This post is very long, so skip it if you don't want to dive deeply into the
issue of 87A repairs. Hopefully, the information will be useful to some of
you.
The reliability and "repairability" of 87A amps depends on the year of
manufacture. My understanding is that the early 1990s models have a number
of design flaws that almost guarantee major troubles. However, I don't know
the exact years during which these problem units were made. I think it was
1990-1993, maybe even including 1994. Does anyone have more precise
information?
My 87A was made in mid-1995, and I believe that Alpha had just made a number
of important design changes that year. But I remember the salesman at Alpha
saying, "These things are hot rods and you should expect a few problems." No
doubt, he had fielded a lot of complaints from owners of earlier models and
wanted to give me a heads-up.
I love my 87A, have run it hard for 17 years, and it has been very solid.
That's not to say it's been completely trouble-free. I've had numerous
issues over the years, all of which I fixed myself without shipping the amp
back to the factory. That was only possible because people at the factory,
mostly Glenn and Brad, gave me the information I needed.
The first problem showed up soon after I bought the amp -- the plate choke
relay wasn't closing when moving between the low bands to the high bands. It
happened when moving in one direction (can't remember which), but not the
other. It turned out to be a firmware bug, and Alpha was able to give me a
patch to fix it (that English fellow who was President for a short time
actually handled my problem himself!)
I believe the next problem, which also showed up in the first few years of
ownership, was high input SWR on 15. That can sometimes be fixed by
adjusting the calibration of the band switch via a firmware patch. Alpha
talked me through that one and the problem went away. Evidently, the factory
calibration hadn't been correct or something got jostled during shipment to
change it slightly.
The next problem came up maybe 5 years after buying the amp. An op amp in
the Ip meter circuit was slowly going bad. It started with the lowest LED
staying lit all the time. Eventually more LEDs at the low end stayed on. It
was a completely benign problem, but annoying nonetheless, and it looked
like eventually the Ip meter could become useless. I traced the problem to
the op amp in question and replaced it. It was an obsolete part and Alpha
was the only place I could get a replacement. They had bought a bunch of the
ICs before they became unavailable, but I wondered what would happen when,
one day, they run out. This is likely a problem for lots of 87A parts, which
is why we 87A owners need to be concerned. I'm sure RF Concepts has parts
and boards on hand, but it's not an infinite supply. I wouldn't be surprised
if they're mining parts from trade-ins that are so bad they can't be
refurbished for "certified pre-owned" resale.
The next problem was a bad RX PIN diode, probably a couple of years later. I
don't remember now what the symptom was -- I believe it was a fault that had
something to do with PIN diode bias. It was a common enough problem that the
factory was able to point me to the faulty PIN diode, which I replaced. The
RX PIN diodes are relatively inexpensive. Some are easy to replace without
removing the T/R board, but others require removal of the board, which is
doable but not trivial. The one I had to replace could be done from the top
of the board. I've not had any further problems with the RX PIN diodes.
The next problem occurred somewhere between the 5 and 10 year mark. It had
nothing to do with the amp, but revealed flaws in the fault detection logic.
It's a long story I've posted about before, but the summary is that I was
getting cathode current faults on one tube and replaced it. Over a long
period, I got random faults of all sorts, and the faults increased in
frequency. Most of the faults pointed to the power supplies. I tore the amp
apart looking for bad components, even to the point of removing parts from
both power supply boards, but couldn't find the source of the problem.
Eventually, the amp started blowing fuses. That's when I realized the
problem was being caused by arcs from a gassy tube. I had kept the old tube,
and when I replaced the new tube with it, the amp worked perfectly (and
still does, with the original 1995 tubes.) Turns out there was nothing wrong
with the old tube -- I must have seated it improperly after removing it for
some other reason. Alpha helped me to get a warranty replacement on the new
tube from Eimac. It sits on the shelf as a spare.
What this exercise taught me is that the firmware fault detection can be
unreliable in certain cases, such as an arc causing the power supply to be
pushed beyond limits. The firmware fault detection code isn't
interrupt-driven -- it operates in a loop, checking each sensor in turn.
That means if multiple sensors report bad values, it's not guaranteed that
the earliest-occurring fault will be reported. The classic case of this is
when the transceiver output spikes above the drive level limit. You don't
get an "Excessive input power" fault, you get a "Gain mismatch" fault.
That's because the latter fault is checked before the former fault. The
reported fault is accurate, but it doesn't tell you what caused the problem.
Dick Erhorn himself confirmed this firmware flaw to me decades ago -- he had
noticed it too and had reported it to the factory. It's probably impossible
to fix. My sense is that no one on the planet knows how all of the firmware
code works and no one is cable of rewriting it. More important, I'm sure
there's no code space left to add interrupt handling, and it would probably
require quite a bit more hardware any substantial mods to both power supply
boards.
BTW, as I did the troubleshooting of the power supply boards, Alpha talked
me through doing a couple of simple upgrades to the LVPS. I think they
consisted of adding a jumper wire and maybe replacing an IC. I don't recall
the benefit. I wish RF Concepts would publish a complete list of 87A factory
upgrades, like Elecraft does for the K3.
One of the more interesting problems I had was a bad contact on the wafer of
the input band switch. I believe the contact was marginal from the day I
bought the amp, and had been giving me random but infrequent fault 17s (gain
mismatch, once again.) The frequency of this fault was very low to begin
with -- maybe once every few months. Eventually I noted that it happened
only on 10m, partly explaining the lack of frequency. But the frequency of
the fault increased very slowly over the years. I think about 5 years ago or
so, the amp started faulting every time I transmitted on 10m. Eventually I
was able to trace it to the input band switch by patching the firmware to
move the band switch to the 12m position when 10m was selected. The input
network is either identical or very close on those two bands, so I figured
it should transmit OK on 10m from the 12m band switch position. Turned out I
was right, and this pretty-much proved the input band switch had a bad
contact. The switch was a little tricky to replace, but not too bad. It cost
$75 for a new switch. Problem solved.
I have had one major issue with the 87A that cost a fair amount of money to
fix, but the cost was only indirectly related to the design. Back in 1997,
shortly after putting up a new 110' tower, I had a really bad lightning
surge. Long story I won't repeat here, but to took out many devices in the
shack and home, with a repair/replacement bill of $10,000 (covered by
insurance, thankfully.) The surge traveled through the RS-232 port board in
my computer to the serial port on the 87A.
Amazingly, the amp worked OK after the surge, but I couldn't communicate
with it. I replaced as many of the support chips around the CPU as I could,
including the MAX-232 serial interface chip, but eventually determined that
the CPU was not communicating. The surge had burned out the serial I/O port
on the CPU. I later found out that the CPU used in the 87A has fused I/O
ports such that the chip can survive even if one of the internal fuses
blows. Most CPUs would be fried by a surge like that. In fact, that's what
happened to my Acom 2000A CPU, which had to be replaced for the amp to work
at all.
However, as you might expect, I could not communicate with the 87A. Many 87A
owners don't even use serial communications with the amp, and might not have
bothered fixing it, but my amp is located remotely in the basement, and I
need to communicate with it. Besides, serial communications is necessary for
getting the fault history, patching firmware, etc. I called the factory, and
found out that replacing an 87A CPU isn't simple. Oh, it's easy enough to
remove and replace the chip. The problem is that the chip contains
amp-specific calibration parameters that are set by the factory after
manufacturing. This isn't limited to the band switch positioning, which I
could have programmed myself. It also includes voltage calibration
parameters for all of the sensors, and I didn't have the right equipment or
knowledge to do the setup. It really has to be done by a factory technician.
I'm not sure, but Brad may be the only person on the planet who knows how to
do it.
One possibility in these cases is to dump the current parameters from the
existing CPU if it's still alive. Mine was, but since the RS-232 port was
fried, the parameters couldn't be dumped! There was one hope: back when my
amp was made, they recorded the calibration parameters. With that data file
in hand, I could have installed a new chip and uploaded the parameters.
Unfortunately, the parameters could not be found at the factory. I believe
that through all the changing of ownership of the company over the years,
the computer with calibration data from some 87As was wiped or lost. Bottom
line: the CPU had to be replaced and the amp had to be recalibrated. This is
why I said the high cost of solving the problem was indirectly related to
the amp design. Note that the Acom 2000a suffers from the same problem.
The charge to fix the 87A was $315, which included a new CPU and the labor
to recalibrate. I think the price might have included a new CPU board, to be
installed if necessary. I've forgotten if it was necessary. The real killer
was shipping the amp to the factory. I think the estimate was $300 each way,
or $600. So I was looking at over $1,000 to repair the amp. But by great
good fortune, the president of my contesting club, K1RX of YCCC, had
arranged with Alpha Radio Products (the company at that time), to put on an
"Alpha Day" in New England. Alpha flew Brad to New Hampshire and he setup
shop at a club member's house, repairing Alpha amps for our club members. I
got on the list and was able to drive my amp 1.5 hours to southern NH for
Brad to fix. It was not a trivial exercise, and I concluded that the $315
charge was fair for the parts and labor involved. Problem solved.
Other than the lightning strike, the most serious and costly problem I've
had with the 87A occurred a couple of years ago. It was a burned out TX PIN
diode. These PIN diodes are very expensive -- $125 each -- and you can only
get them in small quantity from RF Concepts. I don't know if they are
actually manufactured anymore. The problem began with the amp faulting while
transmitting during a contest. I think it was a fault 8 or 9, maybe a 10. It
faulted the same way a couple of times, when transmitting, then began to
fault immediately after warm-up, without transmitting. The faults (again, 8
and 9, and I think 10) pointed to either a problem with the PIN diode bias
supply or the TX PIN diodes themselves. The factory gave me troubleshooting
info and I was able to trace the fault to an open TX PIN diode. I replaced
it, and that stopped the faults during warm-up, and I was able to transmit
OK, but as soon as I increased power to 1500W, I got a fault (I think it was
9, but it might have been 8.) I had noticed that before replacement the
resistance and voltage drop of the new PIN diode and the remaining old PIN
diode were quite different (lifting a contact of the old PIN diode before
measuring, of course). Assuming that the old PIN diode was damaged, possibly
due to whatever had caused the other PIN diode to blow (lightning?), I
ordered another PIN diode and replaced the old one. Unfortunately, that
didn't help -- the amp still faulted at 1500W.
While on the phone with Glenn talking about the problem, Brad happened to
walk by and overheard Glenn's end of the conversation. He said, "Have him
check L1 on the T/R board. It may be arcing." Sure enough, when I checked, I
saw soot on a screw near L1. When I lifted L1, I could see that the rubber
insulation pad underneath it had decomposed over the years, and there was a
sizable divot in the PCB where arcs from L1 had pierced the PCB. These arcs
had blown one of the TX PIN diodes, and might have damaged the other one. I
removed L1, rewound it, put a new pad and other insulation under it, and
reinstalled it. Haven't seen any problems since. Evidently, the failure of
insulation under L1 is a known problem, but I don't know whether it's
limited to a specific set of manufacturing years or production runs. The
problem cost me $250, though half of that wasn't actually necessary. I do
have a spare TX PIN diode on hand, however.
In the course of troubleshooting this problem, Glenn suggested that I
replace the zener diode chain on the HVPS. That's a cheap and easy fix, and
among other things it is believed to correct a long-standing inter-band
interference problem (e.g., 10m causing interference on 20m.)
I'm pretty sure that's the sum-total of problems I've had with my 87A. It
seems like a lot, but we're talking about more than 17 years of ownership.
At least I've never had to ship the amp back to the factory! As I said, I
love my 87A. I have an Acom 2000A, which I really like, too, but I use the
87A for all my main transmitting activities, including the run radio during
contests. I figure the fewer cycles on the 2000A vacuum relay, the better
(I've had to replace it a couple of times, though not because of overuse.)
That said, I worry about the long-term future of my 87A. It seems almost
inevitable that it will hit the wall someday and parts will not be available
to fix it -- or RF Concepts will drop support. I think it's amazing that the
company still supports the amp at all -- mine is nearly 20 years old. I
can't think of any other company that would continue to support products
that are obsolete for that long. But unlike other obsolete ham gear, the
complexity of the 87A will eventually make finding replacement parts
impossible. We're seeing that with transceivers like the Kenwood TS-940,
etc.
Resale is becoming a difficult issue, too. Unless I sell it to some
unsuspecting ham who doesn't know the outlook (which seems unethical to me),
I can't get fair value for it on the open market. My only other option is to
trade it in on a 9500, which I don't want. I've heard rumors of problems
with the amp. And even if those have been taken care of by RF Concepts, the
9500 is expensive, runs and expensive tube, and has a lot of features I
don't need. If I could sell the 87A for the same price as a trade-in, I
would probably get another 2000A -- it's cheaper, the Russian tubes are
cheaper and still available, it's still in production and it's still
supported (very well, I might add.)
But in any case, if I have to give up the 87A someday, I will miss those
lovely PIN diodes. Maybe by that time solid state will be a truly viable
option.
73, Dick WC1M
> -----Original Message-----
> From: Bob Henderson [mailto:bob.5b4agn@gmail.com]
> Sent: Tuesday, July 10, 2012 5:24 AM
> To: amps@contesting.com
> Subject: Re: [Amps] Alpha 87A owner
>
> On the other hand, I have owned and abused my 87A for more than 18 years
> now. Aside from a couple of small problems easily corrected in the
> shack, it has been a rock solid investment. When I have called upon
> Alpha for support they have always been A1.
>
> I have been wondering why folks would trade an 87A for a 9500.
>
> Bob, 5B4AGN
_______________________________________________
Amps mailing list
Amps@contesting.com
http://lists.contesting.com/mailman/listinfo/amps
|