Strange PCIe to M.2 NVMe Adaptor with "supercapacitor backup"

[DaveLTX]

Someone in china obviously thought "hang on I could do a board with supercap backup on a adaptor!"

From a EE standpoint it makes not much sense as, how would a SSD react when the pcie is cut off? If they do still write with power on and logic off then maybe yes but I am apprehensive... Oh and that Optane QLC hybrid is dead before it went in.

 

[broadband]

M.2 2280 is not Enterprise class... just ordinary converter i think.

 

[DaveLTX]

M.2 2280 is not Enterprise class... just ordinary converter i think.

Enterprise 2280 do exist. But I think they made this to emulate the capacitor backup on those 22110 ssds but of course, they have different firmware as well. That's what the supercap is for at least on paper

 

[rootwyrm]

From a EE standpoint it makes not much sense as, how would a SSD react when the pcie is cut off? If they do still write with power on and logic off then maybe yes but I am apprehensive... Oh and that Optane QLC hybrid is dead before it went in.

Amusingly, this theory (but not this implementation) actually might work. It comes down to the host board and NVMe controller.
If we assume a well-behaved host board, then at power loss the NVMe should simply halt all operations for want of instructions. So all the battery is actually doing is preserving cache state. A well behaved host should command cache flush at initialization, and a well behaved controller should have a 'timeout' that flushes cache to disk if no instructions are received in a set period of time.
Similar concept to what AMI's been doing for decades with MegaRAID, except MegaRAIDs only preserve cache (for obvious reasons.)

But I doubt very much that's an actual supercapacitor, or one with enough capacity to handle anything longer than a brief power blip. Hell, to be entirely honest, that looks like it's just a CR2032 case. And it's probably not even actually hooked up.

 

[DaveLTX]

Amusingly, this theory (but not this implementation) actually might work. It comes down to the host board and NVMe controller.
If we assume a well-behaved host board, then at power loss the NVMe should simply halt all operations for want of instructions. So all the battery is actually doing is preserving cache state. A well behaved host should command cache flush at initialization, and a well behaved controller should have a 'timeout' that flushes cache to disk if no instructions are received in a set period of time.
Similar concept to what AMI's been doing for decades with MegaRAID, except MegaRAIDs only preserve cache (for obvious reasons.)

But I doubt very much that's an actual supercapacitor, or one with enough capacity to handle anything longer than a brief power blip. Hell, to be entirely honest, that looks like it's just a CR2032 case. And it's probably not even actually hooked up.

I will be doing some testing but however, it is an actual super capacitor!
I recognize it because I have bought it for power smoothing for a DAC before
It is attached to the 3.3V rail of the M.2 slot with backward diodes to stop it from back feeding into the pcie slot so yes, it is very much attached and measures way larger than 220mF (My capacitance meter is unable to measure anything larger than 220mF)

 

[Kahooli]

is U1 just a fat linear reg or a buck? Layout is... lame for a buck.

 

[DaveLTX]

is U1 just a fat linear reg or a buck? Layout is... lame for a buck.

It is. LM2596... man those are terrible for noise (others have far more advanced bucks) but it's a SSD so it doesn't matter and technically you don't even need it, PCIe slot has 3.3v and I assume its to draw from 12v

 

[newabc]

By the way, google search "pcie to m.2 nvme adaptor super capacitor site:aliexpress.com" will find they are all m2.2280, no 22110.

 

[broadband]

I think 22110 no need supercaps external indeed... so 22110 is enterprise M.2 with embended supercap

 

[DaveLTX]

I think 22110 no need supercaps external indeed... so 22110 is enterprise M.2 with embended supercap

View attachment 25956

Yes. 22110 have PLP built into them. Sometimes you can find PLP on 2280

 

[rootwyrm]

It is. LM2596... man those are terrible for noise (others have far more advanced bucks) but it's a SSD so it doesn't matter and technically you don't even need it, PCIe slot has 3.3v and I assume its to draw from 12v

No, it matters. Because the SSD will behave differently if it sees a voltage drop or spike. I'd at least want some smoothing on the 3.3V. But no, it does not need 12V. M-key only uses 3.3V.

However, loss of SUSCLK or failure to assert DEVSLP is likely to cause consumer controllers to get confused. For this to actually work, it should at minimum assert DEVSLP to assure state preservation.

 

[DaveLTX]

No, it matters. Because the SSD will behave differently if it sees a voltage drop or spike. I'd at least want some smoothing on the 3.3V. But no, it does not need 12V. M-key only uses 3.3V.

However, loss of SUSCLK or failure to assert DEVSLP is likely to cause consumer controllers to get confused. For this to actually work, it should at minimum assert DEVSLP to assure state preservation.

From experience, they are noisy in the audio band as adjustable regulators but as fixed regulators they are stable.