MOSFET test circuit using LEDs to turn MOSFET ON

Thread Starter

SamR

Joined Mar 19, 2019
5,031
Some time back I found a MOSFET testing circuit that used 2 LEDs on the gate of the MOSFET. When exposed to light the LEDs actually generated enough voltage to turn the MOSFET ON. It was the first and only time I had ever encountered a standard red LED used as a photodiode. Does anyone know this circuit? I cannot find it anywhere and of course, I forgot to make the notes I should have when I found it. I guess I should just grab some parts and start experimenting... What I do remember is that only 1 LED did not work and had to have 2 in series. Thx, Sam.
 

Thread Starter

SamR

Joined Mar 19, 2019
5,031
Found it:
Was looking in the wrong places, it was a utuber of all things.

Yes, it uses a low V input BS170, pretty neat trick and the BS170 can handle 500mA 60V 830mW.
 

ronsimpson

Joined Oct 7, 2019
2,986
Some where I saw a application note for making a night light. The LED produced light and measured the amount of room light.
In the day time, the off LED made some voltage, which told the electronics to keep the LED off.
At night the LED was turned off for 1mS to measure the room light. The human eye could not see the small gap in the light. In 1 or 2mS the voltage on the LED went from (1.5 on voltage) down to a low level but took too much time to get to 200mV. So they measured the slope of the voltage. If the room was dark the slope was fast, light = slow slope.

I think I have seen this done with a PIC and some other 8 pin micro.
 

Wolframore

Joined Jan 21, 2019
2,609
I would use a better mosfet, that one has large threshold tolerance 0.8-3v and really needs high gate voltage and has high rds. Here’s my go to for low voltage low cost Si2302. Not sure if it has a through hole version.
 

Wolframore

Joined Jan 21, 2019
2,609
I brought this up in class and was brushed off as it’s not how it’s designed to be used. What’s interesting is the narrow bandwidth. This is how fiberoptics work and you can multiplex data based on frequency and have multiple signals on the same cable. Now I remember why I found this interesting.
 

Thread Starter

SamR

Joined Mar 19, 2019
5,031
I'm still way down on the learning curve for MOSFETs but the Si2302 certainly looks interesting for low turn on and I do have a bunch of 3 pin adapter boards. Seems like I remember Mimms taping 2 LEDs together to use as an optoisolator. Anyways it had been pulling at my memory as being an interesting idea and couldn't remember where I saw it. Now to get out my notebook before I forget it again.
 

OBW0549

Joined Mar 2, 2015
3,566
One thing to keep in mind when driving MOSFETs with LEDs is that while an LED may produce a volt or more when brightly illuminated it isn't going to provide much current; I measured on an assortment of LEDs only a couple of microamps when lit by a bright desk lamp with the LED directly in contact with the bulb.

What this means for MOSFET driving is that the LED isn't going to be able to turn the MOSFET on very quickly because it will take time for that small output current to charge the MOSFET's gate capacitance. So the FET will spend considerable time in the active region, neither fully on nor fully off, and depending on what it's switching it could be dissipating significant power during that time. And turn off will be even slower than turn on.

So beware.
 

Analog Ground

Joined Apr 24, 2019
460
Checkout "Photovoltaic MOSFET Driver" to get an idea of some issues with driving MOSFETs with light. These components use a photodiode but I bet the same issues apply for an LED acting as a detector. Here is an example,
http://www.vishay.com/docs/83469/vom1271t.pdf
1) For higher voltage, stack 2 or more
2) Turning off is an issue because of the diode.

Post #11 notes the low current. However, these devices operate in the microamp range. Doable?
 

Wolframore

Joined Jan 21, 2019
2,609
How about using an opamp as comparator let it hard switch at some point. I don’t mind the delay but I can see issues if driving high current
 

OBW0549

Joined Mar 2, 2015
3,566
Post #11 notes the low current. However, these devices operate in the microamp range. Doable?
Sure it's doable, provided you don't need high speed; if you do, you're probably outta luck.

For example, the data sheet for the IRLZ44 MOSFET lists the total gate charge @vgs = 5 volts as 66 nanocoulombs. With a charging current of 2 μA, it takes 66E-9 C / 2E-6 A = 33 milliseconds to deliver that charge. That 33 milliseconds could be anywhere from insignificant to grossly excessive, depending on the application.
 

Bernard

Joined Aug 7, 2008
5,784
I just had to try it. One green LED shunted by 10 meg. drove a FET- FDD6530A, LL, SMT, to 30mA thru white piranha LED. On @ 6 in. full on @ 2 in. from LED desk light. Battery box 4 AAA.
 

Bernard

Joined Aug 7, 2008
5,784
Added a 10 ohm load in parallel to LED,; total drain = 360 mA, S-D about 150 mV, R hot, FET cold, & LED bright. Light source small LED flashlight @ 2 in.
 
Top