TLV1702: Problem with the comparator TLV1702

Hello Pekka,

From what you describe, it should work. The TLV1702 should work fine on 7.6V. It is a R-R input, so as long as the input voltages are between -4.8 and 2.8 it should function.

Shorting the inputs does cause the output to toggle, so the pull-up is working and the output is not being "held-up" externally. 100K is on the high-side for a pull-up, but not unreasonable if you wish to save power and speed is not an issue. What is on the other side of R1? What is it driving? What are the actual output "low" and "high" voltages? Can the devices down-stream of the output handle -4.2V?

How is the 0.142V generated? Have you measured it and confirmed it is correct?

Do note that the typical 10M input impedance of most DMM's can load a resistor divider causing the voltage to read lower than calculated - so be aware of that.

Same for the positive input. Is it really zero?

The 1702 and 339 are bipolar input devices, so they have bias current flowing OUT of the input (flowing out from V+ towards ground through the inputs). So on the positive input, you could see 500uV (TLV1702) or 2.5mV (LM393) generated across the 100k (R12). But this is not enough to cause the failure, unless the inputs were somehow damaged.

I would recommend a LMC6772, which has a CMOS input with femtoamps of bias current - which would eliminate any issues with bias current (and truly a "high impedance" input). It also eliminates the bias current injection that may also interfere wit the reading.

Can you provide the entire schematic - at least the parts containing the input signals and where the comparator output goes? There is something being overlooked.

Hi Pekka,

the full schematic didn't come through. Please post again by using the "Insert File" button (paperclip) in the header of input box.

Kai

Kai

Kai the symbol was quite different and  I think it was a microphone, that is why I paste the file, but it doesn't succeed.

Pekka Ritamaki

.

Now we can see it, Pekka :-)

Kai

Hello Pekka,

I see a few issue.

The low batt detector, IC5B has the input range violated on both inputs. The specified input range is Vcc-1.5V (Vcc-2V at cold). Your V+ supply is 2.8V, which would make the valid input range up to +1.3V at room temp, and +0.8V at cold. You are applying 2.4 and 2.75V input signals. You are squareley in the dead zone and I would expect failures. You are probably not noticing because the output generally goes low when both of the upper input ranges are violated - and low means good batt. See section 8.2.2.1 in the LM393 datasheet. This should not affect the other channel, tho.

Again, I would recommend a rail-to-rail input device, such as the LMC6772 or TLV1702 to eliminate the voltage range issue.

You are relying on the clamps on the inputs of the logic gates to handle the -4.8V "low" voltage. That is ~450uA on the negative supply per input when low.

Have you monitored all the supply voltages? Can the negative supply support that extra current?

You may want to review the LM339 family application note, particularly section 2.3:

 https://www.ti.com/lit/an/snoaa35a/snoaa35a.pdf

Paul Grohe

Thank you again your good review.

I already mention the power operation was not right at IC5:B.

I modify this according the drawing kuntg.jpg. 

I doesn't  change the resistance change operation .

===

You  mentioned the power supply filtering.

This was quite good operation.

I added two 4.7uF capacitors to the power supply.

This was a great movement to better working.

Zero ohms worked now well! 

I added two 4.7uF capacitor to +/- power supply.

It has a great effect,

I changed R12  100k to 10 k. This has no effect.

I changed the output of the comparator  (R2) from 100k to 2k2 resistor  

It had now effect.

===

Then I checked 100 ohms measuring device.

I doesn't worked well. 

It showed 140 ohms at high  ranges ( decimal point was removed)

Then I checked a little lower value 71.8 ohms. 

It showed 319 ohms at higher range, which was not right.

The output of LM393 was not high or low, but about 0V.

==

Then I checked 1k resistor.

It worked well, the display was 1002.

The LM393 doesn't worked well,

I should give +Vcc or -VCC, not between.

You  intestine LMC6772.

This device is in my own supply (TME), but they have not made an order, so they have 0 pcs. 

==

Doesn't LM393 work as a comapator? 

I have tied feedback resistor from the comparator output to -input. 

It was 6M8, a 1M and 220k.

I didn't helped. but it was before I added the power supply capacitors.

Maybe I have to try once again.

I read the https://www.ti.com/lit/an/snoaa35a/snoaa35a.pdf document.

Regards

Pekka Ritamaki

Pekka

it may be best to simplify what your circuit is and exactly what result you are seeing that is not correct.  It is ok if you draw it freehand and upload.  I also do not see any screen shots from an oscilloscope showing the problem.  would it be possible to show us the improper result that you are observing in that fashion.

Lastly, you mentioned applying negative feedback to the comparator.  what is your intent with that?  

Sorry I cannot provide more insight into the problem but hopefully we can help resolve once it is more clear.

Thanks for your patience as we try to resolve this issue.

Chuck

Chuck

The comparator feedback was a solution!

I put a 1M resistor from comparator output to positively input.

I increase the capacitor value to 4.7uF, but I am not sure if it need.

The range works now at 80 ohms to higher range.

It works also from higher range to lower range a little more resistor, about 90 ohms.

The higher range is little bit slow. but I can fix it.

==

Why I do not use oscilloscope?

I have TEK TDS420 scope, but it is broken.

I  can't use a clear  switch to open the scope.

It is used to broke ( 30 years!), 

I succeed to get the other switch set , which was right clear button, it doesn't work either.

I added a separate . clear switch, but this doesn't help.

==

I put my last schematic on. 

Well, now it woks.

I do some testing,

Thank you for your suggestions.

I can't see oscillograms for the power supply.

Regards

Pekka Ritamaki

Chuck

I need still to make circuit board, manufacture it, purchase the components, place the components, test it, apply to the customer.

Let see if everything works.

==

I changed  the feedback resistor from 1M to 470k .

==

Many thanks

I think I can't live without s scope.

Regards

Pekka Ritamöki