Better heated bed PCB design?

Electronics is not my specialty so perhaps this is a dumb question but here it goes.

I have a 6" Prusa i3 from MakerFarm which came with their PCB heated bed, apparently using Josef Prusa's MK1 design with MakerFarm silkscreened on the top. It uses resistance heating and a very long trace path which heats up when you pass current through it. The amount of current is regulated by one of the MOSFET parts on the RAMPS board with a heatsink attached. Right?

The problem is this: the bed does not heat evenly. It seems like a simple problem of thermal distribution. I use 1/8" plate glass over the heated bed. The center of the bed (where the thermistor is located) gets at least 15C hotter than the edges, with an even greater difference toward the corners. I'm measuring using a multimeter + thermocouple taped to the glass.

I hear this often: you won't get flat prints from ABS, there will be warping without a heated chamber. Looking at the large variance in surface temperature, I'm wondering if there might be a way to design the board with unevenly spaced traces (of even width).

It's also possible the board I have is poorly manufactured - there are apparently some boards out there constructed from plated rather than clad copper. Or perhaps the MK1 design was meant to use with kapton tape covering rather than glass?

Thoughts?

If traces are equidistant then i guess its normal for the middle to heat up more than corners because its heated up from all sides, sort of speaking it doesnt get so much convection with cool air since all nearby air is already heated up. If the middle would have a big gap, perhaps it would somwhat level the temperature. Or another idea would be to have traces more spaced (from each other) in towards the middle of the board. Overall, probably would be ideal to have them gradientally spaced towards middle, incrementally, so the heating effect would be better in the corners and lesser in the middle. At least this is how i see this thing. Just to be extra explicit, the copper traces need to remain equal width otherwise would probably burn at their smallest width. The space im talking about is the wasted space in between the copper traces.

Yes, I've considered putting something around the sides to reduce heat loss through convection. I've also experimented with putting aluminum foil (shiny side up) on the non-contact wood platform below the PCB to reduce heat loss on the bottom.

That's pretty much what I was thinking; increase spacing between traces toward the middle slightly (where the thermistor is taped). I do understand the need to keep the trace width perfectly even throughout, otherwise the thinnest part will have higher resistance and will heat much faster (I think the Joule heating goes something like the square of the resistance).

Thanks!

On a second thought, maybe it should be even less in the center than margins. One because the center aslo gets heated plastic deposits, and also you were measuring it static, but when bed is moving the corners loose even more heat - dinamically during printing. So perhaps have the center even colder than margins.

But to print abs, its ok anyway. I have been printing in abs with perhaps even a bigger difference, for couple of years and only very big parts have warping issues. In which case i put a towel around the printer leaving on the top open. Also what does increase warping for me (a huge difference) is air draft(?) - current, e.g. if i open window or a door, especially in the winter. So i would think a chamber would help avoid this even if its not really activelly heated, its just better for air currents. I guess it still makes convection but with sort of heated air (instead of cold one), so the thermal effect is less significant.

I think the next thing I should pursue is building an enclosure. I expect that it will get hot inside there and I should be able to provide some airflow for the electronics.

But with regard to measurement, I actually taped a a thermocouple to the glass and measured it while printing, so it was dynamic. Your point about movement most likely causing more temperature drop by convection on the edges is well taken - it's possible that it all becomes less of an issue if enclosed and therefore protected from drafts. Building an enclosure is much easier than etching a PCB so I can try that, measure again and compare.

Many of my prints have been in ambient 35C (it's been a hot summer and no AC in my garage) and they still warp if slablike (covers a large area which extends beyond the center of the bed and more than around 5mm in thickness). So I'm inclined to agree that an enclosure should make a big difference because there must be a huge amount of heat lost through convection. But twiddling the spacing of traces may still be worth pursuing.

When you enclose a printer, you should place as much of the electrics outside of the enclosed area as possible since most things don't like temps greater the 75C and prefer even lower temps, but the actual printing likes temps closer to 105C.

Yes, I would definitely not want to run electronics over 50C ambient, less if possible. I had no idea just enclosing the printer could get things that hot but now that I think about it, it makes sense. Running electronics above 70C ambient is usually reserved for test-to-fail scenarios or if you're trying to handle extreme conditions like automotive dashboards.

There's no easy way to avoid getting the stepper motors hot but as far as I can tell they should operate reasonably from 0 to 100C. Even with a 110C heated bed I don't see the air temp getting over 80C or so. On my Prusa i3 I can see how I could exclude the X, Y and Z axis stepper motors from the inner part of the enclosure (or at least their housings).

I'm sure others have already gone through this and there are probably some good enclosure designs out there already...

Found the "essential reading" links for this topic on the wiki:

Heated bed general discussion and applications
Traumflug's page on theory behind heated bed design, with formulae

The discussion on using metal surface for heated bed gave me the idea of using a thin layer of a good thermal conductor like copper or aluminum (expand too much) under glass to improve heat distribution.