Hello,
as it happens, I have a design with LT1074HV on my table and I am having trouble achieving stability over a wide range of operating conditions.
The input voltage range for that design shall be lets say 35 to 45 V and the required output is variable from 4 to 28 V, output load up to 3.2 A. In the design, there is a 50 uH inductor on the #52 toroidal powder core, as suggested in the datasheet and an MBR1060 schottky diode, low ESR output filter cap 1000 uF with ~30mR ESR.
Variability of output voltage is done by a current sink (VCCS) stealing current from the FB pin node, hence increasing the output voltage proportional to the current sunk from the FB pin node. That works fine. However, as some specific output loads and voltages (duty cycles), I am getting instability. The symptoms are also magnified with temperature - the more temperature, the more susceptible to oscillation at higher load. The duty cycle simply starts to get noisy and wildly changing between cycles, resulting in audible hiss. (it is not an single frequency tone, like a classic oscillation).
I have a feeling, that it may be caused by a switching noise induced into the FB or VC pins. The layout is not the best, but the switching current loops are clearly separated from the rest, although there is some not the most awesome FB pin running in parallel with the output leg of the inductor and under it. There is definitely a room for improvement there.
However, I am not sure, if I am dealing with incorrectly designed compensation components, bad layout or both. The compensation network was previously tuned using LTspice, and some values that worked well enough were experimentally found. The available literature for the LT1074 unfortunately never mentions how to design the compensation network componets, we only know the gm of the OTA and that it is very asymmetric (different sink and source currents).
How would you proceed in designing the compensation network for LT1074? Apart of the obvious simulation and trial and still error? Is there even any simple analytical solution to it? I would like to make sure, that the instability problem is not caused by the loop being designed completely wrong.
I also think, that the 50 uH inductor is too little for the input-output voltage range, as there is a substantial ripple current; although still withing spec of the LT1074, the #52 core is not very happy about it and heats up slightly, wasting unnecessary power there.
LT1074 datasheet here:
https://www.analog.com/media/en/technical-documentation/data-sheets/1074fds.pdfsome design app note:
https://www.analog.com/media/en/technical-documentation/app-notes/an-44fa.pdfThank you for any suggestions.