Need more inputs shoe-horned into an installed PLC

[itsmoked]

Has anyone run out of inputs on a PLC and had to cobble up something to fill the need?

I have a PLC limited to 8 I/O modules and I already have the 8 and they're about 90% utilized and so I am not allowed any further modules. (Click with 8 modules already.)

I'm running two BIG compressors with all operational commands coming down from the PLC. But, I have only single high-side and low-side pressure sensors on each of the two compressors. If one of those sensors take a slow dump I could run the compressors into destructive running areas. I want to add some redundancy -triple redundancy- so I can discover a failing/drifting sensor and warn the staff while continuing with the winning voters while a replacement is hunted up and installed.

I was wondering if I could REASONABLY add a second CLICK CPU and then, via either Ethernet or serial, regularly Modbus over these extra eight analog inputs in as alternative values for voting and use?

Anyone tried this?

Keith Cress
kcress -

http://www.flaminsystems.com

 

[xnuke]

Seems to be costly to use a PLC for this, but not too bad since you're using a Click to begin with. Doesn't AutomationDirect make some sort of remote I/O system with Ethernet or serial comms that you could use with the Click? Remote I/O is usually a little less expensive than a new PLC.

xnuke
"Live and act within the limit of your knowledge and keep expanding it to the limit of your life." Ayn Rand, Atlas Shrugged.
Please see FAQ731-376 for tips on how to make the best use of Eng-Tips.

 

[3DDave]

Can you use a multiplexer?

 

[itsmoked]

Hi X; There are a couple of dozen other inputs and outputs so I don't see anything cheaper really. I considered switching PLC types to one that can take remote racks but that switch is very costly plus I'd have to re-write the thing. I don't think ADirect has any remote system as that would allow people to use inexpensive Clicks instead of being forced into their much pricier Productivity series.

That's actually what I'm going to try using the second Click for. I'll load it up with two 4 channel A-IN units and and have the CPU reading them and passing them across once a second and call it a day... I hope. LOL.


Dave; Are there commercial ones? They'd probably work if they'll ship analog inputs with fidelity.


Keith Cress
kcress -

http://www.flaminsystems.com

 

[FreddyNurk]

There may be the option of using fieldbus IO from Wago, Beckhoff, Phoenix Contact or similar. If you can use Modbus then you can probably get away with it.

Whether it ends up cheaper in terms of equipment cost compared to another Click complete with IO, I'm not sure though.

EDMS Australia

 

[xnuke]

I just looked; AD does have remote I/O. It's under the Field I/O link on their menu. The prices look higher than the cost of another Click with two 4-channel AI modules. Seems like the second Click with Modbus RTU or TCP comma would be the way to go.

xnuke
"Live and act within the limit of your knowledge and keep expanding it to the limit of your life." Ayn Rand, Atlas Shrugged.
Please see FAQ731-376 for tips on how to make the best use of Eng-Tips.

 

[itsmoked]

Thanks Freddy.

X; Are you referring to the 'Protos X' stuff? Yeah, not inexpensive..


I spent a couple of hours last night wading around in the CLICK stuff and found they've finally written a very nice manual for it. An entire 30 page chapter on "Communications". It shows very simple inter-CPU communications are possible via any of the three comm ports. I also found they sell a CLICK CPU with Ethernet called an 'Analog Ethernet CPU' that comes with 4 analog inputs! That means I only need one CPU and one analog module which makes it even nicer. I can also use the second comm port on the second CPU to Modbus out to two condenser fan VFDs.

Thanks.

Keith Cress
kcress -

http://www.flaminsystems.com

 

[controlsdude]

Just remember to clear the data if your communications goes down between PLCs so that the system will stop or stay at a predetermined fault state for recovery.

 

[roydm]

I had a similar problem with a shoebox Modicon, I was able to multiplex analog inputs x 8 using a relay output module.
You need to maintain the analog loops while not reading as they take a while to settle down.

It sounds as though you have decided to off-load the analog inputs, if they are used to trip the compressor perhaps consider a hard wired trip handshake rather than relying on communications.

Actually why not just spring for some reliable pressure transmitters, any industrial transmitter will be as reliable as the PLC

 

[itsmoked]

Hi Roy. I did spring for 8 pressure transducers and two 4-channel PLC 4-20mA modules. It should work well.

Keith Cress
kcress -

http://www.flaminsystems.com

 

[auto_eng]

You lost me at slow dump!! Are you trying to be safer or more reliable? I am curious what your voting scheme will be, 1oo2 or 2oo2?

 

[HCBFlash]

Golly Kieth, you're usually the one giving us the answers! Now c'mon, I'd bet one of the first PLC jobs you did that's worth remembering was,,,, somebody that wanted ("needed") more I/O! (On a Modicon?)

We knew you'd get it, and were just, uh, well,,, let's say "indecisive" at the moment. Cheers!

.

(Me,,,wrong? ...aw, just fine-tuning my sarcasm!)

 

[itsmoked]

auto_eng: Slow dump => drift with age or outright failure.

Are you trying to be safer or more reliable?

Nothing safety oriented. Two things:
1) Ability to run with a failed part. Using the usually shelved spares, actually in service, to keep the system(s) running.

what your voting scheme will be

I'll install them.
Run both systems on their same original transducers and study the two new ones. Make some educated decision on a difference between the three sensors that should be deemed a problem. When detected it will notify the car manager and disregard the odd-man-out. One of the the remaining two sensors will be designated "in control" and the system will continue until a spare sensor is installed.

HCBFlash;
All the goodies are supposed to show up Monday! So the fun can start. Unfortunately the first thing is running 1" rigid conduit all over under the car to separate the VFD stuff from all the signal wires.

Keith Cress
kcress -

http://www.flaminsystems.com

 

[auto_eng]

Keith - Trust me, I know all about "slow dump" ...I'm new to the board although been in the industry awhile.
If I understand correctly, you are installing 8 additional PTs to allow for a TMR configuration on your field instruments? Interesting approach, we have moved away from the "more is better" philosophy and use a variety of risk/failure analysis methods to determine design requirements.
Best of luck!

 

[itsmoked]

Hi auto! Welcome aboard.

These rail applications can be tricky because we're often in a moving train in the middle of frikken no-where. These cars often are dome cars with many dozen kW of glass since a large amount is perpendicular to the sun. Without refrigeration a car can become uninhabitable in literally minutes. Redundancy and agility in the control systems is required to prevent really expensive or massively embarrassing events from occurring. Imagine a small company has paid $30k for a five day fly fishing excursion and in the middle of Montana the car goes nova. It will always be in an Amtrak train but it isn't an Amtrak business. You suddenly need them to take your passengers... Or, they'll set you out on a siding in a car that's not habitable! The car has only guests and a car manager with very little technical abilities.

Nasty.

Hence the redundancy. Two power sources, Amtrak Head End Power (480V) or on-board generator.
Two refrigeration systems. Both feeding isolated halves of all evaporation coils. Each with their own condensers.

Currently they are both being controlled by a single PLC. Bad. Both are being controlled by single high-side/low-side pressure transducers. Bad.

We need backups. Since you can't step out at 90 mph you need them pre-installed. But wait! You can't know which one is bad with only two units when you can't run a meter and touch them. So the 3 up voting.

Keith Cress
kcress -

http://www.flaminsystems.com

 

[auto_eng]

Interesting! Good luck!

 

[HCBFlash]

Well Kieth, I'll go along on most of that, but if you've got the CPU and the memory space, you know as well as any of us that verification of state/condition by a different means improves reliability. Certainly there are multiple affected IO, the status of which is effected by your pressures enough to be able to sense trouble by way of cycle timing, ratios or ranges of operation times, repetition, etc. Redundancy has it's place, but also increases fallible component count. It'd be interesting to review this after a few seasons,,,,, kinda....

.

(Me,,,wrong? ...aw, just fine-tuning my sarcasm!)

 

[itsmoked]

Hi Flash. While in theory I agree, in practice in this situation it would be unworkable. The systems are very dynamic. If this was a fixed location with fixed insolation, slow moving ambient, and fixed loads I'd be all over it. But, in this service each compressor has three separate evaporators in very different load areas. The entire system can be near sea level in a 100F ambient and in an hour at 7,000 feet moving thru snow drifts. As I mentioned, every evaporator is two separate evaporators with the possibility of the neighbor running or not running. The vibration, twisting and stresses frequently create refrigerant leaks that would shift the system in negative ways. There would be so much information coming out of this system that it would be pretty hard to analyze and use. True, the more complex the information, likely the more encompassing the info could be... But it would probably be a year long engineering study to fish out any goodies.

I got the VFDs in today so I can ultimately control the head pressures. Got control of the misters for disaster control. All the pressure transducers are in and - will never leak. Now if the flaky cable supplier (Teflon twisted pairs) who's failed to ship for over three weeks now, will get his head out of his nether region I can wire the sensors in.

Keith Cress
kcress -

http://www.flaminsystems.com

 

[waross]

I imagine that the job is locked in now, so this is a general comment, not a suggestion.
I would have considered separating protection and control. One Click for the protection of each system, and a third Click for overall control of both systems.
Any comments?
Yours
Bill

Bill
--------------------
"Why not the best?"
Jimmy Carter

 

[itsmoked]

From one PLC emergency install..

To the two PLCs with some time to install them.

Not sure where I'd put a third one Bill. I still have to install two touch screen status displays.

I think the three PLCs scheme doesn't apply here in the case where all the systems are split so functions are covered and re-covered by the two PLCs. If one dies you still have most functions covered. Also this isn't 'life safety' so you probably don't need a PLC to call a complete halt and stop the hurtling truck or something.

Before you guys harass me about the wire dressing this is a work-in-progress..

Keith Cress
kcress -

http://www.flaminsystems.com