Anybody here familiar with some of the stability quirks and parasitic issues with this type of design?
I believe I'm having a damping problem with the passive power parts. I'd like to SPICE the loops but I don't know how to linearize for phase/gain plotting. Know any sources?
Details Hysteresis controlled Offline Output 100VDC@2Amps Input 120VAC
Thanks
If you have a semi working model, you can measure the gain and phase of the system. You normally start by making the servo control loop way slow. You can then inject a small signal and measure its effect.
Exactly how you are doing the hysteresis controlled version of the converter may make this hard to do. You may have to first figure out what the expected ripple on the output is and replace that with a generator injected into the comparitor.
Making a Cuk stable can be extra tricky because there are two complex pole pairs, a couple of zeros and some extra phase shift at the high end that doesn't have a gain decrease to go with it.
Stability may be over rated. A "hit and miss" regulator can never be truly stable but they perform quite well. A nice stable waveform looks good on a scope but makes it harder to pass the FCC. Adding noise to the switching frequency spreads the RF out and lowers the peaks.
-- -- kensmith@rahul.net forging knowledge
If you have a semi working model, you can measure the gain and phase of the system. You normally start by making the servo control loop way slow. You can then inject a small signal and measure its effect.
Exactly how you are doing the hysteresis controlled version of the converter may make this hard to do. You may have to first figure out what the expected ripple on the output is and replace that with a generator injected into the comparitor.
Making a Cuk stable can be extra tricky because there are two complex pole pairs, a couple of zeros and some extra phase shift at the high end that doesn't have a gain decrease to go with it.
Stability may be over rated. A "hit and miss" regulator can never be truly stable but they perform quite well. A nice stable waveform looks good on a scope but makes it harder to pass the FCC. Adding noise to the switching frequency spreads the RF out and lowers the peaks.
--
--
kensmith@rahul.net forging knowledge
Hello Kensmith
It\'s that ---L---|---C---|
|
switch
If you have a semi working model, you can measure the gain and phase of the system. You normally start by making the servo control loop way slow. You can then inject a small signal and measure its effect.
Exactly how you are doing the hysteresis controlled version of the converter may make this hard to do. You may have to first figure out what the expected ripple on the output is and replace that with a generator injected into the comparitor.
Making a Cuk stable can be extra tricky because there are two complex pole pairs, a couple of zeros and some extra phase shift at the high end that doesn't have a gain decrease to go with it.
Stability may be over rated. A "hit and miss" regulator can never be truly stable but they perform quite well. A nice stable waveform looks good on a scope but makes it harder to pass the FCC. Adding noise to the switching frequency spreads the RF out and lowers the peaks.
--
--
kensmith@rahul.net forging knowledge
.
Hello Kensmith
Thats for the info..
It\'s that classic cuk config that I\'m stuck on.
Stability analysis with amplifiers..is easy...A bode plot is
in&out sin wave frequency, amplitude and phase at an operating
point.
But with switchers...It\'s like 2 operating points and then
there\'s a duty factor too.
I\'ve heard that the cuk can be transformed for spice to get a
bode plot..
Some sort of averaging theory to linearize the circuit for spice
simulation. The switch and diode are replaced with linear
"boxes".
I wonder if I have to do this for the hysteresis controlled cuk?
|-R-C----|
------L------|---C----|------L---|
D A |
pulses in>>---Mosfet diode load
S C |
| | |>>triangle I ....Bode Plot??
Gnd Gnd Gnd
This document just came across the LTspice users group on Yahoo that sounds like it might be applicable to your problem: Advances in Averaged Switch Modeling and Simulation
Thanks for the info Kensmith.
I'm new at posting..A reply went out before I completed writing.
Neglecting the controller propagation delay and transient times for the moment..
It's those poles and zeros in the Cuk top. I'm wondering about..
I've made bode plots of amplifiers.. Set an operating point and sweep a sine...easy.. Look at the gain and phase and one can determine stability.
But the Cuk...it's like 2 operating points (on&off) switching a RLC network with a duty and frequency. With a hysteresis controller the timing is formed by the output inductor current banging into the hysteresis limits. (Hysteresis control current feedback.)
I've heard that Cuk Topology (and like switchers) can be tranformed so that spice can make a bode plot. Something to do with replacing the catch diode and switching Mos with functions so that the switching is averaged.. I'm not sure but I think the whole network is warped into some linear approximation so that there's one operating point and a sweep can be done. (I've noticed some spice programs are using blocks for buck and boost..) ...And this is where I lose it...How to sweep a Cuk?
While writing this I did notice the post about LTSPice and averaging ..so I'll check that out. Thanks
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.