Network Analyzer - Repair

jesse-farrellJesse Farrell wrote 06/15/2023 at 22:34 • 4 min read • Like

I’ve got an AP instruments network analyzer that stopped analyzing. The system is comprised of a USB interface adapter and a main unit. The only symptom is the absence of a power LED on the main unit. 

Instead of checking continuity of the cable, (because who am I kidding that’s no fun) I decided to open up the interface adapter.

Luckily, they gave us nice test points we can check right off the bat. All the low voltage rails (12V, 5V, 3V3) look good… but the one’s near the main connector are all zero volts. Maybe it waits for a PG signal before enabling 18V, +7V, -8V.

Next, I did some poking around the board. Seems like one of the main switchers is a UC3845N. VREF and VCC look good, but the switching node "OUTPUT" is held at GND. I’ll dig a bit deeper to see what the other pins are at.

After poking the other pins, it looks like COMP is being used to disable the device. Its being held low (ish) which is stopping the device from starting. Here’s a blurb from the datasheet. This COMP pin is being held low by a XILINX chip…. ☹. That complicates things.

 I’m not sure what this chip needs before it will enable the main switcher. Checking the cable, I confirmed that the 12V, 5V, 3V3 rails are NOT sent to the main unit. This means the XILINX chip isn’t waiting for a handshake from anything on that unit (since it won’t be powered).

The suspect chip is a CPLD (XILINX XC9572XL). This isn’t something I can debug.... er atleast not easily.


My plan is to remove the XILINX's ability to disable the switcher. Luckily there's a diode leading to COMP, so I can simply remove it (D7) and the switcher should work... what's causing the XILINX chip to assert this disable pin though?

And voila, we have voltages. Nothings too accurate, but I was suspicious it wouldn’t be. Shown below is the +18V rail.

Now I’ll try plugging it in, but first I need the software to try and talk to it.

All seems happy. I can’t test the full feature set (I need more BNC cables), but I can at least see that the system is talking ect. I’ll do a final test tomorrow in the lab. (Tested and looks good).