New Project with Mesa 7I96 and THCAD-05

acourtjester wrote:

Not sure if I can get to the power supply at this point as it is below the shelf that has the 7I76E and THCADs.
 

Assuming the blue box'd power supply in the picture is the one you are using for OHMIC, then it looks like it is not an isolated power supply (it looks like it has a ground hooked up to it). Further, if it has a spot to hook a ground to it, then it definitely is not the double insulated isolation class II type power supply that is recommended. This is dangerous.

rodw wrote:

Respectfully, I spent several weeks working through this with Peter Wallace from Mesa. The intention is clear as conveyed in the circuit diagrams published.

I don't think anyone is trying to take away from the work Peter and you did. But from my experience (and that of a few others) with THCAD Ohmic in the last year+, there is room for improvement.

If the intention was to prevent electrolysis, then both Peter's circuit suggestion, and Phill's circuit would accomplish that.

One advantage of using the diodes, especially if they are well placed, is that it keeps the cutting voltage/current out of the cabinet. This can also prevent catching someone's wiring on fire...

It also prevents voltage accumulation in the THCAD circuit. I've seen this issue, and I know at least one other person has as well. Not really documented on the forum, but if you do enough cutting, and monitor the voltage of the THCAD circuit, you may see that the voltage in the circuit steadily creeps up over time. The other guy I am referring to does a lot of cutting of thin gauge material and it's to the point where he disabled the THCAD ohmic circuit entirely because the voltage would be hovering near the threshold and sometimes trips the sensor prior to, or during a probe (before contacting the material, and with a dry torch).

One advantage of the relay as shown in Phill's circuit is that it makes the voltage instantly zero, and instantly available. Whereas when hooked to the power supply it requires time for the capacitors to drain and fill.

Its not as straight forward as that. We are measuring voltage with the THCAD, not just a switch closing. Its very important that the power supply is outputting 24.0 volts and the recommended power supply has an adjustment screw to allow this. You need to calibrate that with a multimeter.
If you add 2 diodes into the circuit,  you would see a voltage drop of around 1.2 volts. (typically there is a 0.6v voltage drop across a diode). You would need to recalibrate the power supply and possibly the THCAD scaling resistor to account for this. You would also likely have to adjust the thresholds in the hal file.

The voltage should fall to instantly zero when the probe breaks contact. If it doesn't, then the only possible cause is some sort of short in the nozzle that allows voltage to flow where it shouldn't. If thats not resolved, probing is likely to be problematic.

If the voltage is too close to the threshold, you can change the threshold in hal. Ideally, you would allow adjustment of this from within your config if it was a major issue.

There is nothing in the circuit like a capacitor to store electrical potential so the only way I could see that voltage could slowly increase over time is if there was some kind of drift in the power supply (or a short in the torch).

Don't forget that Ohmic3 (not used in Toms Config here) was deigned to do the impossible and that was to  allow ohmic sensing on a water table. Thats long been acknowledged as problematic and many industry sources recommend NOT using it on water tables. The effectiveness of Hypersensing is dependent on a lot of environmental factors some of which have been mentioned above. 

I have long bypassed the optional mains relay as I found there were some issues in my setup including the long time discharging capacitors on power off.

Whilst mileage for others may vary, I have frequently done cutting jobs with over 1500 pierces on a water table with no probing issues. I feel the performance is improved over the earlier relay circuit I used.

My concern is that there are many circuit diagrams on the forum here and some of them of dubious quality have been used by many people with the mistaken belief that they are the gold standard. There are versions of the THCAD and relay circuits in the official Linuxcnc documents, its best we defer to them.

In terms of design, Ohmic sensing does not require a probe-enable signal because the additional isolation is not required. But to remain compatible with Plasmac, it does accept this signal and will not turn on a probe signal unless its present. A much cleaner design could be achieved if the probe-enable was removed from the ohmic component. 

The THCAD circuit is probably more costly than the relay circuit but it requires fewer components and arguably more reliable. If correctly implemented, there is no risk of fires in the wiring. Properly configured it performs flawlessly where other systems don't. I can understand Phill added a relay when testing it becasue he had the relay in his control box. In a new build, the temptation to add these components should be ignored, this is just adding more expense with unnecessary components which do nothing other than to reduce MTBF.
 

The THCAD cannot "accumulate voltage" it has nothing with a time constant greater than 1 ms. If you see a gradually increasing voltage its most likely related to leakage currents. These may be caused by moisture or accumulation of conductive particles.
Moisture and DC bias voltages can cause electrolytic corrosion which will present as a slowly increasing leakage current if moisture is present and create conductive paths that remain even when dry.

In an Ohmic sensing circuit it may be that the sensing current is too low (the current sense resistor has too high a value)
The higher the sense resistor, the more susceptible to leakage current the Ohmic sensing circuit will be.

PCW wrote:

The THCAD cannot "accumulate voltage" it has nothing with a time constant greater than 1 ms. If you see a gradually increasing voltage its most likely related to leakage currents. These may be caused by moisture or accumulation of conductive particles.
Moisture and DC bias voltages can cause electrolytic corrosion which will present as a slowly increasing leakage current if moisture is present and create conductive paths that remain even when dry.

In an Ohmic sensing circuit it may be that the sensing current is too low (the current sense resistor has too high a value)
The higher the sense resistor, the more susceptible to leakage current the Ohmic sensing circuit will be.

Perhaps I should better describe what happens... If you HalMeter the voltage for the OHMIC circuit while a program is running, what you will see is that the voltage spikes full when the torch is on. Each time the torch turns off, it goes back down to near zero. As the cuts progress, each time the torch turns off, the voltage returns further and further from near zero. Eventually, it hovers at say 22 volts continually, with occasional spikes to 23v, which can be a problem when your high threshold in the ohmic comp is set to 22.5v...

As I remember, turning off the power supply and turning it back on will remedy this condition.

Also, adding the diodes and disconnecting the sense leads (via the relay in the circuit) will prevent it in the first place.

That suggests leakage currents as I mentioned. That is if you measure the voltage across the sense resistor with a DVM, you will see that the THCAD is correctly measuring the voltage which is registering the increasing leakage currents.

rodw wrote:

Its not as straight forward as that. We are measuring voltage with the THCAD, not just a switch closing. Its very important that the power supply is outputting 24.0 volts and the recommended power supply has an adjustment screw to allow this. You need to calibrate that with a multimeter.
If you add 2 diodes into the circuit,  you would see a voltage drop of around 1.2 volts. (typically there is a 0.6v voltage drop across a diode). You would need to recalibrate the power supply and possibly the THCAD scaling resistor to account for this. You would also likely have to adjust the thresholds in the hal file.

Correct, and if the voltage drop is accounted for, it still works as intended.

If it doesn't, then the only possible cause is some sort of short in the nozzle that allows voltage to flow where it shouldn't.

That is not correct. At least not with a Hypertherm. Also, the OHMIC THCAD will read above full voltage while the torch is on during a cut, when it is not shorted, and not touching the metal.

Here is video evidence:

If the voltage is too close to the threshold, you can change the threshold in hal. Ideally, you would allow adjustment of this from within your config if it was a major issue.

When the voltage accumulates and gets near full voltage the power supply is capable of, that leaves little room for adjustment.

There is nothing in the circuit like a capacitor to store electrical potential so the only way I could see that voltage could slowly increase over time is if there was some kind of drift in the power supply (or a short in the torch).

The other person who has this issue has a Hypertherm as well. I am willing to say "it's a Hypertherm thing", but also, I truly don't know what causes it. I do not have a shorted torch, and the power supply holds its set voltage perfectly.

Don't forget that Ohmic3 (not used in Toms Config here) was deigned to do the impossible and that was to  allow ohmic sensing on a water table. Thats long been acknowledged as problematic and many industry sources recommend NOT using it on water tables. The effectiveness of Hypersensing is dependent on a lot of environmental factors some of which have been mentioned above. 

I'm sorry, but ohmic3 does not allow ohmic sensing on a water table. A properly sized resistor in the proper place does. With the 24K resistor in place, my OHMIC-THCAD will detect the top of the water drop, far from impervious to water. The only thing that changes this is changing the value of the resistor. Moving averages do not make this sensing method impervious to water.

My concern is that there are many circuit diagrams on the forum here and some of them of dubious quality have been used by many people with the mistaken belief that they are the gold standard. There are versions of the THCAD and relay circuits in the official Linuxcnc documents, its best we defer to them.

Even the circuit in the Plasma Primer (I assume that's what you are referring to) is not correct. It does not show the 24K resistor necessary to even begin to make the whole thing water tolerant.

The THCAD circuit is probably more costly than the relay circuit but it requires fewer components and arguably more reliable. If correctly implemented, there is no risk of fires in the wiring. Properly configured it performs flawlessly where other systems don't. I can understand Phill added a relay when testing it becasue he had the relay in his control box. In a new build, the temptation to add these components should be ignored, this is just adding more expense with unnecessary components which do nothing other than to reduce MTBF.

I only recently moved to the relay method and I guess my mileage is varying; it's awesome so far. I don't agree that THCAD is more reliable. Now that I have experienced both for myself (though admittedly, I have far less time on the relay method), using a THCAD seems like an unnecessarily complicated/costly method to do something that is fairly simple. That doesn't even factor in all of the headaches to dig up the right ohmic component from the right thread, compile it, edit the hal file correctly, etc.