~ OT, any idea to have parallel 8 bit non volatile ram ?

[legacy]

hi
i am thinking about a tiny board with a so called "zero ram NVram, Nonvolatile SRAM" onboard: something like 32Kbyte/8bit ram to store basic scripts.
I have a 3.3V system so i'd prefer a 3.3V SRAM

Quote:

32K DS1230W 3.3V 256K Nonvolatile SRAM
128K DS1245W 3.3V 1024K Nonvolatile SRAM
256K DS1249W 3.3V 2048K Nonvolatile SRAM
512K DS1250W 3.3V 4096K Nonvolatile SRAM
1M DS1265W 3.3V 8M Nonvolatile SRAM
2M DS1270W 3.3V 16M Nonvolatile SRAM

What do you think about NvRam ? Do i'd better use FeRam ?

For example 32K MB85R256F, or 128Kbyte FeRam in package tsop from Ramtron International Corp.

also, is there any free samples pick up ? I am a student, i tried to use my university email account but there was no way: it appears to me that Maxim has stopped samples for hobby due the fact the sample program has been abused by people, ordering samples, then selling them off on eBay or forums.

Personally, I don't think the sample program should be for "hobbyists.", I think if you want a part? go buy the part! I have bough the 90% of the components of my SBC project (6809 SoC on Xilinx FPGA), i miss only the NvRam chip which is incredibly expensive and very difficult to be find for me.

[GARTHWILSON]

One thing I used to have against the NV RAMs was that the 10-year battery was not replaceable. It looks like they've taken care of that now. The other things are the slow speed (100ns) and the high cost. I believe Maxim and others do have ICs to make any SRAM non-volatile (using your own separate battery) but you'll have to see how they affect the access time. Alternately, you can make your own circuit. I could email you a scan of a super-simple method that was in the "Ideas for Design" section of one of the industry magazines years ago, if you PM me your email address. [Edit: See viewtopic.php?f=4&t=573&p=32004#p32004 for the diagram.] It uses a 3.6V NiCd battery that's trickle-charging any time the computer is powered up. With a small modification, you could use a non-rechargeable battery like alkaline or lithium. A capacitor can keep the RAM alive while you replace the battery every so many years. This all means of course that you can't program the RAM externally and then plug it in to your board, since it won't have any voltage to hold it while it's out of the socket.

[clockpulse]

I have one Dallas DS1225Y NVRAM manufactured in 1986 and the battery is still holding.
It was in circuit with 5V power applied for quite a few years but the last 15 years or so it's been sitting around with only it's built-in battery power applied.

I have a few of the Dallas 1024 kbit that are about 13 years old and they are all still ok with battery.
The speed on those are 100ns but I'm using one running at 10mhz and it works fine.

The specs seem to be pretty conservative, especially if they are used at normal room temperatures.

[legacy]

i wonder if 100nsec could be a problem for my fpga@50Mhz, in case i will add a lot of "wait state" cycles to stabilize the data transfer.

[legacy]

@GARTHWILSON
thanks, you have an MP

[PaulF]

Battery backup chips for RAM devices usually just provide a constant source of power for the RAM, switching cleanly between battery and mains power. They don't affect the chip access times because they don't get involved in the chip select logic. Usually, they provide a logic output to indicate that battery power is in use, you can feed this into the chip select logic to inhibit the device when on battery power. Whether this will increase the access time depends on the design of your chip select logic.

I like these devices because you can use any RAM device, they can feed a real-time clock device as well and you can change the battery when the device is mains-powered without any problems.

[bogax]

PaulF wrote:

Battery backup chips for RAM devices usually just provide a constant source of power for the RAM, switching cleanly between battery and mains power. They don't affect the chip access times because they don't get involved in the chip select logic.

Not strictly true (depends on what you're talking about, of course).

For example, Dallas DS1210 has a /CE propagation delay on 20ns (max.)