Solenoid pump driver

Hi All - first post, zero PLC experience, I'm an EE and think I understand the concept. Feel free to disabuse me of that notion ;^)

I'm part of a startup and we're building a prototype system that requires some controls. Also the project budget is personal funds (not made of money).

Part of the system involves injecting a tiny, carefully metered amount of fluid into tank and the rate depends on some variable in the feedback loop. For example we might be adding acid to a tank to lower the pH then read the pH and adjust. There is a calculated 'base rate', and then over long periods of time (many minutes), some 'tweaks' to the rate. I can fo the controls-theory part of that, but it's not very critical.

Anyway we've spec'ed some solenoid micro-pumps that require a 150ms minimum on-time on-time (12 and 24vdc versions available at 0.32A, 0.16A respectively) at are spec'd at 0.32A or 0.16A respectively. The EOM states you shouldn't leave the solenoid 'on' too long as it degrades pump volume accuracy. So I'm pushing for a 0.25s on-time and something in the ballpark of a 20second off-time (adjustable).

As an aside, I'd really have preferred to buy some peristaltic pumps so we could adjust the motor rate, but the amounts of fluid are too small (micro liters per minute) and not not easily dilutable (enzymes).

--

OK, I haven't done circuit design in a long time, but driving a fractional amp Low DC voltage solenoid sounds like an over spec'ed SSR (like a MOSFET type that will switch DC, maybe 5A, 60vdc rating) and a 'flyback' snubbing diode. Some of the inexpensive DC SSRs have on/off times under 1ms, and need maybe 10vdc or better 10ma inputs to switch. I imagine a common digital out will handle that.

So how to I drive this sort of repeated, fixed-width one-shot from a PLC ? I smell timers & counter - is that right ?
This is probably a no-brainer for you old hands, but I don't understand the PLC features well enough and would like some 'direction'.

Any advice on budget PLC for this app ? The prototype system has ~9 thermocouple inputs, 2x analog sensor inputs, 16 bit, and drives 2 of these pumps and another much slower relay. We'l LIKE to access the data & state from ethernet - maybe modbus/tcp.

thanks in advance,
-Steve

Hi Steve,
Yes you are correct in Timer. Most PLC's will scan in micro seconds.
I would suggest a Click PLC for your job. It is very cost competitive. A PLC with ethernet (Modbus TCP), 12 thermocouple, 2 analog in will be ~725.00 USD.
https://www.automationdirect.com/adc/Shopping/Catalog/Programmable_Controllers/CLICK_Series_PLCs_(Stackable_Micro_Brick)
https://www.automationdirect.com/clickplcs/index
Click PLC System BOM
C0-11DD1E-D 1
C0-04THM 3
C0-04AD-2 1
C0-01AC 1

Regards,

WOW - sincere thanks. I've been eye-balling automationdirect parts but with 5 or 6 families of plcs to choose from and then all the I/O - you saved me hours.

Hi Steve,
Here are some links that will take you through programming of the Click PLC. It will start with the hardware, software, communication, etc.
http://accautomation.ca/click-plc-system-hardware/
Here is the video series of the Click. This provides most of the same material as the website.
https://www.youtube.com/playlist?list=PL3y71jAPOdZAltYHLp8Z3aR961EqQUawt

Hope you find this helpful.
Regards,

you could use any cheap PLC, or even cheaper an arduino or even a simple 555 will work.

shooter said:

you could use any cheap PLC, or even cheaper an arduino or even a simple 555 will work.

Click to expand...

Any cheap PLC will not talk Modbus/TCP, I think.

The reliability and maintainability of such homebrew circuitry isn't a good feature, and we intend to get some experience with the sorts of control hardware that we would use in the full production system. That will certainly include a basic PLC for the 'safety' system at least.

We might use RaspberryPi3s, even on the DIN rail, *but* we've decided to only use the USB & ethernet interfaces to avoid any hardware lock-in & reliability issues. We could then easily replace these with a more industrial computer module in production. These would be used to adapt Modbus to MQTT for example. OTOH we could access the Modbus/TCP from a regular server as easily. Does that make sense ?

We *could* use a cheaper no-enet PLC and connect a pi3 to the rs485 modbus/RTU via USB. The cost savings isn't much.

I do have a followup noob question.

I've found I can get 2/3/4 wire pt100 RTDs at a reasonable price, and the interface parts (for example the AD Click parts) aren't very different in price.

But in the production system I think we'd want transmitters for either thermocouples or RTDs.
What are the advantages/disadvantages of transmitting 4/20ma vs 0-10v for analog points ?

Both claim to be able to detect an open circuit (broken wire).

I would always, always go with 4-20mA over 0-10V where possible.

Current loops are much less susceptible to inaccuracy due to voltage drop over long distances, or induced voltage from external sources along the route. Detecting a broken wire on 4-20mA is as easy as telling the difference between 4mA and 0mA; detecting a broken wire on 0-10V relies on configuring some "smart" hardware to give you the message somehow. You can sometimes scale the voltage to give you 2-10V instead of 0-10, and then it's nice and easy again, but really, the 4-20mA has already won IMO.

The only benefit I can see to using 0-10V over 4-20mA is that it's easier to test in the field without disturbing things. You just need a voltmeter. If you need to test a current loop, you have to break into the loop with a milliampmeter, which may not always be possible if it's process critical. But then, if you're dealing with process critical, there are clamp-on meters designed to do 4-20mA - they're not cheap, but they are surely wrought by the gods themselves if you've got a lot of 4-20mA current loops where you work.

ASF said:

I would always, always go with 4-20mA over 0-10V where possible.

Current loops are much less susceptible to inaccuracy due to voltage drop over long distances, or induced voltage from external sources along the route. Detecting a broken wire on 4-20mA is as easy as telling the difference between 4mA and 0mA; detecting a broken wire on 0-10V relies on configuring some "smart" hardware to give you the message somehow. You can sometimes scale the voltage to give you 2-10V instead of 0-10, and then it's nice and easy again, but really, the 4-20mA has already won IMO.

The only benefit I can see to using 0-10V over 4-20mA is that it's easier to test in the field without disturbing things. You just need a voltmeter. If you need to test a current loop, you have to break into the loop with a milliampmeter, which may not always be possible if it's process critical. But then, if you're dealing with process critical, there are clamp-on meters designed to do 4-20mA - they're not cheap, but they are surely wrought by the gods themselves if you've got a lot of 4-20mA current loops where you work.

Click to expand...

If you know the input impedance (250 ohms is a typical input), you can measure the voltage across it and calculate the 4-20ma signal...

Gene Bond said:

If you know the input impedance (250 ohms is a typical input), you can measure the voltage across it and calculate the 4-20ma signal...

Click to expand...

Doesn't help you in the field

Not read the entire thread, maybe you could look at using something like a chemical dosing pump. Very simple construction, low flow and can be controlled by using 4..20mA and more fun stuff.

We use Grundfoss. Not sure how much they are, but perhaps it is something interesting.

We use them for chemical dosing into boilers, incredible how much pressure these things can generate.

4-20mA is a regulated current circuit - the current is the same at all points in the current loop. A corroded terminal box connection which adds some resistance to the loop will not affect the loop's current value, until the corrosion resistance is so high the loop's power supply stalls.

Voltage on the other hand, is dropped at every connection,

Having worked in process for years, 4-20mA is THE standard.
30 years ago that was also true, but then I worked in labs for a couple years and in the labs there were lots of voltage signals 0-1V, 0-5V, 0-1mV.

I'll take 4-20mA any day, it's incredibly robust.

we use the Fluke 773 for measuring current loops without breaking the circuit... works very well!

http://en-us.fluke.com/products/clamp-meters/fluke-773-process-calibration-tool.html

It also allows insertion into the loop, and measurement or drive signals.