Install Second SDRAM without Digital IO board

[Supersonic]

Hello everyone! I have recently replaced the analog IO board with a second SDRAM bank to test PSX core, I have not yet put the digital IO board because I do not have it, after this operation I have noticed instability in the cores, I experienced sudden resets and crashes.
Looking at the schematics of the digital io board there is an IO expander that communicates via I2C with FPGA, since I2C pull-up resistors are on the board itself, could it be that this instability is due to the fact that I2C-SDA signal is floating and communications gets corrupted reporting erroneous states?

Anyone else experienced this kind of problems?

Is there a way to eventually deactivate the communication with the IO expander? Or maybe configure SDA pin with internal pull-up?

I own this De10 nano board configured as Mister FPGA since 2018 with the same power supply and never had any problem.. I even attached the oscilloscope on the 5V rail to check for any voltage problems but everything seems OK, it never triggered for voltages drop..
SDRAM bank 1 is OK, I tested it with Jotegos test cores and the official SDRAM tester.

Thank you.

(Sorry for my English)

[aberu]

1. The SDRAM boards need to be facing opposite to one another with the "THIS SIDE FACES OUT" facing outside, on both of them.

2. You may need to switch the large dip switch for dual SDRAM. - https://misterfpga.co.uk/wp-content/upl ... sition.png

[caad]

Also, none of the IO boards are required for normal operation.

[Supersonic]

Hi aberu and caad, thanks for your answers; the SDRAM banks are inserted correctly and SW3 is in ON position, the problem arise in this condition. If i turn OFF SW3 everything works fine, thats why I tought it's something related to I2C communication with IO expander.
SDRAMs are also OK I tested both, one a time in "slot 1" for 2 hours each.

Haven't watched source code yet, but I think that if SW3 is ON the framework will try to communicate with IO expander and if it's not present and pull-ups are missing maybe problems could arise.. just my speculation..

Hopefully today I should receive PCBs for my DigIO board and re-test everything..

[Supersonic]

Resolved. The problem was on the 3.3V rail, it was, in certain conditions, overloaded. My two SDRAM 128MB banks were the version directly powered from 3.3V; i modified them adding a voltage regulator as in sdram xds v 2.9.

[crispycritter911]

Would you have a picture of your modification? I just built a second RAM Stick and am curious as to how you did it. I haven't tried mine yet. I am running a 2.4 and a 2.5 PCB's with 128MB's each. I am currently bench testing the newer module.

[Supersonic]

Hi, I'm sorry, I don't have it anymore, I used it as a temporary solution.

[lomdar67]

Sorry to pick this up again.

I have a V2.5 SDRAM stick and wanted to order a second one. Is this voltage problem a general problem with V2.5?

So if I wan't to run any setup with two SDRAM sticks do I always need version 2.9 sticks?

Regards,
Michael

[Xbytez]

So if I wan't to run any setup with two SDRAM sticks do I always need version 2.9 sticks?

You can run mixed 128MB memory stick versions without issue from my experience.

[FPGA64]

I run a 2,5 and 2,9 128 mb stick without issue

[Supersonic]

Sorry to pick this up again.

I have a V2.5 SDRAM stick and wanted to order a second one. Is this voltage problem a general problem with V2.5?

So if I wan't to run any setup with two SDRAM sticks do I always need version 2.9 sticks?

Regards,
Michael

I don't remeber wich PCB version were my ram sticks, but with both installed i had problems with many cores, after modification everything was back ok and now with 2.9 everything it's ok. probably, on my mister, it was enough to have just one of the stick "self-powered" just to unload a bit the 3v3 regulator on the de-10..
it could depend by many factors SDRAM ics current absorption or even 3v3 mister regulator performance..
i'd choose a 2.9 version..

[legacypixels]

The de10 3V3 regulator is a LTC3612EUDC#TRPBF rated for 3A. It is unlikely that this is your issue.

[cb88]

Perhaps not enough bypass capacitance or some sort of ground loop issue without a local regulator on the board.

Thinking of making a dual SD ram setup to fit into a portable laptop like case... so sounds like I'll run into that.