[Amps] Pin Diodes

[W0UN -- John Brosnahan]

At 11:19 PM 3/7/2006, Rob & Terri Sherwood wrote:

>It is interesting that Alpha threw in the towel on PIN diodes with the
>upcoming 9500.  Tech support told me the yield on a batch of diodes was
>terrible, possibly as low as 30%.  That is a shame, as my Alpha 89
>(manual-tune version of the 87A) has PIN diode switching, and is a
>wonderful QSK amp.  I don't believe an amp with vacuum relay T/R
>switching can be considered a full QSK amp.   Yet Alpha advertises the
>99 and the 9500 as full QSK.   I run my 99 in semi-breakin only. How
>long can a vacuum relay last in full QSK mode?   There must be a maximum
>keying rate and a typical cycle life for anything mechanical.  Are they
>banking on the fact that semi-breakin is usually adequate?

Hi, Rob--

I, too, was disappointed to hear that the 9500 was going back to a vacuum-
relay-based QSK system since I have been very happy with my Alpha 87As
and have never had any problems with PIN diodes after Alpha sorted out the
issues for the Alpha 86 T/R switch update.  It sounds like the PIN manufacturer
has a process control issue, not unlike issues that Kilovac and Eimac have
had on their products from time to time.

The vacuum relay that is used by Alpha and many others for QSK was originally
a Jennings design and later Kilovac manufactured it under license to Jennings.
But Kilovac had some real production issues for a time and their 
infant mortality
rate was as high as 75% in some cases -- for example, K0RF's experience with
the 80/75 meter switching on his first 2L 80M beam.  Three of the 
four relays were
DOA.  And my own experiences with a 100 or so relays that had DOA rates
well in excess of 50%.    Ultimately, Kilovac redesigned the relay 
and the later
version was not only more reliable than the Jennings but was also faster.

Personally, if I am using OLD (NOS) relays I would tend to use Jennings.
But if I am using new relays then I would tend to use Kilovac.  With older
Kilovacs I tend to run them in a test setup for a few days to sort out the
potentially bad ones.

I was also concerned about relay lifetimes when run in full QSK mode by an
active ham.  I forget the spec on lifetime but it is was something like 1 or 2
million cycles.  At one point (a decade or more ago) I did the 
calculations that
showed that a VERY active CW contester would use up the relay lifetimes in
a couple of years or so.  I don't know how the infant mortality rate 
is used with
the operating lifetime calculations.  If the infant mortality rate is 
included with
the lifetime spec then it would tend to bias it to a lower value and once the
DOAs are removed then the good relays may last a longer time than the
average lifetime spec indicates.  It is certainly true in my experience that
the relays last a lot longer than 1 or 2 years, even for very active 
contesters.
In any case, it IS a consideration and the cost of PIN diode TR switching
systems is not that much different than vacuum relay based systems.  But
infant mortality issues can be a real problem for a manufacturer  -- if the
TR components fail during testing and burn in then the rework is an expensive
cost to bear.  But if they don't fail until they arrive at the 
customer's QTH then
there is also the shipping costs on top of the other expenses, not to mention
the bad public relations issues.

Eimac has had similar issues with the 8877 family of tubes.  When moved from
CA to UT there were issues with the production for some time and when the
product line was ultimately moved back to CA there were again issues.  And
the original pulse-rated version (3CPX1500A7) had problems because it was
specified to hold off a higher voltage than it was being tested for, 
because the
test line was the same as the non-pulse version.  Once Eimac changed their
test procedures on the pulse version things got a lot better.

My personal experience with PIN-based TR switching is that it is very reliable,
but that you need to exercise a little care in the antenna system by making
sure that all of the antennas are at DC ground before they are switched onto
the amp.  To me, it seems like the PIN failures are "biased" towards areas of
high lightning and in many cases are due to static charge build up on 
non-grounded
antennas.  When an antenna is not grounded you can get a large static charge
on the antenna that is discharged through the amp when the antenna is selected
by the antenna switch.  Although there is an RF choke at the output of the amp
that provides a DC ground I have never calculated what the impulse energy is
that spikes into the PIN diodes when an antenna is selected.

My own personal design experience with REAL QSK involves PIN diode TR
switching at the 3 or 4 uS speed for wind-profiling radars at the 50KW pulse
level.   Although the design is somewhat different for a narrow-band QSK system
using an ATR type switch I never saw failures unless something else went wrong,
like a tube flash over that put a large spike into the TR PINS.   I 
attribute a lot
of this success to the fact that the antennas used for these radars were at DC
ground and were never switched in-line like ham antennas are.

Just some thoughts to add to the discussion.

--John  W0UN