear.
Currently the response between PWM and resistance is very non-linear which mean i don't have very fine control at one end. But that is a separate issue.
Vovk_Z:
Is there a strong need to filter a diode input current? Isn't it easier to filter optocoupler output?
There is an another way to filter - use active opamp integrator instead of passive RCRC.
If optocoupler non-linearity makes inconvenience there is a way to negotiate it with two identical optocouplers and opamps. I can't find right example now. If you go this way then you can make an active filter (active integrator) simultaneously.
Oh, here it is. Use a photoelectric-FET optocoupler as a linear voltage-controlled potentiometer.
brabus:
The attached schematic offers a good view of the components needed ando also a simulation of the waveforms.
In red the diode current, with no appreciable ripple.
You can adjust the values to match the desired dynamic.
There is no turn-on current rush either. The response is also nicely linear thanks to the emitter resistor.
Navigation
[0] Message Index
[#] Next page
Go to full version