Hello,
I'm using the LM3478 to convert from 5V to 28.4V. The used design equals the following (we use some caps from other vendors, but the values are right):
https://webench.ti.com/appinfo/webench/scripts/SDP.cgi?ID=1E347F9F43037E96
The system works perfectly fine when working at normal ambient temperatures. However, when I heat up to an ambient temperature of about 65°C, the regulation starts "hiccuping" and thus generating high current peaks on the input voltage.
We use another LM3478 to generate a 40V voltage out of the same 5V, this also starts hiccuping with high temperatures, but with less high current peaks.
The blue line here is an unconnected probe, so you can ignore that one. The yellow is the DR pin of the 28.4V converter, the green one is the DR of the 40V converter.
When disconnecting all output load except the Cout capacitor, the behaviour is the following:
From now on, the blue line is the current on the 5V input voltage. The yellow line is DR pin of the 28.4V converter.
As there is basically no load, the converter only has to work for short amount of times to account for the losses on the capacitors.
At room temperature, this regulations starts normally with the expected frequency and a normal duty cycle
When warming up, however, this starting does not happen in a decent frequency, but the DR goes permanently to HIGH. Thus, it creates a current overshoot.
The warmer it is, the longer this "stuck-at-high" phase takes, and the current overshoot also gets higher
Zoomed out in time, the regulation leads to a current overshoot of more than one ampere at the input voltage.
Do you have any previous measurements, or ideas where this might come from?
