Using Two THTIC Boards with EASERCAT-7000-DIO+I2C – Are Both I2C Inputs Designed for That?

[Smith3rman]

Hi,

I'm working on a LinuxCNC-based plasma setup and considering using two THTIC boards for material sensing – one for arc voltage monitoring and the other for ohmic/hypersensing.

The EASERCAT-7000-DIO+I2C has two I2C inputs, and I'm wondering if it was part of the original design to support dual THTICs simultaneously via I2C. If so:

Are there any specific considerations regarding voltage scaling (e.g., resistor values or divider networks) when using one THTIC for ohmic sensing?

In particular, would it be beneficial to adjust the voltage scaling components for better resolution, or do you think the existing range (up to ~200V) with the limited ADC resolution is still accurate enough for reliable hypersensing?

The idea is similar to what's described in this LinuxCNC forum thread on hypersensing, where precise detection of voltage rise is used to sense material contact without relays or diodes.

Any guidance or clarification would be super helpful, especially if this setup was already anticipated in the hardware/firmware design. Thanks for your time and for making this open-source!

Best regards,
Smith3rman

[MetalMusings]

It's not part of the design, but it can easily be done.

• I2C devices are connected in series, the I2C port on the E7000 is fine, no hardware changes needed.
• Use different I2C addresses for the ADCs. Depends on which ADC you use how that is done.
• Modify the firmware (quite easy, mostly cut and paste) to support multiple units.
• Add PDOs (and SDOs) for the second (and third etc) I2C device.
  The limiting factor would be the data read time from the I2C devices, each somewhere 100 usecs, so a few can be had but not ten.

The voltage divider should give around 2 V to the ADC when cutting (arc voltage around 100V). The ADC has a range of 0-4V (0-5V) and one wants to be somewhere in the middle.
I adjusted resistor values after my plasma cutter that gives out full arc voltage. If your arc voltage is already scaled down, adjust resistors to give 2V at cutting voltage. You can also double-check power dissipation from the resistors, that's the reason for those high 100's kOhm resistor values. Also finding resistors that can handle 500V or so, one need to go with the larger size. If you have say 5V out, you can use a few kOhm 0805s.

Don't know or don't understand the difference for hypersensing. The ADCs are quite sensitive in this config, 11 or 12 bit resolution already. The only risk I can think of is to have too high voltage on the ADC input. There are limits, around 6V as max. I have placed a zener diode in an attempt to bleed off any voltage higher than the zener diode value. But not really tested. If that works reliably, you can adjust resistors to a smaller range maybe 0-30V (I don't know as I said) and let the zener diode bleed off the high voltages when cutting. If that's beneficial for ohmic sensing.

I would think that the THTIC combo with E7000 certainly is quick enough and have the precision needed. I have used it for many hours in my own setup and it works great. It just works.

[Smith3rman]

Thanks a lot for the thorough explanation!

That clears things up really well — great to know that using multiple THTICs over I2C is totally doable with just different I2C addresses and a small firmware tweak. It’s also good to hear the E7000 handles this setup with no hardware mods needed.

Just to add to the context:
I think what’s often referred to as “ohmic sensing” is the more basic method using voltage detection through opto-isolated relays — basically a digital signal indicating contact. In contrast, what I understood as “hypersensing” (like described here) uses continuous voltage monitoring to detect the ramp-up as the torch touches the material. This method, especially with ohmic.comp in LinuxCNC, allows setting both an upper and lower threshold for strong hysteresis, making it much more robust and noise-resistant.

With the 12-bit ADC, even when scaled for ~100V arc voltage, the resolution should still be totally fine for detecting the ramp-up and touch event accurately — so no changes needed to the scaling for this use case. Only customization needed is the firmware support for multiple I2C addresses — and since I only need two, that fits perfectly.

Thanks again — really appreciate the help and the great work on this project!

[MetalMusings]

Isn't it just to add a power source and measure with the existing Arc Voltage reader?

From one of the schematics.
There is already an isolated power supply, let's add one more 24V. And bring to a regulated voltage.
One would need to know when to activate the power supply and when not to, but there seems to be an Ohmic Sensing output pin one can use.

[Smith3rman]

Yes, I think you're absolutely right — if the plasma voltage wont destruct the Voltage regulator U6-LM7815, but i think in this case even after the 300k resistor there is a high possibility that the unisolated regulator may break, i think in the datasheet there is even mentioned that there shouldnt be a feedbackward voltage.

Adding another isolated 24V power source, which is isolated and regulated in one part makes perfect sense, especially since the rest of the ADC path is already isolated and protected.
A solution for the isolation part would be this part: https://recom-power.com/en/products/dc-dc-converters/dc-dc-converters-regulated/rec-p-RSH2-0524S!sH2.html?4
And as you mentioned, using the Ohmic Sensing output pin as a control signal for enabling the power source seems like a clean and logical approach, but i think in that case it was mentioned because of one of the mesa boards they used but i think i can easily use one of the output pins of the Easercat 7000.

[MetalMusings]

Was a quick hack there with the schematics.
So how to connect/disconnect? A small relay could work. Nothing is time critical with that. Can control the relay with an I2C IO expander. Then everything is on the TIC board (new design), controlled with I2C from E7000. Or bring in an extra output signal from the E7000.
The DC-DC converter you link to has only 50 mV p-p max ripple. I have seen worse. Maybe that is good enough so one don't need the voltage regulator at all?
Unfortunately my plasma cutter is one of those where the torch shield is already connected to ground so I can not immediately use this method. I need to disconnect the shield and work piece during piercing and I can do that with a 200A contactor for the work piece current. Or unknown current if disconnecting the torch ground, for the blowback start. Just guessing that ground current is less. I don't want to make modifications to the plasma cutter, staying on the outside.

[Smith3rman]

Yeah thats why i send that Dc Converter here, it also has an integrated I/O for switching On/off i think, so in total a good fit. I think i will try it, but it will take sometime. I ordererd the PCBs yesterday 😄 and also i need to buy the parts, so some time will go down the river....

[MetalMusings]

Now I see the on/off ctrl. Interesting. And this allows 200V on the outputs?
Will be a good solution if so. Minimal changes needed.
So which of the TICs did you go for, the MCP3221 or ADS1014?

I am going to do something with the ground wire, just ordered a torch connector kit.
That gives access to the ground in the torch.
Looking at the wires, they are just regular wires so the ground current must be reasonable,
meaning a relay size must be reasonable, and activation current as well. Maybe this DC-DC converter
can be power supply for relay control. If 80 mA is enough for the relay.

[Smith3rman]

Good question only saw one sec. ago that it's only tested for a minute and not continuous, I will write them an email to ask for that.
If I got an answer I mention it to you.

I am ordering right now both TICs, but I think I go the MCP route.

[MetalMusings]

The performance of the two is more or less identical, both works.
Don't forget to solder in a programmed eeprom, that's mandatory. As soon as it works and is seen in TwinCAT, reprogram with the actual esi file.
I'll wait for the adaptors to show up, need to know the arc current during arc blow back start.
Good luck.

[Smith3rman]

I am in the moment on the creation of an ordering list for the PCB parts, therefore I got some questions regarding the used leds. With the given Resistor network the LEDs shine not as much as they could do, is that still enough and if so which one did you used?
If I got the list ready I will send an link in here for reordering...

[Smith3rman]

Here are the lists:
EaserCAT7000 (still missing the Resistor Networks and LEDs)

THTIC1

THTIC2

[MetalMusings]

Comments
On load capacitance for the crystals. You have selected from the schematics I guess, select from the crystals' specified load capacitance 22 pF and 30 pF. Afaik.
I used no-name components for resistors, capacitors and leds. I had sets of components and I picked the ones I needed. There are no special requirements on those components, "anything" will work.

Led current.
Outputs. There is 10 mA which switches on the TLP 292 good. I have the Arc-ON signal go through the TLP 292 without extra relays or anything.
Inputs. 4 mA. Like you say this can be higher, 10 mA for example. 23V over there would mean 2k3 as the sum of RN1 and RN2, or both of them 1kOhm for 11.5 mA. If you like that, you can do that. It does work with 4 mA as well though. Feel free to select any led you like, there must be leds there though. (At least in the field voltage polarity you will use, the design support both polarities).
Later. Quickly double-checked power dissipation it's around 1/8 W over one of the 1k resistors, that's a lot for those tiny components.
Maybe it's better to stay at 4 mA or maybe go halfway to 10 mA.
Those are the concerns. If you feel you have control over this and want to select different resistors, go ahead. It's getting a bit late and I am not sure I have missed something.

The RJ45 connector WILL NOT WORK, wrong pin-out. Search for J0011D01BNL (pulsejack), the datasheet show it in detail. Equivalent connectors are ok, but have to match the pin-out and the internal magnetics setup.

Edit. I'll just add that the other components look ok.

[Smith3rman]

Thank you for the Input for crystals!
For the parts i choosed i did mainly look for availability, if someone in the near future want to use the same order as mine.

My concerns were that the color LEDs might be too dim and that the TLP292 could have issues with the low LED current (4 mA). Since it seems this isn't the case, I plan to leave the parts for the inputs as they are.

Regarding the outputs: Based on my calculations, with LED1 (1.25V) and LED2 (1.9V), the voltage across the resistor is only 0.15V, resulting in 0.15 mA, which is within the LEDs' ratings, so perfectly fine.

However, I can imagine that at around 1 mA, the brightness of the LEDs at RN12 might be very low. Therefore, I think it would be a good idea to reduce RN12 a bit to make the LEDs brighter. I would change the resistor to 220 Ω to increase the current slightly and improve brightness.

Regarding the RJ45: I changed that last second because I thought it's the same and cheaper. Good that you caught it!

Thanks again, and I'll proceed with the changes as discussed.

[MetalMusings]

I wouldn't worry to much about the LEDs, they are mostly used for verification during commissioning.
D25-D28 are not used, I put them there as a debug help. I always regret not having a diode to switch on when something happens, so there they are. But not used in operation.

I don't see how you come to 0.15V? If you are running something else than 24V field power, you probably should change resistors.

[Smith3rman]

I kinda overthought it and assumed they were in series with the TLP292... but yeah, even then, with a 1.9V drop from a 3.3V supply, maybe the resistors are a bit much—though honestly, it's probably just me being paranoid. 😅 I’ve had moments in the past where the LEDs barely glowed when they were supposed to, so now I'm second-guessing everything. Might just throw in 220Ω for RN14 and RN12 and call it a day—then I can stop worrying whether they’re shining "enough." 😄