Some Wire Bonding...

HP spectrum analyzer frontend mixers.

Spectrum analysers Mixers are the number one on the top list of user destroyed devices (by far).
On these HP mixers very few have just the diode destroyed, the vast majority has both diode and IF bipolar transistor destroyed.
Here you see one mixer where a new transistor was applied and bonded to a new diode junction.

An YIG oscillator from a TEK 492/4/5/6 spectrum analyzer.

This oscillator had a defective output amplifier and produced less than 0dBm output power with a lot of ripple.
Here you see the output GaAs FET replaced by a more recent Harris HMF33000 (the original was a non cataloged (AFAIK) Avantek type inspired on the historical TC300 from HP).
Die attachement was eutetic and all wirebonding was with my beloved wedge bonder.
On the right you see a close-up of the replaced FET.

When the big ones goofed, YIG from a TEK496P.

This oscillator had a defective output amplifier and was repaired as on the example above.
However I noticed that the output FET had excessive drain supply and that was the presumable reason for its failure. Looking for the cause I discovered the voltage regulator (a 7808 but in chip) was missing his ground wire bond connection. It was missing but not because it evaporated or got lost, it is beyond any doubt that this connection was never there as both the chip pad and the gold substrate have no signs of any wire ever being attached. Only visible mark is a one time probe needle contact the same as on the other IC pads. Needless to say that this voltage regulator chip has no gnd on the back side, furthermore all Avantek YIGs I opened so far had the 3 connections on the regulator chip as required.
Clearly Avantek goofed on this one... at least quality control did. I just wonder how the output FET survived that long with about 11.8V on drain instead of the 6V that should have been present there (the FET has an Absolute Vds maximum of 8V trusting the datasheet).

YIG oscillators from HP8563's/HP70000's spectrum analyzers and alike.

I have seen many of these YIG oscillators fail. They exist in several different HP analyzers as many models employ the same or very similar oscillator, even if they have a different HP ref number all end up to have exactely the same oscillator hybrid circuit inside (despite some variations on the bias/alc board).
All of them, without exception, have a failed oscillator transistor being all the rest perfectely ok. The transistors look however good DC wise but strangely the chip seems to have developed some kind of moist on the bonding pad. It is indeed a puzzling semiconductor failure.
Here you see a new transistor applied.
Last picture on the right is a close-up of oscillator section while the substrate was siting on the hot work-holder of the bonding machine, picture taken right after it has been wire-bonded.

Dual YIG oscillator (HP) wide band coverage from 2 to 20GHz.

This is a more modern design, already incorporating MMIC amplifiers on both oscillator sections.
The low band uses a bipolar transistor to cover 2 to 11GHz which is an extreemely wide coverage for a bipolar oscillator, followed by a wideband TWA (traveling wave MMIC amplifier). The high band uses a GaAs FET that in this case is the input FET of a standard MMIC amplifier ...interesting design.
This one had a problem on the low band oscillator section. A small deterioration in YIG sphere properties resulted in slightly less Q on the lower frequencies that was enough to stop the oscillation at bottom frequency range. Replacing the YIG sphere followed by extensive retuning brought it back to life over the full operating range.

Broad band amplifier 2-22GHz (26.5GHz) from a HP sweep generator.

This job turned out to be quite interesting as the original TWA (traveling wave amplifier) chip could not be sourced at a reasonable price. However replacing it with a Celeritek or Hittite chip proved to be possible after considering the similarities in chip performance and supply voltage and bias. A Celeritek chip was applied and the results were excellent (as good as original or even better). Note the Celeritek chip is slightly smaller than the HP original and some care was needed to make a connection to the chip without adding to much deviation from the 50OHm impedance potentially critical issue at the top of the frequency range (26.5GHz). This was achieved by using a ribbon connection to the chip that was tapered to match the chip bonding pad.

Tek. 492/6 spectrum analyzer frontend mixers.

These mixers had a rework using a gap-welder to attach the beam lead diodes and to attach the ribbons that connect the quartz (or alumina, on recent models) substrate to the soft-substrate that connects to the main substrate on the mixer body.
Here you see several stages of the workout.

A nice MMIC amplifier (17dB / 6-18GHz / p1 18dBm).

This amplifier uses an HMMC5618 MMIC from HP and both assembly and outside case where homemade. Substrates, plain 50 ohm lines, were recovered from old and incomplete hybrid assemblies I obtained surplus.

Any comments: Luis Cupido