microprocessor controlled SMPS

[Joerg]

Hi Frithiof Andreas,

Thats because you worry too much: Open some consumer item and see how it's
done, f.ex. the Washing Machine - in mine there is a two-phase drive with a
single-chip micro driving IGBT's straight through a set of opto-couplers
with very little "glue" on it, there is some current and voltage sensing and
I assume that there is a short-circuit turn-off in the drivers too.

The rest is code - that controller stuff is not what fails - last time it
was the carbon brushes in the synchronous motor it uses. Easy to fix once
one found the interpretation of the flashing lights on the net.

Looking at consumer goods is always a great idea. That's how we all
learn cost efficiency. Just one word of caution: These manufacturers do
not have to worry as much as some of us have to. If, for example, the
washer errors and floods your basement they might say that is your
problems, not theirs. Then to add insult to injury the insurance company
might surprise you with the statement that this isn't covered.

Elevators, cars, industrial gear, medical devices and the like are
another matter.

But even on consumer gear there are examples of poor failsafe strategy.
Remember all those house fires caused by plugged up dryer exhaust ducts?
Many of those were attributed to poor failsafe structure. A reasonably
designed overtemp monitoring can prevent that.

Assuming you are in Scandinavia this one may hit home a bit: In the US
we have lots of house fires because of over-fired wood stoves. A simple
bi-metal damper control could reduce the air intake upon exceeding a
certain temperature. Doesn't need electronics. Some stove have that,
many don't.

Regards, Joerg

 

[Frithiof Andreas Jensen]

(Takes drink of his double strength brew)
How dare you suggest I'm jumpy!

No code would be done, if not for Caffeine!

I disagree. On a fast micro in the flyback topology, handling the short
circuit case in software is not very hard at all.

I think we misunderstand each other - I am thinking on the case when one of
the switches fail in a multi-phase or push-pull topology yielding a short on
the *converter side*. A short on the load side is different, and of course
should be handled in software with confidence.

How fast you charge up to the working voltage is usually not the issue.
It is the terminal impedance and how the circuit reacts to a change in
load that really matters.

You cannot regulate the output voltage faster than you can move energy
to/from the storage caps - so in practice your frequency response end up
being power limited. Reduce the capacitance and the loop can be faster - but
the supply then has to handle a higher peak power to keep the output steady.
It's a tradeoff.

 

[Ken Smith]

No code would be done, if not for Caffeine!


I think we misunderstand each other - I am thinking on the case when one of
the switches fail in a multi-phase or push-pull topology yielding a short on
the *converter side*. A short on the load side is different, and of course
should be handled in software with confidence.

Oh! that's what you mean.

I was refering to a load side short. Since nothing in any of my supply
designs ever fails, I hadn't considered the posibility.

You cannot regulate the output voltage faster than you can move energy
to/from the storage caps - so in practice your frequency response end up
being power limited. Reduce the capacitance and the loop can be faster - but
the supply then has to handle a higher peak power to keep the output steady.
It's a tradeoff.

I think this is where you misunderstand.

Imagine a 1000V supply that has to maintain that voltage to within 1mV.
(Yes I know this is extreme) Assume that the output cap is 1uF. Imagine
that it is up and running and the load suddenly increases by 1mA. The
output voltage will be ramping down at a rate of 1V/mS. The circuit must
react within 1uS or the 1mV regulation requirement is not met.

BTW: There really are supplies that must be this stable in the short
term.

 

[The Real Andy]

I am, because I want all the speed I can get - and I like asm for some
reason

Sick man. I hate asm, but there is no way I am going to buy a c
compiler for every micro that i use.