Image Map Image Map
Page 1 of 5 12345 LastLast
Results 1 to 10 of 48

Thread: Flash based - MFM Drive Emulator

  1. #1

    Default Flash based - MFM Drive Emulator

    I've been thinking about hard drive solutions for my obsolete MFM needs [PDP-11] a lot lately.

    At first, it seemed replacing the entire MFM setup with a SCSI controller / drive solution, would be easiest, but it seems this is not to be. The market has swallowed up all these cards [which were expensive to begin with] and now speculators are asking ~$800 for the controllers only.

    Just now I started looking for MFM-IDE solutions, and most of these seem focused on attaching MFM drives to IDE controllers. Not what I want.

    Thinking about it, a $20 USB thumb-drive [memory stick] has more storage than my old systems would ever be able to use, is cheap and provides implicit compatibility with my current PC based systems.

    So what I really need, is an MFM-USB adapter, that allows a memory stick to emulate an MFM Hard drive of yesteryear.

    Even at low quantities, it ought to cost under $100 and be infinitely less obsoleteable because it's memory is based on FLASH, not rotating media.


    Anyone else like this and want to comment or contribute?

  2. #2
    Join Date
    Jan 2007
    Location
    Pacific Northwest, USA
    Posts
    25,232
    Blog Entries
    20

    Default

    Take a look at Phil Pemberton's Disk Ferret--it may have some of what you need.

    Another tack would be to take a look at some of the floppy emulators (e.g. HxC) and soup them up with a microcontroller that's 10x faster and has the capability as a USB host (for the pen drive) (say, an ARM). Of course, you'd need to provide the appropriate logical interface and there would be a pile of programming to do.

    Either way, it's certainly doable, though I'm pessimistic about the $100 price.

  3. #3

    Default

    Thanks for the reference...

    What Phil's done there is actually harder to do, although it involves a lot of the same elements. [MFM & USB] Looks like he's taken the same kind of hardware path too [I didn't find a schematic though].

    Phil has to power an actual MFM drive... we don't. We only need enough power to operate 1 USB device and the MFM interface to the controller. We take in regulated power intended for the MFM drive we're replacing, which ought to be plenty.

    I figured a PIC would need to be coded to emulate the MFM interface, [this I can handle] and then something to translate the logical requests to a format that could be stored and retrieved from the memory stick. Plus another device do the USB. [I'm hoping off the shelf here]

    As for price... it's all a matter of quantity. I figure a small PCB, 3 PICs at $10 ea... under a hundred seems doable. I need to re-establish my relationship with a PCB prototyping house. I have the software to make schematics and layouts. I may be able to breadboard a prototype and do the PCB only once. Hard to say.

    I expect the hardest part will be understanding the MFM stream to emulate the DEC RQDX3's expectation of RD53 and RD54 drives. I may need to make an MFM "sniffer" to watch the RQDX interact with real drives in order to figure it out. Fortunately, I have XXDP ZRQ?? sources to guide me somewhat. I also have RQDX1 and RQDX2 controllers here, to expand upon compatibility tests once it works.

    Accessing the memory stick ought to be dead simple.

    It's not a "trivial" project, to be sure. But with so many of us "hackers" out there that these CQD-220 cards are becoming sought after, and original MFM drives expiring, it seems like there ought to be enough of us interested to make a go if it. After all, SCSI won't be around forever either.

    I'll start combing PIC app notes for MFM examples... who knows, I might get lucky.

    Otherwise, I guess I'll dig out some old PC/AT schematics. Come to think of it, I had a buddy who did consulting work for QUANTUM and MICROPOLIS... I have his system here. It has all his schematics and layouts. Might be help there too.

    Of course, I need to add this to my list of other s__t going on in my life. What the heck... gotta have a little fun somewhere.

  4. #4
    Join Date
    Jan 2007
    Location
    Pacific Northwest, USA
    Posts
    25,232
    Blog Entries
    20

    Default

    A PIC, IMOHO, is fairly underpowered. While it can be pressed into action for a floppy, you're dealing with 10x the data rate on an MFM drive (15x if the drive's RLL).

    If you want to try, take a look at what it takes to play a USB host rather than a USB slave for your USB pen drive. It's not trivial, although I suppose you could try to use a Vinculum to do it.

    Personally, I'd do the work with an ARM-based CPU. Faster clock, bigger registers, better addressability and some have UHC capabilities built in.

    But I still think you're overly optimistic and you might be over-estimating the demand for such a beast.

  5. #5
    Join Date
    Mar 2011
    Location
    Atlanta, GA, USA
    Posts
    1,123

    Default

    I've been thinking about a similar project recently but at a controller level rather than drive. The problem with vintage drives is that they are 'vintage' with worn bearings and sludge for lubricant. My need is more in the PC realm so I was thinking about what it would take to decode the ISA bus and emulate a ST-506 or vanilla IDE interface with a FPGA and a MCU on the backend.

    What you're suggesting is actually simpler and could probably be done using a Cypress FX2 chip tailoring it's GPIF interface to the MFM drive standard (which auto-bulk transfers in/out on the USB side). Is there a resource where I can read up both the electrical and protocol standard for MFM drives?

  6. #6

    Default

    Quote Originally Posted by eeguru View Post
    My need is more in the PC realm so I was thinking about what it would take to decode the ISA bus and emulate a ST-506 or vanilla IDE interface with a FPGA and a MCU on the backend.
    When it comes to IBM PC's and XT's, the XT-IDE project is a proven solution. An inexpensive, simple, easy to build ISA IDE controller that works with any normal 16 bit IDE drives - including compact flash adapters, etc.

    The problem is really with other platforms with more complicated bus structures - where designing a controller from scratch would be too complicated. For example, my AT&T 3B2. The disk controller is part of the motherboard - and can only talk to MFM disks. Sure, a SCSI card exists, but it's too hard to find, and even then it's not all that great of a SCSI controller. A bridge allowing me to use an IDE disk in place of an MFM drive would be ideal.

    I definitely do believe a modern microcontroller is up to the task of decoding and emulating an MFM disk drive, and using soemthing more reliable (IDE or flash memory) for storage. The MFM interface is fairly simple, since the drive is completely unintelligent (step, step, goose). A microcontroller would simply need to keep track of whatever track/sector the host computer is looking for, and read/write to it over the data line. Granted, in application it's much more complicated, but still... I do believe it's doable.

    -Ian

  7. #7
    Join Date
    Mar 2011
    Location
    Atlanta, GA, USA
    Posts
    1,123

    Default

    Quote Originally Posted by RetroHacker_ View Post
    When it comes to IBM PC's and XT's, the XT-IDE project is a proven solution. An inexpensive, simple, easy to build ISA IDE controller that works with any normal 16 bit IDE drives - including compact flash adapters, etc.
    Thanks for the link. Though I was also planning on EMS/XMS emulation backed by DDR connected memory. And I'm learning Verilog at the moment, just bought a hot air rework station, and looking for exercises

    The main problem I see with any standard MCU is the 5 MHz serial bit rate of MFM (and 7.5 MHz for RLL). You can't expect most MCUs to be able to keep up with that. Even if you serialize it on the USB host (as you would with even a Cypress FX2), it's going to be pretty taxing.

    I supposed you could try and work some 8:1 LVDS transmitter/receiver chips and some level shifters in to some of the work.
    Last edited by eeguru; March 14th, 2011 at 02:18 PM.

  8. #8

    Default

    Seeing the comments that have ensued from my first post, I'm not sure we're all talking about the same project any more, as you're responding to one another's posts.

    I'm talking about a device to make a USB memory stick "look like" a familiar MFM drive to an ancient hard disk controller card.

    To break this down a little more....


    • 5mhz signaling rate is not sustained. MFM drives have substantial seek times, relatively high rotational latency, and low block sizes. Consider 5mhz a "burst" rate.


    • The result being that the 5mhz / 200ns time is well within many PICs. Specializing the task into several devices makes both the timing and programming simpler.


    • I had no intention of doing RLL drives. My primary interest is in providing DEC QBUS controllers with suitable solid state replacements for the MFM drives they expect.


    • There is no point in using todays IDE drives in this instance. They have so many times more storage that it is totally wasteful. Even today's memory sticks [at 1 GB] have hundreds or thousands of times more capacity than these older systems know how to use. Even the smallest, cheapest memory stick has excess capacity.


    I consider this approach far less expensive than designing a QBUS controller. I've done that before, and it presents substantial difficulties and costs. I expect it would end up costing far more than a number of QBUS SCSI cards at $700 each.

    It sounds from your posts, that others are thinking in collateral terms and at least, we could share information if not more.

  9. #9
    Join Date
    Jan 2007
    Location
    Pacific Northwest, USA
    Posts
    25,232
    Blog Entries
    20

    Default

    Sounds good, let me know when you have a prototype for testing...

  10. #10

    Default

    Quote Originally Posted by Chuck(G) View Post
    Sounds good, let me know when you have a prototype for testing...
    Ha ha.... I suspect there is irony in that suggestion.

Bookmarks

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •