Something comparable to ZVP3306A. Make the resistor values 10X larger and see if the LEDs stay off when SW3 is off.

 

The circuit is far more complicated than it needs to be, and R7 can be expensive. If I were you, I would think about changing it.

Been there, done that...
It's an automotive retrofit...

 

The circuit is far more complicated than it needs to be, and R7 can be expensive. If I were you, I would think about changing it.

Per post #11 all changes must be doable on the existing board. And in the world of automotive electronics, a pot, even a weird one, is considered more reliable than a transistor.

ak

 

The circuit is far more complicated than it needs to be, and R7 can be expensive. If I were you, I would think about changing it

How does he end up saving money with your advice? Changing the design means he would have to buy a transistor to replace a potentiometer that he already owns.

 

Then you may want to articulate exactly what you CAN change.

At this point, the only thing you can change are the LEDs and the resistors associated with them?

If that's the case, I suggest that you use a) low brightness LEDs - they tend to start to glow under higher levels of current; b) put leds in serial so they are close to 12v; c) consider current regulating devices (diodes or otherwise) in serial with the leds -> those devices will exhibit high R before their set current, thus limiting the off-state current / glow.

But it makes sense for you to reaffirm the limitations of the project first.

 

Then you may want to articulate exactly what you CAN change.

At this point, the only thing you can change are the LEDs and the resistors associated with them?

If that's the case, I suggest that you use a) low brightness LEDs - they tend to start to glow under higher levels of current; b) put leds in serial so they are close to 12v; c) consider current regulating devices (diodes or otherwise) in serial with the leds -> those devices will exhibit high R before their set current, thus limiting the off-state current / glow.

But it makes sense for you to reaffirm the limitations of the project first.

Limitations were specified above. Setting LEDs in series will mean that the 20 ohm resistor will have less turn-up/turn-down ratio. Right now, the 20 ohms will allow drop 0 to 5 volts based on 10 x 220 ohm resistors downstream. Lots can be changed on the existing board but we have to be careful in monitoring where the OP has headed and that he arrives at an intermediate point before we start talking about new directions or he will quickly get lost.

 

Then you may want to articulate exactly what you CAN change.

Admittedly, it was a bit like pulling teeth...

OP didn't want to disclose that it was an automotive retrofit for fear that the thread would be closed. It took a number of posts to discern the OP's intent...

 

Sorry about "pulling teeth thing" guys, old forum rules denies all talk about automotive things, didn't know that rules was changed, when I made first post about this. You all have been real good help to me figuring this out. I'll try ZVP3306A and 10x R1 R2 next week, as electronic component stores are closed at weekends.

Originally bulbs are fitted with red rubber condom filter thingy on them (don't know proper name for that thing, sorry, lol ), all panel lights in red at night, needles are also red originally. What I did is converted it to glow nice and white, that's why there is need for high brightness white LED's, result is like this, at day:

...and at night (needle color change is not a separate red LED, but a simple trick with color filter made with red permanent marker):

Problem with LED's not turning off looks like this:

In my country there is a Audi community forum, where people talk about Audi cars and show off mods, repairs etc. LED's allows to change dash panel to any color, just remove red filters from dials and solder whatever color LED in.

R7 rheostat looks like this, you can't scrap it or change it, because it's acts also as push button extension, when trip computer with auto-check is installed over it, its mounting screws are circled red (not my picture):

...and the back panel with PCB on looks like this (not my picture):

The schematic in first post SW3 is car headlight switch, witch also switches on all interior panel illumination, SW4 is ignition on +12V line (DIN72552 number 15) switch, so as ignition is on, if headlight switch is off, the clock, outside thermometer and trip computer LCD light on. When headlight switch is on, all illumination brightness ir regulated trough rheostat R7. LED1 actually is two leds, for the clock and for thermometer, trip computer has separate active circuitry in it, that doesn't take power from Q1 transistor.

 

I'll try ZVP3306A

Any general purpose P channel MOSFET will work. Vds needs to be 20V or more, current can be 100mA, on resistance not important, you don't want the threshold voltage to be too low or high and Vgs needs to be 20V+.

Thanks for the additional details. It's nice to know what we're working on.

 

Did the test using ZVP3306A and 10x resistance values, no cigar. Ended up with 10M & 4M7 for R1 and R2 (yes, megaohms), that did the trick on test board. Now I'll go and get instrument panel out of the car and try on car.

 

No, still not working. Apparently we are missing something. With 10M & 4M7 there should be no current going trough to light any LED, but they still lights up dimly.