Concept: Multi-Op Boot switch

[Pegsat]

I say "Concept" because I've not seen this before. I made a pic of what I am considering, but I wanted to run it by (hopefully) technical people who would know why this would not work.

The closest thing I could find is by ICY DOCK, but it has 4 SATA ports, or another that has one eSATA port.

I don't want to plug/unplug as my MB seems to only have two Hard Drive ports for booting.
Here's my plan...

[diif]

Re: Concept: Multi-Op Boot switch

The boot order for the hardware is set in the BIOS/UEFI.
That would need changing every time you were booting to a different drive.
What is it you are doing that requires lots of changing of the OS ?

[stj]

Re: Concept: Multi-Op Boot switch

that's a shit idea, other than the sata errors, you wont be able to cross-acess the drives.

just add some ports with a pci(e) card.

[ratdude747]

Re: Concept: Multi-Op Boot switch

Or use GRUB2 or the like to boot... The only reason I'd see a use for a hard switch (which would probably need to be a digital multiplexer to maintain good connection) would be in high security systems (like the sort CIA/NSA/M16 agents would use) where physical isolation is critical. And I'm sure they already do this in some form.

[Pegsat]

Re: Concept: Multi-Op Boot switch

The boot order for the hardware is set in the BIOS/UEFI.
That would need changing every time you were booting to a different drive.
What is it you are doing that requires lots of changing of the OS ?

The boot order would not matter as the boot would see only the ONE drive that was powered up. What I am doing is running XP for one reason, Win7 x86 for another reason, Win7 x64 occasionally because many programs I have don't run on it, and then another drive for screwing around with (such as trying Linux versions) that can be reinstalled when needed.

that's a shit idea, other than the sata errors, you wont be able to cross-acess the drives. just add some ports with a pci(e) card.

I don't want to cross access the f'ing drives. Try reading some.

Or use GRUB2 or the like to boot... The only reason I'd see a use for a hard switch....

I don't want any other drive being recognized when I boot. I don't want them to cross communicate. I don't want to have any accidental f'ups by having another drive in the system I am working with.

This leaves me to plugging and unplugging drives before booting. That's the reason for the freakin' question. I just want to turn the switch to which drive I want to use and not have to go buy a new motherboard with 4+ bootable sata ports. If I had a bunch of money and space, I'd just have four freakin' computers and eight freakin' monitors. But no, One computer, two monitors one keyboard and mouse.

From my own research, I'm not the only one who has been looking for a TRUE sata switch, not just a power switch.
I can build the circuit from scrap, which I guess to get my answer, I'll just have to do it and see if the other drives will/will not drag the data (like in a short circuit) from the powered drive.

[Pegsat]

Re: Concept: Multi-Op Boot switch

Oh, and sure, I could buy a pci card, but there would be no guarantee the bios would decide to boot from it having the only powered drive. (Another "would have to buy to try thing.)

[stj]

Re: Concept: Multi-Op Boot switch

well maybe you should do some fucking research about SATA signalling before posting shit about switches.
it's low voltage and high frequency - it does not even like having the tracks different lengths or away from shielding - and you want to put it through a fucking wafer-switch???

buy some pullout trays.

[goontron]

Re: Concept: Multi-Op Boot switch

This won't work. Too noisy. RS232, maybe, Serial ATA? No. As it is i can't get an extender card i built to work above SATA I, With SATA III devices.
With noisey wafer switches? No way in hell. They won't even do 3v3 TTL at 1 Mhz reliably.

You could, theoretically get away with switching the power rails using a wafer switch, but only with more resilient OSes like Unix/Linux, and only if its not the boot drive and it's a hot-swap controller or you reboot and switch before POST. Even then, it may hang the SATA controller on the board if its not hot swap capable.

Keep in mind you could use logic-level switching. or even proxying, where you have a powerful microcontroller (EG. ARM like WD drives use) and have it receive and send all the ATA commands between the MOBO and drives, thus it could switch between the two drives, transparently deal with reinitializing the MOBOs controller (again, needs to be hot swap), and manage the SMART information. Really, this isn't feasible unless you plan to manufacture them. The R&D and firmware design and implementation would be insane, and to get fast enough microcontrollers cheap enough you would have to buy them in large-bulk orders.

Also, getting SATA certified to sell them isn't a cheap task either. Simple, Not cheap. https://www.sata-io.org/certified-logo-program

With all this said, if you don't plan to manufacture and sell them, it would be cheaper getting a PCI-E SATA card, or getting a new MOBO. This is assuming your board doesn't hotswap, but i've only seen one hot swap capable controller with only two ports....

[Th3_uN1Qu3]

Re: Concept: Multi-Op Boot switch

Here is some study material: https://cdn.badcaps-static.com/pdfs/...59f91ea673.pdf

A mechanical switch most definitely won't work. An electronic switch that can be commanded by a button or mechanical switch could maybe work. Try looking for HDMI switch ICs. HDMI is a bit lower bandwidth than SATA but it might work.

[stj]

Re: Concept: Multi-Op Boot switch

it wont work, sata is bi-directional.

[Th3_uN1Qu3]

Re: Concept: Multi-Op Boot switch

That is one thing i didn't take into account. Either way, look for some high bandwidth analog switches...

But tbh, this project could be more trouble than it's worth, and getting a board with more SATA ports would be a lot more feasible.

[Pegsat]

Re: Concept: Multi-Op Boot switch

Thank you both goontron and Th3, you both deserve the "I'm a: Professional Tech" label.

My design was downright dirty, I know..., the lines would be soldered in parallel, very close- side by side, as in the diagram.

I wasn't using a switch for the sata (as the last sentence indicated), relying on straight jumper from all ports, but then only one drive powered at one time... would be sending/receiving data.

The thought being, only one drive could communicate to the sata input on the motherboard. (Since the other drives are powered down.)

The concerns were that the non-powered drives might suck something out of the powered drive, AND that the noise might be leaked out or drawn in screwing up the data.

^I keep repeating the same things because the answers given didn't actually address that.^
Also thanks for the read, I will have a look at it.

[goontron]

Re: Concept: Multi-Op Boot switch

^ That's because we can't answer that for sure. Some drives may have capacitors to ground, or internal voltage references or pullup/pulldown resistors that could cause issues, while others may not. And that can change by revision of the model as well.

[Th3_uN1Qu3]

Re: Concept: Multi-Op Boot switch

No drives i've encountered have capacitors to ground. On the other hand, the impedance of the lines and the receiving IC (and inherent parasitic capacitance to ground) are enough to bring the lines down if you wire in a drive that isn't powered up. That is why a lot of mainboards wait a bit on SATA detection if you have forgotten to plug power into your drive.

The parallel thing *might* work with two drives of which one is powered down, but you have near zero chance of getting that working with more than two. You will need active switching.

[goontron]

Re: Concept: Multi-Op Boot switch

^ This.

Also, I have. Early batch 1tb Seagate 7200.12. Real low serial number. 8000s I think.

[Pegsat]

Re: Concept: Multi-Op Boot switch

Many thanks to you guys.

No drives i've encountered have capacitors to ground. On the other hand, the impedance of the lines and the receiving IC (and inherent parasitic capacitance to ground) are enough to bring the lines down if you wire in a drive that isn't powered up. That is why a lot of mainboards wait a bit on SATA detection if you have forgotten to plug power into your drive.

The parallel thing *might* work with two drives of which one is powered down, but you have near zero chance of getting that working with more than two. You will need active switching.

That was the tech answer I was wondering about. I didn't realize that MB's had that delay for that reason. And parasitic capacitance sounds like what I feared about the concept.

Thank you for your serious assistance. (Both of you.)

[goontron]

Re: Concept: Multi-Op Boot switch

^Im not sure if what th3 is describing is the spinup wait, or not, but yes, computers do have a 500ms or so wait for the sata lines to initialize...

[Th3_uN1Qu3]

Re: Concept: Multi-Op Boot switch

I meant that some boards will still try to detect something when only the SATA data cable is hooked up, without power, because they see a load at the end of the line. Hence why it may happen that in this case, the motherboard pauses for a few seconds before figuring out there is no drive detected.