Created Date: April 02,2018

Created by: metaltech

I 've done a bunch of searching to learn, but still not satisfied that I know what I'm trying to do. The application: PLC dispenses a liquid via solenoid valve. Monitor the weight as volume builds, shut off valve when target is reached. The hardware isn't bought yet, so I'm flexible, based on suggestions here. This will be a production device, not a one-off for lab work, so cost does matter. Planned on CLICK with built-in analog, C0-12DD1E. Only need one channel of analog input, and 3 discrete outputs. An HMI will handle setting the target weight, etc.

I'm intending to use a beam-type load cell, 1 kg max capacity, rated output 1mV/V, excitation 10VDC (15VDC max), 350 ohms nominal (I tried posting a link, but it was 20 pages!) I didn't realize how small the output was, so learned that an amplifier is necessary. I intended to use analog voltage-in, but realized the 12-foot run from the amp to the PLC might cause problems with noise, accuracy, etc. I'm now thinking 4-20mA, although I'm ignorant about it, thus this posting. I 've done reading, but still not understanding how to calibrate everything such that zero weight (output is at 4mA here?) can be read in the PLC, and scaled so max weight of 1kg generates max amplifier output of 20mA. Or, am I totally misunderstanding? EDIT: I think I see that in the programming now.

So, the questions:

1) Am I right to think that current input is more appropriate than voltage for my application? (Even though my limited knowledge can likely program the voltage easier.)

2) Does anyone with experience in using load cells have a favorite brand, etc. to recommend? I 've seen cheapy Chinese on ebay for $15 and good ones for hundreds. I need to keep in under $150 or so. I'm assuming the higher priced ones are actually worth their higher price.

3) Same goes for amps. The prices range widely. Any feature I should consider mandatory?

4) Are there "gotchas " like matching the load cell to the amp? I 've noticed different specs for excitation voltage and input resistance, in particular. How critical are these matchings?

5) Is there any advantage to using an analog input module rather than the PLC's built-in analog? Looks like 1 bit more resolution for the module. I need to be accurate on the weights within 10 grams or so, not super critical.

Thanks for all answers and recommendations. I'd also be grateful for links to tutorials, etc. that might leave me more enlightened.

metaltech

Created Date: April 03,2018

Created by: Adisharr

Current output is going to more noise immune and should also compensate for wire resistance changes. You'll see somewhere the maximum loop resistance and you should be fine. Make sure to use shielded cable (potentially twisted-pair) and only shield one end unless otherwise stated.

Strain gauges are usually more accurate at their top level so you would have to do some testing to see how accurate it would be at the low end. I would have a few known weights to check for accuracy. The analog card can do scaling right on the channel which makes it a it easier.

1. Current is good

2. I 've only used a few and they were all over the place brand-wise. They were also expensive (>$ 800.00)

3. Get the best one you can afford (lowest noise, highest resolution)

4. Usually the mfg. will recommend an amp. with their strain gauge. In the case of a Chinese mode, I'm not sure if they have that option.

5. Use the module and get the extra bit.

Other here's probably have more experience, the models I used were very sensitive and a couple times were ruined by a weight dropped on them.

Created Date: April 03,2018

Created by: Tinker

Something to consider is that the CLICK's built in analog is not super fast 50mS (1/20 of a second) conversion time, maybe not a problem in your application. You didn't say how fast you are filling, but if it were, say, a half second fill time you'd be limited to something like 10% accuracy, if the filling takes 20 seconds, then AD conversion time would only result in 1% error. The module is a little faster (but only a little faster and it would increase you cost a lot)

If one was experienced in circuit design, one might consider using an analog comparator and using a analog output from the PLC to provide the reference value, then the output of the comparator would go to a discrete input, conceivably even an interrupt, though that might be excessive.

EDIT: I might mention that a BRX has a much faster (and higher resolution) AD converter, however, it also has a higher price tag

Created Date: April 03,2018

Created by: metaltech

Adisharr~ "Strain gauges are usually more accurate at their top level so you would have to do some testing to see how accurate it would be at the low end. I would have a few known weights to check for accuracy. The analog card can do scaling right on the channel which makes it a it easier. "

Yes, I was going to try calibration at several weights, as the actual fills will be varied. Thanks for the tip on the analog card.

Tinker~ Good point about processing time. A full filling would likely take 7-10 seconds, but there will be some partial fills for only 2 seconds or so, so your point may be very important. I wasn't aware of the comparator, I'll look into that. Thanks for the comments.

Created Date: April 03,2018

Created by: OkiePC

I agree about the use of a 4 to 20mA signal conditioner (load cell amplifier).

As for your logic, I would plan for some test runs in which you calculate the amount of overrun to expect so that you can act early to close the valve and hit the target. You could do this manually or program the PLC to do it automatically. New equipment may be quite repeatable and then drift after a break in period, so I tend to make this calculation a part of the ladder logic that can be selected. Say your error at the end of a cycle is outside of a deadband, then recalculate this compensation adjustment. Have a way to disable this automatic calculation so that a one-off screw up caused by extenuating circumstances does not jack up your control for the next cycle.

A couple of the weighing systems I have worked with where there is no analog control will use two separate discrete outputs...one for a fast fill...the other for a slow fill, the idea being to minimize cycle time and maximize accuracy.

Created Date: April 03,2018

Created by: MikeN

I'm intending to use a beam-type load cell, 1 kg max capacity, rated output 1mV/V, excitation 10VDC (15VDC max), 350 ohms nominal (I tried posting a link, but it was 20 pages!) I didn't realize how small the output was, so learned that an amplifier is necessary. I intended to use analog voltage-in, but realized the 12-foot run from the amp to the PLC might cause problems with noise, accuracy, etc. I'm now thinking 4-20mA, although I'm ignorant about it, thus this posting. I 've done reading, but still not understanding how to calibrate everything such that zero weight (output is at 4mA here?) can be read in the PLC, and scaled so max weight of 1kg generates max amplifier output of 20mA. Or, am I totally misunderstanding? EDIT: I think I see that in the programming now.

From what I remember load cells cannot use a standard analog interface, the voltages are too small and sensitive. You need a load cell input card to connect the cell to a PLC, but I dont think Automation sells any. At least they didnt last I checked. Also load cells output voltage that varies based on load, so you cant use a current input type. As that doesnt read voltager and instead measured the current variation to get the value, which is oposite what you have with a load cell.

It may be possible for you to get a conversion device that you can mount near the load cell that takes the mv input from the cell, and converts it to analog ma signal. Or if your load cell has a partner amplifier specifically for it that amp may have an analog output you can use. Then you can send that new signal output to the PLC on an analog 4-20ma card, rather than the load cells direct mv signal.

Created Date: April 04,2018

Created by: metaltech

Okie~ great info, really useful. I'll keep that strategy in mind when I get into the programming. I'm using an HMI and recipes (this device will blend two liquids in a desired ratio), so I may incorporate a hidden screen for calibration or compensation, to adjust for those changes over time you mention.

MikeN~ right, the load cell puts out next to nothing. That's the point of the amplifier. I 've seen amps with 0-5 or 0-10VDC outputs, or 4-20mA outputs. A great amount of them have both voltage and current outputs. That brings the signal level up to something that that is useful to the PLC input.

Created Date: April 04,2018

Created by: MikeN

I looked briefly at products from Automation again just to see if they had anything that would take a load cell input directly, turns out that it seems all the thermocouple input cards I looked at can be used for load cells. For Productivity, BRX, and Click. Probably all the other PLCs as well.

They also now sell a digital panel meter that takes load cell inputs, can perform scaling directly in the display meter, and will provide a 4-20ma output to a PLC.

I think the thermocouple input cards taking load cells directly is pretty cool though. Can do whatever scaling and averaging and logic you want on it directly in the PLC without having to go through a second piece of hardware first.

https://www.automationdirect.com/adc/Reviews/Catalog/Programmable_Controllers/CLICK_Series_PLCs_(Stackable_Micro_Brick)/Analog_I-z-O/C0-04THM?showReviews=true

Created Date: April 04,2018

Created by: metaltech

MikeN~ Thanks for the comment. I had run across posts in forums when I was trying to gain knowledge, and saw people using TC modules with the load cell connected directly. I don't feel confident that this will work for us, as the load cell is about 12 feet of wiring away from the PLC. So, I expect that such a low-voltage signal would get lost in noise or degraded too much to be useful. That's why I think I need to amplify it close to the load cell before going to the PLC input.