Are CMOS RAMs noisy?

[Nessus]

Do CMOS RAM chips have noisy data lines on output?  Or, are there anything special (pull-ups, etc) that need to be done?

 

I've wired-up my ROM-RAM cartridge on a breadboard to test the circuit.  It runs Fred Kaal's Ed/Assem IV great, but that's with the RAM disabled.  I can generate the RAM and EPROM bank-select and address strobes, but my data test fails.  The program writes a word to the RAM with a MOV R1,*R1 instruction, and then immediately reads it with a C R1,*R1+ instruction.  I use a LogicPort analyzer to check the RAM accesses and the compare fails, frequently after ten or so cycles.  There are a lot of data transitions during the read cycle, as if the data lines are noisy.

 

Here's a picture of the breadboard:

 

I realize that breadboards are noisy, and the RAM chips have been laying in my parts box for more than a decade.  As I recall, they are 50ns cache chips I salvaged from dead computers.  I'm also thinking of changing the RAM to some newer chips, except that I can't find new 64KB chips, which is what I designed for.

 

K-R.

Edited May 20 by Nessus

[Gary from OPA]

Adding the bypass or decoupling caps is needed this helps prevents issues like you are seeing. Just add a ceramic 100nf cap across the power lines of each chip.

 

 

[matthew180]

Static RAM?  Probably not noisy, they are 100% combinatorial logic inside.  They will only be as noisy as the frequency at which you are accessing the data, but they won't be any more noisy than the address and data bus itself.

 

SRAM is not like, say, a buck regulator or something that has an internal frequency generating circuit.  Use typical bypass, and make sure you compensate for breadboard noise (although the limits of that have been proven/pushed pretty far.  See the RadicalBrad 6502 mega computer on 6502.org forums).

[Nessus]

This may have been a false alarm, I'm sorry to say.  I've been fighting this thing for days, when it occurred to look at the bus data on the logic analyzer traces.  Here's the routine I used to check the RAM:

 

  0158
  0159            *********************************
  0160            **  RAM data read/write test.  **
  0161            *********************************
  0162

 99/4 ASSEMBLER

VERSION 1.2                                                  PAGE 0004

  0163            DATEST
  0164 609A 02E0         LWPI >8300        * Set workspace pointer.
       609C 8300
  0165 609E 0300         LIMI 0            * Disable interrupts.
       60A0 0000
  0166
  0167 60A2 C820         MOV  @>6000,@>6000  * Run from Bank 0.
       60A4 6000
       60A6 6000
  0168
  0169            D#LOOP
  0170 60A8 0201         LI   R1,>7C00     * Initialize address counter.
       60AA 7C00
  0171
  0172            D#LOO1
  0173 60AC C441         MOV  R1,*R1       * Send data to RAM.
  0174 60AE 8C41         C    R1,*R1+      * Check data.
  0175 60B0 1604         JNE  D#STOP       * If no match, STOP.
  0176
  0177 60B2 0281         CI   R1,>8000     * At end of RAM?
       60B4 8000
  0178 60B6 16FA         JNE  D#LOO1       * If not, continue...
  0179 60B8 10F7         JMP  D#LOOP       *  ... else, start over.
  0180
  0181            D#STOP
  0182 60BA 10FF         JMP  D#STOP       * Stop here!
  0183
  0184                   END

 

The read-data matches the data previously written, even though the analyzer triggered an error.  I had set it to trigger on >60BA on the address bus; on the cartridge it was >00BA with CART_SEL low.  I looked at the analyzer trigger and figured it found a brief time where the bus switching had the >60BA.  After changing the trigger to look for >60B8 followed by >60BA and set the select and DBIN strobes correctly, the analyzer didn't trigger.  When I manually triggered the analyzer, the read-data still matched the previously written data.

K-R.

[Nessus]

Gary from OPA, thanks for the advice about the bypass caps.  I had previously added a 10uF tantalum cap on the breadboard power bus, but had forgotten to put bypass caps on the chips.  They should be the little yellow thing sticking up among the rat's nest of wires.

 

Matthew180, thanks for the links and info about breadboards.  I spent a couple of hours this morning looking in to it.  Most fascinating!

 

As I mentioned above, I think I found the problem.

K-R.

[+Ksarul]

BTW, here is a current source for the 64Kx8 SRAM chips. They sell 50 of them for under $45, delivered.

[Gary from OPA]

3 minutes ago, Ksarul said:

BTW, here is a current source for the 64Kx8 SRAM chips. They sell 50 of them for under $45, delivered.

Sadly, I ordered a set of 3 for my apedisk99 build back at the end of April, and they ended up in Scarborough, Yorkshire in England on may 13th instead of Scarborough, Ontario in Canada. So tough luck for me.

[Nessus]

49 minutes ago, Ksarul said:

BTW, here is a current source for the 64Kx8 SRAM chips. They sell 50 of them for under $45, delivered.

Thanks!  I'll keep it in mind.

I've tended to shy away from ebay (always thought they had a 'shady' reputation), but was pleasantly surprised with my EPROM eraser order.

K-R.

[Nessus]

47 minutes ago, Gary from OPA said:

Sadly, I ordered a set of 3 for my apedisk99 build back at the end of April, and they ended up in Scarborough, Yorkshire in England on may 13th instead of Scarborough, Ontario in Canada. So tough luck for me.

That's a drag!  I hope they corrected the problem!

Reminds me of a couple incidents.  My post office is in a town called Naples, in the US.

A couple of years ago I was ordering some parts over the phone, and the person at first thought I said Maples, like trees, until I have her my ZIP code.  Apparently, her computer had a database that corrected it.

The other one involved the clerk at my post office.  She mailed a letter from her town to a friend in (my town) Naples.  Her friend received the letter a month later, having gone to Naples, Italy!

K-R.

[Gary from OPA]

2 minutes ago, Nessus said:

That's a drag!  I hope they corrected the problem!

Reminds me of a couple incidents.  My post office is in a town called Naples, in the US.

A couple of years ago I was ordering some parts over the phone, and the person at first thought I said Maples, like trees, until I have her my ZIP code.  Apparently, her computer had a database that corrected it.

The other one involved the clerk at my post office.  She mailed a letter from her town to a friend in (my town) Naples.  Her friend received the letter a month later, having gone to Naples, Italy!

K-R.

Not yet, I am still trading back and forth emails with the Chinese orange speedpak service over how they shipped it to the wrong Scarborough.

[+bent_pin]

18 minutes ago, Gary from OPA said:

Not yet, I am still trading back and forth emails with the Chinese orange speedpak service over how they shipped it to the wrong Scarborough.