Great news - I was able to hack my 3400 128MB module to work on the Kanga! 
It surprised me that the 3400 128MB module didn't work - considering that the 32MB Apple module worked fine on both machines, and both modules had 60ns rated chips. Apart from the Kanga having a G3 inside, the only major difference between the two models is that the 3400 has a 40Mhz bus, and the Kanga 50Mhz. When the 128MB module was installed in the Kanga, it would either not detect at all, or every now and then Mac OS would report a memory error at startup; then it would freeze within a couple of minutes.
So it got me thinking that perhaps the 128MB module was running at a slightly faster latency to work at optimal bandwidth in the 40Mhz bus 3400, and the latency needed to be slightly reduced for compatibility @ 50Mhz bus in the Kanga. Looking at the 128MB part - which is very generic - two solder pads R8 and R9 (with zero K resistors - or shorts) suggested to me that something could be adjusted. I didn't have much to lose, considering it was a bit of a downgrade to keep the 64MB Kanga over the 144MB 3400 - so I switched both resistors to their other position on the PCB (well, I switched one by one but the same issues occured - needed to relocate both in the end).
And, success! 160MB fully detected and rock solid on the Kanga. I can now attempt to install OS X onto this machine - which will be a sight to see.
Pictures below detail the location of resistors R8 and R9 (this is the "after" picture, ie. for Kanga compatibility), and a Mac OS screenshot.
Hopefully others can use this information to max out their Kanga with more common 3400 RAM!
JB
It surprised me that the 3400 128MB module didn't work - considering that the 32MB Apple module worked fine on both machines, and both modules had 60ns rated chips. Apart from the Kanga having a G3 inside, the only major difference between the two models is that the 3400 has a 40Mhz bus, and the Kanga 50Mhz. When the 128MB module was installed in the Kanga, it would either not detect at all, or every now and then Mac OS would report a memory error at startup; then it would freeze within a couple of minutes.
So it got me thinking that perhaps the 128MB module was running at a slightly faster latency to work at optimal bandwidth in the 40Mhz bus 3400, and the latency needed to be slightly reduced for compatibility @ 50Mhz bus in the Kanga. Looking at the 128MB part - which is very generic - two solder pads R8 and R9 (with zero K resistors - or shorts) suggested to me that something could be adjusted. I didn't have much to lose, considering it was a bit of a downgrade to keep the 64MB Kanga over the 144MB 3400 - so I switched both resistors to their other position on the PCB (well, I switched one by one but the same issues occured - needed to relocate both in the end).
And, success! 160MB fully detected and rock solid on the Kanga. I can now attempt to install OS X onto this machine - which will be a sight to see.
Pictures below detail the location of resistors R8 and R9 (this is the "after" picture, ie. for Kanga compatibility), and a Mac OS screenshot.
Hopefully others can use this information to max out their Kanga with more common 3400 RAM!
JB