[Prev][Next][Index][Thread]

Re: MMD (not another acronym!) --> HV Diodes and Failures

Mike:
	My high-power HV diodes typically puncture when they fail,
so they become somewhat shorted (to a meter), rarely hard.

Since many HV diodes are really stacks of diodes internally, I
will see the internal resistance which is normally (good diodes)
say >400Mohm reverse and ~75Mohm forward become 5Mohm (or less
or more) one way or even both ways, indicating that some or all
of the stack can fail, but not conduct hard.

Small HV diodes (1ma sort) fail both shorted and open.

When the high power diodes fail in TC use, (full wave bridge)
I'll suddenly see a dramatic decrease in TC output and a
very large rise in AC input current.  My Variac will sing
almost immediately.

I highly recommend never potting or encapsulating these high-power
diodes as an array unless you are using a string of say low voltage
1KV diodes. Then, only encapsulate one set (20KV worth for 10KV-15KV
operation) and use four of those to build the bridge.

I've never had more than one leg of the bridge go out at a time,
so this will let you replace just the failed 1/4 portion.

As far as I can tell, the RC networks are independent of the
operational and failure modes ... you have to have them for series
capacitors regardless, and if AC gets imposed into the MMC array,
the RC certainly won't hurt but can hardly help.  Other than that,
I was a little confused by the statement:

> Multi-string MMD's seem to be impractical for size reasons, and
> silly to boot as the 1kv, 6A diodes would seem to be sufficient
> for any system under 84 kva ; )

This implies that you believe the tank capacitor to be optional ...
which it is not.

As far as diode failure and stress on the caps, the only good thing
here is that if a leg of the bridge fails, the bridge output will
drop (so the MMC won't get stressed) and the input voltage to the
bridge will be effectively loaded down if not partially shorted,
so the bridge output drops even more, protecting the capacitor
bank even more.

One thing you didn't mention though was the limiter resistor
between the caps and the bridge.  You *MUST* put some sort of
limiter into the circuit or the current surges are certain to
blow the diodes out.  This will also help prevent cascade failure
problems if the caps or the diodes fail.

As to the over-rating on the capacitor, I've measured a full 50%
negative ring after discharge present on the cap bank, so 10KV
will generate a full swing of +10KV, -5KV (or 15KV Peak to peak).
That means you really need 20KV minimum to run a 10KV system or 2X.

Once you get everything in tune and the runs are predictable,
then you can push the margin down to 1.5X, but out-of-tune
coils produce very high reflected power which can be RFI
offensive as well as damaging to components.

*---------------------------*
* Kevin Ottalini            *
* WhoSys / Who Systems      *
* High Voltage with Style!  *
* ottalini-at-mindspring-dot-com   *
* Often in Another Reality  *
*---------------------------*

----------
> From: Tesla List <tesla-at-pupman-dot-com>
> To: tesla-at-pupman-dot-com
> Subject: MMD (not another acronym!)
> Date: Tuesday, March 07, 2000 5:49 PM
> 
> Original Poster: "Mike Nolley" <mnolley-at-mail.slc.edu> 
> 
>     As preparation for the diode bulk buy, I have a few questions for the
> learned ones about diode failure.  Do hv diodes fail open, or closed?  If
they
> fail open, then a single-string MMD would not catastrophically fail, if
it were
> encapsulated--  It would just turn off.  The problem with this is related
> to the
> encapsulation--the diodes would be hard to replace.  If on the other hand
they
> fail closed, I think it would be fairly likely that a single-string MMD
would
> fail quickly, if it were rated close to supply voltage.
>     My second question is, how much should the MMD be over-rated--*given*
the
> use of an R-C network (which I presume any diode-user would need by
default).
>      Multi-string MMD's seem to be impractical for size reasons, and
silly to
> boot as the 1kv, 6A diodes would seem to be sufficient for any system
under 84
> kva ; )
>       Thanks for any, and all responses.
>                     --Mike
>