Hi,
We use your LM51551DSST in an isolated DC-DC converter with optocoupled feedback into the COMP pin. With no other load than the feedback system, the regulator is able to provide the correct DC output. However, there are stability issues as the power load increased. At present I can pull some 10W out of it. Go above that and it shuts down. The designed power is in the 13-15W range.
Q: What are the known causes for the LM51551DSST to produce very short pulses on the GATE pin? This question baffles us and we can't debug the rest of the design while it remains.
In our design the clock frequency is set close to 250kHz. At that rate there is a rising edge on the GATE pin. A great number of the pulses are very short (200ns). We have two previous revisions, one that displays short pulses, another one that never does. We have yet to tell the engineering difference that cause stability in one, not the two other ones.
As load current increases (using RND 320-KEL 103 300W programmable DC load), the -broad- pulses in my plot widen. The narrow ones have the same width, although they are somewhat rarer. Eventually, with increasing load, broad pulses grow so wide that the become united with the next narrow pulse. At this stage the regulator goes out of regulation, the output voltage drops and occasionally jumps up a little. By reducing the load close to 0W the circuit will start up again. The input voltage is not greatly affected by these mode changes.
With increased load I would instead expect the narrow pulses to widen, not the wide ones.
We have considered the following as causes of narrow pulses, and would greatly welcome your input since addressing the items on our list has not solved the issue:
- GATE series resistance is too low (going from 10Ohm to 68Ohm brings improvement, not elimination of short pulses)
- Improved decoupling of input and output (different decoupling show different behavior, not elimination)
- Power-on with or without load (there have been brief runs of even-width pulses, they are hard to replicate)
- LC filtering and series L (again, improvement not elimination)
- Instability on CS pin (hard to distinguish due to ground lead pick-up, looks similar to GND measurement)
- Feedback issues (COMP is DC with overlaid pulse noise pickup. It rises with increasing load power, from 1.5V@0W to 1.9V@10W)
Best regards,
Børge
This is what GATE looks like on a scope plot.
