[Test-Equipment] Viable 8640B Final Amplifier Replacement Developed

[Steve Harrison]

At 03:30 PM 2/26/2007 +0100, Jose V. Gavila wrote:
>Hello!
>
>There is some hope :-)
>
>http://www.shelbrook.com/~ve7ca/Tst864.htm
>
>Best regards,
>
>JOSE

Just got done replacing that final amplifier. I would have tried using a
MAV-11 except I discovered I don't have any more  ;o((((((  I don't have
the URL now but ran across another 8640 final amplifier replacement using a
2N5109, a device I've had lots of experience with. But I don't seem to have
any more 2N5109s in stock, either  ;o((((((  Lots of 2N3866s but those roll
off shortly above 200 MHz or so (although they'll put out up to 3 watts in
class C and over a half watt in class A).

Looking through what I DID have, however, revealed a MWA-130 (and more
importantly, another one to "save" for a spare  ;o). A yahooey search
eventually pulled up a data sheet and lo and behold, the thing's rated at
14 dB gain and a 1 dB compression point of +18 dBm; and best yet, it's
speced for DC through 1 GHz. This is just about what's really needed for
the final amp; it'd be nice to have several more dB gain as well as output
power rating of, say, +20 dBm; but beggars can't be choosers, eh? The 8640
original final amp is rated to have 16 dB gain from 400 kHz through 550 MHz
at something like around +18 dBm output power at the 1-dB compression point.

The MWA-130 runs on only 5.5 volts, however, requiring a pretty large
dropping resistor from the +44.6 Vcc supply in the 8640. At about 65 mils,
this required 600 ohms that would dissipate a bit over 2 watts.
Fortunately, I found a pair of 300 ohm, 1 watt metal film resistors. Three
mix number 43 beads over the "hot" side of one resistor sufficed to
decouple the output of the amp from the power supply. I was able pry off
the cap of the original 8640 final amp device after some application of
heat from the soldering iron. The wafer also came unsoldered; it must have
been held on with low-temperature Indium solder. I cut a 1" diameter piece
of dual-sided 0.028" PTFE-glass board to fit in the center of the original
final amp device's mount, then used an Exacto knife to cut traces for the
MWA-130 implementation. Due to the required RF output detector for the AGC
amplifier, this circuit consumed more real estate than you'd expect,
leaving little room for the -130 amplifier. Initially, I simply soldered
one edge of the -130's case to the board ground plane then dead-bugged the
input and output leads to the board. But that left the -130 sticking up way
too high to reinstall the fancy cover for the amplifier module, which is
NECESSARY in order to prevent excessive RF leakage for weak signal
measurements (the 8640 is already a bit marginal in leakage for sensitive,
low-noise receiver sensitivity measurement). After I'd confirmed the -130
worked, then i chopped a hole in the board (which, by now, was solidly
soldered onto the gold-plated final amp module's mounting package). Laying
the -130 upside down in the hole, top of the case against the original
mounting flange surface, left JUST ENOUGH clearance to be able to use thin
copper tape to make connections between the -130's input/output and the
board. The two 1 watt power resistors had to be mounted off the flanged
package; I laid them against the back wall of the amplifier enclosure in
the hope that they would radiate their heat onto the finned amplifier
enclosure rather than heating up the insides of the whole amplifier
enclosure, turning it into an oven. I really need to put a thermister
inside there to check just how hot the insides really gets after an hour or
two. If over, say, 120 degrees, I may have to move one of the dropping
resistors elsewhere outside of the enclosure, as was done by one of the
guys who first tried the MAV-11.

The end result: good output power to at least +10 dBm with harmonics still
limited by the internal harmonic filter, above about 170 MHz. Below that
point, harmonics drop further to -40 dBc and output power can be raised to
at least +13 dBm down to about 68 MHz. Below there, output power can be
raised to +18 dBm. My spectrum analyzer has a blown 40 dB input attenuator
so I was limited in my harmonic level measurement since I was overdriving
the spectrum analyzer at power levels above about +10 dBm.

The MWA-130 seems like it'll do the job fine at the +10 dBm level and even
higher below about 65 MHz or so. It doesn't appear to be gain-limited. I
did notice that when output attenuation is zero, the level meter will
wriggle a bit with various output loads, indicating the -130 IS somewhat
sensitive to output loads. At +10 dBm output, this isn't a problem since
there's 10 dB output attenuation.

I've got several other devices I'd like to try out, including a cupla
Motorola CATV modules that have a bit more gain. While they're designed for
75 ohm systems and have a low frequency limit of 5 or 40 MHz depending upon
the device, they may still prove useful.

Another great candidate is a cascaded pair of MSA-0520 amplifiers from
Avantek or Amplifonix. I've got a few of those and may try them, too.
They're capable of much more output power, up to 200 mW, as well as being
flat from a few MHz all the way past 2 GHz. Their main limitation is their
mounting method: they are chips on a small ceramic substrate that must be
pressed against a heat sink for cooling. Also, each -0520 device only has
around 9 dB gain, IIRC, so a cascaded pair will be necessary. However, this
isn't necessarily a bad thing because they also need a bit of negative
feedback to flatten the low end response. This will also, of course, reduce
the overall gain across the band. I've played with those back in the early
'90s, so do rember a bit about their limitations.

Steve, K0XP