next topic »

[on:]

Hello Forum.

I believe I have bitten off a little more than I can chew with my first Arduino project.
Most of my project is complete as far as the logic goes but my elementary knowledge of analog circuits has a
gremlin rearing its head.


Here's my projects premise:

I'm using the arduino to sample the voltage between the torch and work piece on my CNC plasma cutting table and
using that voltage to adjust the height of the torch. DTHC (digital torch height control)
The plasma cutter is constant current and as the torch introduces heat into the metal it warps and varies the distance (as well as the subsequent voltage of
the torch height relative to the material.
I'm trying to keep the voltage constant by continuously adjust the steppers on the z-axis (torch axis).

Here's my setup thus far:

    * probes from the work clamp(+) and electrode(-) send roughly 160-200vdc to a 50 amp bridge rectifier - I'm following someone elses lead on the rectification
    * rectified signal is then passed through a simple voltage divider
    * signal is then run through a low pass filter consisting of a 330 ohm resistor and 100uf/25v electrolytic capacitor
    * MCP3201 12 ADC sends value to arduino via SPI  

When arduino gets the value it runs some simple if/else functions based on height preferences relative to the material being cut.
When done with the calculations it encapsulates the control information and voltage information into modbus holding registers and sends them to my CNC control software to be read and acted upon.

Here's my problem:
As a beginner my understanding of the analog side of things is extremely limited.
I know my plasma is very noisy and has high frequency signals that are more than likely making there way into my ADC.
The reference voltage that I get from my ADC jitters.
It's not a huge variation but it's more than a couple volts and that equates to more slop than I want to deal with.
It also limits the speed at which I can cut.

I've been reading about two stage Pi LC low-pass filter because that's what commercial units use with success.
I just purchased an oscilloscope and could acquire some data but not exact sure what I'd be looking for just yet.

Does anyone have experience with these filters and willing to share some knowledge?

I've been studying these circuits and trying to visualize what is transpiring underneath but I'm in over my head.
Here's a pic of the circuit on the left and what I think is happening on the right???

This is the torch and workpiece:

[Reply #1 on:]

It sounds like you need a multi-pole low pass filter to more agressively filter out high frequency nioise.

You can do this either in the analog domain or the digital domain. For the latter check out the filter library someone posted here recently http://arduino.cc/forum/index.php/topic,52681.0.html

For analog filtering an active filter is needed for multi-pole designs really - or inductors.  RC filters cannot create arbitrary filter pole designs as the poles interact too much.  Active filters use op-amps.

[Reply #2 on:]

It sounds like you need a multi-pole low pass filter to more agressively filter out high frequency nioise.

You can do this either in the analog domain or the digital domain. For the latter check out the filter library someone posted here recently http://arduino.cc/forum/index.php/topic,52681.0.html

For analog filtering an active filter is needed for multi-pole designs really - or inductors.  RC filters cannot create arbitrary filter pole designs as the poles interact too much.  Active filters use op-amps.

Thanks for the link Mark.

I'd like to handle this in the analog realm and keep the processing load as light as possible if at all possible.
I know that the circuit that commercial THC sensor units use incorporate inductors.
Think the next move is to see what the plasma is actually putting out as far as a waveform goes.
Any advice as to how to acquire the waveform?
My oscilloscope probes are rated to 600v but I'm vaguely remembering reading about not using the oscilloscope ground when testing welders/plasma with high frequency start.

[Reply #3 on:]

Also, divide the problem..

- First use some known-quiet voltage sample like a 1.5v battery and make sure readings are stable.. There may be so much HF noise that Arduino and it's power source are being affected. If that's clean and stable then look at waveform where you are sampling..

Interesting problem.. Had trouble initially sampling data from Plasma Deposition system years ago... Lots of energy about!

Regards, Terry King  ..On the Red Sea at KAUST.edu.sa
terry@terryking.us

[Reply #4 on:]

Yes, I think there will be a lot of RF flying about - that's the first thing to filter before anything else (opamp circuits will misbehave in odd ways if high levels of RF are on their inputs).  Definitely worth having an RF choke in line with the signal as first line of defence. And ceramic filter capacitors will perform well at RF.  Whole of low voltage circuitry needs to be in a grounded metal box too!

[Reply #5 on:]

I skiped all the post after the first one

Um u could go the low tech way And use training wheels on the touch head . Set a stop so it can only go so low. And use a spring to push it down.That way it will allways stay at the same distance from the metal. Also as a side note if ur warping the metal theres a few things wrong with ur setup. At my last sheet metal job we take turns running the burn out table same idea of what u have but the table is 16x8. burning out certin metals is toxic so we'd take shifts.

That plasma cutter only moved up and down for the thickness of the metal that's it.
If the metal is warping u have a heat/speed/air flow problem toomuch psi wount cut right and ull have to go slower.. The only metal that would warp for us is very very thin alluminue(sp?lol) it more or less tin foil. We would just put small weights on it as it was cutting it out.

It sounds like u have a heat setting problem or speed problem cause only very very thin stock should do that.

[Reply #6 on:]

I hope I didn't piss off any one with my post. It's just I use these machines for years. Went to school for this trade ect. So I have a great understanding how these machines work and so on. For the most part the big burn out tables CnC controlls And the cutter controlls are not linked. Kinda true it just tell the tourch when to be on when to be off. That's why I suggested it was a setting on the machine . Like I said the metal shouldn't warp if u got those settings set up right.

But I have used hand held cutters that the only thing u could adjust was the air psi. So if that the way ur machine is I see why u would need to make it move up and down and why the metal is warping. I should of asked about ur machine befor putting my 2¢ in.

As a side note what will u be useing this machine for ? What type of metal are u cutting?

[Reply #7 on:]

Hi. How is it going with the torch control project?

[Reply #8 on:]

Note the date of the first message.

Pete