Why does testing with multimeter fix my circuit issues?

Question

I am using this heater along with this thermocouple and this amplifier and am just trying to be able to set and hold a consistent temperature. I have the heaters connected to a transistor circuit and I'm running a simple PID loop using a Raspberry Pi so that when the temperature is lower than the setpoint, the pin connected to the transistor switches to HIGH and the heater heats up and vice versa.

Here's my schematic:

I'm having a really weird issue however where when I first start running the program and set a temperature (that's higher than room temp), I see the Raspberry Pi pins pulled high as expected, but the heater's temperature doesn't change (i.e. they're not getting any/enough voltage). HOWEVER, when I go to measure the voltage difference across the two heater wires using my multimeter, it suddenly starts working (before you ask, the multimeter pins don't touch anything other than their specific wire lead, and those don't touch each other or anything else at any point). It doesn't matter how long I wait before I set the temperature or before I measure the voltage, it always just triggers everything to work perfectly.

I've tried using other multimeters and the same thing happens. I have zero clue how measuring a voltage with a multimeter does anything to impact the functionality of my circuit and don't really know where to go from here. Any ideas?

Answer 1

You have a nice simple circuit, but real-world design needs some extra touches, to prevent problem like yours. I advise:

(1) Gate resistors - Use much lower resistance for R1 and R8, to lower their nodes' impedance, make them less susceptible to interference. I don't see why they have to be 100k. The RasPi should be fine driving 10k.

(2) Input resistors - Add series resistors to the MOSFET gates, small resistance like 470 ohms, to low-pass filter interference. and slow down turn-on (acting with MOSFETs gate-channel capacitance) just a bit.

(3) For connectors to/from MOSFET module & heaters, etc. Use more than one pin for the GROUND, If the currents being drawn are more than a few milliamperes. The contact resistance of one pin may not be low enough to prevent signal-ground-bounce.

(4) Heaters - are they pure resistance? They may have inductive property (if wire wound design). The the turn-off may cause inductive voltage "kick" on the MOSFET drain, and (thru internal capacitance) feedback to its gate & interfere with the turn-off. Some MOSFETS have an internal diode to protect them from this, but if yours do not, then add them (across drain & source, with diode "pointing up" to drain, for N-channel MOSFET).

(5) Gate capacitors - Add a small capacitance, like 47pF, from each MOSFET gate to ground, placed physically close to the MOSFET. This could further help for reasons in (2) and (4) above.

Regards, RS