LMV7239: Non inverting comparator with hysteresis oscillating

Hello Stef,

DO NOT place a cap between the input pins. This can cause several problems - particularly with hysteresis. Basically, you are slowing it down and defeating the hysteresis - which can actually cause chatter or "bursting"

I would increase the 10nF capacitor across R15 to up as high as 1uF (ceramic). There can be bias current transients on the inputs when the output transitions, so the reference needs to be be a low impedance point.

Please realize that the LMV7239 is a high speed (75ns) comparator and should be treated as a high speed gate. There are large current spikes on the supplies as the output transitions (5V in 2ns!), so supply bypassing and layout is critical. You have done pretty good, but I would put R17 at a right angle to R16 (below R18) to prevent output coupling back into the positive node. I would also try paralleling a 1uF (ceramic) cap across C26 to help with the supply transients.

I would also sprinkle a few more ground vias around the circuit to stiffen-up the grounds - like around R15 ground.

Hi Stef,

Paul gave great advice to help your issue. There was one typo and the 10nF capacitor across R15 should be replaced with a 1uF capacitor (not R1).

It is unfortunate that the simulation models don't capture real world performance. One additional option for your is to increase the hysteresis for your design.

For your 1kOhm resistor on the output, I would also consider increasing it to 10kOhm to slow down the output edge rate.

Best Regards,

Joe

Applications Engineer

Linear Amplifiers Business Unit | Comparators Product Line (CMPS)

All information in this correspondence and in any related correspondence is provided “as is” and “with all faults”, and is subject to TI’s Important Notice (http://www.ti.com/corp/docs/legal/important-notice.shtml)

Paul, Joe,

Thanks for the feedback.

I added 1uF 0603 X7R caps over R15, and R26. However the circuit still oscillates when the "temp sensor" voltage is between 1.040V and 1.065V
I noticed that the oscillation stops when I measure pin3 with a 12pF oscilloscope probe, however the hysteresis is significantly degraded as you mentioned.

It seems that I made a poor choice of comparator for this application? It will be used to buffer and read an analog temperature sensor with relatively high output impedance from an external module. So I have a slow signal and use a high speed comparator. The hysteresis must remain 100mV since this translates to a 10 degree centigrade temperature difference. The output should drop to 0V when input is 1.1V and go back to 5V when the input drops below 1.0V.
Perhaps I should just switch to a much slower comparator, do you have any suggestions which one might be most suitable for temperature measurement applications?

Thanks,
Stef

Hi Stef,

I think the problem is that the input of your comparator circuit isn't driven by a fast amplifier which can fully absorb the fast edge of comparator output signal when the output toggles from high to low. Via R18 a fast signal is fed back to the input and the input reacts with it by increasing the voltage at pin 3 instead of decreasing.

So there are some remedies:

1. You heavily speed up the driver of temperature sensor signal to be able to absorb the fast edge coming via R18.

2. Or you feed the temperature sensor signal to the -input of comparator instead to the +input with the fast hysteresis feedback.

3. Or you take a slower comparator.

Can you show a schematic which includes the temperatur sensor?

Kai

Hi Kai,

I do not have a schematic of the temperature sensor circuitry, since it's an output from a module that we bought from a 3th party supplier. 
From past experimentation I learned that it has a relatively high output impedance, and that buffering is necessary.

Your argument that the signal source might be the problem did set me on the right track of verifying that part.
Instead of using the temperature sensor, I am using a R&S HMP2030 lab power supply to test this circuit independently. Apparently the output ripple of this particular PSU is way out of spec, at around 200mV.

This means that the comparator is not oscillating anymore after adding the 1uF caps on R15 and C26.
What I perceived as oscillations is actually the comparator behaving normal but triggering on input noise that is larger then the hysteresis of the circuit.

I added a 1uF cap between at the R17 input to GND. This + increasing the power rail and reference voltage buffer capacitance seems to have fixed the issue. The circuit now behaves as I had intended.

Modified schematic and PCB: 

Thanks everyone for the advise on how to improve the stability of the circuit, I will use your suggestions to improve this and future designs!

Stef