I am trying to use the LT8608 as a positive to negative voltage converter as described for example linear technology design note 1021.
The simulation in LTSpice works fine, producing -6V from 15V supply. Necessary output current is less than 1A.
But the prototype isnt working, it immediately draws a high current from the power supply and the LT8608 controller is destroyed. I already checked the voltage ratings of the capacitors and inductor.
When I change the controller from LT8608 to LT8609A, the same prototype is working fine.
Is there any major difference between these two controllers which is relevant for this behaviour?
What could be the reason that the LT8608 is drawing such a high current right at the beginning?
Many thanks in advance
Schematic in LTSpice:
according to the mentioned DN1021 a Schottky Diode might be helpful to prevent unwanted voltage spikes at StartUp.
In general the possible output current for this 'negative Buck' approach might be smaller than for the regular buck schematic. This depends on the Vin and Vout setting. Hence the LT8608 might be too small for the required 1AThe LT8609A has a bigger Power FET inside than the LT8608, hence it's capable of more current which might be in your case necessary for the increased current at start-up.
First try the Schottky with the LT8608, if this won't work just go with the LT8609A.
thank you very much, the Schottky and additionaly the "Reverse Vin Protection" of the datasheet were necessary to solve the problem.
But now I get the next problem:
The device is synchronized from an external source, the level shifting of the sync signal is done as described at DN1021 (series capacitor).
At 200mA output current, the SW signal looks good (duty cycle about 0.3, freq = 500kHz) but at 400mA or above, the signal consists of a longer pulse and two short pulses. See the attached pictures.And the LT8608 and the inductor are getting pretty warm.
I've read that it works with pulse skipping mode, but only with light loads?The maximum calculated Iout(neg) (as described in DN1021) is about 1,05A, so I think it should work at least up to 500mA.
Please make sure the external clock signal is equal to or higher than the RT resistor set frequency. Otherwise the controller might have weird behavior.