we’re using the MPQ5069 to control a few 24V devices in our system. The switch behaves properly for most of the devices except for the 24V / 8A loads that we have.
Initially, the switch would fail at all times when the slow start was set to >20ms. Enabling auto-restart caused the chip to become really hot thus leading us to think that it was due to too much power neededin to be dissipated during slow-start. We then disabled auto restart and soft start set to its minimum (1ms) after which the switch would work better.
We are still however experiencing extensive periods of time where the switch simply faults. The switch will sometimes recover by itself. Our duty cycle is really low (we turn on/off every few minutes).
While the switch is rated for 36V, most of the parameters in the datasheets are at 12V. Should I be expecting significant changes of its caracteristics / behavior when used at 24V / 8A ?
At 8A the chip is dissipating about 1.1W (Power Loss vs. Load Current Graph)… Do you have heat sinking to cover this? You may want a temperature probe in the center of the chip for testing.
Also consider watching the timer pin, if it is ramping up after powerup, then you are already in fault condition.
Yes, it’s heatsunk. Based on the power loss vs load current graph, and the efficiency one, does that mean that Rds is irrelevant? Its spec is 7-12mOhm however if we’re at 8A and worst case Rds, that would give 768mW, not the 1.1W or more shown in the graph, and better than 99% efficiency. Where is the disrepancy coming from?
You are neglecting switching losses, which are a significant portion of the total losses in the switch at that frequencies. Very low Rdson doesnt guarantee unusually high efficiency (as you mentioned, in the realm of 99%). Switches are not driven by magic, and the sole driving of the switch generates losses. In addition, you have inherently “lossy” Cgd and Cgs (because of hard switching) to [dis]charge hundreds of thousands times per second.
In out case, we aren’t switching much. We only switch every few minutes, at most so in this case shouldn’t the losses due to Rds be dominant? And the switch is in failure mode, it doesnt not even turn on the first time.
Not at all, because every cycle consists a turn on loss, then conduction loss, and a turn off loss. Your effective duty cycle of the converter as a whole, has nothing to do with chip losses while actively switching. Even if you deliver high power for 1s every 10s, then during this 1s, total chip losses will be as shown on the mentioned graph; but the average loss over time will be 1/10 of that value.
But this chip is not a converter, it’s a switch. MPQ5069. Unless I’m really missing something.
How did you end up with this issue?
We’re designing a similar circuit around this chip, and the lack of support / resolution your post received has me concerned that we should look for a different solution.