• Please review our updated Terms and Rules here

Project Request: USB 5.25" Floppy

Raven

Veteran Member
Joined
Mar 7, 2009
Messages
2,752
Location
DE, USA..
There is a huge demand for these. Google the phrase "USB 5.25" Floppy" and hundreds of hits with people hoping and pleading for one to exist abound.

The way I see it, two methods exist to do this:
- Interface a common parallel controller (Microsolutions 3.5" external, etc.) to USB via a converter and then write a driver.
- Create a USB floppy controller that provides a 34-pin header for use with a floppy drive, possibly in a pre-fab 5.25" enclosure ready to accept a 5.25" drive (or a 3.5" drive, why stop at 5.25".. this would allow combo drives).

I don't have the resources or experience necessary to design or build this, which is why I labeled this a project "request".

If you doubt the desire for this item, just wait for the posts to pour in seconding my request - lol.
 
This might be one way to go:

http://www.arstech.com/item--usb2isa.html

It is a USB to ISA adapter. Not cheap at $149, but once you have it, it seems you could plug in your own floppy adapter card (or other things, too).

The description of the software makes it sound like it could be configured for most ISA devices, and there's also an SDK (costs more) if you need to do something different.

They also have a customized build of Dosbox that is supposed to be able to access the ISA card in their adapter (builds for Windows and Linux).
 
Hi! You are free to request all you want but if you truly desire for this to become reality you'll almost certainly have to do it yourself. I doubt very much there are builders capable of pulling off such a project sitting around idle just waiting for this idea to come along.

Chuck G was considering a similar project at one time and I volunteered to help with the PCB aspects. I have no idea what's become of it although I haven't heard anything about it in quite a while so I assume its been set aside.

A practical step to making this happen would be to design a system and publish a schematic. Build a prototype PCB and test it out. Publish some test code. It is certainly possible to do. However, I doubt most builders are going to defer existing projects just to work on a USB floppy drive when there are other items of interest remaining. A request with no personal investment (time, money, effort, skills, etc) is not very compelling.

Thanks and have a nice day!

Andrew Lynch
 
I've got my prototype sitting off to one side. There's a pile of software to write for it and right now, I don't have the time. Maybe someday.

Phil Pemberton has an FPGA-oriented project (Disk Ferret, discussed on cctalk) that might also do the trick. But it involves a bunch of SMT, so I doubt that it's a good candidate for DIY.
 
Hi! My offer still stands for making a PCB which may allow offloading some of the work for SW development to other builders if you have "proof of concept" code already. As I recall you had some success reading various disk formats. However my plate is pretty full at the moment as well so I don't plan on diving into a major development project either. Is the hardware stable enough to even consider a prototype PCB? I suspect this project is going to sit fallow for a while.

Thanks and have a nice day!

Andrew Lynch
 
Hi! My offer still stands for making a PCB which may allow offloading some of the work for SW development to other builders if you have "proof of concept" code already. As I recall you had some success reading various disk formats. However my plate is pretty full at the moment as well so I don't plan on diving into a major development project either. Is the hardware stable enough to even consider a prototype PCB? I suspect this project is going to sit fallow for a while.

Yup probably. Yes, I have "proof of concept" prototypes, but all involve something that I don't like; i.e. parts count too high, SMT devices, etc. I'll probably take another run at this toward the end of the year using an Atmel 89xx uC, as a high-pin count one can be gotten in PLCC (through-hole). The prototypes do just fine reading and writing FM and MFM.

At any rate, those are my thoughts right now.
 
I am willing to put money into this project, and any skills/time/etc. that I am capable of - I have minimal electronics and soldering skills and no ability to write chips or code for them, however, unless x86-compatible.

Perhaps you could post some schematics of your prototype designs, Chuck, and let the community take a crack at reducing part count, etc.? I don't know that anyone here who might be interested in this project would be willing, but perhaps they are and this would speed things up?

Just so you all know, I'm not trying to be pushy or anything by starting this thread - I just wanted to make it clear that there was a demand for this device, and I am one of the ones who keep running across a need that there's no product to fill.


This might be one way to go:

http://www.arstech.com/item--usb2isa.html

It is a USB to ISA adapter. Not cheap at $149, but once you have it, it seems you could plug in your own floppy adapter card (or other things, too).

The description of the software makes it sound like it could be configured for most ISA devices, and there's also an SDK (costs more) if you need to do something different.

They also have a customized build of Dosbox that is supposed to be able to access the ISA card in their adapter (builds for Windows and Linux).

I have seen this, and considered it, but had abandoned the idea because I figured one would have to write their own drivers for a given card under Win32 or Win64, etc.. If, as you say, they have a customized version of DOSBox that allows access, perhaps I will re-investigate those products. Thanks.

Edit: Also, it looks like since last time I saw these products, they devised a nice, only slightly more expensive, 7-slot variant: http://www.arstech.com/item--usb2isasl.html

I cannot, however, find any mention of a custom DOSBox - could you link me to that?
 
Last edited:
In that link to the 7 slot it says "We currently offer support for DOS under Widows (actually Windows correct :D ) operating systems.". Is that a reference to DOSBox I wonder?
 
In that link to the 7 slot it says "We currently offer support for DOS under Widows (actually Windows correct :D ) operating systems.". Is that a reference to DOSBox I wonder?
I don't get it. That solution looks like it'll cost >$300 by the time you're done; why not just take an old (free or next to) small form factor PC box and network it?
 
It says it has to use the same IRQs and resources so if you had a floppy controller it might well conflict with the one in the host system. I am starting to think this is the wrong product for this particular case.
 
Well my root problem is the lack of a floppy controller on my motherboard (and very likely my future motherboard), so that shouldn't matter.

Edit: Yeah I just checked and IRQ6 is free.. unless -6 is the same thing as 6.. didn't know we could have negative IRQs now.. :eek:
 
I don't get it. That solution looks like it'll cost >$300 by the time you're done; why not just take an old (free or next to) small form factor PC box and network it?

Because I'm obsessed. My main box must read 5.25" floppies - on it's own.

I can get it to do this now, via a cannibalized hackjob of a Backpack 3.5" converted into a 5.25" but there are two problems:
- If I want to mount it internally (why not, neater than a hackjob anywho) I'd want to power it from the PSU somehow, and I don't know enough electronics to do that.
- It requires a VM with parallel port passthrough (Parallels, ironically, is the only one with this functionality that actually works) to access the drive.

This is not really good enough, but it's the best I can do under my own power. The only real issue here is a lack of driver support. I can't locate someone to write up an NT-compatible driver for the damn thing though, even though we have the source code for the driver readily available. If someone could modify the Backpack driver source we have and compile it into a working NT-compatible driver for parallel Backpack drives, I would pay them to do it. Unfortunately none of the languages I'm versed in can be used for this except maybe assembler, but using ASM to write an NT driver doesn't sound reasonable, plus I'd have to start from scratch because the original drivers (I believe) are in C.

If someone can come up with a proper USB 5.25" 1.2MB floppy drive that just gets plugged in via USB and recognized (possibly with driver installation from a disk) as a 5.25" FDD with no issues, I am willing to pay quite a bit for such a unit.

I was planning on upgrading my motherboard almost solely for the purpose of gaining a floppy header (and maybe some more SATA ports, a PCI or two) from a mATX to ATX board, however, I only want Intel boards and I can't locate one that uses DDR2, 1333mhz FSB, Core 2 Quad, at least 1 x16 PCIe and a floppy header.

This is such a maddening thing. If I had the knowledge Id've created this three hundred times over but I don't even know where to begin - I constructed part of the housing for my external 5.25" Backpack out of cardboard ripped from a box - I have no physical construction capability...

I have considered a Catweasel on and off for about two years now. They cost enough so that it's been prohibitive, but I might be able to afford one soon. The side benefit of being able to read exotic formats (like C64) is quite nice, but I primarily just plan to use it as a damned PCI floppy controller (which should really exist in it's own right!). I have concerns, though, as to whether or not this will work for such a simple task, or if it will relegate me to using odd third party software just to read and write a file to the drive, or if it will provide them with drive letters, which would be fantastic. Either way that's better than what I have now, as at least it wouldn't have to involve a virtual machine.

I've probably spent a good 20 hours total over the past few months thinking this situation over in my head and this is the sum of that---^

Edit: Ah yes, then there's the other solution.. dualbooting. I have been considering a dualboot for DOS compatibility for some time now, to a 32-bit Win7, but perhaps I should look further back to 98SE or the like, where there's direct hardware access and Backpack drives will work. There's a patch to allow 98SE to work with any amount of RAM (and actually use much of it.. it refuses to install or run with 1.5GB as it stands) but it costs money.. Perhaps it's about time I dropped money on that, it's less expensive than most of the alternatives I've come up with.. Do Backpack drives work under Linux? That's another consideration...
 
Last edited:
Because I'm obsessed. My main box must read 5.25" floppies - on it's own.
...
If someone can come up with a proper USB 5.25" 1.2MB floppy drive that just gets plugged in via USB and recognized (possibly with driver installation from a disk) as a 5.25" FDD with no issues, I am willing to pay quite a bit for such a unit.
...
What's the difference? OK, it'd be a little bigger, but it'd be a 1.2MB drive in an enclosure that you plug in via ethernet and it'd be recognized with no drivers necessary. If you've got some room in your main case then all you'd need is one of those PS splitters and presto, it's an internal drive/interface.

Some people are just sooooo fussy...

;-)
 
I'm playing with the idea of interfacing an AVR using V-USB to a Backpack floppy. It should be pretty easy and the Backpack has a real (NS 8477) controller in it with registers that can be addressed directly. All you need is a little "glue" logic that twiddles the parallel port lines.

Cheap and simple--and if you're one of the fortunate souls to have taken advantage of the $0.99 eBay on Backpack floppy drives, you're really there on the cheap.
 
@Chuck:

I bought a stockpile of them. They're still on eBay, btw.

Anyway I have about 8 of the things, one of which is a very messy 5.25".

If you come up with that solution that would let it hook up via USB I will most certainly be a customer for ya.

I would like to submit a request that if you do persue this, perhaps look into making a mechanism to power the backpack board from the PSU as well, so that the whole thing can be mounted internally? Even if you don't do this, I'll still readily buy, btw.

@Mike:

I did try to mount a second motherboard in my case, couldn't fit one. If you recall I'm the guy who linked you those splitters, and the reason I know about them was from THIS VERY QUEST - lol. I was so obsessed that I wanted to weld two cases together (or to be more accurate, get someone else to) and hook them up to one power button, before I decided that *that* was a bad idea (at that point it doesn't even feel like one computer).

I also tried to mount my modern mobo in a huge full AT case but couldn't find a way to mount it or anything. I know it could be made so, so I tried mounting a baby AT motherboard in the top section of the case to no avail, rendering the whole thing moot. Again my lack of construction capabilities hinders me.

I don't generally talk about my insane escapades into 5.25" floppy land - I realize it's a subject that really makes me look crazy. :p
 
Last edited:
Well you have stuff like what this kid made out of meccano:http://www.youtube.com/watch?v=keYC0Vi3-08

I have a vintage meccano set so I wondered if they still sell it. There are some suppliers that sell to retailers and I found this place in the UK. The place in the UK only opens for business once in a blue moon so not very practical right? Has anybody else found a reliable source for meccano so that drive enclosures can be made?
 
I'm playing with the idea of interfacing an AVR using V-USB to a Backpack floppy. It should be pretty easy and the Backpack has a real (NS 8477) controller in it with registers that can be addressed directly. All you need is a little "glue" logic that twiddles the parallel port lines.

Cheap and simple--and if you're one of the fortunate souls to have taken advantage of the $0.99 eBay on Backpack floppy drives, you're really there on the cheap.

I got a couple of 'em from the guy on the 'Bay last week.
Mine have an FDC37C78. (which looks like another 82077 superset along the lines of the PC8477).

/Bill
 
Seems like the best solution to mounting old drives would be a FUSE filesystem (Linux and BSD variants including Mac OS X) or this under Windows:

http://dokan-dev.net/en/about/

This would avoid the need to write kernel drivers, since the part of the code that talks to the odd peripheral is in user space. There are language bindings for about 20 langs.

There's already a FUSE driver for C64 disk images, just as an example.

Some work to do but once done, your funky drive looks like anything else.

Also, a problem in a user-space driver should not crash your whole box.
 
Back
Top