Voltage sensing switch with deadband?

[Gashmore]

I need to maintain pressure on a hydraulic system within a narrow +-3% band for some tests but the commercial precision pressure switches are way out of my budget . I have a 4-20ma pressure transducer that I can drive at 24V with a maximum 700 ohm shunt resistor (Say 680 ohms). I need a switch that will turn the pump on when the voltage drops below 4.35V and off when it reaches 4.45V.

I have been experimenting with an LM1458 dual op amp and some 10K trimmer pots I had on hand and can get a pair of LED's to light as the input voltage goes high or low outside the .1V range but I am having a brain freeze on how to latch a relay on when the low op-amp turns on and off when the high op-amp turns on. I can change the logic by reversing the op-amp inputs but any way I do it the pump will turn off as soon as the pressure gets within the range. There will be no deadband so the pump will cycle rapidly. Ideally the pump should cycle for maybe 10 seconds every 4 or 5 hours.

Any suggestions?

 

[ericgibbs]

I need to maintain pressure on a hydraulic system within a narrow +-3% band for some tests but the commercial precision pressure switches are way out of my budget . I have a 4-20ma pressure transducer that I can drive at 24V with a maximum 700 ohm shunt resistor (Say 680 ohms). I need a switch that will turn the pump on when the voltage drops below 4.35V and off when it reaches 4.45V.

I have been experimenting with an LM1458 dual op amp and some 10K trimmer pots I had on hand and can get a pair of LED's to light as the input voltage goes high or low outside the .1V range but I am having a brain freeze on how to latch a relay on when the low op-amp turns on and off when the high op-amp turns on. I can change the logic by reversing the op-amp inputs but any way I do it the pump will turn off as soon as the pressure gets within the range. There will be no deadband so the pump will cycle rapidly. Ideally the pump should cycle for maybe 10 seconds every 4 or 5 hours.

Any suggestions?

hi,
Configure the LM1458 as a limit comparator.

EDIT: as the LM1458 is a OPA its not suitable... Look at the LM393 dual comparator.

This is an example:

 

[Willbe]

hi,
Configure the LM1458 as a limit comparator.

EDIT: as the LM1458 is a OPA its not suitable... Look at the LM393 dual comparator.

This is an example:

And, if the readings change slowly the OP may want to add some hysteresis, maybe 1/10th or 1/20th of (4.45v - 4.35v).

 

[ericgibbs]

hi,

This circuit can be set for a 0.1V deadband

 

[Willbe]

turn the pump on when the voltage drops below 4.35V and off when it reaches 4.45V.

Reading this more carefully I see that you need 4.45 - 4.35 = 0.1 = 100 mV of hysteresis. You can do it with one comparator.

You can use an opamp as that comparator; the output slew rate is not as high but you may not need a high slew rate anyway.

What's your relay coil specs?

 

[Gashmore]

AH HA! I knew there was a simple way to do it. I have some 393s on hand. Somewhere around here I have some 6V2s also if I can find them.

I have several relays on hand. I was considering an Omron G2RL-14. Coil reads about 350 ohms.

 

[Willbe]

Hysteresis calcs can be weird.
Let's say Vref into the - terminal is 4.40v
The signal goes into the + terminal through a, say, 1kΩ resistor and the output swings exactly 12v, assuming your relay has a 12v coil.

As the signal in swings from a high value down to almost 4.35v you want the input pin to see almost 4.40v. 4.40v-4.35v = 50 mV across 1kΩ = 50 µA, and since the + terminal doesn't draw current this same current must go through the feedback resistor.

The comparator hasn't fired yet so the output voltage is still 12v.
(12v-4.4v)/50µA = 152kΩ.
Now, the signal in drops just below 4.35v, the output switches to 0v, and so the + terminal now sees 4.35v(152)/(152 + 1) =4.32v.

For the Vref terminal to see 4.40v again, the signal in has to rise to 4.43v (not 4.45v!). Now I remember why I don't like hysteresis calcs!

You can make the output signal sense correct by running the relay either down from +12v or up from ground.

 

[Roff]

How precise and how stable (with temperature) does this need to be?

 

[Gashmore]

How precise and how stable (with temperature) does this need to be?

It has to be pretty tight. There can be some variation but the rig will run 24/7 unatended for 2 months so what ever I do has to be pretty reliable. The application is to test the "creep" or long term elongation of a high tech rope under constant static load. The entire test stand is enclosed in an insulated box held at 80F and the load will be a constant 10% of breaking strength (2200 lb +- 100lb). The hydraulic cylinder pressure has to stay between 215 and 235PSI.

An LVDT supplies the elongation to a data logger and a pressure transducer supplies the tension. A second pressure transducer controls the hydraulic pump. The data logging part worked well for the dynamic load tests but I can't start the static test until I figure out how to control the pump. Also there is no money in this project so it has to be done on the cheap.

 

[Roff]

Here is the most complex Schmitt trigger I have ever designed. I'm embarrassed to post it, but it will be very temperature stable, and should be almost perfectly independent of the value of Vcc, within the range I specified. It is very easy to adjust. If you adjust R6 first, the other adjustments are noninteracting. You can select the polarity of the output with a jumper, and the output will drive a MOSFET if it doesn't switch too frequently. It will also drive a BJT.
I have been accused of posting some Rube Goldberg designs. If someone else can come up with something simpler with the same features, I'll crawl back under my rock.

 

[on1aag]

Hi Gashmore,

Here's another possible solution to your problem.
It's a little bit less Rube Goldberg like.

on1aag.

 

[Roff]

Hi Gashmore,

Here's another possible solution to your problem.
It's a little bit less Rube Goldberg like.

on1aag.

Much better.
I think you need to swap the set1 and set2 inputs on your schematic.

Signed,
Rube

 

[on1aag]

Hi Ron,

You're right, there's is an error in the schematic.
But the simulation was ok ! The top resistor, 30k, goes
to the top trimpot and the wiper of this trimpot is
connected to pin 6 ( set 2) of the ICL7665. (The short
yellow wire below the trimpot.)
The short grey wire below the lower trimpot goes to
pin 3 (set 1) of the ICL7665.
I knew that my post might bounce back, as it did.
Thanks for the correction Ron, I hope I'm off the hook now.

on1aag.

 

[Roff]

Hi Ron,

You're right, there's is an error in the schematic.
But the simulation was ok ! The top resistor, 30k, goes
to the top trimpot and the wiper of this trimpot is
connected to pin 6 ( set 2) of the ICL7665. (The short
yellow wire below the trimpot.)
The short grey wire below the lower trimpot goes to
pin 3 (set 1) of the ICL7665.
I knew that my post might bounce back, as it did.
Thanks for the correction Ron, I hope I'm off the hook now.

on1aag.

I found the error by simulating on LTspice. I'm too lazy to analyze it on paper.
Nice solution!