UC300 >> Analogue Input Specs

[Robertspark]

Balazs,

Could I ask what the Analogue Input Specification are for the UC300eth and UC300usb (presume they are the same of similar)?

The UC300eth manual was not clear what the voltage limitation was for the analogue inputs, although there is 12v available on the 16pin AI/O IDC connector on the standard 5LTP interface board.

Maximum Analogue Input voltage?
Samples per second? (is this loop or buffer speed dependent, screenset)
Is this one shot or averaging sampling?
sampling resolution (10bit, 12 bit, 16 bit)?

Thanks for the guidance & time in advance as always,

[would probably be useful to list the analogue output spec's too, and have both added to the manuals]

[cncdrive]

The full range the ADC can read is 0-12Volts.
The ADC has 12 bits resolution.
The ADC has about 20kHz analog bandwidth now, because of the filters built into the -5LPT motherboard.
The ADC max. sampling rate is 18Msps for the UC300ETH, I think it is 2Msps for the UC300USB but will have to look that up.

The sampling is done with avaraging a few samples and also the sample and hold time is adjusted to fit the bandwidth of the -5LPT to make a more precise sampling with a longer SH time thus again averaging.

[Robertspark]

Thanks Balazs, that's impressive and on paper it could do THC, given the maximum open circuit voltage of the 1:50 voltage divider is 7 volts (for the hypertherm PMX 45.... suspect all the other hypertherms up to 105 will be similar).

12 bit resolution (2^12) = 4096 bits, therefore 12/4096 = 0.002929v/bit.
therefore for a 7V FSD = ~2,389 bits
given the cutting voltage is around 75 - 160v ~85v (~580bits range)...

suggestion if looking to use THC internally would be to amplify and offset the signal which would improve the control range.

.... hence my suggestion for 2 ADC input channels, one with a low resolution, and a second with an amplified offset cutting voltage range

[cncdrive]

Yes, the resolution is enough I think and also the bandwidth of the -5LPT analog input channels should be enough.
What about the sampling times?
I mean what sampling time would be optimal?

[Robertspark]

disclaimer to anyone reading this .... I don't work in the plasma industry and this is a hobby and is gleaned from various sources of information, if you know better / more definitive information, please share it, educate, guide and correct please

Sampling time, two things at play here....

1) feedrate > the highest feedrate that I've seen listed anywhere was 600 inches / minute (although I'd be surprised from my own build if there are many machines actually able to plasma cut at this feedrate [not rapid motion, cut!] because of the acceleration requirements to get good quality cuts on corners / changes in direction at the lower end of the cnc market.... lets face it these are not more likely not $100K machines running uccnc / uc controllers...

600 inches / min = 10" / sec
15,240 mm / min = 254 mm / sec

edit: eg. hypertherm powermax 105, production settings, unshielded consumables, @ 45A stainless steel = 500 ipm from the manual, powermax 125 will be faster, although I suspect most uccnc / uc controller plasma users will be in the 45 - 65 or 85A range, and those in the higher ranges are likley to be cutting thick plate at a lower equivalent speed to us smaller plasma owners (mine lives at about 5m/min, but I'm sure others cut good cuts faster than I).

2) voltage > the plasma cutters are a constant voltage power source.... although the current will vary ..... technically both will vary in my opinion as the power is not going to remain constant at ~6kW [for a PMX 45] the whole of the waveforem given they are fed from a sinusoidal 50hz (in the UK / most of the world) , bridge rectified to DC, complete with capacitors for DC ripple control, power supply (I've not put an oscilloscope on my PMX45 to check out the waveform).

PMX45 has 2 x 5600uF capacitors in parallel (from the service manual, 230v/ single phase variant) , I calculate the ripple at ~28v, but the DC supply voltage will be ~ 324v, and full draw current of 30A, cutting voltage after the IGBT [dc] would be ~132v.... so maybe it's a very smooth waveform... I'm just not sure and willing to be educated but ....

although it appears I'm plugging hypertherm, they seem to be the only manufacturer that list comprehensive operator and service manuals, and the answer to the voltage lies within the service manuals wiring schematic diagrams....
Powermax 45, the power supply is IGBT, with a switching frequency of 36khz (last page, service manual).... hence it gives you an idea of the waveforms you're likely to get. Although some of the lower end ones I suspect are still transformer power supplies.... my old Powermax 350 I think ran at 100hz (could see the grooves in the plate cut as ripples..... although I've still got something similar with the PMX45 although fainter and much closer together).

I may if I get a chance see if I can get a plot on the scope and publish it for an idea of the waveform, Keith, Terry or anyone else passing may already have one they wish to share

[cncdrive]

Hi Rob,

Thank you for the info and sharing your ideas.

My thinking was very similar about the sampling time as yours.

The 2 main things I thought about were to get an idea about the max. cutting speeds people need and what precision do they expect and then make a calculation for a sampling time which could respond to the THC signal change within less than the required tolerance. Ofcourse this is a course calculation only, because we could not see an exact related result, but could get some basic idea about the sampling time needed.

The second is exactly what you mentioned, the PWM frequency, because we will sure need only fractions less sampling time than that, so it is the upper limit/constraint for sure.

As with everything else on the world we should find a good mid-value, because with too fast sampling we would measure the switching noise.
And with too slow sampling the reaction of the Z axis would be too slow.

Unfortunately currently we do not have a plasma unit to measure on, but even if we have one I'm afraid that different models have different properties, so measuring out one unit might not give a fully clear view.

My other idea is maybe making the sampling time adjustable would be the best idea, but then still the question is what is the absolute maximum that is required?
A few hundreds of Hertz or 1kHz or more?
It is an important question, because if we want to implement this functionality we should clearly see this to be able to even plan,
because the different sampling time requirements may require different type of implementations in code.

[Robertspark]

If you have a look at reply 167 here, the sampling rate is about 20khz. (This is the same one as I have)

https://www.machsupport.com/forum/index ... #msg221333

The razordance is 1khz , post 102 here....

https://www.machsupport.com/forum/index ... 0.100.html

I know it's a lot of pages.... but if you can spare the time, there are some real gems in that discussion... not all the pages or posts are thc but quite a few are in the middle.

With the likes of terry, Keith sterling (creator of the razordance), and hood.

[cncdrive]

Thanks Rob for collecting these threads for me. I will read them.

For now with my current knowledge in this topic which knowledge is yet far from complete I think that the 1kHz is probably closer to what is needed than the 20kHz.
I think that the 20kHz value would probably introduce more error in the readings than what we want to see as it is too close to the PWM frequency you've mentioned previously.

[Robertspark]

I would suggest maybe a little more than 1khz, the reason behind this is as follows...

If you're designing for the maximum cutting velocity of 600 inches / min (10"/sec or 254mm/sec) 1khz sample rate would provide one sample per 0.254mm

I've not been able verify the following (but I'm sure someone can chime in and advise) other than from the following source:

A 1% change in voltage (100 to 101 volts) is equal to several thousands (typically .015 or more) of arc gap change so the THC must be able to see and act on a small change in a large number of volts.

mid page 6/57 https://candcnc.com/images/Manuals/LCTHC-Manual_REV9.pdf

The guidance I've found from around 2-3 documents from THC manufacturers is the gap per volt is in the region of 0.015 to 0.020 " per volt (0.381mm to 0.508mm per volt)....

Now the trick comes in if you have a fast enough moving Z axis..... if your sample rate is 0.254mm travel distance, and lets say your hysteresis is 1 volt (for simplicity, as the calculations can be adjusted accordingly) ..... your cut height is using a hypertherm 45 is ~ 0.08" (2.0mm) to 0.02" (0.508mm) [45A & 30A accessories]

hence at cutting speed you have only 4 to 1 volt of permitted error before ***crash*** [minimum setting of hysteresis on the Proma is 1Volt.... and it uses dry contact relays for thcUP, thcDN and thcOK....]

Edding cnc (default settings below) use 0.5V hysteresis... Neuron from the website 0.25 volt (0.125mm resolution), hence 0.5mm / 1volt ties up as well http://neuroncnc.com/

At 600"/min travel speed, the maximum bevel you could deal with would be

If the sample rate was increased to 2khz, at 600"/min, the travel distance per sample would be 0.127mm travel distance per sample.

I would suggest, given there needs to be some filtering and hysteresis.... aim for 4khz.... but if it was adjustable... then it would allow for tuning relative to your Z axis response. a little, and also to your plasma cutter, but you're probably not likely to require anything above that considering the majority of your customer base would be in the 2/3rds 600"/min cutting speed at best (400"/min 10160mm/min) to 50% 600"/min (300ipm 7620mm/min)..... hence at 4khz sampling you are sampling at 0.03175mm ..... which is very high is you're setting at 0.25v hysteresis (0.125mm Z axis travel) because you would get 4 samples within the hysteresis window at a matched Z axis feedrate to the X and Y blended.

It you input your Z axis speed into the calc, then you could consider this as part of the filtering along with the blended feed rate

This always impresses me....https://www.youtube.com/watch?v=nDj5SSle-ng

--------------------------------------------------------------------------------------------------------

There is an interesting bit how edding CNC look at plasma control with some default settings here (not seen them before tonight):
http://www.edingcnc.com/upload/files/PlasmaUserManual.pdf

[cncdrive]

Hi Rob,

OK, I think upto 4kHz sampling is managable.
Also we could use higher sampling rates and avarage the samples, but I think that going too fast with the sampling and avaraging would cause the result less precise than if we sample less with the Sample and Hold channel left open for a longer period.
I think the max. 4kHz sounds reasonable...

The hysteresis we could also make setable with 1bit resolution, which is based on your calculation with the current -5LPT motherboard electronics would be 0.14655... Volts per bit. (Volts of the plasma arc)

By the way, you've mentioned that the Hypertherm has the 1:50 Voltage divider, is that a standard circuit built into the Hyperterms or is an option?
Also a question if that signal is isolated from the high voltage parts or needs to be isolated?