Problems with LM319

[Baldin]

I'm having problems in getting one half of a LM319 to work correctly for a over curent detector.
The current sensing resistor consists of 3 x 0.022 Ohm resistors, which should trigger the circuit at a bit more than 16 A. The + input of the Comperator, is connected to a voltage divider 100k/1k to give a 120 mV voltage.
Vcc2 is 12V.

Problem is - input is 0V and + is 120 mV but the output is 10,4 V!!

Doesn't the LM319 work as a window detector this close to ground?
Is it because the differential voltage is two low??

Whe removing R56 the output goes low (makes the diferential voltage 12V).

The Class-d output has not been mounted yet, so this is not the cuase of the problems.

After some experiments I don't find the LM319 too easy to work with


Anyone familiar with this comparator, that could help??

[consort_ee_um]

I think the problem is that the common mode input voltage range of the LM319 does not include ground. You need to use the LM339 or one of its dual equivalents.

[Genomerics]

Yes, like what he said. The LM319 input common mode range does not include ground.

If you look at the data sheet, I haven't found a 'good' one yet, you need something like 1-2V. The internal schematic shows an NPN differential pair with an NPN current mirror driving it's bottom.

That adds up to at least 2 VBE drops or 1.2V if it is going to work.

You might try two resistive dividers from your 12V rail to the inputs to pre-bias them up to 2V. Tie one to ground and the other to your sense resistors.

You'll have to do some sums to set up the proper trip level.

Otherwise there are more modern comparators, as fast or faster, whose input common mode range includes ground.

DNA

[Pafi]

The simplest solution: increase Rshunt to twice, and connect to a base of BJT with a pre-bias of 300 mV!

[Baldin]

Thanks guys.

This is one more time where simulations does not match real life

Bad thing is I already made the PCB ... with SMD


Oh well as long as it's fun and as long as I learn something

..... think a rail to rail op-amp would have don better in this place.



Have also been wondering: I have made the Rsense of 3 x 0.022 Ohm 1W SMD resistors, mostly to avoid wire wound resistors, with their added inductance.

But maybe such a small coil in in the PSU line would actually help dampening ringing and EMI.
SMD in this place is not optimal as SMD components can't handle any real power. I had to pay almost 2$ per SMD resistor!!!!

On the other hand, maybe the small bars Eva is using would be even better .... wonder where these can be obtained!

[zilog]

Originally posted by Baldin
Thanks guys.

This is one more time where simulations does not match real life

Bad thing is I already made the PCB ... with SMD


Oh well as long as it's fun and as long as I learn something

..... think a rail to rail op-amp would have don better in this place.



Have also been wondering: I have made the Rsense of 3 x 0.022 Ohm 1W SMD resistors, mostly to avoid wire wound resistors, with their added inductance.

But maybe such a small coil in in the PSU line would actually help dampening ringing and EMI.
SMD in this place is not optimal as SMD components can't handle any real power. I had to pay almost 2$ per SMD resistor!!!!

On the other hand, maybe the small bars Eva is using would be even better .... wonder where these can be obtained!

They are called OAR-3 (for 3W versions) and are obtainable from Farnell, RS etc. The problem is the immense cost of obtaining them, but if you decide on getting some, I might be interested in 4 pcs of 40-50mOhm.

[Baldin]

Pafi:
Not sure I get your proposal right ..... by pre-bias, do you mean to tie emitter to -300mV?

If Rsense = 0,015 Ohm then a current of say 20 A would give a 293 mV and together with the 30mV would be just enough to trip the output (probably a little higher current would be needed)!

Se picture. That's what you meant?

But then I have to get the -300mV first, and as it's a full bridge I did not intend to use any negative PSU!


Zilog:
Thanks for the link to RS ...... well as I was using 3 times 1W SMD resistors .... the OAR thing will only cost a third!!


consort_ee_um and Genomerics:
Well now that I look closer, the data sheet actually states that the input range using 0-5V should be between 1 and 3V .... so for 0-12V I probably ned to go up to the 2V you are suggesting.
Genomerics, I'm not totally sure what you mean .... could you draw a small picture?

[BWRX]

Baldin, aren't there any other comparators that have an input range down to ground that you could just drop in? That'd be the easiest since you already have the boards made.

I looked up the package for the LM319 but it's quite strange that it's not an SOIC8, or is it?

I thought it would be and you could use something like the LM2903, but it looks like you would need to do some extra wiring to get it to work.

[Pafi]

Baldin!

At low temperature there will be a little higher current limit, but as soon as temperature rises, VBE will be lower, so current limit too.

[powerbecker]

Originally posted by Baldin
I'm having problems in getting one half of a LM319 to work correctly for a over curent detector.
The current sensing resistor consists of 3 x 0.022 Ohm resistors, which should trigger the circuit at a bit more than 16 A. The + input of the Comperator, is connected to a voltage divider 100k/1k to give a 120 mV voltage.
Vcc2 is 12V.

Problem is - input is 0V and + is 120 mV but the output is 10,4 V!!

Doesn't the LM319 work as a window detector this close to ground?
Is it because the differential voltage is two low??

Whe removing R56 the output goes low (makes the diferential voltage 12V).

The Class-d output has not been mounted yet, so this is not the cuase of the problems.

After some experiments I don't find the LM319 too easy to work with


Anyone familiar with this comparator, that could help??

Baldin,
try to do a level shift.
Heinz!