A few questions about a homemade Sdram Shield

Hello to all,

I wanted to experiment with the V3 mojo with the sdram shield and realized it takes a little too much money and time to receive it.

So I made the sdram shield manually by modifying the original.
As I could not find the same memory chip inserted on the official shield.
(I turned the house over to find an old pc133 sdram bank and I’m sure there are but I couldn’t find the cards :-/ ).

So I unsoldered one from an old hd (K4S641632K-UC60) :smiley:
4 x 1,048,576 words by 16 bit → 8Mbyte if I’m not mistaken (16 bit words are found everywhere)
Organization Row Address Column Address
4Mx16 A0~A11 A0-A7

It is a little different from the original it has 8 extra bidirectional data pins and
two data mask control pins and has one less addressing bit.
but I think it is compatible and now I’m editing the example file to do some tests.

My questions are these:

  • there is a possibility of frying :angel: mojo if I am wrong in setting the doors
    bidirectional data of the sdram?

  • is it possible to use it in 8 bit format without modifying the example file?

  • I have never used sdram and on some schemes they are inserted
    25 or 33 OHm resistors in series usually to the data pins.
    In fact on the HD card are inserted at least the 16r on these pins.
    I believe it is used to eliminate disturbances (stub resistor).
    Do you think they need a test shield? seen that on the original
    there are not?

I powered the card without the mojo and it seems to work, by measuring the voltages on the pins they are floating or at Hz since the chip is not initialized and it absorbs a few mA.

Fabio

Hello to all,
Test shield sdram update.

I can’t get it to work. :huh:

I have modified the project so that it is compatible with the sdram chip in my possession but I cannot get any meaningful data.

It could be that I don’t see all the possible variations of the modules for this type of memory.

The mojo has not suffered any damage, I did tests on it. :slight_smile:

I checked the connections on the pcb and everything is correct.

I would not like the chip damaged during desoldering. :huh:

The only drawback is finding a complete datasheet.
On the net you can find the synthetic one but it is very meager in terms of information, not even to compare with the counterpart Micron.

It seems Samsung no longer has a decent datasheet on file describing the operation of the sdram chip in question.
I found a better one but it refers to a previous generation of such memory where the architecture is practically the same but some timing varies.

The thing I can still do right now is to check all the timing by comparing the Micron datasheet with the Samsung one.

If anyone of the forum wants to help me I can provide both the project and the schematic.

I made the card with kicad and it is very similar to the official one, they vary only some pinouts.

Fabio
[hr]
I checked the times and these are respected.
Always if I have not made mistakes in the verification but it does not seem to me.
Samsung memory is slightly faster than the micron.
Only the Tsh = 1n compared to the Tsh Micron = 0.8n

Interfacing with SDRAM without a full datasheet is likely to be very difficult. Each RAM chip has its own timing requirements and trying to guess these will be a pain.

All the example files and tutorials are designed around the specific SDRAM chip we used to sell.