I the cheap nema 17 are good, I have two printers still in development, one's on and off working because I tried cutting corners, one is just a disaster. My FabLab owns a Stratasys uPrint SE and I use that as my standard. If I can beat that precision and consistency, I've made something good.
I would line to see how precise I can manufacture a part. On my Precision Machining and CNC Automation course it's EASY to get +/-0.005" it's actually pretty easy to get +/-0.002" and our cnc machine is can consistently get +/-0.0002" which is what I'd like to mill out my next printer parts on. Probably going to go with a corexy style with ball screws on every axis.
The point is, I want to get this technology as close as I can to that so I can easily and quickly prototype an inkjet based 3D printer that I'll use the accurate motors I have access to, because I can.
I advise against attempting it unless you have the right equipment. Here's the technique I use to replace driver chips, and it should work with processor chips too. You will need:
Thermocouple temperature measuring device. I use the thermocouple attachment on my multimeter.
Hot air SMD desoldering tool with a square nozzle about the right size for the chip.
Vacuum pick up tool. Very cheap on eBay.
Fine tipped tweezers
No-clean flux pen
Fine-tipped soldering iron
Strips of corrugated cardboard covered in Kapton tape - probably not needed when replacing a processor
Use the strips of Kapton-covered cardboard to mask Molex connectors and other sensitive parts near the chip to be replaced from the hot air. Place the board on the electric hotplate with the thermocouple probe underneath it and adjust the hotplate to get a reading of 100 to 120C on the thermocouple. Wait several minutes for the board to come up to temperature. This makes desoldering easier. Heat the chip with hot air at about 260 to 300C until you are quite sure that the solder has melted all round. Gently lift the chip off with the vacuum pickup tool. As Tony says, it is easy to lift the fine traces off the board if the solder is not completely molten all the way round.
To fit the new chip, apply no-clean flux to the pads. Drop the chip on the pads, the right way round, using the vacuum pen. Nudge it with tweezers until it is perfectly aligned with the pads. Heat the chip with the hot air until the solder is molten again (about 20 seconds), then remove the hot air and gently press the chip down with the tweezers, taking care not to shift it sideways. Remove the hot air and turn the hotplate off. Allow to cool.
Check under magnification that all the pins are soldered down and there are no bridges. Touch up any bad joints using the flux pen and fine tipped soldering iron, and remove any solder bridges using solder wick. Finally, use isopropanol spray and cotton buds to remove surplus flux.
Based off of your datasheets, your Z / extruder motors in particular have an extremely high inductance and would probably benefit from a much higher supply voltage. (Higher than the TMC2660's support) I wish that more (Or any, even) of the single IC stepper motor drivers supported 48V.
If you have continue to have speed issues with the Z / extruder motors at 24V, I'd suggest going with a lower inductance motors for these two - Something like the 34mm one from here:
When picking motors, you essentially want the one with the highest phase current for the torque you require. (Within the limits of the drivers you're using, of course)
There are no gcodes to send SPI signals. I'm not sure that it would be practical to define any, because SPI peripherals vary so much in what data they expect and what they return. Maybe it would be practical for simple output-only devices.
If you are not using thermocouple or pt100 interface boards then the easiest place to pick up the SPI signals is probably the header that the daughter boards plug into.
Interesting! I would be surprised if static electricity could be sufficient to blow the VSSA fuse because I wouldn't have thought there was enough energy in a static discharge. I will research the topic. It does seem very plausible to me that driving filament through a Bowden tube and then extruding it can build up static charge. Perhaps it would be a wise precaution to ground the hot end metalwork to prevent charge building up. To do this safely would require a separate ground wire to be run to the hot end from the ground side of the power supply output.
I just remembered, there was a bug in the GNU C library that caused older versions of PanelDue firmware to freeze sometimes when used with later versions of RRF. If it happened at all, it was duringRRF start-up. Pressing the Reset button on PanelDue resolved it. If you are using an old version of PanelDue firmware, then that might have been the problem.
My 20mm nema 17 gets quite hot probably in that range, it's not an issue but you don't want printed brackets softening so how about a mk8 extruder aluminium heatsink bonded to the back. Runs fairly cool now barely any mass added.
Good idea DJ - quieter than a fan too.
See http://reprappro.com/documentation/ormerod-2/wiring/#Fitting_the_cover_and_wire_routing for how RepRapPro did it.
This is nice as you see what it could look like.
The interesting image from that page:
And its description:
"All the wiring looms use shielded cable, to reduce the effects of Electro-Magnetic Interference (EMI). It is important that the shielding makes contact with ground, by earthing them through the metal case. Each wire has a slot in the case; bend each cable so it sits in the slot. Make sure that there is not sideways pressure on the housings, where they are plugged onto the pins of the Duet; they can loose their contact with the pin, and make a faulty connection."
So the light comes on without the cables attached as well. There's nothing close to the back of the board at all. And I can't detect any short in the bed wiring either, cables and connections look fine and resistance is 1.8 Ohms, which I guess is normal. Also , I didn't touch anything related to the bed when installing the second fan. Oh, and I run the fans on the internal 5V supply, if that might be a clue?
I have a multimeter but I don't know what to look for if I should check the mosfet and driver chip.
I have seen else where on the forum that you've said that a 2.8A motor is reasonable for a wifi board. When you say that, is that at the upper end of the limit, or would the board just about cope with a 3A motor?
So it was just the trigger level? You could in theory just lower that?
Also I'm interested in the outcome of this. I've got issues with 3Dlac and the IR sensor.
Well I'm pretty confident that the switch will work - that's what it's designed to do. However, it is a contact switch so you can't use it directly on the bed unless you use some mechanism to deploy it. That is to say, you can't have it permanently lower than the nozzle because it will be scraping on the printed part, and you can't have it higher because the nozzle will touch the bed before the probe does. My plan is to use it above the hot end mount with the nozzle itself being the probe. Check my sliding mount thread to see what I mean.
So it's the sliding mount part that could screw up the repeatability. I've nearly got a working prototype but I have to change the entire XY carriages to test it.
When it first happened the E0 stopped working, but the bed continued to work. Stupidly I moved the E0 thermistor leads to the bed connector and that stopped wortking too. I assumed I blew the input because the bed was still working? Anyway I just tested the fuse and it was open, I jumped it and with just USB and a thermistor connected all the temps are woking? I'm still a little puzzled why I thought the bed temp was working when I first damaged it if the fuse was already blown. Thank You
Looks like your connection to Duet3D was lost, please wait while we try to reconnect.