Help designing Resistance Meter

Thread Starter

driller

Joined Nov 24, 2009
13
I am attempting to utilize a Linear Resistance Meter circuit I found at: http://english.cxem.net/izmer/measured8.php. Rather than making it as a "range-able" meter, I am fixing it at the 10K Ohm range. The resistance I will be measuring will be from a digital potentiometer but that's not important right now. For now I am just using a fixed resistor.
Has anyone had any experience with this circuit or could you provide some insight? I would like to use a digital LED display to show the resistor's value but am having some issues with the output voltage of the circuit to the display. Could you recommend a type of digital display to use?
 

someonesdad

Joined Jul 7, 2009
1,583
I can't provide any experience, but I would ask why you wouldn't just use e.g. a cheap digital multimeter to make the measurement -- a $3 Harbor Freight meter would probably make the measurement you need and with very little construction. :p
 

Thread Starter

driller

Joined Nov 24, 2009
13
Without going into too much detail which would only add to confusion, it will end up being a panel mount display on a control box I have made so I was looking to customize it.
 

wayneh

Joined Sep 9, 2010
17,496
I'd still just hack apart the working DMM and customize it to your heart's content. I believe they're labelled CenTech and are often free with any other purchase at Harbor Freight. Battery included. I probably have a half dozen of them. ;)
 

#12

Joined Nov 30, 2010
18,224
I had trouble with segments dropping out when I did that (butchered a Cen-tech). If you can leave the front (with display) assembled, you won't have that problem. I wound up spending $10.50 for a ready made, panel mount meter. Just add 5 volts and declare your decimal point.

ps, that cxem.net design is temperature sensitive. (The readings will drift with temperature changes.) Make a choice depending on how accurate you really need it to be.
 

Thread Starter

driller

Joined Nov 24, 2009
13
Thanks. I have a couple of those DMM's and considered it but I don't like the fact that it limits my digits and decimal placing (i.e. 6800 = 6.80 in the 20K range setting). It would work but I would rather be able to see the full value. Attached is a basic block diagram of what I'm attempting, which is basically a DMM.
 

Attachments

#12

Joined Nov 30, 2010
18,224
It seems to me that you could make a constant current source of 1.00 milliamps and put that current through the potentiometer. The result would be 10.00 volts if the resistance was 10.00 k ohms and likewise, 1.00 volts if the resistance is 1.00 k ohms.

Make sense?
 

Thread Starter

driller

Joined Nov 24, 2009
13
I think that is what the first half of the Linear Resistance Meter circuit (link to circuit in first thread) is supposed to do. I will look more into it... Thanks
 

#12

Joined Nov 30, 2010
18,224
You're right. IC1A and Q1 are trying to be a constant current source. My problem with that is that the diodes used to make a fixed voltage can change with temperature.

I would change the 2 diodes to a resistor and use a j-fet at its zero temp-comp point to make a constant current device from the new resistor to ground. (I'm going to call the new resistor Rnew, now.) Then you have a stable current through Rnew, and IC1A and Q1 will work better. Do you know how to do a zero temperature drift current generator with a j-fet?

What I'm trying to get at is that if you only want to measure one resistance, you can eliminate a lot of parts. Just pour 1.00 milliamp into the test terminals, connect the 10k pot, and the fixed current will cause a proportional voltage. That circuit you offered can be a lot less complicated and a little more accurate
 

Thread Starter

driller

Joined Nov 24, 2009
13
"Do you know how to do a zero temperature drift current generator with a j-fet?" No I don't.

I did some testing and realized my "constant" current changed when I changed the resistance being measured. It will be a variable resistance from project to project. Maybe I'm not understanding this application...
 

#12

Joined Nov 30, 2010
18,224
Look at page 3 of the J201 document. Top left, transfer characteristics.

You will see that all 3 temperature lines cross each other at about 2ma and negative 3.25 volts.
(ps, ignore the 1maMax label I put on that file. It's for a different circuit.)
Give the j-fet a positive supply voltage on its drain terminal. A 9v battery will do just fine. When you use a resistor from the source to ground and ground the gate, the idle current will be diminished. You start by trying 3.25V/.002A = 1625 ohms. Then use a hair dryer while watching the current through the j-fet with your current meter. Adjust the resistance until the current remains stable when the hair dryer is applied. If the current increases with heat, you have too much resistance. You can get the current to remain stable within 10 or 20 ua from room temperature to "too hot to touch". That's 1/2 to 1 percent change for maybe 70 degrees F. When you use this circuit without the hair dryer, you get way better than 1% temperature drift.

One of the limitations is that the j-fet needs a few volts of "headroom". That is, Vcc must always be a few volts higher than the voltage across the source resistor.

With the circuit you provided, you can set Rnew to drop 1.00 volt and use that for the reference for IC1A and Q1. You use up about 3.25 volts in the source resistor and still have 4.75 volts (plenty) of "headroom" for the transistor.

ps, the circuit you provided uses a 9 volt battery. That will not get you a 10 volt display on a 10k pot, but you already knew that.
 

Attachments

Thread Starter

driller

Joined Nov 24, 2009
13
@ #12> Can you clarify what you described in post 9? Are you saying to use the j-fet in replace of Q1? Is Rnew replacing the diodes making the reference voltage for IC1A at the negative input? I guess what I'm asking is are you offering a new circuit entirely or how are you suggesting I change mine?
 

#12

Joined Nov 30, 2010
18,224
Replace R1 with an n-channel j-fet (like a J201) and a resistor (somewhere between 1k and 1.5 k), configured as a constant current sink.

Replace D1 and D2 with a resistor.

That is all. Replace 3 parts with 3 parts and get better than 1% stability.
 

#12

Joined Nov 30, 2010
18,224
Excellent!

Now you do the hair dryer adjustment to find out exactly what current your particular j-fet wants to idle at, then choose a resistor for R1 so it causes the amount of voltage you want (at the current the j-fet is allowing).

The voltage you choose to be across R1 will also be across R3, and that is how you set 1.00 milliamps to go into the potentiometer.

In case you didn't know, j-fet idle current is the sloppiest specification I've ever seen. Every one of them wants to idle at a different current! I found that 1.2k was the right resistor (for the j-fet I had) when I did this.
 
Last edited:

Adjuster

Joined Dec 26, 2010
2,148
This exercise with JFET characteristics is no doubt most educational, but nowadays there are also plenty of band-gap reference devices providing accurate voltage standards. Substituting a bandgap device of around 1.2V in place of the two silicon diodes in your circuit should give a stable result.

You might still want to consider replacing the output device with an FET though, as the present BJT circuit has a temperature drift risk from varying alpha. Just be careful if using a very small JFET though if you want currents of more than a few mA, that your wanted current does not exceed IDSS. The gate junction could then turn on, certainly ruining the accuracy, and possibly cooking the FET.
 

#12

Joined Nov 30, 2010
18,224
@ adjuster
I expected somebody would chime in with news about modern references long before now. I have no objection. I was only being educational about a way to provide any voltage you want using cheap, hobby parts. Almost any small j-fet will work for this purpose, and they beat the pants off a diode for temperature stability. Great for a one-off circuit. Unnecessarily expensive to calibrate each one for a production run.

ps, you're right about using a j-fet for Q1. Completely eliminates base current from the equasion. A J201 would work in that spot. (I predict an order for J201 transistors.) It would require reversing the inputs on the op-amp.
 
Last edited:

crutschow

Joined Mar 14, 2008
34,281
If you replace the diodes in the reference circuit with a accurate shunt regulator reference, then the current measurement will be stable with temperature.
 

Ron H

Joined Apr 14, 2005
7,063
I would do it like this. The LT1009 is not cheap ($2.60 at Digikey), but it has a simple ±5% trim scheme, and is very stable with temperature.
I substituted the LM358 (cheap, but OK for this app) because the CA3240 has a common mode range the begins 4 volts below the positive rail, so it wouldn't be guaranteed to work with a 1.4 or 2.5 volt reference. LM358 CMR begins at 1.5V below the positive rail.

You could use LM336 as a reference (cheap), but it has a tempco trim scheme (requiring a pot and 2 diodes)) that requires that it be set at 2.490V, which means you need another pot for calibration. Tempco is not specified without the pot and diodes.
LM4121 is another possibility, but it might be hard to find in a thru-hole package.
 

Attachments

Last edited:
Top