> They are just plain under-designed. There are eight
> 450V caps in the supply.
> With a no-load plate voltage of 3600V that gives a 0%
> margin.
The capacitors were initially 600 volt foil, and they were
required to have a certain leakage at 525 volts at a
specified temperature in a test at the manufacturer. That's
one of the things you can have done when ordering thousands
of caps factory direct. The original caps would be fine for
years at 4000 volts. I have an AL1200 and an AL1500 that
have been running on the 220V tap ever for at least 17 years
now. That's a long time for a "marginal" component.
I can't speak for the parts they are buying now, but the old
ones weren't any worry at all sitting at 3600V. The only
recorded failures EVER in the original caps from Centralab
and Mallory were when a bleeder/equalizer opened. They
virtually never failed.
For miltiary
> equipment we typically design anywhere between 40% and
> 60% margin.
Actually that's a bit foolish for electrolytics unless they
are subject to surge voltages. There is very little life
degradation operating them right at rated voltage.
You can see in this technical paper by a manufacturer:
http://www.nmr.mgh.harvard.edu/~reese/electrolytics/tec2.pdf'
where Nichicon says:
-9-4 Applied Voltage and Life
The degree that applied voltage effects the life of the
capacitor when used below the rated voltage is small,
compared to the degree that ambient temperature and
ripple current effects life. Therefore, when estimating the
life of a capacitor, the voltage coefficient to the applied
voltage (Fu) is calculated as 1. An example of the test
results is shown in Fig.2-16.
This is why, with 600 volt foil, a 525 volt test rating, and
a 450 volt working rating the capacitors were perfectly
fine. Of course I can't speak for anything they are doing
now, but i hope they would still be holding to the original
specs.
> When I pointed that out the Ameritron service rep his
> response was that
> "Well the loaded voltage goes down to 3300V so there's
> plenty of margin",
> which is bull&*%$.
Actually it isn't. It's a combination of ripple current and
temperature that has the largest effect on life.
> In intermittent amateur service the typical duty cycle is
> 40%
> so the caps are seeing 3600V more often than not. The
> caps began to arc
> internally, and then blow the balancing (bleeder)
> resistors.
You have that exactly backwards.
If the bleeders are bad, that is why the caps are failing.
Not the other way around.
The largest single problem with manufacturing when I was
watching Ameritron was the bleeders. The largest root cause
of bleeder failures was handling of the resistors in
assembly. It was a constant fight to keep employees from
setting boards on top of each other before the caps were
mounted. A second major cause of bleeder damage was
screwdrivers. If the assembler used an air driver without a
shield the blades would often scar a resistor.
The second largest failure was a backwards cap.
To reduce failures the resistance of the string was measured
before caps were installed. The electrolytic assembly was
tested in a HV fixture that applied 1000 volts and measured
leakage current as well as voltage across each capacitor.
There was a duplicate divider that would trip a light if an
electrolytic was backwards or a bleeder was open.
I'm not sure how many of these proceedures are still in
effect in Starkville, but it would be a grave mistake to
order random parts from distribution, to not use guards on
tools, and to not test the assembly before use.
> In addition to that, they are listed by CDE as 2000
> hour MBTF caps. At an
> hour and a half of use a day - three years and they're
> done. CDE has
> 500V and 550V caps in the same value at up to 5000 hour
> MTBF. The
> cost of using those would have been peanuts compared to
> the total cost of the
> amplifier and would have made a much more robust design.
As I recall the better foil was used in the design. There
was quite a delay one time when they couldn't get the better
foil. The extra testing and better (higher voltge) foil cost
something like 60 cents per capacitor.
73,
Tom
(who actually knew what was going on with that assembly
prior to MFJ ownership)
_______________________________________________
Amps mailing list
Amps@contesting.com
http://lists.contesting.com/mailman/listinfo/amps
|