LM393: how much negative input voltage at non-inverting input (and how much current) can cause damage?

Hello Nitish,

The Abs Max is 50mA, but that is a lot...We recommend 10mA or less.  It takes about -1V to get 50mA. By then, the output will probably be incorrect.

Please see the LM339 family application note, it covers reverse voltages in section 2.6.

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

There is the possibility of an inversion, or incorrect output, when the output is pulled below the ground pin. This can occur as low as -400mV depending on temperature. I would not use the circuit as shown.

Instead of dividing down the negative supply, you could use an "upside-down" voltage divider to a positive potential (instead of ground) that level-shifts the negative voltage to a potential above 0V.

Hello Paul,

Thank you for your kind response and the link to the app note, as well as your suggestion on how to modify the circuit.

Section 2.6 appears to indicate that the damage occurs when current increases to several mA:

"Noticeable nanoamp currents will start to flow when the input is at -0.3 V, and will increase to several tens of milliamps as the diodes start to conduct".

In our circuit, however, the max current out of the non-inverting input would be 19.3uA due to the high valued resistor, 130kohm.

Also, we are unable to understand what problem an output inversion would cause.  The pull up resistor value to 3.3V is 510k and the load resistor (not shown) is 1Mohm.  All these high valued resistors shouldn't allow any damaging currents.

Unfortunately, we don't have the liberty to change the circuit at this point.

We would like to make sure that we are identifying the problem correctly.  We would appreciate your further response.

Kind regards,

Nitish Agrawal

Hello Nitish,

Even though large currents are not flowing at -0.5V, the reverse voltage can cause the input stage to malfunction. At -0.5V, the current will be about 2uA

At this votlage there is a risk of the output can randomly flip states (no longer stays "low"). This behavior will vary device to device - so just because one device "behaves", there may be some that do not.

The specified input range is down to 0V, with an abs-max of -0.2V. TI recommends staying within the specified voltage range.

You will not be "damaging" the device, but you are at the risk of improper operation. The output can randomly flip, or even not respond.

Is this normal during operation? Or during a fault condition?

Hi Paul,

Thank you for your support yesterday.

Can you recommend a pin to pin as well as SO-8 package compatible alternate for LM393 that won't latch up in an undesired state (or get damaged) when experiencing negative input voltage?

We don't necessarily need its output to make sense when it experiences the negative input voltage.  We just want it to resume normal operation after the negative input condition disappears.

Kind regards,

Nitish Agrawal

Hi Paul,

We have another question regarding your comment yesterday:

You mentioned that the present circuit won't "damage" the LM393 device but the output may randomly flip or not even respond.

Would you mind clarifying whether the LM393 comparator would resume normal operation when the negative input voltage was removed?  In our product, the LM393 is not resuming normal operation once the negative input is removed.

We ask this because we want to ascertain the correct reason for LM393 rendered unusable in our product, so that we can address that issue.

Kind regards,

Nitish Agrawal

Hello Nitish,

It should recover to normal operation, but it may take a few more microseconds - longer than the normal propagation delay.

Is this a DC voltage? Or an AC signal or pulse? Are there any transients on the supply or output?

There are several devices you can try, but no device will comfortably handle a -500mV input.

You can try the new LM393B, or the original LM393-N (National Semi version). Or for a different architecture, try the TLC393, TLC372 or TLC352.

Unfortunately, you are operating in an characterized area, so you will literally have to figure out which device performs best in your circuit.

I would recommend that you see if you can adapt the "split divider" that is shown in section 2.9.2.1 of the appnoote. This allows the negative swing to be clipped and divided...though your division ratio is pretty low...but it may help.

Hi Paul,

Thank you for your response and your recommendations....we appreciate it.

The negative voltage is a DC voltage.  There are no transients...however, the negative DC voltage does appear abruptly.

All parts you have suggested state an abs. max rating of -0.3V on the input pin per their datasheets.

We found OPA2237....it can handle -0.7V.  Can you recommend something that will handle up to -0.9V?

We would prefer not to re-do our our layout as we are already in production.  A simple part swap would be the best.

Kind regards,

Nitish Agrawal

Nitish

As Paul explained, we do not expect any damage to occur when the input abruptly changes to -4V (-0.5V at the input pin after the resistor divider).  The output state could change state as a result of this negative input voltage but the device will recover from this condition.  Unfortunately, the largest negative abs max input voltage we have to offer is in the -0.5V range.  This is on the TLV1702 but this dual comparator is only available in the VSSOP package and you have a SO8 footprint.  Based on all the information provided, it is best that you remain with the LM393. Changing to an amplifier that is not intended to be used as a comparator will have other deficiencies when under normal operation.  So we would not recommend that as an option.

Chuck

Hi Chuck,

Thank you for your response.

In our case, it looks like LM393 is not recovering, at least in one of the units we have.

What do you think of OPA2237?  Its datasheet says it can handle Vss- - 0.7V on its inputs....and is available in SO8 package.

Kind regards,

Nitish Agrawal

Hi Chuck,

I just realized that OPA2237 is an opamp, not a comparator.

Kind regards,

Nitish

No worries.  Your design challenge is a difficult one.  Unfortunately, I am not sure if you will find a solution that does not require a change to the hardware.

Chuck

Hello Nitish,

One thing you could try is to increase the values of the divider resistors to lower the available current, which will then clamp earlier.

Try doubling the R63 and R64 values (keeping the ratio) and see if it still latches. Or re-calculating the resistor values to increase R63 (~250k) to lower the available current.

Hi Paul,

Thank you for your response.

We can definitely increase the values of the divider resistors.  Would you mind clarifying what you mean by "...which will then clamp earlier."?

FYI, this particular LM393 comparator circuit doesn't encounter slow moving input voltages.  It is mean to detect a sudden change in the input.

That is why we were considering using an opamp like OPA2237, which can handle at its input (-)0.7V below the negative rail.

Also, the LM393 comparator output presently drives an NPN transistor's base in our scheme...so, if the LM393 is replaced using OPA2237, then the non-logic type voltage output of OPA2237 is not a concern.

Given this background, do you still think that OPA2237 wouldn't be a good choice?

We would appreciate your response.

Kind regards,

Nitish Agrawal

you will need to discuss the amplifier on the amplifier forum, this forum is for the comparators and we are not as experienced as our amplifier counterparts on the OPA2237.

in regards to changing the resistor divider, I believe Paul's point is that this will change how much negative voltage the comparator sees at the input.

Chuck