Suitable Comparator Needed

Hi there,
I've been trawling through datasheets without finding what I want, or I'm failing to see it if it's there.
I want a DIP Quad Comparator, for sensing and indicating polarity in a voltmeter assembly. (Ok, singles or duals if necessary.)
I want to use a 9 to 12 volt power supply, but I want to be measuring +/- 25 Volts.
I am having difficulty in finding one that combines these last two requirements, as most appear to be limited to having the maximum differential input fairly close to and within the supply rail levels.
If you know of any suitable chips (or a neat workaround), I would greatly appreciate the information. Thanks for any help you can give.
Circuit below.

Hi!
You can 'step down' or devide the input voltage using a circuit called a voltage devider. This circuit consists of two resistors connected in series and a split in the middle, see the crudly drawn paint circuit attached. The output voltage can be calculated very easily using only ohms law. The current though the circuit equals the voltage devided by the resistance of both resistors. The next step is to calculate the voltage of the output, this is equal to the voltage drop of the second resistor so it can be calculated by deviding the total current with the resistance of the second resistor. Putting these 2 steps together creates the following formula:
Vout=(R2/(R1+R2))*Vin
By calculating the circuit correctly you can step down the input voltage to a voltage usable by the opamp.
Regards,
Boi

My first thought was a voltage divider as well... When Frank mentioned Quad Comparator in a DIP package the "LM339":http://www.ti.com/lit/ds/symlink/lm339.pdf came to mind.

Yes it is pretty much the 'standard' comperator. If the application isn't too critical (e.g. doesnt need to be that precise/temperature stable) i would also reccomend this chip since its cheap and easy to find.

One more thought on the voltage divider. Frank may want to use a potentiometer for one of the resistors so that he can calibrate out any error due to resistor tolerance.

Yes, but make it both a resistor and a trimmer, so it's not so sensitive--otherwise, there will be too much change with each twist of the diddle-stick (actually, in this case, a simple flat blade screwdriver will do, but I couldn't resist digging up that old term ;)

The idea is to get the two fixed resistors in the "ball park", and then take around 2% of the lower resistor's value [this assumes the other resistors are 1%, as they probably should be], adjust that value to the nearest standard trimmer value, and then subtract 1/2 of that trimmer value from the lower resister value and put the trimmer in series with the lower resistor (see attached diagram). Thus, the trimmer can be used to "dial in" the correct ratio between the upper leg of the voltage divider and the lower leg.

Note: when choosing a trimmer value, consider that most trimmers have a 20% tolerance. So, design for the lowest case value -- i.e. if it's a 500K 20% trimmer, the lowest value will be:

500K * (1 - 20/100) = 400K


Let me know if you need an example.

OK, you twisted it out of me. Here's an example:

Say you want a divide by two function with a 2 meg input resistance. Nominally, each resistor is 1 Meg with 1% tolerance. To achieve something closer to 0% tolerance, add a trimmer, thusly:

# Take 2% of the lower resistor value: 1M *0.02 = 20K
# Chose a trimmer value that is closest to that value. You might be able to find a 20K trimmer, but a more common value is 50K so lets go with that.
# Let's say it has a tolerance of 20% (as it probably will), so the lowest value is 50K * 0.8 = 40K
# Subtract half that value from the lower resistor value: 1M - (40K/2) = 980K
# The closest 1% standard value is 976K, so we'll go with that. Because the trimmer we are using has, worst case, twice the needed range, it will easily cover the 4K variance between our nominal value (980K) and the real world value (976K).

See attached diagram for the result

Hi all, thanks for your input.
I have carefully studied the NS LM139/LM239/LM339/LM2901/LM3302 datasheet, but it says:

"Differential input voltage range equal to the power supply voltage" and:

"Positive excursions of input voltage may exceed the power supply level. As long as the other voltage remains within the common-mode range, the comparator will provide a proper output state. The low input voltage state must not be less than −0.3 VDC (or 0.3 VDCbelow the magnitude of the negative power supply, if used)"

So it appears that using a differential input significantly greater than the power supply voltage would be outside its spec, and particularly cause problems if the input is negative polarity.

Obviously I had considered a simple voltage divider, but I reckoned three separate resistors, the two comparator inputs across the centre one. Then the precision of the values would be irrelevant, as only the polarity would count for the comparator inputs, but this might reduce sensitivity to low level input and affect input impedance.

My project is a quad multi-voltmeter to sit sit at the back of my test bench, to measure multiple circuit parameters without having several DMMs cluttering up the space. The (cheap & cheerful) 2V & 20V DC panel meters I have obtained work perfectly either way round, but don't show polarity !
So I want to add LEDs by the terminals to indicate this, but with a comparator to maintain high input impedance.

The meters run on 9 -12V DC or AC, and, as a precaution, are DC isolated from each other by capacitors in the supply lines.

(Image & circuit below)

Trying again to make the image load.

The image needs to be one of the three file types: GIF, JPEG (.jpg), or PNG.

otherwise it won't display.

How about something like this:

http://en.wikipedia.org/wiki/File:Stacked_Villard_cascade.svg

driven by a 555 timer. Use that to provide pos and neg 25 volt rails to the comparator. That way, there will be no concerns about the input voltage exceeding the supply rails.

OR

how about using a diode to block negative signals, then feed that into a comparator on a single positive supply. Because of the extremely high input impedance of the op-amp, the forward drop on the diode should be fairly low -- and perhaps a Schottky diode would be better (though, the high reverse leakage might be an issue). With the inverting input tied to ground, and the non-inverting input tied to the diode, it won't matter if the input voltage goes higher than the positive rail, your comparator output (according to the spec you supplied) will go high for anything that goes high enough to exceed the forward drop of the diode (which, again, should be rather low). What forward drop there is, should provide just enough gap to prevent false positives.

If you need this to indicate a polarity change at as close to zero as possible, then you will need to generate a negative voltage (using something like what I suggested above) and then, through a very large resistance, bias the input negative just enough to overcome the forward drop of the diode. That is, if that negative bias can be above -0.3V, and I think it can.

AND, for that matter, because of the high impedance involved, that negative voltage could be supplied by a small battery (perhaps a lithium coin cell with a voltage divider to get the sub -.3 voltage)

Trying again !

What are you measuring?
Is it just polarity and differens between 2 and 20 Volt?
Then an easy way is just to four led and four resistors.
If you are think its okey with both diods light at 20 Volts.
For example red and orange for -2 and -20
Green and Blue for + 2 and +20.
If you need help with the math just ask, but specify the leds Vf.

Hi Steve,
Thanks for your help. I already have the transformer to supply the four meters, and I am trying to keep things as simple as I can.
My thinking is, that surely my problem can't be all that unusual, and probably there already is a basic comparator that will do what I want, which is simply to accept both positive and negative input signals somewhat greater than its supply voltage without additional circuitry.
So I was hoping someone knew of such an animal.

Hi Per,
Thanks for your message. The meters are for general purpose electronic testing. I want to maintain a high impedance at the voltmeter terminals, therefore I don't want to run the LEDs directly from the measured voltage. Using a comparator with a high input impedance seems to be an obvious simple solution. But as I have indicated above, the basic spec of the LM139 and similar doesn't allow a significantly higher input signal, both positive and negative, above the supply voltage.

In thinking about this more, it occurred to me that what you want is a zero crossing detector, so I Googled that and came up with this:

http://allegromicro.com/en/Design-Center/Tech...Sensing-Techniques-for-Power-Electronics.aspx

Or this (but you'll still need at least -0.7V supply to the comparator [or perhaps only -0.4V if the comparator can take -0.3V below the negative supply rail on the input):

http://www.circuitstoday.com/zero-crossing-detector

But, I don't know of a comparator that can take voltages on the input that go largely below the negative rail (but I've never looked for such a thing).