Constant heat wire cutter?

[moreil]

I've built a CNC wire cutter (http://homecnc.org/wire) , but have had a little difficultly cutting very thin wing cores ( < 5mm thickness) as the leading edge gets melted by the wire.

The problem appears to be that the wire is moving in foam that's already been heated so it continues to heat up causing very poor precision.

So to that end, I'm thinking about building a circuit to maintain the wire as a constant temperature. I was planning to measure the resistance of the wire (as a proxy for the temperature) as use an AVR to pulse-width modulate the current so as to maintain a (near) constant resistance in the wire.

Something like this:

http://www.homecnc.org/images/thumb/8/81/2006-09-03_140757.png/1024px-2006-09-03_140757.png

The current and voltage across the wire is measured by the AVR (using the 10-bit 15k samples/sec ADC onboard), which then turns off the FET for a while when the appropriate resistance is reached.

Any comments? on the circuit?

Something I'd like to do is add a isolated interface so the temp can be set from the parallel port, but I've never used an opto-isolator before. Would the parallel port supply enough power to run an opto-isolator?

See http://www.homecnc.org/index.php/Constant_Heat_hot_wire_cutter
for the details I have to date.

[Cheesehead]

Use an op-amp as a buffer, perhaps?

[moreil]

Use an op-amp as a buffer, perhaps?


I should explain that I'm particularly worried about not wanting to share a ground with the computer. There's no guarentee that the ground for the computer and the hot-wire power supply will be compatible, so some sort of isolation is a good idea...

[Chippie]

If you are using a dc power supply of sorts to power the cutter, then it should be isolated from the mains anyway...(via a transformer perhaps)
just leave the 24vdc floating...i.e neither side is referenced to earth....

then there should be no issue...

[moreil]

If you are using a dc power supply of sorts to power the cutter, then it should be isolated from the mains anyway...(via a transformer perhaps)


I agree with the theory, it's just that when it actually comes to plugging this into my $3000 laptop, I get a little nervous... :)

I think my heart is set on opto-isolation. and I should probably learn how to do it properly anyway. :)

[Andrew McGregor]

Isn't USB supposed to be isolated?

[Chippie]

Configure the parallel port to generate a pwm signal that feeds the diode in an opto-isolator, use the transistor section to control the power reg section feeding the wire.....should be doable and gets around your anxiety... ;)

[moreil]

Configure the parallel port to generate a pwm signal that feeds the diode in an opto-isolator

So a parallel port can definately supply enough current to power the diode in the opto?

Note that using the computer to generate the PWM isn't terribly good. (That's what I do at the moment). The problem is that the temperature of the wire varies by more than 200 degrees depending on if it's just entered the foam, or if it doing a vertical cut or a horizontal cut, or a tight bend etc etc. Makes it rather difficult to do precise cuts. So the point of this exercise is to get a closed loop temp control.

[MatC]

Definitely a worthy aim to get temp control going :)

Why is IC2D output shorted to gnd? Not good. Output won't be at gnd just because you short the inputs together.

USB isn't isolated in any shape or form that I know of.

Parallel port (a few mA I'd think) probs won't drive an isolator, which generally needs 20-30 mA if I recall. Simple resistor and npn transistor to amplify and away you go. Use a battery to drive the opto for starters. Building (and testing) an isolating board before you throw 24v near your PC is def a good idea.

[moreil]

Why is IC2D output shorted to gnd? Not good. Output won't be at gnd just because you short the inputs together.

Good point. I should have left the output unconnected.

Updated circuit: (from the main page here (http://www.homecnc.org/index.php/Constant_Heat_hot_wire_cutter#Circuit) as normal)

http://www.homecnc.org/images/thumb/8/82/2006-09-03_172658.png/1024px-2006-09-03_172658.png

I'm actually using IC2D now as a fine current measurement. It's a 200x amplifier, with the base of the range set from a crude DAC (by running a PWM output from the AVR through a low pass filter).

Question: What would the ripple at R10 be if the PWM line is a 10KHz square wave from the AVR?