According to the constraints specified in the data sheet for the LT8330, the SEPIC inverting converter voltage output can be no greater than half that of the non-inverting boost converter. At least, this is what I read considering the limits that can be used in the FBX voltage divider resistors.I need to generate -48V (from +12 in) and it need not supply much current - it's used in an electrostatic application. I've already configured a design for its opposite, a +48V boost, which fits into the FBX range without any worries. Oddly, the LT Spice model, using the standard values on the data sheet for a -24V converter, achieves only -14.5V.Can the LT8330 be extended to -48V out by exceeding the FBX limits and achieve a reasonably stable voltage? (+-0.1v is fine here, plus noise). Or, are there better solutions?
Thanks for the post.
The negative counterpart of the SEPIC is commonly called a Cuk.
The voltage across the coupling capacitor on a Cuk is VIN + VOUT, but the equivalent capacitor voltage on a SEPIC is VIN. That is why a conventional SEPIC can be designed for a higher output voltage than the Cuk counterpart. The FBX pin on the LT8330 doesn't limit either positive or negative output voltage .
The -24VOUT application example from the LT8330 datasheet runs fine in LTspice. I took the LTspice example for the -12VOUT Cuk from the LT8330 web page and modified it for -24VOUT according to the datasheet example circuit. The image below shows the results.
There is an easy way to make -48VOUT using the LT8330 - just use a voltage doubler as shown in the LTspice simulation image below. The idea came to me from one of our field application engineers, who probably got the idea from one of our (brilliant) customers. Due to the topology, the circuit is not too electrically noisy to begin with, and the ferrite in the schematic makes the circuit even less so. The circuit is also nice since there is no direct path from input to output (as on a boost), so a short on the output will not directly pull down the input.
I briefly considered a doubler, and especially given the circuit charges an element and draws virtually no current. In the end, I moved to the LT8570 which, fortunately, has a very nice and apparently tested design for -48V.
Thanks for the suggestions!
Thanks for the update.
For others who may use this post, my colleague suggested the addition of a resistor to the doubler circuit as I show below. The resistor sometimes is needed when C4 dumps charge to C6 at the first SW turn-on. If the combined impedance of the SW, L3, D2 and the capacitors themselves is too low, then the SW current may be high enough to hit protection the first time SW turns on, which will trigger a re-start. Worst case is at higher VIN and constant-current load.