On Fri, 19 Jul 2002, Roman Black wrote:

>Peter L. Peres wrote:
>>
>> >You need *at least* 680v
>> >insulated fets.
>>
>> That's the second reason why you use a buck regulator followed by a H
>> switch polarity inverter.
>
>I don't think that is a good idea either. Now you are
>talking a switching semi good for 3.5kW @ 350v, a buck
>inductor good for 3.5kW, and a h-bridge good for 10A
>and 680v minimum. We recently had an AC engineer on the
>list saying that the AC mains is typically exposed to
>1.6kv spikes of considerable energy, your suggestion
>is exposing the h-bridge semis directly to this.

?! Switching semis for 3.5kW at somewhat high voltage are not hard to get
but if you had paid atention to my reasoning you would know that the
switcher would be rated lower than this. I'll just remind you that P=U^2/R
at all times, and the buck regulator will supply max power with the pass
switch open when this happens, accounting for minimum losses. Also a buck
regulator does not care very much about power, more about current. The
latter is a balmy 15A or so peak. This is why high voltage is interesting.
A Rdson = 0.1 ohm FET switch (10 1ohm devices in parallel) will only
dissipate 15W total at this time (neglecting switching losses). The H
bridge built with IGBTs will also dissipate only a few watts, and an
active buck diode (more parallel FETs) will also burn a little. A coil
that handles 5kW in this regime probably requires careful study for
saturation (and probably field compensating windings) and a big core, and
it won't run at high frequency, and that's about it.

>Then you have the cost of the parts involved (not even
>including the 28x 12v battery stack and maintainance).
>The "typical" setup of a 3.5kv toroidal transformer and
>push/pull lwo-voltage semis (probably about 36v etc)
>has so many advantages. I worked in a factory and was
>responsible for maintenance on a *large* series battery
>stack, it's not pretty. They develop differing
>characteristics, which also change over time, and need
>constant attention.

Yes they need charge equalizing. Not so terrible if planned in from start.
Remember that they are cheap auto batteries.

>> >Now look at typical HV switcher, this will use a 1500v 6A
>> >MINIMUM rated switching transistor. To switch a
>> >100v rail!
>>
>> They are somewhat reliable, and I say this on the grounds that the vast
>> majority of TVs out there did NOT visit your shop(s) within their useful
>> life. The high specs on the LOPT driver are due to the way you need to
>> drive the H deflection yoke (it has to be half-resonant or else you do not
>> get the required current to deflect the beam in a 120 degree crt).
>
>Sure, i'll confess to being jaded. :o) But semi's
>switching at high frequencies are continually exposed
>to thermal and magnetic stresses. They DO wear out.
>I'm real curious what voltage overhead YOU would suggest
>for semis switching 340v in high frequency PWM??

In a buck converter scheme as I was proposing 500V devices for all
switches would be enough. Then you could have identical devices everywhere
and just buy two rails of them from the manufacturer (great savings).

>> I could never understand why they used high frequency H deflection and low
>> frequency V deflection. The other way around would save 25% or more
>> energy, cost, etc. (the H:V ratio is 4:3 in TV).
>
>They chose it that way to save on lines per frame.
>Your way would make it 33.3% more lines per frame.
>Even with the lower number of lines they still had to
>interlace...

Huh ? The H/V resolution ratio of the eye is about the same ? More like
they had technological trouble with sync separators and power supplies and
they wanted the H and V frequencies as far apart as possible.and one of
them on the mains freq. so minimize noise bars.

Peter

--
http://www.piclist.com#nomail Going offline? Don't AutoReply us!
email listserv@mitvma.mit.edu with SET PICList DIGEST in the body