Insane or Possible

Tell me if you think this is possible what if you could buy a ssd card right say a 64gb and make it user upgradable you unscrew the card and the card itself has easy to snap or solder or screw on nand chips being that is a 64gb card their would be one slot filled but lets say there were 7 empty slots available this way you would only have to buy the nand chips when needed

 

It could be done, but I don't think any of the current devices have anything you could work with.

The chips are surface-mount (not socketed) and wave-soldered or baked to flow solder micro-balls onto the leads and the pads. The factory controller chips vary from capacity to capacity, even though the board is mostly the same. So, adding chips to the available pads would go unrecognized by the controller (which wouldn't be looking for the added chips).

If you notice a difference in benchmark results on SSDs of same model line but different capacity, it is because the internal controller chip is different (specific to memory of each unit). Change of memory needs a change in controller.

There may be some exceptions regarding controller chip finding added capacity chips, but the wave-soldering/baked-microballs issue is still going to be a hinderance to almost everyone. How many of us actually have a solder-bath or a reflow station immediately available?

From the manufacturer side, production is already streamlined for SMD. Also, if they made it easy for users to upscale, they wouldn't make any money selling the higher capacity units.

I like the way you think, though.

 

thanks just thinking out loud

 

That is a good idea though. Snap in NAND chips to upgrade. I can see it happening in the future, and why not? It would actually likely make the companies MORE money because they'd probably make more money on the NAND units themselves over a full SSD unit. Like RAM modules, even if it only had like 1 or 2 spots to add another 64GB or 128GB modules inside. Then you could always add more capacity later when you can afford it, or becomes cheaper.

 

I agree, it's a great idea. I hope we see it happen some day.

@rekeyroberts: keep thinking out loud. Figure out a way to weasel yourself a job somewhere in the hardware industry. Make us some better toys.

 

This idea is akin to having hard drives you open up and swap the platters on.

 

Yeah, but I suppose that since SSDs don't have moving parts, replacing NAND modules would be much easier than swapping platters (in theory). I imagine that the platters must have very specific balances and tolerances and won't tolerate smudges too well.

 

No it's not. To open a mechanical hard drive, you must be in a very clean environment, without any dust, etc. Otherwise, your hard drive can be ruined. There are many risks associated, which is why you read about so few attempts to remove the platters and case. With an SSD, there are no dangers associated with removing the case.

 

From an enthusiast's perspective, this would be awesome for upgrading. Whenever a faster SSD came out you could just buy a new "barebones" SSD (controller + RAM + PCB + plastic shell) for maybe $40 and move all of the NAND from your current SSD over.

From an SSD manufacturer's point of view, this would be terrible. Controller makers would still be able to recoup their R&D expenses, but there'd be no profit for anyone further upstream. NAND manufacturers would also be hard hit since people would just keep carrying over their NAND until all of the P/E cycles were used up.

 

Actually the companies would lose money because nobody would buy this.

With socketed flash modules, how do you expect the drive to remain in the 2.5" form factor? The boards of 2.5" SSD's look pretty crowded already. There is no space for anything else.

If all the flash sockets are not populated, performance will take a nosedive. With 1 socket populated, your performance will be worse than a mechanical hard drive. Much worse. Why do you think all those very small sized SSD's are so freaking slow?

Secondly, if the drive's firmware has to be overly basic to accommodate virtually any type of flash memory you could plug into it, how do you think that will affect performance? It will be something that will run fine with everything, but will not be tweaked for maximum performance, otherwise it will have too many compatibility issues.

And this doesn't help anybody save money. The price of one of these socketed stand-alone controllers would be much greater than the cost of a normal controller as part of a non-upgradable SSD. The socketable flash memory would have a higher price tag too since the packaging would be more costly.

Upgradable SSD's serve no purpose whatsoever. You need to upgrade all you flash memory at the same time unless you want horrible performance, but then you are stuck with an older controller. Not only that but all these parts would cost a premium compared to a non-upgradable SSD. The whole concept just does not make any sense.

I meant akin to the level of silliness.

 

Come on Qing Dao where's your sense of optimism? It's a plausible path, just not with the way it's designed today.

 

Honestly, I foresee the day when computers will not be unlike the computers in Star Trek with isolinear chips, only the chips will have different modules. Like you would have RAM modules, CPU modules, graphics modules, etc, and you can just hot swap them, and they'd all be inter-compatible.