Stepper EM noise

I'm building a 3d-printer. I use linuxcnc and a Gacko 540 to control all the steppers. To control the heated nozzle i've built a controller using an arduino uno, AD595 chip, and the LCD shield from adafruit. The arduino reads in the temp from the tc, and switches on/off (using PID) a fet to supply power to the heating element on the nozzle. This all works quite well until i turn on the stepper motors. I can turn on the 540 with no motors connected, and all is well. As soon as i plug in at least 1 motor, and turn on the 540, the TC reading begins to jump all over the place, thus making control of the nozzle non-existent.

The arduino is powered from a laptop. The heater is powered from an external power supply. The 540 is controlled from a second computer, and the steppers are powered from a separate power supply as well. It certainly feels like its a noise issue from the motors. I've replaced the motor wiring with shielded cable, and have tried filtering the output signal from the ad595, but still the interfearance persists.

and advise/suggestions would be greatly appreciated.

thanks

How far away is the thermocouple sensing point from the AD595 input terminals? Does the wire run in a bundle with the extruder or other stepper motor cabling? If so, it is possible that noise is being picked up by the thermocouple cabling, and so filtering the AD595 output may be ineffective.

I think your absolutely right on that suggestion. I moved the ad595 circuit to the nozzle head, and ran a shielded cable to the controller board... and that seems to have minimized the problem. It also looks like i need to run shielded cable to my steppers (which i probably should have already been doing). I think the long t/c extension wire was acting as an antenna for the EM. I'm going to replace the remaining wiring today and give it a wirl.

thanks for the thoughts!

well that didn't work. The noise went down... but still not to a level where the temp control will reliably control. The odd thing is that with one motor plugged in. i get the interference, but if i run that motor the interference goes away, and the t/c measures fine.

Just thinking out loud here - I'm guessing that one lead of your thermocouple is 'grounded'. I wonder whether wiring it up to the AD595 in a way that keeps the differential inputs rather than grounding one lead of the thermocouple might improve things, by keeping any induced noise as a 'common mode' signal that would be rejected by the differential input.

Do you have access to an oscilloscope? I wonder if you could identify with greater confidence where the noise is entering the system by using the scope. Of course, sometimes the scope probe itself can introduce noise, and you may have trouble figuring out what 'ground' to connect to your scope probe.

Using a common ground for shielding is probably difficult to do in your case. The G540 is optoisolated, so you could probably establish a common ground on the output side of the G540 for the stepper shielding, any limit switches, and the like. But then what do you do with the metal frame of your machine, the stepper motor housings, and so forth? The Arduino's ground is likely to be common to the 120 VAC power ground, which would also be the ground for the PC and the input side of the G540. Whether or not the heater power supply and nozzle share that ground is anyone's guess.

But if you could characterize the noise better, perhaps you could knock it down with some kind of low-pass filtering, maybe a ferrite bead or two, for example. The thermocouple signal is bound to be a slowly-changing one, whereas any noise is probably going to be relatively high in frequency.

I've looked at the various signals, and there was some faint noise on the TC in to the 595, and lots of noise on the output. which would make sense. I put a low pass filter (100k R and 1nf Cap) and that smoothed out the signal. But now as i turn on each stepper there is an offset induced. So if the temp is 11 deg c with no motors on, with (1) motor the temp jumps to 13deg, and with (2) motors on temp jumps to 18deg. I'm assuming the offset is a result of a stronger EM and thus the high freq signal will impact the filtered signal (?). I'm no electrical engineer so i need to do some more reading on filters and such.

Where in the circuit did you place the low-pass filter?

I'm wondering whether a ferrite bead on the input might be better than an RC filter.

it's on the output from the 595. I tried a ferrite bead on the t/c input wire, and it didn't make a difference. It was a bead from a monitor cable, so it seemed to be oversized. I don't know if that makes a difference or not.

Ferrite beads may have different characteristics that could impact their effectiveness in a particular application. Although I am not certain, I would not be surprised to find that the diameter of the hole in the bead may have an effect. Maybe you could try wrapping two turns of the thermocouple wiring through the big hole to increase the effectiveness.

Have you looked at this AD595 application note? http://www.analog.com/static/importe...tes/AN-369.pdf

They have a section about 'noise suppression' that includes a test for determining whether the noise is entering via the thermocouple leads or via the power supply. Also note that figure 4 shows an RC filter on the input terminals.

I looked through the spec sheet, but never the app notes. Thanks for the link. That was very helpful.

So i did the tests suggested, and the noise is definitely coming through the tc extension wire. I tried wrapping the cable around the core several times, but it didn't have an affect. The notes suggest using shielded tc extension wire, so i'm going to try that. I have to order it though. I will try some of the RC filtering on the tc input, but with such small voltages i don't know how effective that will be.

An RC filter's effectiveness isn't dependent on voltage. Make the time constant about 10mS to 100mS; the frequency that is interfering is 40kHz so the filter will have terrific attenuation for that noise. The reason the problem "goes away when the motor is moving" is because the drive switches much less frequently when a motor moves. Also be very scrupulous about how you wire the thermocouple to the AD595 because noise can also be the result of ground-loop currents.

Mariss

Attachment 209770So i put the filter on the incoming tc signal as suggested and it looks like that fixed the issue. I can control the nozzle temp within 2 degress with all the motors running. So thank you all for the suggestions, advice and education. I'm posting a pic of the the first print out of the unit. not great... but considering i haven't messed with any of the slicing software programs settings, i don't think it's too bad.

I used a 1k Resistor and 0.2uf cap on the +/- legs, and a 10uf cap across the +/- t/c wires. I haven't done the math to confirm these fall in the values suggested.... it's what i had at arms length.

anyways thank you for all the help on this.

200nF and 1K gives a 200uS time constant. That should get you 48db attenuation for the 40kHz noise which should be plenty. The 10uF cap really doesn't do anything (no harm, no good) because your TC resistance is so small (<< 1 Ohm). Make sure you place the RC network as close as physically possible to your amplifier inputs.

Mariss

Glad to hear that things are working...