Continue to Site

Welcome to EDAboard.com

Welcome to our site! EDAboard.com is an international Electronics Discussion Forum focused on EDA software, circuits, schematics, books, theory, papers, asic, pld, 8051, DSP, Network, RF, Analog Design, PCB, Service Manuals... and a whole lot more! To participate you need to register. Registration is free. Click here to register now.

Stability Analysis of DCDC-Conveter using PFM

Status
Not open for further replies.

pancho_hideboo

Advanced Member level 5
Joined
Oct 21, 2006
Messages
2,847
Helped
767
Reputation
1,536
Reaction score
732
Trophy points
1,393
Location
Real Homeless
Activity points
17,490
We can apply PSS/Pstab Analysis of Cadence Spectre for DCDC-Converter using PWM.

However we can not apply PSS/Pstab Analysis for DCDC-Converter using PFM.

If you will do stability analysys of DCDC-Converter using PFM, what method do you use in simulation ?

If I use state-space-averaged model for PFM, how can I define gain ?
 

In case of doubt, derive the parameters in transient analysis.
 

PFM is still fundamentally based on duty cycle variations, so state space averaging applies the same as it does to PWM. In the small signal sense, all the pole and zero locations of a converter will be the same whether it uses PWM of PFM. But the input matrix B may change a bit.
 

A key question whether the system is hysteretic (comparator,
over/under gating of pulse train) or linear (error V-F converter).
The hysteretic style can be unconditionally stable, the V-F
linear has the classical control loop stability to tune. V-F style
can probably be behaviorally approximated to allow AC analysis.
Bang-bang hysteretic would be more of a head scratcher.
 

Right. However period is not constant.

Not same.
The beauty of state space averaging is that the model doesn't depend on the switching period, just the fraction of the switching period spent in each state (the duty cycle). Whether the duty cycle is varied by controlling period of pulse width is irrelevant. For example, take a dcdc converter which is operating with fsw=10khz and Ton=25us (d=0.25). If I perturb it so that its fsw stays at 10kHz while Ton changes to 40us, then its state space averaged model will have the same response as if I perturbed its operating frequency to 16kHz while its Ton stays fixed at 25us. Both describe exactly the same change in duty cycle (0.25 to 0.4), and that's all that matters to the state space averaged model of the circuit.

The only practical difference between pfm and pwm is in the gain of the PWM modulator which takes some control signal (call it u) and converts it to duty cycle d. For a PWM modulator where period T is fixed while Ton is proportional to u by some gain (call it α), then d given by d=u*α/T. For a PFM modulator where Ton is fixed and fsw is proportional to u by gain α, then d is given by d=u*α*Ton. In another PFM case where Ton is fixed and (Toff^-1) is proportional to u by gain a, then d is given by d=Ton/(Ton+1/(u*α)). In the latter case the modulator is nonlinear, so for small signal analysis you just have to calculate its small signal gain at whatever bias point you've chosen. Whatever the gain of the PWM modulator is, it basically gets absorbed into the input matrix B.
 

I agree with mtwieg that there may be value in treating it as pwm because its a form of PWM just happens to vary switching frequency too. Does switching frequency have an impact on loop gain or stability - probably not much. So you may not need to simulate that or can verify that independently.

I'll also point to the value of a tool like PSIM which does AC sims by churning through time domain simulations behind the scenes. If the circuit simulates time domain then it can plot the AC curve just fine with any modulation scheme you've come up with.
 

Status
Not open for further replies.

Part and Inventory Search

Welcome to EDABoard.com

Sponsor

Back
Top