Thermocouple card used with load cells?

Hey guys. I'm looking for an inexpensive way to integrate 4 load cells into a project. Right now, the project is using an Arduino with a HX711 amplifier / converter and it has been having numerous issues.

My load cells are 50kg, though I'm going to change them out for 10kg beams. Excitation voltage is 10-15v, I suspect I'll send it ~14v through a good regulator. Output is 2mV/V.

This is my first real step into load cells, so as I understand the specs above, if I send it 14v, I would get 28mV on the output.

I'm looking at using a Automation Direct Click for this project with a 16bit C0-04THM module. The module has a 0-39.xx mV range. If I'm understanding this correctly;

16 bit = 65536 \ 39mV range = 1680 units per mV. 1680 * 28 = a usable range of 47051. 10000g \ 47051 = .21g resolution? This would be with a 10kg cell. With a 50kg cell, it looks like I'll get a 1.05g per unit resolution.

Are there going to be any issues with this method? I'm not after big accuracy. The weights to be placed on the scale will vary in 15g increments and are known values.

My concern is I don't know what is going on internally with the thermocouple module. I assume I can scale this to whatever I want, though I'm not sure if the load cell (and the thermocouple module) are linear or non-linear. Linear scaling is easy, non-linear I'll have to learn how to do with the Click. I've been pampered and have had a non-linear scaling instruction available to me with the Productivity line.

Thoughts?

Scaling: You will have to do some filtering and have a method to "graph and display" the weight every so often, but after the weight is solid on the scale, the curve will pretty much be linear to the edges of the functional range of the load cell.

Your 16-bit total is correct, however you need to see if you are using a signed integer or not. If the integer is unsigned, you have the leftmost bit to utilize as a sign placeholder and the other 15 bits used as an integer value, which effectively cuts your value in half.

I do a TON of work on scales and have always wanted to use a PLC card to make my own scaling app. That would be the bee's knee's. We actually have a loadcell card for a SLC500 but I don't have time to testbench with it.

This is just a design thought:

If you are already using alternate electronics, you will want some way to combine those signals ensuring that you get very similar mV signals back (aka trim each load cell). You CAN do this with software, but since you already have the ardunio installed it may be easier to do all trimming and combining at the ardunio and run an analog output back to the PLC for your actual scaling software.

Thanks for the reply. From my experience with other AD analog products, their 16bit cards are signed. On all of the 0-10v and 4-20mA cards I use (on Productivity), I always get a value of 0-65536.

The Arduino's are going away entirely. The few units in the field with Arduino's in them are getting retrofit to the new package that I design. Something is causing the amplifiers or the load cells to just up an die randomly on some or all channels. I have two back in the shop right now, each has 4 load cells. On one of them, 3 cells appear to be dead, there is no signal coming from them (as measured with a Fluke 289 that is capable of reading mV signals). The other, 2 of 4 cells are dead. I built that particular one myself. On initial startup, all 4 cells were working, I saw the data in the serial monitor with my own two eyes. On the next reset, scales 3 and 4 were dead.

I'm eliminating ALL Arduino's on any future product. Even the Ruggeduino's have been problematic.

These should be pretty easy to implement in ladder logic, as long as the physical hardware will work. On startup, wait 30 seconds and tare the scales. There are 4 individual scales. From there, they are only looking for values of 30, 45, 60 or 75 grams. I was going to give them a +\- 15%. I was going to use a pair of compare contacts, if the weight falls in between that +\- 15% of expected weight for ~1500ms or so, it sets a bit.

The scales / timers will be manually reset by a game operator.

Hmm, there are only a few things that take out loadcells. The most common I have ran into at my area of work is moisture getting into the cells and improper grounding, but I don't know what kind of load cells you are using.

It may behoove you to just utilize an actual scale setup. For $1500 you can get an Avery-Weightronix ZM303 weight indicator and trimming board.

You hook all 4 loadcells up to the board, balance them and the weightronix can fire outputs for each setpoint you have, or you can use data coms from it.

Make sure that your shield wires from the load cell are only grounded on one side (this can cause current to flow through parts of electronics you don't want it to.

Brandon_K said:

I'm looking at using a Automation Direct Click for this project with a 16bit C0-04THM module. The module has a 0-39.xx mV range. If I'm understanding this correctly;

16 bit = 65536 \ 39mV range = 1680 units per mV. 1680 * 28 = a usable range of 47051. 10000g \ 47051 = .21g resolution? This would be with a 10kg cell. With a 50kg cell, it looks like I'll get a 1.05g per unit resolution.

Are there going to be any issues with this method? I'm not after big accuracy. The weights to be placed on the scale will vary in 15g increments and are known values.

My concern is I don't know what is going on internally with the thermocouple module. I assume I can scale this to whatever I want, though I'm not sure if the load cell (and the thermocouple module) are linear or non-linear. Linear scaling is easy, non-linear I'll have to learn how to do with the Click. I've been pampered and have had a non-linear scaling instruction available to me with the Productivity line.

Thoughts?

Click to expand...

The Click will SCALE the input for you. You may manipulate the scaled (DFxx) value later, but there is NO 'exposed' unscaled COUNT on Click.

sparkie said:

Hmm, there are only a few things that take out loadcells. The most common I have ran into at my area of work is moisture getting into the cells and improper grounding, but I don't know what kind of load cells you are using.

Click to expand...

Load cells are very sensitive to three things:
Mechanical overload. Regular load cells are only rated to 25-50% overload.
Mechanical shock. Even the load cells isnt subjected to a very high static force, an impulse with a high peak can break the strain gauges inside. I have experienced that a load cell for 500 N got damaged from being dropped on the floor.
Welding currents (!). Since load cells often hold a weighing vessel suspended with no other contact from the vessel to ground, then any welding on the weighing vessel will mean the current has to pass through the load cell which will destroy it. Always have fat earth straps directly over the load cells from vessel to base, and even then also instruct everyone to not weld when the load cells are in place.

sparkie said:

For $1500 you can get an Avery-Weightronix ZM303 weight indicator and trimming board.

Click to expand...

The Laumas TLS that we use and can attach up to 4 load cells directly cast approx half of that.
There is a newer version called called TLM8 that is even better, but we havent gotten around to actually trying it out.

I have installed numerous load cells and have always had to be aware of the following.

Load cell support and mounting. You must install the load cell evenly, no torque, and not allow the cell to touch the frame, only the load cell mounts.

in any application, load cell shock can be 6-10 times what you are weighing.

please define your application in more detail.

james

JesperMP said:

Load cells are very sensitive to three things:
Mechanical overload. Regular load cells are only rated to 25-50% overload.
Mechanical shock. Even the load cells isnt subjected to a very high static force, an impulse with a high peak can break the strain gauges inside. I have experienced that a load cell for 500 N got damaged from being dropped on the floor.
Welding currents (!). Since load cells often hold a weighing vessel suspended with no other contact from the vessel to ground, then any welding on the weighing vessel will mean the current has to pass through the load cell which will destroy it. Always have fat earth straps directly over the load cells from vessel to base, and even then also instruct everyone to not weld when the load cells are in place.

The Laumas TLS that we use and can attach up to 4 load cells directly cast approx half of that.
There is a newer version called called TLM8 that is even better, but we havent gotten around to actually trying it out.

Click to expand...

Mechanical overload and shock are kind of a given, but you are right I should have mentioned them. Those strain gauges have some pretty small/sensitive solder points.

I'll keep that type of indicator in mind. I work in a packing plant, so our scaling systems almost all use full-blown indicators because we check our scales often and daily. Our entire business is based around them. We have a LOT of in-motion scales too, so it is often better to pay for some specific-purpose stuff that is rated for wash down. We use 180 degree water at 600psi.

Where welding is an issue for us, it is on a steel rail that mounts a monorail high accuracy in-motion type scale, so we really don't need good solid grounds around our cells. Perhaps once a year we have a PSU on an indicator taken out by welding, but then again I have went through the plant and taken care of all grounding and shielding (shielding landed at both ends is asking for trouble).

I also second James as to cleaning and blowing any debris off of the surface area of the load cell mount. Clean all threads as well, and if it is in a harsh environment a bit of anti-seize compound does wonders the next time you do the job.

Anyway, is this a multi-load cell scale or is it 4 individual scales with a single loadcell. I'm thinking that since you are weighing such a small amount of material it is 4 single load cell systems. In that case you can get cheap indicators that either a)fire outputs when a certain weight is reached or b) give you a signal to control based on weight. Weigh Tec has some pretty nifty little devices that are as tough as they come for fairly cheap, but once again I'm thinking you want to get away from specific purpose electronics. It is a LOT cheaper for an indicator to take a hit and need replaced than a load cell or thermocouple card. Utilizing a regular 4-20mA or 0-10V analog card you have a much cheaper piece of equipment.

If you are going to read the mv, (which is fine) I would suggest that you also read the excitation voltage and calculate mv/v if you can, or at least verify that excitation stays as it should.

Plus keep in mind that Thermocouples are not linear but strain gauges are.

RussB said:

Plus keep in mind that Thermocouples are not linear but strain gauges are.

Click to expand...

I believe the ai module should take care of this. He should be setting it up for mv not a linearized temperature

g.mccormick said:

I believe the ai module should take care of this. He should be setting it up for mv not a linearized temperature

Click to expand...

The title of the OP is "Thermocouple card used with load cells?". Guess I should have read the whole thing more closely.

Thanks for all of the input guys. I'm going to call Automation Direct in the morning and run it past them before I finalize on a prototype plan as I have an alternate plan as well, which I think may actually work better (see below).

For those asking about the application, this is for a game in which players have to put a certain amount of volume (in this exact case, with pre-filled test tubes) into beakers which are mounted on beaker stands, which sit (hinged) on a load cell. In future use, we'll allow the customer to fill with their volume of choice as we've had a few customers ask about using water to fill the beakers.

There is a video of it with it's current control pack here; https://www.youtube.com/watch?v=9QXCh9YOmxk

That one happens to drive a pair of 1w UV LED's to reveal a set of clues written in UV ink. Other models have a hidden box with a maglock, drawers, etc. The principal is the same on all of them however.

During some research today I learned more about piezoresistive force sensors, like this one from TekScan; https://www.tekscan.com/products-solutions/force-sensors/a401 They're simply a 2 lead variable resistor based on weight applied. The resistance varies from 100k ohms @ no load to 200 ohms @ 10kg. After 100g of applied force (looks to be around 8k ohms) the scale becomes very linear all the way up to the max of 10kg.

So, with that sensor and a simple voltage divider circuit, I think I can get away with feeding ~20v in, put a 100k resistor in for a divider and I should get a pretty good 0-10v output, to which I can scale in any way I want. The sensors are cheap, ridiculously easy to mount and don't have some of the environmental variables that load cells seem to succumb to like temperature variables. And since output is linear, I can easily put a tare function in and only need to worry about weight added. As a bonus, a 0-10v card is half the cost of the thermocouple card.

The biggest drawback to the above is that the card is only 13 bit resolution. So I went from having 47051 to play with to 8192. That said, since I'm able to pick the resistor value for the voltage divider, I can tune the circuit to fit my needs and make full use of the resolution.

sorry with a resistor as divider the best is to have 50%. It would also mean the line is not linear anymore.
a Thermocouple card has no exitation same as a mV input card. and you will need it to get your resolution.

The cheap solution with the HX711 is working better when you have some diodes added to the circuit, to prevent overvoltage anywhere.