TTL -> NMOS -> TTL

[Bjarni Juliusson]

Hi everybody.

I was building a little circuit out of scrap TTL logic I had lying
around, when I needed a small amount of RAM. I found a 1K chip
apparently manufactured in 1974, and figured it was probably TTL. At
first it worked fine, but I soon noticed the data output would flicker
when the addresses changed. I touched the chip, and it flickered some
more. I suppose the problem is marginal input levels. A googling
revealed that the chip (a 2102) is in fact NMOS, not TTL as I first had
thought. Because I'm still learning this stuff, my question is: how do I
properly interface this NMOS device to the TTL devices making up the
rest of the circuit? Do I pull up? In that case, what values are
appropriate? Any help greatly appreciated.


Bjarni

 

[PeteS]

First, *all* SRAMs (the 2102 is a 1k x 1 SRAM) will have indeterminate
data during address changes assuming CS and OE are active. The reason
is not all the addresses change at the same time for 2 reasons:

1. The internal thresholds of the address latches of the SRAM will have
varying levels.
2. The address generator will not change all outputs at the same time,
nor with the same slew.

The 2102 is a slow (by today's standards) device, by the way.

NMOS does need solid inputs (although many of them had thresholds that
were suitable for TTL drive). The 2102 is a 5V device, so make sure you
are using 5V drive and power. I can't find a data sheet with a cursory
search and my old data books from that period are long gone,
unfortunately.

If you want to pull the inputs, I suggest about 47k pullups (but bear
in mind this will slow things down more as the early NMOS devices had
significant input capacitance). Try 10k if you want to speed things up
a little.

Also keep in mind that the output doesn't have very much drive, so
don't try to directly connect it to a LED for example (a good rule of
thumb is <200uA for reliable outputs).

All such devices have various timing parameters such as valid input
address to valid output data (which may vary depending on the gate - OE
or CS). If you have the data sheet, study the timing requirements.

Cheers

PeteS

 

[Bjarni Juliusson]

PeteS wrote:

First, *all* SRAMs (the 2102 is a 1k x 1 SRAM) will have indeterminate
data during address changes assuming CS and OE are active. The reason
is not all the addresses change at the same time for 2 reasons:

1. The internal thresholds of the address latches of the SRAM will have
varying levels.
2. The address generator will not change all outputs at the same time,
nor with the same slew.

The 2102 is a slow (by today's standards) device, by the way.

Yes, but this was so slow that I could see it.

NMOS does need solid inputs (although many of them had thresholds that
were suitable for TTL drive). The 2102 is a 5V device, so make sure you
are using 5V drive and power. I can't find a data sheet with a cursory
search and my old data books from that period are long gone,
unfortunately.

I have made sure the supply is 5V. I had a problem with that earlier
actually, but changed power supplies. I couldn't find any datasheets either.

If you want to pull the inputs, I suggest about 47k pullups (but bear
in mind this will slow things down more as the early NMOS devices had
significant input capacitance). Try 10k if you want to speed things up
a little.

OK, I will try this when I get home. The speed is noncritical as this
circuit reads keypresses and only needs to run at a few kHz or so. Right
now I'm clocking it by hand.

Also keep in mind that the output doesn't have very much drive, so
don't try to directly connect it to a LED for example (a good rule of
thumb is <200uA for reliable outputs).

The output is buffered by a TTL device before driving the LED.

All such devices have various timing parameters such as valid input
address to valid output data (which may vary depending on the gate - OE
or CS). If you have the data sheet, study the timing requirements.

I doubt strongly that it is a timing problem. The flickering is clearly
visible to the naked eye and occurs whenever I touch the top of the chip
with my finger.

Thank you very much for your reply. I will try pulling up the inputs.


Bjarni

 

[Bjarni Juliusson]

I went home and tried pulling up the inputs. Unfortunately, I couldn't
reproduce the problem before I added the pullups, but I'll leave them in
and see if the problem comes back when I've dragged my feet sufficiently
on the rug. Thanks for your help anyhow.


Bjarni