Bevor several years I bought two of the well-known faked (?) Agilent 82357B USB to GPIB Interfaces. It was clearly a kind of fake or black market selling. The serial number at the box did not fit to the actual serial number of the unit, the functional test certificate was missing a serial number and there was also no signature, therefore it is a completely useless sheet of paper.
But these units were much cheaper than the genuine Agilent 82357Bs and good enough for my home lab.
Both interfaces were in full working condition when they arrived from China.
But after 2 month the trouble began.
The first interface suddenly failed, the Keysight connection expert software recognized an active unit at the bus, but could not communicate with it. I only could see some instrument with its address, but it did not answer to any command, not even to "IDN?". It seems that the data transmission was broken somehow. I replaced the 82357B with the second one, everything was back working. I contacted the seller for a refund, but he wanted that I send back the unit to China. Too much effort for me, I did not want to do that. So, I agreed with the seller that I buy two further interfaces, what I wanted to do anyhow and I got a third for free as a replacement.
After another 6 month the second interface failed in a similar way. Every connected unit gave a GPIB error message while trying to connect and communicate with the device. Now It looked like a problem with the control lines of the GPIB bus.
I replaced it with a third interface that had arrived meanwhile. This one worked in the last two years and did not shown any problems at all until now. The other two 82357B are still in their unopened boxes.
There are a lot of discussions about these interfaces and I have seen several messages about fake 82357Bs that has stopped working after some month in use, like mine.
The two interfaces were laying in the shelf together with other unfinished repairs for about two years.
Now, whilst the corona crisis, I have enough time to complete some old unfinished repairs. I have repaired several instruments now, that had laying around in a shelf since years. And I found these two broken interfaces again somewhere in a box with electronic junk. Scrapping or give it a try to fix it?
I decided that I have enough time, so I gave it a try to repair!
The interface consists mainly of a Cypress microcontroller for the USB handling, an Agilent branded GPIB controller, driver chips for the GPIB data and control signals and two sps voltage regulators for the internal 5V and 3.3V supplies. There is also a PLD or FPGA from XILINX with some glue logic between the microcontroller and the GPIB controller, that's all.
The microcontroller does not contain any firmware, only a bootloader. At every start of the Keysight connection expert software, this software is transferring the firmware via USB into the microcontroller. After the transfer is competed the Error LED at the 82357b distinguishes and the green READY led stays on. That was the case for both of the broken interfaces, so it seems that the microcontroller was ok.
That was the first important cognition, the microcontroller is working.
Second was testing the internal voltages, both (5V and 3.3V) were ok.
How to proceed now?
My first assuming was that there is a problem at the GPIB bus side and that at least one control or data signal is missing. It also could be a problem with the GPIB controller or the Xilinx chip, but that was not very likely.
Next step was to check the data and control signals at the GPIB connector with a scope or a logic analyzer. But this means much effort, so I first tried a much easier way to test the GPIB I/O part. I checked every data and control pin with a component analyzer. You get a characteristic signature for every port, like at the picture.

I started from the upper left GPIB connector pin and checked every control/data line against GND. Actually I did not really expect to have success with this simple kind of testing, but to my surprise I could not see a signature at pin nr. 4, only a flat line. This could not be, the GPIB contact was connected to a pin of one of the driver ICs. The reason was an interrupted trace between the IC and the connector. But even with a microscope I was not able to see any interruption. Anyway, I patched the trace with a piece of thin copper wire, connected the Interface to my lab computer and a hp power supply and the interface was back working! That was much easier than expected.
I did the same with the second failed interface, same problem, but here the connection between GPIB pin Nr. 5 and the other driver IC was interrupted. And what do you see when you have a closer look to the picture: Yes, the connection to Nr. 4 was already patched in the factory! Whoever made this board were aware of this issue!
I assume these interfaces were rejected from Agilent/Keysight and were intended for scrapping, or they are fakes with poor circuit board quality.
I don’t know, but now I know how to fix these little bastards in future, if they fail again.