Target voltage detected Target Device ID (0x77a0000) is an Invalid Device ID.

After using a PICKit3 successfully for a month, with two different installations of MPLABX, all of a sudden we're getting this message:


Target voltage detected
Target Device ID (0x77a0000) is an Invalid Device ID. Please check your connections to the Target Device.
 
We have tried MPLABX 4.15, 4.20 and 5.0 and two different PCBs containing a PIC24HJ256GP610A.  This reported Target Device ID appears to be one bit off from the correct ID, which is 0x77b0000.  
 
There are many examples here of people who are getting a zero ID, but not something as odd as this.  We've spent two days checking voltages and signals and uninstalling/reinstalling various versions of MPLABX, but we keep getting the same error message.  MPLABX will go ahead and erase the firmware, but then fails on trying to program new firmware.
 
Any ideas?

Our client is using the same board without any difficulties.  We have tried 3 different PCs running MPLABX, two different PICKit3's, and new cables.  
 

 
 
 

Thanks, qhb.  We didn't do the design, and it hasn't changed in two months . . . Anyway, hard to understand how we get a target ID off by one bit every single time in over 50 attempts on 3 different systems, 3 flavors of MPLABX and 2 different PICKit3's.  We are in Colorado and our client without problems is in CA, so we were attributing the problem to altitude.  In any event, I see what you mean:  Section 2.5.5.

PICkit3 has internal 4.7K pull-downs. Combined with your pull-ups, they create voltage divider. When the line is released by both PIC and PICkit3, the voltage is stuck in the middle. Therefore, don't expect reliable communications. If you have problems with your particular board, it's a good idea to remove the pull-ups.
 
When programming, PICkit3 sends lots of commands. It actively drives the PGD line all this time. Then it releases the PGD line (which goes to VDD/3 because of your pull-ups) and waits for the response from the PIC. The bit you're losing is the first to come out of the PIC (id is sent LSB first). It comes out as '0' instead of '1'. Could this be caused by the previous VDD/3 period? Unlikely, but given the "right" combination of voltage levels and sensitivity of receiving gates it certainly may happen.
 
Of course, there could be other reasons, such as bad contacts, bad soldering, or damaged chip. But it's a good idea to start troubleshooting from the obvious mishaps.