Thanks for the assistance John. I think I've fallen into the trap of using a
crappy breadboard thus resulting in sloppy connections ending up with a
sagging supply voltage. The bench supply is more than adequate and well
regulated.

I have been playing with a protection circuit that I need to develop to
protect FET's used as output devices.
I want to have their drive circuit turn them off when a certain current is
reached to protect them etc.

The source to the comparator is derived as below:


(+)--------------------------------------------------------|

|

[LOAD]
Source ---------------------- |

__|
---||
__ FET

|

|
(-)---------------------------------------------------------|

NOTE: There is no limiting resistor in the FET circuit, so it is vunerable
to short circuiting.

There are 8 FET's driven from PORTB of a micro. I want to use the signal
from the comparator to feed back to the micro to turn the offending port bit
off, and use another output port to indicate the fault.

Hopefully once I get rid of the crap breadboard the problem will go with it.

"John Fields" wrote in message
...
On Sun, 10 Sep 2006 00:59:17 GMT, "Farticus"
wrote:

Using a comparator (one section of a LM3900) I connect the (+) input to
a
source voltage via a 1K resitor.
The source is obtained from a series load resistor, so that as the load
varies so does the source voltage to the (+) input.

The (-) input is connected to the wiper of a 5K pot. The comarators open
collector output via 2K2 to pos supply.

---
It's not clear how you get your source voltage, but your comparator
should be wired like this:

V+--------+---------+--------+
| | |
| | [2k2]
| | |
| +---|--[1M]--+
| | | |
VIN--[1K]-------+--|+\ |
| | ------+--VOUT
[5k]--+--|-/ LM3900
| | |
| | |
GND-------+-----+---+-----------GND

Note the 1 megohm resistor used for hysteresis.
---

The idea is to compare the two inputs so that a signal can be detected at
the output of the comparator depending on the level applied to the (+)
input comared to the setting on the (-) input. Normal comaparator
operation.

All unused terminals grounded, the chip supply is decoupled.

The problem I am experiencing is that the voltage on the (-) input is
also
varying slightly in sympathy to the varying voltage on the (+) source
input,
enough so as to prevent the comparator from operating when set to small
values, ie: 0.8 volts.

Example:
The (-) input is set to 0.8 volts.
With the equipment output circuit running at normal load the applied
source
voltage on the comarators (+) input is 0.5 volts.
When the output load is increased the source voltage at (+) input
increases
accordingly, but so to does the voltage at the (-) input. The result is
that
the comparator does not "compare" at the expected point( 0.8volts) - if
at
all. Sometimes the preset voltage on (-) keeps lifting to a level above
the
source with the result that the comarator does not trigger.
Does anyone know how to overcome this "voltage creep"? What should I do
to
prevent it?

---
The only thing short of a wiring error that comes to mind is that if
your supply voltage is varying with the load, then your reference
will be varying also. The way around that is to use a real
reference which will stay stable as the supply varies, or to stiffen
up the supply, or to regulate the comparator's supply.


--
John Fields
Professional Circuit Designer