LMV331: Osculation noise on output when in low state

Gary,

Have you tried moving units from board to board to see if the issue follows your board vs the unit? The soldering in the pdf you sent me doesn't look the cleanest and could potentially be a reason for the oscillation.

Regards,

Jonny

In addition to Jonny's comments, I am a little concerned about the value of R125.  Using a 1k ohm pullup resistor is a little on the low side.  Would it be possible for you to change the value to a 10k?  Also, it would be recommended to add some external hysteresis to your circuit since the input appears to be slightly noisy at least in comparison to the voltage applied to the inverting input.  I am not sure that will totally explain what is happening but these are good recommendations regardless.

Chuck

Hi Gary,

I agree with Jonathan. The soldering looks absolutely terrible! And that the oscillation "comes and goes" also indicates that your board suffers from a soldering damage. But the soldering damage need not only to have occured with the LMV331 but also with the passive components. I remember an overheated ceramic cap wihich showed a very similar "oscillation".

Well, I don't aggree with you that all the other signals are "clean". If you carefully look at the yellow scope trace you will notice that the positive input signal is superimposed with noise spikes which exactly coincide with the noise spikes seen in the output signal:

This, in turn, is an argument that we have no oscillation here, but rather ground noise or possibly ground bounce. But it can also mean that the output transistor of LMV331 is overstressed by a too high output low current. This would explain why the noisy spikes are not seen on every board, because the sinking capabilities of LMV331 show some manufacturing tolerances. So, I agree with Chuck who recommends a higher pull-up resistance.

You can easily see whether it comes from a too low pull-up resistor. Solder a 10k pull-up resistor on the board and bridge it for a brief moment with a 2k2 pull-up resistor. Take a leaded resistor for the test. If the noise spikes at the output increase whenever you bridge the resistor, then a too low pull-up resistor is the cause of your problem.

Finally, I would also recommend to improve the ground routing. You should not have this much noise in your circuit.

Kai

Thanks Kai and Jonny

Gary

At this point you have several suggestions and we will await your feedback to see if you need any further assistance.

Chuck

These are good suggestions. I will give them a try and get back to you. I have checked the solder joints and they do seem good. The board is conformal coated which makes the solder joints look worse than they are, but I agree not the best I have seen. On other boards I see the same system noise, but the osculation noise never appears. I have found that when the scope probe is on the output of the LMV331, I can start and stop the output oscillation just by turning my soldering gun on and off. I suspect a noise spike coupling into the scope probe from the soldering iron enters into the output pin and provides the stimulus needed to start or stop the oscillation. Sometime the oscillation will remain for days while at other times it may disappear in seconds. Its possible this chip was damaged slightly by ESD and is now more sensitive to this issue. 

I will try the resistor change first and get back to you with my findings. 

Hi Gary,

hhm, can it be that your scope probe is damaged? Looks like a bad ground connection or a damaged cable shield.

Have you tried to measure with a spring probe?

Kai

Another good suggestion from Kai.

Gary

In regards to your ESD question, I do not think so but damage to a device is always possible.  However, in this case it could also be ground noise.  There is not really anything in the comparator that would cause the output to oscillate like that when the comparator output is low unless the input was very close to the switching threshold.  In your case, you have an overdrive of at least 1V, so that is not the case.  Therefore, it is likely that you have some sort of ground noise on your board or are picking up noise from some other source.  When the comparator output is in a logic low state, we are turning on the output transistor which holds the output to ground.  So if the ground plane has some oscillation on it, the transistor will just pass that to your probe.  If you use a probe configuration as Kai suggested and measure from the comparator ground to the comparator output, it is possible that the oscillation goes down significantly or completely.  Please try this experiment and report back.

Thanks

Chuck

Hi Gary,

Can you supply us with a photo of the top of the TL331? The datecode is composed of dots above the pins on the top side - so it is sometimes easier to take a photo. Or, if you have the original reel packaging, please take a photo of the reel label.

Are you seeing this on one or more devices?

Sorry I have been slow to respond, but I finally found some time to revisit this problem today. I have actually been working on getting more of these back form the field. 

I replaced the output 1K pull up resistor with a 10K pull up and was still able to get the oscillation, but at much lower amplitude. The first scope shot above is with the oscillation present and the second is without the oscillation. When I first power up the board there is no oscillation. I can get it started by quickly powering on and off my bench-top heat-gun which I believe creates a noise spike via the scope probe and causes the op-amp to oscillate. 

With the 1k resistor the oscillation is large enough in amplitude to cause communication errors in our system. This unit was in a customers home and would periodically cause the communication errors and annoy the customer. When the service guy would show up the unit would work fine. The unit would work for several months with no issues and then start generating errors for several days. I was able to visit the customer myself at just eh right time to witness this oscillation. I replaced his defective board and brought it back to our lab. I was able to consistently get the op-am to oscillate by using the heat-gun method mentioned above.

I do not see this noise on GND or anywhere else on the board. I have the scope in Peak detect mode. I know the oscillation is real because it was causing comm errors when present without the scope connected. I also cut the trace going to the microprocessor just in case it could be coming from it's port pin. I have looked at several other known good boards and have not been able to duplicate the oscillation issue. I even took a good board and conducted 15K ESD testing on the inputs in hopes of making a good part bad, but was not successful. 

I am waiting to get more units from the field. Several customers have their units connected to the cloud and I can detect this comm error remotely. I am trying to determine just how common this specific problem can be. If I get more samples of this issue, I will then try doing part swaps to see if the problem follows the op-amp. However, I am a little reluctant to try this until I have more samples. 

. 

Hi Gary,

Do you mind if we contact you at your registered email address?

We would like to start a return process, and those details so not need to be public.

In direct contact - closing thread.