Tom,
http://www.cde.com/tech/CDE_Lytic_Reliability.pdf
First let me say I think you are a great guy and I've read your posts
before, you seem to really want to mentor people.
Doug calling me an armchair quarterback, not knowing all the facts and not
knowing what I am talking about (I seriously doubt Doug would have ever said
that to my face) has encouraged me to look for a CDE webpage I was previously
on when working on a linear HV supply before.
OK, coming from a simpleton like myself I would never in a million years
build a power supply with a working DV voltage rating the same as my power
supply voltage on a Linear amplifier HV power supply. My statement about
Ameritron quality stands unchanged (yes the QSK5 may be a great unit, I used to
own the internal model with my AL-80A, but they need to document well radio
interconnects to help some radio users from destroying theirs)
OK, enough of me talking based from practical experience and common sense
from my end and lets look into a Cornel Dublier website.
I found a really good website from CDE (Cornel Dublier) that I was on about
a year ago and lucky enough to find it again. Reminder, CDE capacitors are the
ones we are talking about in these Ameritron linears. They should be the
authoritive source.
Here are some quotes from the Cornell Dublier (CDE) article and some
potentially oversimplified conclusions:
1) quote-"The MIL handbook basically starts from the maximum rated conditions
then states the incidences of random failures diminishes by half for each 20%
drop in applied voltage and for each 20 degree celcius drop in temperature".
Now they discounted some of the items in the MIL handbook, partially based upon
their stated increase in reliability in recent years but later said "we do not
disagree with the voltage derating factor" and "failure rate is proportional to
the cube of the ratio of the applied to rated voltage". They also make alot of
references to large capacitors (linear amplifier HV supplies included?) when
discussing this. CDE states "the rate of random pre-wearout failures of a bank
of capacitors is equal to the number of capacitors time the failure rate of a
single capacitor" (no mention of bleeder resistors, tolerance of bleeder
resistors). Being that the HV is not regulated, ripple is a player as well in
this as well and they discuss ripple (maybe
ripple and bleeder resistor tolerances are a main player as well?)
2) At the beginning of the article they talk about the life model of the cap.
Quoting part of the article "Vr is the rated DC voltage. The rated DC voltage
is defined such that the peak applied voltage should not exceed this value".
Immediate conclusion on that part is that if an Ameritron has a 3600VDC
supply with unavoidable ripple, the peak voltage will exceed the 3600vdc by
half the amount of ripple present. The Ameritron has a 3600vdc supply with
exactly 3600vdc working voltage of capacitors. Line fluctuations aside, the
peak voltage is exceeding the working voltageand shouldn't according to the CDE
article. Also, nobody on this reflector has ever experienced line fluctuations
in AC before have they? Any positive line fluctuations would cause the Vr
(rated DC voltage) to be exceeded by even more. I have seen one too many
varistors blown in areas of poor power due to line fluctuations.
73,
Mike
Tom W8JI <w8ji@w8ji.com> wrote:
>Also, for the price on their high end units, there really
>is not an excuse for them to use 2,000 hour caps vice the
>5,000 hour caps that are not much more. Read my earlier
>email about how you can expect 3 years before the power
>supply is beyond it's rated life expectancy.
We can read it, but we should be aware it isn't correct.
I'm not sure if MFJ is buying 2000 hour MTBF caps or not,
but the 2000 hour number is NOT the expected mean time hours
before failure in operation. It appears to me that is what
you are doing, and that what you are doing is absolutely
wrong. ****Everyone needs to understand this very well.
Whatever the raw number, 2000 or 5000 or whatever, it is NOT
the expected life. That is a base number used in a MTBF
formula.**** In other words a "2000 hour MTBF" in the
component rating does not mean that is the operating MTBF.
That is simply very wrong. Anyone who reads component data
sheets or application notes would know that.
(Those caps weren't initially speced to be 2000 hour parts,
but that was in the early 1980's and I'm not sure what they
are buying now.)
To put it bluntly, the numbers about expected life in the
earlier email are meaningless. They are not even correct for
a 450 volt part operated at 85 degrees C (185F) internal
temperature because, even if raised by a power of 1 (an 85C
cap operated at 75C internal temperature), the life is 4000
hours.
It's been a while since I have done this but the voltage
derating is a multiplier based on ratio of published to
maximum to actual operating voltage, so a 450V cap operated
at 300 volts has a multiplier of 450/300 = 1.5 while a 450 V
cap right at 450V has a multiplier of 450/450=1. That number
gets multiplied times the 2000 hours, that number doubled,
and that number raised by an exponent related to
temperature.
This is why voltage makes very little difference compared to
temperature, and why the capacitors are placed right in the
inlet airstream and suspended in an open air configuration.
The actual formula you would have to use is:
Lop = MvLb2 ^ [(Tm - Ta)/10] where Lop is the expected
operating life in hours, Mv is a voltage multiplier for
voltage derating, Lb is expected operating life in hour for
full rated voltage at full rated temperature, 2 is just a
multiplier, Tm is the maximum permitted internal operating
temperature in °C, and Ta is the actual capacitor internal
operating temperature in °C.
The capacitors are 85C capacitors. They, unlike most
applications, are in full room temperature air inlet with
all the cold blower air pulled across them. If the voltage
is derated as you suggest, Mv changes by the RATIO of
derating. But if you look at the life equation MVLb2 is
raised by the exponent Tm-Ta divided by 10.
Tm is 85, Ta is 30 if you are in a 86 degree room but is
modified by a temperature correction influenced by case size
(which controls internal heat transfer) and internal power
dissipation while operating. I **recall** internal
temperature was calculated to be around 38 degrees if in
continuous operation with a 2500 watt load on the power
supply.
We would have:
1*2000*2 ^ (85-38)/10 = 4000 ^ 4.7 = 18,800 hours MTBF
assuming the 2000 hour part was used and the steady load was
2500 watts.
Let's say the amp ran 40 hours a week, and the operator
transmitted 25% of the time with a steady carrier at 2500
watts input. That's 10 hours of operation drawn from the
18,800 reserve and since the idle time of 30 hours draws
from a 3500^5.4 = 18,900 reserve we can consider it to all
be from the same reserve. The life does not go up by idle
time.
So we simply have 40 hours out of 18,800 every week.
18800/40= 470 weeks. Your 3 years is actually 470/52 = 9
years when an amp is operated 40 hours a week, but even that
isn't even the time to hard failure. That is the time until
onset of wear out.
The largest single cause of failures is an open bleeder
resistor. That mode totally dwarfs any other failure mode.
People should always have the common sense to check the
resistors when replacing a capacitor, and if one cap fails
ALL the capacitors should be changed. The techs at
MFJ/Ameritron should be smart enough to tell people to do
the right thing and check the resistors, to change ALL the
caps if one fails, and to put the polarity sensitive parts
in the correct direction. A good tech does more than send
out parts, he educates the customer and leans from the
customer.
It's necessary for a tech to walk customers through things
rather than just sending out parts because most people
actually don't understand why things fail. If a tech lets
the customer do simple things wrong two or three times
you'll get an earful about how bad the design is!
I'm also not sure about QC at MFJ. QC is a major PITA in any
environment, and it is worse when no one constantly watches
everything. As I said before, and I'm NOT guessing when I
say this, the largest single source of capacitor failures by
a huge margin is bad resistors, and the largest cause of
resistor failure is handling of the actual parts. This could
all go back to someone not handling circuit boards correctly
at the factory.
73 Tom
---------------------------------
Talk is cheap. Use Yahoo! Messenger to make PC-to-Phone calls. Great rates
starting at 1¢/min.
_______________________________________________
Amps mailing list
Amps@contesting.com
http://lists.contesting.com/mailman/listinfo/amps
|