Hi!
An update, I ordered some more RAM to test with.
Two 16MB and two 1MB.
Same issue with all sticks, card only detects 512kb.

Next step I guess is to try to resolder the connector and see if it fixes the problem.

If anyone else got any ideas I'm happy to hear them!

I also read that you can flash this card to a AWE64, if there any benefits to that except diabeling the IDE?
Maybe thats worth a shot if resoldering the connector doesnt work.

filurkatten wrote on 2022-03-23, 12:48:

Same issue with all sticks, card only detects 512kb.

So this means that 1M sticks are detected as 256K sticks. The only difference between these sticks is that 256K sticks use address bits A0-A8, whereas 1M sticks use A0-A9. So if RAM sound fonts are working fine up to 512KB, you should inspect A9 with extreme prejudice. A9 is at pin 18.

So, I have used the multimeter to trace the ram socket pins and they all seem to go where they should.
Then I decided to resolder the connector but still the same issue, only detecting 512kb of ram.
After that I thought, "well, Ill just give the reflasji g to a AWE64 a shot I dont need the IDE anyway".
It flashed and is now recognized as a AWE64 but still only 512kb is showing.
I have tried combination with the RAM I have at hand, 4x 16MB and 2x 1MB.

Troubleshooting adresslines is kinda new to me, and I dont have that fancy eqipment. My oscilloscop is 40Mhz so I dont think that will help me here.
What chip on the card is responsible for the adresslines? Maybe the pins need some fresh solder.
Any pointers is appriciated!

mkarcher:
Thanks for the reply!
I will have a look tomorrow and follow that trace to see if I find anything suspicious.

Hi again!
A9 seemes to be a good lead!
I tested with my multimeter in continuity mode the adresslines and they all lead up to the resistor arrays.
From there they go down to the buffer chips, all adresses excwpt A9.
The trace looks good but part of it is under the resistor arrays.
I will first try to add some fresh solder to the pin that should be connected to the buffer and see if that helps.

Hopefully I got some time to do that this weeked.
I'll keep you updated!

Adding fresh solder didnt fix the trace going from the resistor array to the buffer so I desoldered the array and checked the trace and it was a really small break between the hole and the trace.
So I ran a wire from the hole to the buffer, resoldered the array and that fixed the problem!

Thanks mkarcher for pointing me in the right direction. 😀

filurkatten wrote on 2022-03-25, 23:05:

Thanks mkarcher for pointing me in the right direction. 😀

You're welcome!

Main takeaway for you: Always describe your issues with as much detail as you did here. The information that 2* bog standard 1MB modules were recognized as 512KB instead of 2MB, but passed the test was the key point that pointed to this specific address line. As the CT3670 doesn't have onboard RAM (so thanks for explicitly mentioning that number, too), I could be confident that it's not a problem with the on-board/SIMM jumper you already took a look at (which would have been my first idea to "SB (AWE)32 only detecting 512KB of RAM"), but I was sure that part of the SIMM RAM was actually tested.