monitoring 24 vdc power supply with analog inputs ?

hi everyone


i have 2 24 volt dc power supplies that i need to monitor the voltage on . using ab slc 5/05 processor . i have a ton of 1746-ni4 analog input cards laying around can they be used for this . does anyone know how to scale 24 volts . i normally use scp but with 0-10 , 4-20 and so on . any ideas would be most appreciated

Yes, you can use the NI4 to monitor your 24V supplies, but you will have to be a little creative. The NI4 will only accept -10V to +10V or a 20V differential, and that leaves you 4 volts short for your 24. You will have to scale your voltage down by using a voltage divider network of resistors and then scale your analog input as if it were seeing the 24V. Use some fairly large resistors so as to not load your supply down. For example, two 10K resistors in series, 20K total, across your supply will give you a 12V drop across each one and you can use 12V as input to your NI4. Once inside the processor, you can use the SCP or SCL or other method to make it look any way you want.

i didn't use NI4 in a while but if i recall correctly (maybe i'm wrong but i'm lazy to check it out...) the card would fault if voltage exceeds 10.3V or so. it may be a bit higher but without checking manual i would think that 12V could be a bit over limit.
Anyway you don't want to be too close to the edge (what if someone cranks the output to something like 26 or 28V?).
you could use standard values like 4.7k and 2.2k to get the range of about 0-32V with 0-10V PLC input. power rating for resistors could be 1/4W or even 1/8W since total power dissipation for both resistors at 32V is only some 150mW or 0.15W. current draw of the divider is only about 3.5mA (miliamp) at 24V.
don't pick too high values for resistors, you want to keep it well below impedance of the card. too small values and the thing will smoke or get hot at least...
you can also configure card as current input 0..20mA. In that case you have one of the resistors already inside the plc card so all you need to do is add one externally (upper half of voltage divider). in this case you would get up to 5x bigger current through divider and at 20mA and some 20V drop, that resistor better be 1/2W or even 1W.

As voltage goes up, resolution goes down. You don't want to look at the entire 24 volt range. Just the last 5 to 7 volts will do. A 24 volt supply running at 18 volts is failed, so you don't care about the 0-18 volt portion.
Use the 10 volt input mode. Figure a maximum of 28 volts, and low end cutoff of 18 volts.
Put a 18 volt Zener diode in series with the line. 1N4112. Mouser, Digikey, etc. Use a 1 uF, 35 volt Tantulum cap across the input to filter the noise from the Zener. If you're stuck going to Radio Shack, 18 volt may be hard to find. You can use two 9.1 volt Zeners in series, or any other combination.
That's a $2.00 solution, and nothing gets hot.

To make it safer, add a 250ma slow blow fuse in series, and a 10 volt, 1 watt Zener across the input. This will protect your card from overvoltage. It probably has an internal Zener, but I doubt that it has a 1 watt rating.
The resistor divider solution sounds easier at first, but heat is a problem. The Zener circuit will run much cooler. Radio Shack (and others) have small prototypings PCB's, and matching project boxes.
The 1N4112 is a 1/4 watt Zener. You could also use 1/2 or 1 watt types.
If the A/D input impediance is high, the Zener might 'sawtooth' in and out of conduction. This will appear as a few millivolts of noise. Add a 10K, 1/4 watt across the input to lower the impediance.

The series string is: +24 - 1N4112 - fuse - +A/D input.
Parallel circuit is cap, 10V Zener (1N4740A), and 10K across A/D input.
Can someone post a schematic for me?

Zeners drift

Forget the zener. Unless a zener diode has a significant amount of current flowing through it, it is not going to drop its rated voltage. Even with a several milliamps of current, the voltage drop will not be stable. With the few microamps of current the Analog Module draws, the voltage drop will be nothing near the rating of the zener. The little you gain with the added resolution will be more than lost with the reduced accuracy.

Use a voltage divider as others have suggested. With the analog Module's approximately 750 K ohm input impedance, you can use fairly high resistance resistors without loosing accuracy. If you use 2 20K resistors is series, the total heat dissipated by both resistors will be .015 watts. I doubt this much heat will cause a problem.

I haven't done this yet, but I have the same goal, and this is my intended approach:


http://www.plctalk.net/qanda/showthread.php?t=20647

Vic, I agree. I wish I could delete that post.

Put the module in 0-20 ma mode. That should give a 250 ohm input impedance. For voltage modules, add a 250 ohm, .25 watt resistor across the input, and set it for 0-5 volts.
All you need is a 1K, 1 watt resistor in series.
Set scaling for 0-25 volts.
Current will be 20 ma at 25 volts and .4 watts.

No, Keithkyll does have a point! Simple resistor dividers are a real killer when it comes to resolution. 99% of the time, loggers will show the exact same value even when the supply swings for 0.1V.

Why not use a 3-legged 7918 regulator (NEGATIVE 18V) and measure its differential voltage (between IN & OUT, but in this case IN being the 0V or common)? I know most people don't have them immediately within reach but you would be amazed when measuring small supply voltage swings.

Not sure what's the card's resolution but let's suppose it's 12 bit: Using the divider panic proposed, 24.0V will result in digital value of 3134dec. If the supply is raised 0.1V its corresponding value will be 3147dec, meaning a raise of 0.415% (just like 0.1/24).

When using the 7918 regulator 24.0V will result in a (differential) measuring voltage of 6.0V, corresponding to a value of 2457dec. If the supply is raised 0.1V the measuring voltage will become 6.1V (as the regulator will keep its output voltage steady to 18V) thus raising the value to 2498dec, meaning 1.669% or 4 times better than the divider setup.

No need to tell you what happens if the card's resolution is nowhere near 12 bit...

A differential voltage of 3V is an absolute minimum so the measuring range will go from 21V to 28V. I've also heard of low power 79L20 types, needing only 1V differential, but haven't found them yet.

I'll try to post a schematic later on.

As promised...




Use of any capacitor is not recommended (spikes or gaps will be filtered out).

Permutations!

phil buchanan said:

hi everyone


i have 2 24 volt dc power supplies that i need to monitor the voltage on . using ab slc 5/05 processor . i have a ton of 1746-ni4 analog input cards laying around can they be used for this . does anyone know how to scale 24 volts . i normally use scp but with 0-10 , 4-20 and so on . any ideas would be most appreciated

Click to expand...

Is it not amazing how we got from here, to where the last post is? I love it!

thanks

thanks for all the help everyone . now i got to get the components ordered .

i hope i didn't miss something but i never saw anything about precission
of the measurement, just monitoring of power supply. but if precission is required, i would
still definitelly look into voltage divider. first of all it's dead simple and cheapest
solution possible. second it's far more stable and accurate than solution using
voltage regulator or zener. also NI4 is 16-bit card, but since we are looking in single
ended application with no sign, that's still good 15-bit or 32767 counts.

idea of using zener or regulator is novel but unless one uses precission voltage reference
and precission instrumentation opamp, results will be just poor. voltage regulators
(specially the old 78xx or negative version 79xx) are not going to give nearly as good results
as simple voltage divider. even datasheet says some 300mV output error at only 25degC
and it gets only worse. i know because i used them a lot.
with given working range of only 7V (21-28V) and error as large as 300mV...
once this is scaled 4x to give us reading in 24V range, errors will be multiplied as well
and be up to some 1.2V! sure the output wont continuously swing up and down +/- 0.6V but still,
this is not even close to precission measurement...

on the other hand (without hair splitting and making this too long), proposed voltage divider
for example with range of 0-32V, and 32767 counts (15-bit), would get somting just under 1mV...

Agreed. I was thinking hot resistors and overvoltage blowing the module.
Later (post 7), I came to my senses and recommended a 1-resistor solution with the card in 0-20ma mode. Will that work on these AB cards? With a 10 volt input channel, that should tolerate a 100% overvoltage condition (at least short term).

The other info posted by Sparkz is still a good solution if he stays in voltage mode. Also good for other applications. Not sure if a resistor needs to be added at the A/D input to increase load. I expect operating temp to be stable, so thermal drift shouldn't too much of a problem. Even if a little drift occurs, the regulator still offers protection.

Sparkz said:

As promised...

Use of any capacitor is not recommended (spikes or gaps will be filtered out).

Click to expand...

Hello there I am very sorry for digging out an old post but I have one questian.

Will this circut work with 12v if I was to scale this right.
Many thanks Steve