RE: insuring the stability of power delivery
This is a new wrinkle! I'm not a distributed power supply designer, but
I'll take a stab at this anyway. Shout me down if I don't know what I'm
talking about.
At 11:02 AM 5/31/2001 -0400, Lynch, Brian wrote:
>In our work with power systems, we have found that to insure
>stability of a system, the closed loop output impedance of the
>sourcing supply must be lower than the closed loop input impedance
>if the load power supply over the frequency band and voltages
>of interest. This concept is also discussed and taught at various
>universities. VPI and MIT are two I have direct knowledge of.
I see this as two problems - startup and steady state.
In startup, the output impedance of the PSE power supply is what it is, and
the input impedance of the PD appears in series with whatever is doing the
inrush limiting, be it in the PSE or the PD. The PSE power supply isn't
likely to oscillate in this case. The line voltage is another matter -
presumably if the limiter was in the PSE, the line voltage could oscillate
(by Brian's theory); if it was in the PD, the line would stay quiet, but
the node inside the PD between the inrush-limiting pass device and the
input of the PD supply could oscillate. The PD designer can beat this by
holding the switcher inactive until the input cap has reached its final
value, or by balancing impedances properly.
By this logic, if we force inrush limit into the PD, the spec can mostly
avoid this issue (at least in startup). If we allow inrush in the PSE, we
would have to define the nature of the PSE Zout and the PD Zin at startup.
Once startup is complete, all inrush circuits are turned on hard, and their
loops go open - they won't oscillate anymore. Now the issue is between the
Zout of the PSE supply and the Zin of the PD supply. I'd argue that the
various pass devices (and the wire, if it's present) will ensure that the
PD impedance (as seen by the PSE power supply) is reasonably high in all
cases. The PSE power supply must have a low closed loop output impedance
(DC and AC), with the actual value dependant on the number of ports it
might need to power simultaneously. With a wide PSE (64+ ports?), that
could still be a challenge.
The spec can't avoid the issue here - I suspect we'd need to spec the
maximum effective impedance over frequency at the PSE port (= the actual
Zout multiplied by the number of simultaneous powerable ports), and the
minimum impedance looking into the PD jack.
In this analysis, I'm taking Brian's assertion about impedances at face
value - I need to check this assumption before we follow this logic too
closely.
Dave Dwelley
Linear Technology