Eprom Speeds?

[+Larry]

I'm getting ready to order some new Eproms, and I'm curious about what speeds other folks have been using. Most of my older Eproms are 200 ns types, but I see that more modern CMOS (e.g. 27C64's) run down to 40 NS or so. Any advantage on the Atari to having such fast speeds? I'm tentatively thinking about 100 ns parts. I've heard a couple reports of problems when using very fast Drams on an Atari (?), but never heard of any problems related to Eproms other than making sure that one used 250 ns or faster. Any thoughts?

-Larry

[Rybags]

I'm not expert in the department but a couple of factoids:

 

- in the old days, RAM upgrades recommended a maximum of 250 ns access time for RAM chips. Which sounds perfectly feasible given that the Atari runs at near on 1.8 MHz, so 250 ns translates to just under half a machine cycle.

 

- the 64Kx4 chips in the XEGS (mine at least) are 120 ns (the suffix 12 denotes 120, 10 for 100 etc)

 

I think the problem running really fast chips is that they actually expect the host to be mega-quick with things like RAS->CAS transitions, whereas the Atari probably does things at a rather relaxed speed by today's standards.

[+Larry]

I'm not expert in the department but a couple of factoids:

 

- in the old days, RAM upgrades recommended a maximum of 250 ns access time for RAM chips. Which sounds perfectly feasible given that the Atari runs at near on 1.8 MHz, so 250 ns translates to just under half a machine cycle.

 

- the 64Kx4 chips in the XEGS (mine at least) are 120 ns (the suffix 12 denotes 120, 10 for 100 etc)

 

I think the problem running really fast chips is that they actually expect the host to be mega-quick with things like RAS->CAS transitions, whereas the Atari probably does things at a rather relaxed speed by today's standards.

 

Thanks, Rybags. I'll go with the 100's (it actually costs more to go slower on some of these!)

 

And BTW, you mentioned in another thread that the feature you liked most about APE was not having to physically change floppies (IIRC). How right you were! By moving all my 720K utility disks to images (and using Tom's Navigator), I've been able to move files around and get a lot better sorted/organized!

-Larry

[MrMartian]

I'm getting ready to order some new Eproms, and I'm curious about what speeds other folks have been using. Most of my older Eproms are 200 ns types, but I see that more modern CMOS (e.g. 27C64's) run down to 40 NS or so. Any advantage on the Atari to having such fast speeds? I'm tentatively thinking about 100 ns parts. I've heard a couple reports of problems when using very fast Drams on an Atari (?), but never heard of any problems related to Eproms other than making sure that one used 250 ns or faster. Any thoughts?

-Larry

 

As far as EPROMs go, there would never be a problem going with a faster one. The speed of an EPROM is the maximum amount of time after you ask for a certain address, until the data the valid. It doesn't matter how long after that you actually read the data.

[Urchlay]

As far as EPROMs go, there would never be a problem going with a faster one. The speed of an EPROM is the maximum amount of time after you ask for a certain address, until the data the valid. It doesn't matter how long after that you actually read the data.

 

Is the same true of DRAMs, or do really fast modern DRAMs need faster refreshing?

 

I already know the Atari 8-bits can use anything from 100 to 200ns DRAM, but would there be a problem with, say, modern 25ns chips (assuming they physically will fit I mean)?

[supercat]

As far as EPROMs go, there would never be a problem going with a faster one. The speed of an EPROM is the maximum amount of time after you ask for a certain address, until the data the valid. It doesn't matter how long after that you actually read the data.

 

Generally, a system which works with a 250ns memory should work with a 2.5ns memory, but there can sometimes be problems. Some systems may require data to be held for a few nanoseconds after a clock edge. In such cases, it may be theoretically possible for an overly-fast EPROM to mess things up.

 

RAM can sometimes have bigger problems. In some systems, a RAM address may change at about the same time as the enable is switched off. Depending upon the design, there may be a brief moment (<2ns) where the wrong address is enabled for writing. A 200ns RAM isn't likely to write data during that 2ns window, but a 20ns RAM very well might.

[DamageX]

AFAIK, nobody makes 5V DRAMs anymore, and the style that the Atari used (16-pin DIP) are probably never rated faster than 60ns.

[MrMartian]

Generally, a system which works with a 250ns memory should work with a 2.5ns memory, but there can sometimes be problems. Some systems may require data to be held for a few nanoseconds after a clock edge. In such cases, it may be theoretically possible for an overly-fast EPROM to mess things up.

 

RAM can sometimes have bigger problems. In some systems, a RAM address may change at about the same time as the enable is switched off. Depending upon the design, there may be a brief moment (<2ns) where the wrong address is enabled for writing. A 200ns RAM isn't likely to write data during that 2ns window, but a 20ns RAM very well might.

 

Yes, with RAM there are many issues that can occur with regards to speed. Since RAS and CAS are used as strobes to latch in memory addresses, as well as selecting refresh on some chips, holding one of them on for a very long time (at least to what the chip is designed for.. ) can result in some other function being selected other than just plain reading an address.