HP 8082A "quick" restoration turns into IC reproduction project (Patreon)

While I was waiting for the replacement ICs that died in a sparky power supply incident last week, I decided to turn my attention to a recently acquired HP 8082A fast pulse generator. I thought it was going to be another quick one-afternoon power supply job. Well, this "does not power up" unit turned out to power on just fine, but has many faults in it. After finding two faulty transistors, I came upon a much tougher fault: one of the special high speed ICs had failed. And it's an HP only circuit, made just for this pulser - basically unobtainium. It's U2 on the bottom right of the schematic (pic 2). The ND and complementary DEL outputs are supposed to produce very sharp 2ns wide pulses, with rise times of less than 1 ns, which are then processed by more specialized ICs in the instrument to add the correct width, leading and trailing edges. I get ND pulses alright on pin 8, but the complementary DEL output on pin 9 is dead.

Luckily a clever chap, Matt D'Asaro, which must be Ken's twin separated at birth, had a similar problem on an earlier generation of this pulse gen, took a die shot of his 1820-0285 IC, and proceeded to reverse engineer it (pics 3-5). Not the same as mine, but clearly a precursor of mine. It does a similar function to the bottom of my failed 5081-3011 IC. Matt proceeded to make a reproduction of his IC with surface mount components and fast 1GHz transistors, and it worked. That encouraged me to follow the same approach.

But all I really need is to make an inverter gate with the weird HP levels and 50 Ohms impedance match, so I can invert ND and get DEL again. You can see how such a gate is designed in the middle of Matt's schematics, using ECL-like circuitry (but not regular ECL levels). So I started from Matt's schematic, extracted the gate, changed a few resistors to get the right levels, and LTspice tells me it should work, as you can see from the screen shot of the simulation (pic 6).

I fired up KiCAD, made a little PCB with RF connectors so I can insert the reproduction gate cleanly in the instrument, and sent it off for fab at PCBWay (pics 7-9). Which is of course more trouble than it's worth, I should just junk the 8082 I got and buy another one on eBay. But what's the fun and learning in that.