Webench LM3671MF-ADJ/NOPB C1 Incorrect?

George,

It looks like the 10p cap was picked in favor of the 6.8p capacitor because of the target voltage rating and the higher cross over frequency. I hope this information helps.

Britt,

Thanks for following up on this. I'm ordering parts for production. Would you say that the 10pF is "better" I'm purchasing 50V, NPO, +/-0.25pF parts.

Thanks,
George

George,

Not really. All I was really saying was that the 10p was chosen by the CC (Circuit Calculator) in WEBENCH® over the 6.8p. There is an algorithm that is used to pick the parts along with weighting based on the user inputs and the function of the design. This algorithm scored the 10p a little higher. Basically, if you look at the Bode Plot, you will have to make the decision for what is best for your particular application. Is the higher phase margin (with a lower Bandwidth) better for your needs? Is there a cost difference? Is there a size limitation? Is there a reliability concern?

All of these are questions that you will need to answer, based on your specific project needs and engineering judgment. I believe that either choice would do well for a typical application.

Please be sure to prototype the power supply BEFORE going to production with it. This is simply a good engineering practice. I would hate to see your design go to production and find out that there was some concern that was not originally captured and the design did not perform to your expectations.

Britt,

Thanks for your feedback. Yes, absolutely everything will be tested but some of the key parts in the design have lead times so we're preparing orders now. I should be able to test next week.

To be honest I haven't messed with the filters too much over the years. Can you point me to good references on what to look for during my testing that these filters will impact? The other major components on the small board (~1"x1") are Bluetooth module, accelerometer/gyro, and a Li-Ion battery charger.

Thanks,
George

George,

Good to hear that testing is planned.

I do not have a good reference for this kind of part where the additional compensation is required based on the output voltage of the device. The datasheet seems to have a pretty good description of what is needed. There is some internal compensation for the device (it appears from the internal schematic) which is in addition to the extra pole/zero that C1 and C2 provide.

Impact of the capacitor choices should be seen in the transient response or in the measured Bode plot. Most of the time, any stability issues can be seen here during testing.

As usual, watch for injected noise from any of the other circuits into the power supply and keep a good board layout to avoid any loops.

Hi Britt,

I've run into an issue that might be related to this issue we discussed back in July. I opened another thread yesterday () but remembered we had discussed this so I thought you might have some good input.

When the board goes to sleep the current consumption is well under 100uA and everything seemed fine on our initial testing but now I've run across some boards that exhibit this behaviour that I described in my other post (current consumption goes from 5mA to <100uA when the LM3671 filter is touched).

Perhaps this is an instability issue with the LM3671 not going from PWM to PFM mode and that can be improved by adjusting this filter??

Thanks,

George