Yes, it's the vibration nose transmitted through the house that I'm concerned about. I do lots of my printing after my wife and children have gone to bed and a couple of them are very light sleepers.
What is your VREF set to now and what did you have it on when you were getting drift? Also could you just confirm you're using A4988s with nema 17 motors?
The IRLB8743PBF is a MOSFET. They are the two switches which are near the heated bed and hotend terminal on the RAMPS board and look like this [
uk.farnell.com]
They allow the pins on the arduino which can only supply 10s of milliAmps to turn on and off large currents and also at voltages which the arduino can't handle.
The MOSFET comes with its own resistance when on, called RdsOn in the datasheet, and this is the number you use to calculate how much heat will be converted from electricity in the MOSFET and need to be dissipated from the MOSFET, using the formula heat power to dissipate (unit Watts) = RdsOn x Current squared (^2 in spreadsheet formula). The MOSFET is only rated to a certain temperature so if you can't adequately dissipate heat the temperature may rise close to this limit and start to damage the component. You can add a heat sink in good thermal contact with the MOSFET (the hole in the aluminium tab is so you can screw it to a heat sink) and although this is standard practice for stepper motor drivers, in the brief period I have been following RAMPS I haven't heard of people doing this for their MOSFETS.
Given that the power is proportional to the square of the current and only directly proportional to the RdsOn, the current is the most significant factor in this issue and I understand that the heated bed currents are much higher than the hotend (and certainly the fans) so we ought to focus on the heated bed first.
You could reduce the current going to the heated bed (through firmware) and insulate it better (underneath mainly!), or accept longer heating times and lower maximum temperature. I have just bought a Mk3 bed with the shiny aluminium side facing up, which will minimise radiant heat loss. Enclosing the print space might allow higher air temperature in contact with the heated bed and reduce convective heat loss but then we're into a balance with print layer cooling requirements etc. The print cooling fan will also have an impact. So what else can we do to reduce the thermal stress on the MOSFET (assuming you also don't want hot components near you or a thermoplastic enclosure)?
The stock RAMPS MOSFETS are designed to be cheap and don't have a great RdsOn, whereas the IRLB8743PBF has a very low RdsOn at the voltages and currents we are interested in (RdsOn varies with current, voltage differences between MOSFET pins and ambient temperature). Here's some good research on the best MOSFET for our needs [
reprap.org]
You will notice that the pulse width modulation frequency is also relevant and I don't know what marlin uses but arduino pwm functions are usually about 400Hz IIRC, so on the end of the spectrum which is good for MOSFET switching. The datasheet attributes which are relevant to the frequency impact on heating are the times to switch on/off, given as rise time and fall time (usually in nanoseconds).
There are a few other factors to consider when selecting a MOSFET for RAMPS, max DC voltage, max current (note this can be limited by the package or physical size of the MOSFET and need a heat sink to reach the level stated in the datasheet - our needs should be fine with a TO220 package without a heat sink) and VgsThreshold, which is the voltage needed from the switching controller to start turning the MOSFET on (for N channel MOSFET) or off (for P channel MOSFET). This voltage is just the start of the turning on/off so a good rule of thumb is to double this to be confident that the MOSFET can be fully switched on/off by your controller. The IRLB8743PBF has VgsThreshold of 1.8V and we're confident our 5V arduino can achieve double that (3.6V) when it switches a pin to a high state.
Sorry if that was too much detail or patronising.
happy
