LM5050-1 noisy output issue

Some additional thoughts on this issue:

With an input of 27.6V (to match up with 2 SLA batteries), the gate waveform is a sawtooth, with the rising edge being slow and the falling edge being almost instantaneous. (I could send a screenshot of my scope if that would help)  This would lead me to believe that the problem is that the chip is momentarily detecting a reverse voltage condition and pulls the gate drive hard low, then the 'condition' goes away, so the gate voltage once again rises to try and turn the FET on, ...  That having been said, the extra filtering I put on (the 100 uF cap right at the output of the FET as well as additional filtering on Vs, should have resolved the problem.  I can't identify a reason for the issue. 

Could this be a defective lot of parts that are hyper-sensitive to noise?  The circuit has this problem with a load of about 200 mA as well as a 1-2 A load.  If the problem were noise, I would think either light loading or heaver loading would affect the way it works.  I have also tested it at roughly 15A (the FETs can handle in excess of 60A, and have <1mohm RdsON), but the FETS get very warm because the only thing conducting is the body diode.

Hi Duane,

Welcome to the E2E forums and thank you for considering our product for your design.

If you can share any schematic or waveforms, it would help us troubleshoot the issue much faster.

You can click the "Use Rich Formatting" text on the bottom right-hand corner of the reply box. Then there will be options such as "Insert Picture" or "Insert File".

However, it sounds like the circuit is being very sensitive (as you have indicated.

We have one app-note on this subject which may be of interest:

However, the main point of the app-note is that adding capacitance to Vout should help (as you have tried).

If this does not resolve the issue, there is another option.

Overall the LM5050 features an extremely fast reverse current response (the spec is a 50ns response time). However, this can lead to nuisance shutdowns in systems with some level of noise.

As a result, we would recommend adding some filtering to the sense lines themselves. Please refer to this image below:

So this would create an RC filter which would help prevent nuisance trips due to noise.

For example, you can try an R of 10ohm and C of 0.1uF to create 1us of filtering which may resolve all the nuisance trips.

If 1us of filtering is too much for your system (you want a faster response time for reverse current shutdown), then you work down and decrease R and C.

Thanks!

Alex

Alex,

This looks like a great solution.  I'll implement it, then post whether or not it fixed the problem.

The FETs were chosen based on the circuit design constraints.  I am driving a motor to control a fire door, so it has to be able to supply >30A without excessive heating.  I have it set to trip at roughly 45A, although typical currents are much lower at around 17A or below.

When the LM5050 is working, it controls the Vds very nicely with around 2.6V on the gates at low currents, going higher as the current increases.  The two FETs are used to achieve the higher reverse-breakdown voltage required, since the FETs breakdown is only 40V, and R117, 118 are used to force any reverse voltage to divide evenly.  Note - the schematic shows a LM5050-2 part, but the circuit is, in reality, designed for an LM5050-1, and both resistors should be 10K.  The circuit, as built, is correct.

Duane

Norbert,

It worked great for me.  I experimented with a smaller cap, but a .01 caused it to oscillate again.  The challenge is that the more filtering you put in, the slower the reverse shutoff becomes, so I had to find the 'sweet spot' for my circuit.  Given the type of circuit you are driving, I would suspect that even more filtering will be required.  You could try something ridiculous (4.7 uF, which is readily available in MLC, for example), and if that stops the oscillation, then it becomes a matter of finding the smallest cap that will do the job.

I also put in the 100 ohm and 0.1 cap on the power (Vs) pin.  If you are using a -2, then that option isn't available to you, but otherwise, it seems like cheap insurance.

I hope this helps.

Duane

Hi Norbert,

We greatly appreciate your feedback and thank you for taking the time to share your experience.

The LM5050 "features" a 50ns response to reverse current, but as you have experienced it can be too fast/sensitive for some applications and lead to nuisance trips. That is where the RC may help.

We added the leakage current note in several places within the datasheet to ensure customers are aware of it, in case they have a battery application where 3mA may be a concern.

That being said, if we end up releasing a new high voltage ORing controller, it will not have these issues.

For low voltage applications, the TPS2410 family of devices do not have leakage issues and the TPS2410/11 havea dedicated filtering pin.
 
Thanks!
Alex

Hi, Norbert,

Interestingly enough, my 5050 circuit is to replace the LTC part.  I have had MAJOR issues with it not starting up every time (3-5% typically), and LTC's tech support was less than helpful.  Hopefully you have better luck.  Perhaps LTC has re-spun their part and it now works reliably, but that certainly wasn't my experience.

FYI, the symptom of the problems I had was that I would turn the circuit on, and every once in a while it just wouldn't turn on, so all the current went through the MOSFET body diode.  My application is in life safety, so unreliable operation simply wasn't an options I could live with.

I wonder if an inductor at the output of the LM5050 ideal diode circuit would help to prevent the issue....

Duane