• Please review our updated Terms and Rules here

Hooking up an 8" floppy to a modern PC

Malvineous

Experienced Member
Joined
Jun 18, 2010
Messages
119
Location
Brisbane, Australia
Hi all,

I just stumbled across this forum and it looks really interesting! The reason I am here is that I have obtained an 8" floppy drive and I would like to see if I can hook it up to a modern PC so that I have a method of reading old 8" disks.

The drive is a Shugart 800-2 and I found the manual on bitsavers.org, which tells me I need to power it with 240VAC (!) and 24V, 5V and -5V DC. Is there anything I need to be aware of when powering up this drive, or can I just get a 24V supply and a couple of voltage regulators and all will be good?

It looks like I can use an FDADAP to connect it up to a standard PC floppy controller (assuming said controller supports 8" drives.)

Does this all sound right? I'd hate to break it because there were some tricks I wasn't aware of!
 
Yup, just make sure that your 24V supply can deliver about 2 amps. I'm assuming that you've got a 50 Hz drive there and not a made-for-USA 60Hz unit--the spindle motor is mains frequency-sensitive--which is what the 240VAC is connected to.
 
Thanks for the quick reply! The AC motor says "230V 60-50Hz, 1800-1500RPM" on it. So I take that to mean running it at 50Hz should produce 1500RPM. I assume that's right because the local power is 240V 50Hz and the drive used to be used locally.

EDIT: Oh and one other thing, does anyone know what the name of the power connectors are on the drive? It'd be nice to get some mating connectors if they're still available so I don't have to modify the drive.
 
The end result with the motor is that the disk should spin at 360 rev/minute. Let's assume that you don't have a weird 230V/60Hz unit.

As to the power connectors, it's still the same as what the OEM manual calls out. For the 6-pin DC power, it's an AMP 1-480270-0 (be sure to get pins for it when you order). For the 3-pin AC power, it's an AMP AMP 1-480303 "Mate-n--lock". Most of the big supply houses (Mouser, DigiKey, Newark, etc.) carry them.
 
Sorry about taking so long to follow this up! I've found all the connectors (many thanks for the info) but my lack of electronics experience is proving tricky with the power supply. The easiest way (I can see) is to buy an off-the-shelf PSU that delivers +24VDC as well as some negative voltage, then use an LM7805 regulator to get +5V and an LM7905 to get -5V. Unfortunately it's proving rather difficult to find a PSU capable of delivering both +24VDC @ 3.4A (to power two drives simultaneously) as well as delivering some negative voltage to feed in to the -5V regulator.

Another option would be to generate the -5V directly off the +24V line, but I can't see any way this can be done with an easy off the shelf IC like a voltage regulator (since the LM79xx series requires a negative voltage to begin with.)

Can anyone suggest an easy (cheap) way that one might go about generating these voltages? According to the manual each drive requires +24VDC@1.7A, +5VDC@1A and -5VDC@0.07A.
 
Re the minus 5 volts.

1. There are voltage regulator chips out there that do positive to negative conversion.
http://focus.ti.com/lit/ds/slts041a/slts041a.pdf is an example.

2. An alternative is to acquire an AC to +5V DC regulated plug pack (wall wart) and treat the postive pin as negative, and the negative pin as positive. That will work if the plug pack contains a 'proper' transformer, not an auto-transformer. I would think that to meet double-insulation standards, etc., that all plug packs have 'proper' transformers, but someone reading this may know better.
 
I don't know if a linear 7805 would supply enough current at +5V if dropped from +24--almost 80% of the power would end up in heat--you'd probably need an LM323K and a generous heatsink.

But why stay stuck in the 1970s? The LM2576 "simple switcher" regulators are inexpensive, efficient and can even be used to get your -5V from a positive supply. Buy one of those cheap 24VDC switchers made for CCTV off of eBay and you're set to go with two LM2576s and a handful of parts.

You could also peruse Jameco's selection of surplus power supplies and probably find something suitable for under $20.
 
Yes, I got it working and I can image 8" floppies, but I am using a Kryoflux to do it. See my announcement on the Kryoflux forum for more detail on the set up.

Given the Kryoflux talks to standard PC floppy drives, there should be no reason why you couldn't connect one up to a normal PC. I don't think a PC floppy controller could read many of the 8" floppy formats out there though, so it would depend on what disks you wanted to read. A Kryoflux will read everything.
 
I don't think a PC floppy controller could read many of the 8" floppy formats out there though, so it would depend on what disks you wanted to read. A Kryoflux will read everything.

Not really true on both marks. Most late PCs with floppy controllers have good succes in reading both MFM and FM floppies. A Kyroflux, like a Catweasel, can be made to record domain transistions, but interpreting these is up to software--if none exists for the floppy you're trying to read, is the floppy being "read"? How about a Drivetec-produced floppy--or a Lisa "Twiggy" ?

There's a program wandering around that will test your PC's controller for its ability to handle FM. No extra drive needed--just use a 3.5" HD drive and get a thumbs-up or down. If a system can handle a 3.5" HD floppy, it will handle an 8" drive.
 
Yeah, what Chuck said. You need an adapter to deal with the different interface pinouts but IMO you've got as good as or a better chance of reading an 8" diskette on a PC than many of the 5 1/4" formats, if only because there were so many more incompatible formats used on 5 1/4 than on 8" disks, many of which followed the original IBM 'standard'.

Mind you, there were certainly some good programs out there to let you read all those different formats.

Info on Dave D.'s site regarding making an 8" adapter and testing the FDC's single-density capability:
http://www.classiccmp.org/dunfield/

Scroll down to software/images.
 
While MFM floppies are common in 5.25" and 3.5" formats, the 8" drive came around before a lot of that was standardised. I believe the sectors on an RX02 disk for instance, start off with FM for the sector header, then switch to MFM (or even MMFM) for the rest of the sector. That had the Kryoflux team scratching their heads for a while! I don't think a PC floppy controller could read that. I think the bitcell width was nonstandard too, so they had to add another option to the FM decoder to handle it. Then of course there's nothing stopping you from writing Amiga style data or some other non-PC format onto an 8" disk (the Kryoflux can do this.) While this was probably never done for real, if it was, a PC controller wouldn't be able to read it. Basically what I was trying to say was that there are so many low-level variations with 8" formats, there's no guarantee it would work, unless you know for sure the disk you want to read is in a similar format to a PC floppy.

With regard to interpreting the magnetic flux/domain transitions to get actual data out, the Kryoflux team are very willing to write decoders for new formats, if you can send them the raw dumps. And of course there are plans to add write support for raw dumps, so even if you can't "read" the floppy, you can still make a very accurate copy of it onto a new disk.
 
With regard to interpreting the magnetic flux/domain transitions to get actual data out, the Kryoflux team are very willing to write decoders for new formats, if you can send them the raw dumps. And of course there are plans to add write support for raw dumps, so even if you can't "read" the floppy, you can still make a very accurate copy of it onto a new disk.

Oh, I've got some doozies. I submit that the people at Kyroflux haven't seen half of them. 8" HS GCR (with no clue as to what the GCR groupings actually are), 8" HS MMFM with funny bit orderings, 12 bit bytes... RX02 " funny" MFM isn't awful at all.

As far as reproducing the floppies accurately, how do you know? Does the KF somehow automatically compensate for quantization errors and oddball write precomp schemes? The world isn't made up exclusively of common formats--there is some very bizarre stuff out there. Before LSI controllers (and commodity CPUs), the human mind really showed its ingenuity. There was no thought given to interchangeability.
 
Last edited:
I can't see how it's any different from a Catweasel or DIskFerret or Deviceside or any other sampler that can be built with at most an FPGA, an MCU and a bit of RAM. A couple of years ago, I posted a DIY version using a fairly low-end AVR 8-bitter. Got as much interest as rotten sardines in a perfume factory.

Basically, the idea is that you time transitions between pulses and record the intervals. Most modern MCUs have "capture" facilities for doing this (and PWM facilities for writing). You make note of index holes along with the sample data.

You can make guesses based on histogram-type data. For example, this histogram:

Code:
  0      0     0     0     0     0     0     0     1     0     0
 10      0     1     8    17     8    21    34    46    36    18
 20     12     7     4     3     1     2     3     0     3    25
 30     11     5     6    11     3     2     4     1     3     4
 40      3     4    10     5     3     4     7     5     6     9
 50    182  1343  3109  3214  1463  1328  1957  1410   433    46
 60      3     1     1     3     2     1     1     0     0     2
 70      1     3     1     2     0     1     0     3     2     2
 80      2     1     2     0     3     0     1     3     1     2
 90      1     0     1     1     1     2     1     2     3     3
100     17   110   247   430   554   694  1317  3235  6292  6341
110   5610  4648  1383   134     1     0     0     0     1     0
120      2     2     1     1     3     3     0   455     0     0

(the clock period is on the left hand colum; the remainder are counts) This tells you that there are basically 2 frequency peaks, meaning, in all probability, that the disk is FM-encoded. You are left to deduce the addressing structure, bit order, data representation and word length on your own--and then, you are left with decoding the filesystem organization. The sample in question comes from a Fairchild Sentry ATE HS floppy. In many cases, you don't even know the content of the disk--and equipment to use it has long been scrapped. All you have is a disk and perhaps the name of the equipment used to create it--sometimes not even that (I recall one episode when retrieving some information for a big aerospace company that I was able to tell them the equipment used to create the data--they had never heard of it!)

The point is that it's often not as trivial as others make it sound. It's fine if what you're dealing with is a common piece of gear, but can get rather involved if no one has a clue.
 
...Basically what I was trying to say was that there are so many low-level variations with 8" formats, there's no guarantee it would work, unless you know for sure the disk you want to read is in a similar format to a PC floppy.
... And what I was trying to say is that there are far fewer 'real' 8" formats (I haven't seen many 8" Amiga disks) than 5 1/4" and that many of them can be read (and written) with a suitable PC, unlike many 5 1/4" formats like Commodore, Apple, etc.

As Chuck can confirm, there are several programs that will read and interpret data on many different format disks as is; not to disparage Kryoflux which is certainly useful for certain situations, but this sort of thing has been done for decades before Kryoflux came along and there are many alternatives, both for reading at the data level and also the flux transition type.

@ Chuck: Any rough idea how many 8" formats there were, and how many 5 1/4" ?
 
Hi

I am hoping the soon to arrive XT-FDC project will be able to help with most of the common 5.25" and 8" FM and MFM formats. The prototype board includes headers for both kinds of floppy drives. That doesn't mean it is complete but should handle most cases.

Chuck did propose a project several months (years?) ago to use an MCU as a FDC as bit sampler. I recall the discussion and also a later discussion to make a virtual floppy drive using an MCU to serve disk images from an SD board. We did discuss implementing the projects in both cases but nothing ultimately came of either.

Where these projects go wrong is that they usually require advanced MCUs which mean larger chip packages and inevitably SMT. It is not that the boards cannot be implemented. Actually with a solid schematic making the prototype PCBs is relatively easy even using multiple SMT devices. The problem comes in that these are specialty devices similar to the Catweasel and Kyroflux. There is just not a lot of demand for them and once they require SMT it really narrows the DIY hobbyist participation even more.

depending on how the XT-FDC project pans out maybe we can use a new model for community projects. I am willing to support most any project but it has to be a hobbyist friendly to support enough PCBs to make the project economically viable. However even if we could just get the prototype boards made, built and tested, etc then at least we would get some real information whether the projects are *technically* viable.

Its worthy of some community discussion. Spread the work and spread the risk is essential though. Unless you are loaded with a ton of disposable income no one is going to sponsor projects "out of hide"

Thanks and have a nice day!

Andrew Lynch
 
Back
Top