# Having problems understanding LM339

Discussion in 'Electronic Basics' started by happyhobit, Jul 13, 2003.

1. ### happyhobitGuest

Here is my problem.

I'm using an LM339, the Vcc = 15.25 VDC and I have an LED connected between
Pin 1 and a 2K resistor going to Vcc

I have 13.870 VDC going to Pin 7, the (+) input, and I have 13.860 VDC going
to Pin 6, the (-) input. The LED is off.

I increase the voltage in Pin 6 to 13.862 and the LED goes on.

Pin 6 is still 8 mvolt below Pin 7. Why does it turn on? What am I missing?

Jay

2. ### John PopelishGuest

You are missing the fact that real comparators are not perfect. They
have a spec on input offset voltage that tells you how far from a
perfect the inputs can be and still have the comparator indicate a
match.
Here is the data sheet:
http://cache.national.com/ds/LM/LM139.pdf
The typical input offset is 2 millivolts, but the worst case is listed
as 5.

There is also an undocumented effect caused by having unused inputs
left floating that can degrade the offset. Have you tied the unused
inputs to some voltage lower than the positive supply minus 1.5 volts?

3. ### happyhobitGuest

Thanks John,

I tied all the other inputs to ground and now the triggering occurs when Pin
6 is 5 mvolt below Pin 7. I guess it's not bad for a Radio Shack part.

You noted 'positive supply minus 1.5 volts'. This gives me a clue into
another problem I've observed.

A friend using a LM339 with a supply of 15.5 VDC. He connected Pin 7 to the
supply voltage(15.5VDC). When he brings the voltage to Pin 6 above 14.5
volts the output goes to ground.

Is this because the inputs are greater than the 'Input Common Mode Voltage
Range? Is there any problem using it this way?

Jay

4. ### John PopelishGuest

It is probably a reject, but may serve your purpose. 1 millivolt
offset is quite common for this part, at room temperature. You might
try connecting a small capacitor (say, .1 microfarad) across the
inputs and see if the offset changes. This would indicate that your
two voltages are not pure DC, but have some spiky noise that is
shifting the effective input voltage a bit. The output stage switches
on faster than it switches off, so a brief turn on pulse that repeats
before the output can turn off, can keep the output on, even if the
average input voltage is not at the switching threshold.
Yes, exactly. The input circuit only functions between (+Supply -
1.5) and (-Supply - 0.3). The input voltages should not exceed this
range in a well designed circuit.
The offset degradation is one possibility. The propagation delay
through the comparators may be degraded, also.

5. ### happyhobitGuest

Hi John,

Well I tied a .1mfd cap between each input to ground. Now it changes state
within 2 mvolts. 6 mvolts noise. Wow.

Thanks again,

Jay