A-axis thermal runaway

[WayOutWest]

Has anybody else had problems with thermal runaway on the A-axis NEMA14 motor?

I've killed two of them now. The first one went up in smoke before I started pushing the envelope on motor wiring (using Cat5-STP) so that doesn't explain it.

The motor works fine but it gets really warm. If I run for a few hours it gets too hot to touch. Run for a full day and it starts to smell bad. Eventually it stops working.

I have my A-axis motor current set really really low; after the first one died I made sure it was always just barely above stalling level. In absolute terms it's about 1/4 of the current setting for the X/Y axes and much lower than the Z-axis current level.

Like I said, two of them died and my third one started getting really hot. So I bolted some heatsinks to it and now it doesn't even get warm; easy solution but I still wonder...

The NEMA17 motors for the X+Y axis are pretty stock parts, same with the Z-axis NEMA17. But the A-axis NEMA14 is a somewhat specialty motor that I can't find anywhere other than StepperOnline, so I'm starting to wonder if it's not quite up to the same quality standards as the other three motors. Or maybe they never intended for it to be mounted axle-up (steppers seem to vent most of their heat through the side opposite the axle, which faces downward on our machines).

My theory is that the motor's internal heatsinking isn't quite as good as the other three, so it gets hot, which causes internal resistance to go up, which causes more heat at the same current level, which causes internal resistance to go up ----->>>> feedback loop ---->>>> fail. Even one or two copper heatsinks seem to break the feedback loop.

I'll be switching to a NEMA11 A-axis motor when I install my Juki nozzle changer (ye gads Robotdigg shipping is slow!) but I will be using the same NEMA14 motor to drive the auger pump on my flux+paste dispenser (long thread about that coming soon, I found a cheap one!).

[WayOutWest]

By the way, the more I read about how steppers work the more surprised I am that the driver chips don't do more-clever things like thermal measurements using at-rest resistance (using a low-freq signal to ignore impedence from the windings' inductance), missed-step detection from back EMF, digitally selectable current level, and a separate (much lower) current level for holding position rather than moving (higher current). All of these things could be done using ancient CMOS technologies like 180nm which are super cheap to prototype on, the >=110nm fabs hand out shuttle runs these days like they're candy. Sigh, so many projects, so little time...

[Mark Harris]

By the way, the more I read about how steppers work the more surprised I am that the driver chips don't do more-clever things like thermal measurements using at-rest resistance (using a low-freq signal to ignore impedence from the windings' inductance), missed-step detection from back EMF, digitally selectable current level, and a separate (much lower) current level for holding position rather than moving (higher current). All of these things could be done using ancient CMOS technologies like 180nm which are super cheap to prototype on, the >=110nm fabs hand out shuttle runs these days like they're candy. Sigh, so many projects, so little time...

Agreed... sadly most 3 phase brushless motor driver evaluation kits from the likes of TI/Maxim/LT all do back emf step detection for commutation. I'd like to see a lot more 3 phase steppers - Leadshine ones are amazing, 1.2deg step angle but lower torque than their bipolar counterparts - i suspect that the torque at higher rpms is better though.

Have you tried lowering your driver current?


Back in Australia I had to use two very cheap VGA heatsink/fans on each of my NEMA34 steppers


With 50c ambient in my garage, there was only about 20c temperature rise available before the steppers would overheat and the machine would start dancing around like a headless chicken (two ballscrews on the X axis... so when one overheated everything went to hell)

[Mark Harris]

By the way, the more I read about how steppers work the more surprised I am that the driver chips don't do more-clever things like thermal measurements using at-rest resistance (using a low-freq signal to ignore impedence from the windings' inductance), missed-step detection from back EMF, digitally selectable current level, and a separate (much lower) current level for holding position rather than moving (higher current). All of these things could be done using ancient CMOS technologies like 180nm which are super cheap to prototype on, the >=110nm fabs hand out shuttle runs these days like they're candy. Sigh, so many projects, so little time...

As a note on this, I forgot to mention the ST L6470 - http://www.st.com/web/en/resource/techn ... 255075.pdf
I'm thinking of making a board for 8 of these for another project, its a pretty nice SPI stepper driver, and it can detect lost steps plus handle the majority of the motion control for you. Seriously nice single IC stepper drivers with a lot of smarts - at least according to the datasheet, I havent actually tried one just yet. I have a project which needs 12 stepper motors, 8 of which are pretty small current requirement - the 2 large ones (x/y) will be relatively expensive leadshine digital drivers, but i'm thinking the others will all be run with these ICs. The board with these on would be right with the cluster of 8 motors, which will just need SPI and power going to it, radically reducing EMI from not having 8 stepper motor cables running all over the machine.

[WayOutWest]

Have you tried lowering your driver current?

I have my A-axis motor current set really really low; after the first one died I made sure it was always just barely above stalling level.

[WayOutWest]

As a note on this, I forgot to mention the ST L6470 - http://www.st.com/web/en/resource/techn ... 255075.pdf

Wow, no joke; thanks for pointing out that chip. Yeah, that's pretty much most of the features on my "gee why don't stepper drivers do X" list. Digital current control and automatic step-fraction-enlargement during high-speed movement for starters.

[Mark Harris]

As a note on this, I forgot to mention the ST L6470 - http://www.st.com/web/en/resource/techn ... 255075.pdf

Wow, no joke; thanks for pointing out that chip. Yeah, that's pretty much most of the features on my "gee why don't stepper drivers do X" list. Digital current control and automatic step-fraction-enlargement during high-speed movement for starters.

Yeah i found it the middle of last year when I was working at Solarbotics, everyone in the office was like "what are you looking it?" because I kept saying "wow, holy crap" to just about every paragraph. Certainly has a lot of features you dont tend to find in a) a stepper driver that price, b) a single IC stepper driver.

[Covert]

mine gets almost too hot to touch as well. i haven't looked into a fix yet.

[mawa]

Mine at first got hot too. But lowering the current with the potentiometer on the tinyG board as far down as to still have enough torque help. Now the stepper only get hand warm.

[WayOutWest]

mine gets almost too hot to touch as well. i haven't looked into a fix yet.

I definitely recommend putting heatsinks on it sooner rather than later.