Created Date: October 12,2007

Created by: Cap

I 've been playing with temp measuring in a Wine Tank Farm, for a few years now. The Farm has 2) sets of tanks, One at 31 Tanks, And the other at 17 tanks. The Route that I chose was to use an Analog Devices Temp Chip #AD592an (looks like a 3 leg transistor ) and a single analog Volt Input card per each Tank Set. The scheme is to use a relay input from an output card ( in the 450 i'm using )to switch each sensor to the analog card one at a time. The current from the AD592an is droped across a 10k ohm resistor that has a 1,000 MF cap hanging across it. The resistor is used to develop a volt siginal that the analog card can read ( in the 70 deg f range about 2 volt).. and the cap is there to smoth out the electrical garbage that is present. The 'scann ' time for the 32 stations of the large tank farm is about 2 minutes. I chose a slow scann time as I have introduced a lot of electrical noise in the control wires by running the control wires inside of the SAME electrical conduit that is controlling the cooling water selonids for the tank farm.. I understand that it is a No No to due, but when you have your own gun to your head to get the project done, and I had a soultion for the noise problem.. So I did it.

I am currently designing up a tank farm controller that will control and monitor a 128 tank farm.. And I'm looking to do it with a 4 input analog input card, and still have my 2 minute 'scann Time '.. It will use 128 input relays to switch the AD592an to one of the 4 channels of the analog card.. And it will use 128 a/c outputs to turn on the cooling water selonids..

Cap

Created Date: October 12,2007

Created by: FACTSTech

The F0-04RTD is designed so that an RTD can be wired directly to the terminal block and you will get a direct temperature reading in the PLC.

The F0-04THM is designed so that a Thermocouple can be wired directly to the terminal block and you will get a direct temperature reading in the PLC.

The F0-08ADH-1 is a 0-20mA input and the reading in the PLC will be a value between 0 and 65535 proportional to the mA input. This will work if you have a temperature sensor that will output a 0-20mA signal. In order to get a temperature reading you would have to use ladder logic to scale the 0-65535 reading to the temperature range of your sensor.

The question of which module is appropriate is determined by the temperature sensor you will be using.

Created Date: October 13,2007

Created by: AZRoger

Originally posted by Cap:

... I have introduced a lot of electrical noise in the control wires by running the control wires inside of the SAME electrical conduit that is controlling the cooling water selonids for the tank farm ...

I am currently designing up a tank farm controller that will control and monitor a 128 tank farm.. And I'm looking to do it with a 4 input analog input card, and still have my 2 minute 'scann Time '.. It will use 128 input relays to switch the AD592an to one of the 4 channels of the analog card.. And it will use 128 a/c outputs to turn on the cooling water selonids ...

Cap

Cap,

Concerning solenoid noise: Since you are controlling the solenoids with the PLC you can know exactly when the noise will be present. It sounds like you switch at most one (or 4 with the 128 tank solution) at a time, just after you determine that you need to. And this occurs once every 4 seconds. You have at least a couple of "quiet " seconds to do your sampling during each 4 second interval.

Concerning input selection: You can get by with many fewer than 128 relays to switch your inputs if you use multipole relays, such as 4PDT or even more. I'll explain in words here. If a diagram would help, I'll figure out how to get one on the this thread and post it.

Start at your 4 channel analog module. Wire the commons (4) from one relay (call it A) to the input module and you have handled 8 inputs. Connect the commons of a 4PDT (B) to the NO contacts of A and another 4PDT (C) to the NC contacts. Now you have handled 16 inputs. Repeat the process with four more 4PDT (D,E,F,G) and you have covered 32 inputs. Repeat with eight more relays and you have 64 inputs. Finally with sixteen relays, expand the tree to 128 inputs. The total count of 4PDT relays is 31. The number of outputs needed to drive the swithcing logic is 5. You might need another relay or three to act as amplifiers when pulling in the 16 or 8 or maybe even just 4 relays in parallel.

This only works for the inputs as only one (4) of these are monitored at the same time. You'll still need 128 outputs.

This is an interesting application. By visiting each input and output on a two minutes cycle you make sure (a) you don't turn all the solenoids on at the same time (causing a blackout on the farm) or (b) chatter the outputs due to minute differences between samples. Have you built a hysteresis into your ladder logic. For example, pump/valve solenoid on at 72 deg and off at 71 deg? Do you find that many of the tank cooling circuits toggle every two minutes when they 're just about the right temperature? Do you need to monitor the different in temparature between the top and bottom of the tanks? Are there circulating pumps during fermentation? Just curious. I'm not planning to start my own winery -- the stores have plenty to meet my needs. http://forum1.automationdirect.com/board/smile.gif

Roger

Created Date: October 14,2007

Created by: Cap

AZRoger: I like your solution for the 128 Inputs..I'll scamm on that and try to figure out a reason that I would not want to do it..!! Till then I'll lay it out on the board and see how it fits..especially on the 'Bit Math '.. you'll se later...

All three of the tank farm controllers I have built use the same control wire scheme, and that was to 'Heat up ' the plus side of all of the sensors and 'switch in ' the output side of the sensor to the input card. I had a choice when I was laying out the circuit, and I chose not to have a 'Common Input ' and switch each sensor to the Hot Side, as any problems with a sensor could stop the entire tank farm from reading..

I also decided to use only one of the inputs on the A/D card as, the programming would be simpler.

The PLC can handle 32 bit math so one bit represents one tank, and I step through the program 32 times, each time looking at the next bit over till I get to the end, then I recycle and start again.

The control logic was written on paper then entered on a hand held programmer!! yes it was a chore, but it was not written in RLL.. I used mnemonics and and referenced the bit positions directly in the I/O V locations.. I'm and old micro programmer and mnemonics was not a problem..but the typing in on the handheld was a CHORE... I have now graduated to a lap top with DS5.0 ..It's difficult doing ANY changes to the old program, as all of my notes ( and my brain) are in mnemonics,but DS5.0 can not be written in mnemonics, only viewed... and not an X or Y is to be found on the documentation... The next guy to follow will have a hell of a time unless he ( she ) speaks Mnemonics!

I do have a Hysteresis as to the temp of the tank, and the V out registers are my Tank On Bits.. So I don't have to make a separate register for that, I just reference the V location bit position for tank cooling water solenoid as I'm doing the temp calculations.

On the smallest tank of the bunch, it'll turn on or off every 15 minutes or so.. The largest tank will be lucky to change a deg per hour.. and I use a 2 deg spread..

If you haven't figured it out by now I am using the DECO instruction to stand up the bit positions from 1 to 32. All of the math is bit math, and I use a 'mask " that is 32 bits wide to find out what bit I'm working on this time.

There are a few data tables that hold byte data of tank temperature (*10) from 20 deg f to 90 deg f. These are 'legal ' values, as nothing else is reasonable in a winery! Also I have a 'Forced Off ' bit. This is if the wine tank is COLDER than the cooling water, do not turn on. And a 'Capture Zone Bit '. This is used for the 'ALARM ' feature. if the tank gets to temperature, than falls out of the 'Hysteresis Band ',(either over or under temp) then It will send out a Flashing Lite Alarm.

The Display is using a DV-1000 display, it was a bit of coding to get the display to show tank temps, tank numbers, set point temps, forced of and cooling water coil status.. But it came out and is workable.. It had to be done in a 440 or 450 ( at the time ) plc as none of the other plc would allow 32 display lines in the dv-1000.

EW57: The AD592an is soldered to beldon comm wire, with shrink tube over the soldered leads. I have now learned that a shrink tube is available with a 'Ring of Solder ' inside of it for soldering to electronic 'Pins '.. I'll use that one next time. In the first generation of sensors I made, I took a 3 " length of 3/8 stainless tubing and welded a 5/16 ss ball bearing on one end. I then filled the 'tube ' with a potting compound and pushed the sensor and wire into it. I then slid a piece of shrink tube over that and heated it. that held it all in-place till the potting compound set. The sensor was pushed all the way to the end of the tube and is sitting up against the back side of the ball.. The first batch I built 50, for the 17 and 31 tank farm.. Only had one 'Failure ' and that was a sensor that was 'slow to respond '.. I figure that the sensor was glued to an air pockett, behind the ball.

The Sensor is then inserted into a 'thermal well ' that was welded into the tank when it was manufactured.

For the Third system that we made, I hooked the AD592an to the beldon wire, then covered the sensor and wire with adhesive lined shrink tubing. And left the sensors this way. These sensors were then GLUED to the side of the stainless tanks. This was chosen, because the entire tank farm had the thermo wells too high for wine fermentation. And sometimes the wine was not even at the thermo well during fermentation! Rather than move all the wells ( empty the tank first! ) we decided to try this glue approach.. It Worked!.. Took a while to get the proper glue and tape to hold in place till the glue dried.. On a sweating tank... But we finally figured it out..

On the 128 Tank Farm, I get to graduate to touch screen!.. cant wait to figure those rascals out!!

Cap

Created Date: November 07,2007

Created by: EW57

Cap,

I figured out the 10k was necessary to get 10mv/k. I see that you 've got to convert Kelvin to Fahrenheit to get a linear reading & found the necessary conversions (° F = 9/5(° K - 273) + 32), sounds like alot of math fun!

Did you ever try the National LM34? (http://cache.national.com/ds/LM/LM34.pdf)

Thanks again!

Created Date: November 07,2007

Created by: Cap

Did not like the '34 ' because of the needed components.. I just used the 24 dc outof the 440 that we used.. I fuse protected the 'Bus Rail Line ' with some 1/4 amp fuse.. Just to protect the power supply..

For the math.. I was trying to get .1 deg accuracy .. the A/D card was set up for 1-5volt.

looked over my notes (from 97-98!!) and I think I took 16 samples added them all together and then did the math on that TOTAL.. not the avrage of the total.. that afforded me better accuracy. all the math was done in base 16 math, so the 16 samples worked well. My notes show 3.109 volt@ 100F 2.665 volt@20F and 2.554 Volt 0 f.. that's 0.00555 volt/Deg f of 5.68 steps ( counts of the card ) per deg f.

The math that I did allowed me to not only change the 'Height ' of the conversion line known as offset error ( steps to temp display ) it also allowed me to change the angle of the line ( as used in Ph meter calabration) that would give me a gain error correction.. I will post the math from my notes.. it does not imply that I fully understand it any more!!

Temp=(((((Count+Adder)*16)*625)/(4264+divider))-1589880)

Count = total from the AD Card(Read 16 times and added)

Adder= correction factor to 'Bump ' display temp up or down a few counts..(maby!)

4264= I think that is a fixed number

divider= 1422 that is the slope of the line.

1589880= I think this is the number to subtract to get to zero deg f from K.. I think..

and the 'short version ' of this formulia I found was reduced to

Temp=(((Count*16)*625)/(4264+1422)).

Either case you get the idea, do the math on the largest number possable so the 16 reads will average the errors to a better accuracy than possable with a single read. The results were a Times Ten of the actual temp.

Cap

Created Date: October 12,2007

Created by: EW57

Greetings all! For strictly comparative measuring, (measuring from approx 50F-250F) is there any compelling reason(s) for using an RTD/thermocouple card vs. a regular analog input card? My application is for a DL06 & the new 8 channel cards look pretty attractive compared to the 4 channel temp cards while being dramatically less expensive at the same time. My asking of this question alone may have exposed my lack of knowledge on these sensors, so please steer me in the right direction! Thanks again!