How to use optocoupler to control a fan?

Using Arduino General Electronics
1

Hi,

I am wondering if i could use TCDF1910 with a 3.3V Wemos Mini R1, to control a 10 V pwm of a fan.
Problem is i don't quite understand the connection in the attached picture? Probably pwm signal to pin 1 and gnd(-) to pin 2, but is there need for some resistor or what's that in the attached connection table ?

I figured out that when the 10V DC supply for the fan control pwm is coming from the fan, that's connected to pin 4 and pin 6 to fan input? Or what is the correct connection?

Full datasheet of the optocoupler: TCDF1910 Datasheet | Temic - Datasheetspdf.com

Fan is some old EBM-papst model, connection diagram attached too.

ebm-papst.png
ebm-papst.png953×886 86.1 KB

TCDF1900.png
TCDF1900.png694×921 79.4 KB

2

Correct connection is shown on the left bottom side, just replace BJT with mosfet.

Pin 6 (+10V) -> resistor (1k) -> pin 6 (optocoupler) & -> pin 7 (controller).
Pin 4 (GND) -> pin 5 (optocoupler).

arduino pwm pin -> resistor (2.2k) -> pin 1 (optocoupler).
arduino gnd -> pin 2 (optocoupler)

You may need to tweak arduino PWM to get it into recommended 1-10 kHz range.

Update.
After I finished typing, I realized that optocoupler's DS has a typo:
t-on = 0.1 us, and t-off = 0.2 usec - can't be true. Photo-voltaic response time is about 0.1 - 3 milliseconds, it's means you shouldn't use this type of the coupler with PWM. Get 4N25/26/35/37 or 6N136/137.

3

@FantomT,

Not sure how you support your conclusion the datasheet is wrong. Photo diodes can be very fast.

Perhaps I am missing something.

4

JohnRob:
@FantomT,

Not sure how you support your conclusion the datasheet is wrong. Photo diodes can be very fast.

Perhaps I am missing something.

I've read a dozens DS from all major players in the industry, TELEFUNKEN couldn't make 10^3 faster couplers in 1996, that 20 years later no one else could reproduce.

Thinking why it's slow, probably the issue not with phd, but phd heavily loaded by input capacitance of the mosfets. Since phd current is a fraction of the input current, there is just no power to recharge 10 nanofarads or so capacitance in less than 100 microseconds.

5

@FantomT

You are probably correct.

This isn't a device type I've used very often and then not in a time critical application. And I do see it's a very old part.

But for my own edification I looked at what I thought would be typical numbers:

Gate Capacitance of a 2N7000 is 60 pf max. I did not consider the G-D charge nor the G-S charge as it was further than I wished to go.

The Temic spec says the speed test was done at 10 ma input.

So if we assume the transfer function of the diodes is 1000:1 (a pure guess) then we could have 10 µA to work with.

If I assume the FETs need 5V we can calculate dt using the simple formula I = C dv / dt

dt = C dv/I I get 30 µs

OK so if I'm even close the part won't meet Temic spec but its not that bad.

However I do agree your suggested alternatives would be much better choice :slight_smile:

6

FantomT:
I've read a dozens DS from all major players in the industry, TELEFUNKEN couldn't make 10^3 faster couplers in 1996, that 20 years later no one else could reproduce.

Thinking why it's slow, probably the issue not with phd, but phd heavily loaded by input capacitance of the mosfets. Since phd current is a fraction of the input current, there is just no power to recharge 10 nanofarads or so capacitance in less than 100 microseconds.

You are confusing photo diodes with either photo transistors or photo resistors I think

Photodiodes when reversed biased cannot be slow, there's no mechanism for them to be slow,
although lower capacitance variants (PIN photodiodes for instance) can be even faster!

7

MarkT:
Photodiodes when reversed biased cannot be slow

Right, When/ IF reversed biased. And it's not the case with opto-relay.