SRAM based NVRAM

[vlad01]

That's a great point Ant!

I forgot that FRAM was even a thing since the last time I heard news about it was when they were developing the tech and it was thought to fail in the market so development slowed or halted iirc, but they were aiming for DRAM replacement early on in that. Had no idea that it ended up in eeprom and SRAM market space as an actual real product.

This summary appears to show that it's pretty ideal for this.

Looking at the last bit, yeah that's why it never replaced DRAM like they hoped.

Screenshot from 2026-05-01 08-12-17.png

[Holden202T]

i looked up fram ..... $40 just for the chip!

the whole reason i started looking at srams was cause they are like $10 with a controller!

[vlad01]

Yeah actually, I've been talking to Staggie via PM, he found old stuff on this forum where this was already looked into, and the FRAM requires the CE to be clocked, so it won't work on the Delco without more serious engineering, and yeah the cost is horrendous for what it is. Just like I thought that FRAM was dead in the water way back in it's development, that cost reflects that low scalability and presumably low yield out of the fabs. Back when I read the papers on the concept and early samples of the tech and the aim for eventual DRAM replacement (would have been awesome), it sounded like it wasn't going to ever make it as a product. Now that it's a product despite the odds, it only a tiny niche one at that, hence the price tag.

The other Cypress ultra low powered chip I suggested is what he's going with to investigate further. It should on paper last a few years on coin cells we hope.

[antus]

I'd be interested to know why clocking ce is a problem. Is it just complexity? There is a clock signal, or perhaps some very small MCUs that could handle that with very little code if it's not a direct connection, or needs some delay etc. Something like https://www.digikey.com.au/en/products/ ... SF/4368794 (wtf, atmel got bought by microchip?)

[Staggie]

This is just my personal opinion, but yeah pretty much is adding complexity and expense, with the only real gain being no battery required.
Would it be a cool little project just cos, yes, but I feel it would also cause its fair share of grey hairs. If we didnt have to change the way CE behaved i would be all for it but I feel the KISS approach is better for long term reliability in this case.

P.s. that microcontroller is definitely tiny, could certainly think of a few uses for that when dont need all the I/O.

[Staggie]

So a bit of an update on this one, finally got around to putting one together to test the CY62128ELL and I can confirm there is a hell of a power saving, with WE pulled up on Vout it still is only pulling 0.9uA at idle.
Have managed to get it programmed and bank switching correctly, also being detected correctly in TunerPro RT as writeable.
Will give it a run in a vehicle in the next week or two but all looks very promising. 12 years battery life over 2 CR1225's should be enough I reckon.

[vlad01]

Well done! keeping an eye on this one.