Hi all, I finally got a replacement GT mini (thanks Neal!) however there’s some sort of parasitic drain lighting up the LED when it should be off - it’s veeery dim and the drain is pretty negligible but I’d like to fix it if possible. Thankfully the bezel was not glued on the replacement, it was ridiculously tight though, so I’ve taken a peek inside and included some pictures below. I’ve cleaned all the contacts and given the driver a quick once over with some isoprop. I can’t see anything wrong with the driver itself but I’m a huge novice when it comes to the electronics. Can anyone spot anything wrong? Is there anything I can check relatively easily? I’ve seen reports of leaky FETs in other lights, could this be the cause and if so how do I test for it?
Also, I took the opportunity to flash Anduril while I had the driver it out. The LED still lights up regardless of if the switch LED is set to ON or OFF.
Yeah, sounds like a leaky FET. Think the fix was to add a pull-down (or -up) resistor to keep it turned off when there’s no active signal turning it on. It could “float” and just let a slow drip go through the FET. The resistor just cranks the faucet tighter “off”.
Hmm, would swapping the FET would be a better solution? I think I’d rather do that than experiment with resistors. Is it possible to test for leaks with the FET out of circuit?
Lightbringer brings up an interesting solution. Mosfets shouldn’t have their gates floating or at high impedance if they are being used as an on/off switch. Most of the time you can count on the active device (ie: microcontroller) being adequate. But I have seen some situations where you do need a pull down resistor (ie: about 20K resistor from gate to source pin). The value of the resistor isn’t critical. I just like to use a fairly large resistor such that I’m not wasting much power when controlling the gate.
To verify that this is the problem, I’d just hold a resistor with leads from the gate to the source and see if the problem goes away. If it does, then I’d obtain a small surface mount part and solder between those two pins.
To me it looks like the FET already has a pull down resistor on the gate, the one marked 01D ? Try shorting the gate to GND to check if issue is resolved before resorting to anything else. If shorting to GND solves it, check that 01D resistor.
The red arrow in this picture is pointing at the FET gate. Try shorting that with the G tab (indeed a GND point). Another thing you can check is shorting the gate of the 7135 (yellow arrow) to GND.
If the 01D resistor is the gate pull down resistor…………and the mosfet is an N-channel mosfet, then one side of that resistor will be ground and you can just take some tweezers and short across it temporarily to see if the problem goes away. But that assumes that 01D is indeed a pull down resistor. And I highly doubt you’ll be lucky finding this to be the fix. A resistor that is the gate pull-down resistor is in the 10k ohms or higher and they simply don’t go bad…………unless maybe a bad solder joint?
Ground can also be found via:
One pin of the Atmel microcontroller (U2) has a ground pin. You’d have to look at the data sheet for that part to know which pin.
Likely the big area of copper plane surrounding the pcb is ground.
The negative side of the battery is also the circuit ground.
First of all where is the mosfet? To me the part to the right of the U2 silkscreen looks more likely. I just don’t know what that bigger square part is in the upper right side. Whatever part it is, it doesn’t look to have exposed pins. One of the easiest ways to figure out what is going on is to put an oscilloscope on the gate (and other areas) to see what is going on. But you may not have an O-scope.
I think you are right about the picture. I just looked at one image and didn’t get the full picture.
100K ohm resistor is quite a bit higher than I’d use in my designs. Check it with an ohm meter without power. Even though its in circuit, it should read near or lower than what the actual value is. If the problem is this resistor (and its function is to pull the gate to ground when no control signal is present), then by shorting the resistor via a pair of tweezers will put ground to the gate and the mosfet should be off. Tells us what happens after you try that.
My assumptions right now are that its an N-channel mosfet where the Source pin of the mosfet is grounded. If one end of this resistor is not tied to gnd, then that changes things.
Ok, sounds like the actual FET or 7135 is leaking regardless of gate state. There isn’t any other path the current from LED- can take to get to GND on a correctly assembled driver. I guess it could happen if there is a faint connection between LED- and GND somewhere else on a faulty driver, but I wouldn’t know where.
I’ve had 7135s leak like this and had to replace it go get rid of the issue. I’ve personally not had it happen to a FET but from what I gather it very well can. Unfortunately I don’t think there is any way to test the 7135 and FET separately without removing them from the board.
My understanding of a FET + 7135 driver is that the 7135 is in parallel with the FET. Are you soldering skills ok? I’d open up the “OUT” pin of the 7135. That’s the pin in the picture that looks like its not soldered all the way (the one to the left as you look at the 7135). I’d get an Xacto knife to pry up the pin as you heat it with soldering iron. Solder wick would help if you have that. Basically lift the pin just enough so that its not making contact with the board and that will take the 7135 out of the circuit and see if the problem goes away. That wouldn’t fix the problem but it would tell you where the problem is coming from.
This is a bit of a wild shot………….but maybe before you try that, see if there might be a slight stability issue with the 7135 where a capacitor between the “out” pin and the “gnd” pin of the 7135 fixes it. I noticed that the data sheet for the 7135 suggests a capacitor between the “out” pin and ground depending on pcb layout and distance of 7135 from LED. I think that this pcb is ok without one but it might not hurt to temporarily place a 0.1uf cap between “out” and “gnd” pin of the 7135 and see if that does it. You could also take your finger and touch both of those pins at the same time and see if the problem goes away. Just make sure you haven’t walked across the carpet and have any static before doing so. The “gnd” pin is in the middle. Sometimes just a little capacitance is all it takes.
