• Please review our updated Terms and Rules here

Kodak/Verbatim 6.6 MB floppy drive

Chuck(G)

25k Member
Joined
Jan 11, 2007
Messages
39,428
Location
Pacific Northwest, USA
I'll toss some information in here, since I'm very familiar with this drive and the drive that immediately preceded it (the 192 tpi unit found in the Kaypro Robie).

Originally devised by a company called Drivetec in the pre-IBM PC days, this is an example of an embedded servo drive. That is, each track is formatted at the factory is a special servo mark between sectors. Said servo is offset slightly to either side of the track, such that if you can read both servo marks, you know you're right on track. Positioning is done with two actuator steppers via a lever arrangement--one stepper is for coarse positioning; the other is for fine positioning--that is, it adjusts the position of the head once the heads are approximately on-track.

The disk spins at 600 RPM (double speed) and the whole affair is controlled by a 6800-series MCU (which is the reason why it doesn't like fast machines very much). The unit will read and write standard 360K floppies, but at double-speed.

What sent Drivetec into Chapter 7 bankruptcy was that the preformatted floppies were about $15 each (that's in 1970s dollars) and no more durable that a regular floppy. So if you trashed one, it was $15 down the drain. And even at double speed (600 RPM), seeking was slow as was the rotational latency when compared to a standard hard drive.

Kodak was just getting into the computer media business and picked up the remnants of Drivetec for virtually nothing. They brought out the 6.6MB unit that was still on the drawing board when Drivetec went under in hopes that they'd have a captive media market. (Anyone remember Kodak disks and the guarantee that they'd even recover data on one their floppies if it ever failed? Must have been an MBA making that decision).

Because of the embedded servo, I don't believe it's possible to devise a simple formatting program, although you could certainly build a rig to do it if your machine shop skills were up to snuff.

I believe there's a Drivetec manual on bitsavers.
 

Lwall

New Member
Joined
Dec 17, 2013
Messages
7
Hi guys,

I was employee #2 at Drivetec and know (knew 30+ years ago anyway) just about about everything about the the electronics and the formatting of the disks as I developed all the electrical concepts for the drive and designed the electronics for the formatter. I'll try to answer any questions you might have.

Unfortunately, when I downsized four years ago, I tossed all my info about the drives/formatter since I hadn't used it in so many years so I don't have any manuals or documentation.

Questions????
 

Lwall

New Member
Joined
Dec 17, 2013
Messages
7
I've got the reference manual for the 2.8MB one. How was the embedded servo recorded--and in what format?

That must indicate the formatted capacity. The two Drivetec models came in unformatted capacities of 3.3 and 6.6 MB. Marketing hype...

The embedded servo info was written by some pretty heavy duty and precise mechanical gear, which of course, added to the high cost of the media. And the process was fairly slow. Taking several minutes per disk. As noted in another post, the servo bursts were written "straddling" the tracks so when centered over a data track, you would read equal amplitude from each burst. The actual data format was MFM, same as the standard floppies of the day.

One thing I noted in one of the posts, poster said the drive spun at 600 rpm. That was only true when reading 1.2MB or 400/800K disks. The Drivetec disk was read at 360 rpm. The higher bit density and rotational speed meant that data came off the disk at the same data rate as the older 8 inch floppies and twice the rate of the standard 5-1/4 in floppies. We sped up the spindle speed to keep the higher data rate and improve the amplitude we'd get off the older floppies. Remember, our heads were only half as wide so we'd get less signal at the same speed.

The servo burst themselves consisted of a code to tell us which burst we were reading followed by an area of constant frequency that we used to measure the amplitude.

We tried to prevent accidental formatting. We moved the write protect notch to a) prevent writing older floppies; b) make our disks look write protected in other standard drives. We also added logic in the drive to prevent writing that started at the "Index". In a typical format command, the controller waits the the index signal from the drive, then starts writing the whole track. If we see write gate going active at index, we block it.

Given the nature of the servo, you would need a mechanism that is 384 tpi for the 3.3 and 768 tpi for the 6.6 in order to format a disk. I don't see any practical way to do it in a drive. Then there is the electronics, we wrote both sides nearly at the same time to cut down on the format time. That meant we had two sets of format and verification electronics. We used an optical shaft encoder to give us angular position to insure we got the servo burst and data portions in the right place.

Hopefully that helps. I apologize if I misinterpreted your question. Feel free to ping me if you want more. May take me a bit shake out some of the cobwebs.
 

Chuck(G)

25k Member
Joined
Jan 11, 2007
Messages
39,428
Location
Pacific Northwest, USA
I'm familiar with hard disk servo writers--they could be real works of art. I remember one done for Tandon that used a slab of granite and a laser-optical encoder. Interestingly, the positioner wasn't electromagnetic, but pneumatic (we had a bottle of pressurized nitrogen to run it).

Does the MCU (68xx, IIRC) in these things control only positioning, or does it also take part in actual data recorded? In other words, does the drive care if the customer records MFM or FM, GCR, etc.? With the embedded servo, that must imply that sector sizes are fixed at least in physical track length, true?

On the 320, there is a small 8-position undocumented edge connector. What is its purpose?

On the 6.6 unit, there are two banks of 4-position DIP switches. The first bank, like the sole bank on the 320, specifies the drive select. I don't know what the function of the second bank is, can you shed some light?

My blank 3.3 and 6.6 disks were obtained from Verbatim, when they were closing out their Santa Clara operation. Did Verbatim format these or were they processed by Drivetec?
 

Lwall

New Member
Joined
Dec 17, 2013
Messages
7
Our servo writer was pretty much like the hard disk ones but we were able to use a high precision stepper and lead screw. The accuracy was verified with a laser but then we just ran open loop.

The 6805 was used primarily for handling the steppers. Data recording was all handled by the controller. Drive does not care. May be possible to do something with a 1,7 or 2,7 code and get 50% more capacity. It would be tricky, but I'd stick with leaving the header info as is. So, you have to read an MFM encode header and then read/write a GCR or RLL encoded data block. If I had kept any on my old schematics, I could tell you where to get the equivalent a sector pulse. As I mentioned, I tossed everything a couple years back. And yes, the sector length is fixed and determined by the spacing of the servo data.

The small connector was for testing use. Don't remember the specifics.

The second dip switch may have had to do with how we handled the 320 track issue. The floppy controllers only had an 8 bit track register so we had play games. Zero thru 159 on even drive selects, 160-319 on odd, or via and option switch, use another interface line to handle the inner tracks. By the time we got that far, I was running the program at Kodak and some of those design details were handled by other engineers. I got pretty buried in developing an AGC circuit for the read channel and a two stage comparator for the fine stepper.

Most of the disks were formatted by a Kodak group in San Diego. Kodak bought a company down there called Spin Physics. Can't remember if they retained the Spin Physics name or just called themselves Kodak. Verbatim was the marketing and sales arm since most people recognized them media suppliers for a long time.
 

Chuck(G)

25k Member
Joined
Jan 11, 2007
Messages
39,428
Location
Pacific Northwest, USA
Thanks much for information!

Given that the number of blank formatted disks for these things are probably outnumbered by the number of drives still in existence, I've wondered about being able to repurpose them a bit.

A lot of old 5.25" drives developed alignment issues and reading disks written long ago can be a real headache. I keep a few drives around with varying amounts of misalignment built in for this purpose. While it's not possible to do much about azimuth, one can, with a little determination either use a previously misaligned drive or intentionally misalign one to work.

My thought was to directly control the steppers and simply use read success or failure to micro-adjust the position to read old conventional floppies. Do you think that this might be a workable idea?

Again, thanks for the input.
 

Al Kossow

Documentation Wizard
Joined
Sep 1, 2006
Messages
2,817
Location
Silicon Valley
Thanks much for information!

Given that the number of blank formatted disks for these things are probably outnumbered by the number of drives still in existence, I've wondered about being able to repurpose them a bit.

A lot of old 5.25" drives developed alignment issues and reading disks written long ago can be a real headache. I keep a few drives around with varying amounts of misalignment built in for this purpose. While it's not possible to do much about azimuth, one can, with a little determination either use a previously misaligned drive or intentionally misalign one to work.

My thought was to directly control the steppers and simply use read success or failure to micro-adjust the position to read old conventional floppies. Do you think that this might be a workable idea?

Again, thanks for the input.


There are a couple of guys in Philadelphia trying to use one to read Twiggy disks. I sent them some formated media in normal jackets for them to experiment on
a couple of months ago. I also hung onto a few drives from Don Maslin for the same experimentation. It's interesting to learn that the micro is only used for the head
positioner.

I've not pulled a drive apart. How tough is it to get at the heads to clean?
 

Chuck(G)

25k Member
Joined
Jan 11, 2007
Messages
39,428
Location
Pacific Northwest, USA
Head cleaning is easy--just remove the 4 screws holding the shield in place. There's the head, right in front of you. I wish that all drives were that easy to clean. However, when you replace the shield, note that the front and rear edges have a small "lip" bent into them and that you'll need to make sure that the shield's in place before you re-install the screws.

On the PCB bottom of the 6.6 PCB, the second set of switches is labeled DCHG, RDY, 600 and SKCPT. Well, DCHG and RDY are obviously "disk change" and "ready" for output on pin 34. 600 must either enable or disable 600 RPM mode. SKCPT is a bit of a mystery, however.
 

Lwall

New Member
Joined
Dec 17, 2013
Messages
7
Head cleaning is easy--just remove the 4 screws holding the shield in place. There's the head, right in front of you. I wish that all drives were that easy to clean. However, when you replace the shield, note that the front and rear edges have a small "lip" bent into them and that you'll need to make sure that the shield's in place before you re-install the screws.

On the PCB bottom of the 6.6 PCB, the second set of switches is labeled DCHG, RDY, 600 and SKCPT. Well, DCHG and RDY are obviously "disk change" and "ready" for output on pin 34. 600 must either enable or disable 600 RPM mode. SKCPT is a bit of a mystery, however.

SKCPT is Seek Complete. Because there were so many tracks, we provided a means where, with an intelligent controller, you could send the step commands at a high rate, then go off and do other things while the drive was seeking. Polling the SKCPL line let you know when the drive had reached the destination. We even did some slewing. The normal step rate was 3 mSec. If the step pulses came in at a faster rate, we buffered them, then ramped up to about 1 mSec, then ramped back down at the end of the seek. This was probably never used in the PC world as most people used the NEC 765 floppy controller and the max step rate was 2 mSec. The WD chip could issue step pulses at about 800 uSec.
 

Chuck(G)

25k Member
Joined
Jan 11, 2007
Messages
39,428
Location
Pacific Northwest, USA
Ah, "buffered seek", okay, that makes sense.

That's actually pretty old on floppies. At least one 8" floppy that I know of did this (I'll have to look it up). And the Micropolis 1115 5.25" drives did this--no "seek complete" line but rather just dropped READY until seeking was done. Micropolis, in all of their floppies used a leadscrew with multiple steps per cylinder, so seek was pretty slow, compared with the taut-band models. The 1115/1116 used a MOS Technology MCU to handle the job. Ultimately, the slow seek probably killed off Micropolis' floppy business. But Micropolis probably made some of the best-constructed (and expensive) drives in the business.
 

Lwall

New Member
Joined
Dec 17, 2013
Messages
7
Since the head width of the 3.3 is half the width of a 1.2 MB drive, you should get pretty good results without the fine adjust. One of the problems you have with a misaligned drive is noise from adjacent tracks. Not as big a problem with the narrow head.

Controlling the steppers directly is possible. Just have to figure out which phase to use at track zero, to make sure you're at least starting out where zero is suposed to be. We looked at the flag on the carriage, the flag at the back of the course stepper and the which phase the stepper was on. We aligned that combination to the track zero.
 

Chuck(G)

25k Member
Joined
Jan 11, 2007
Messages
39,428
Location
Pacific Northwest, USA
That's very useful, thanks.

One final question; when reading "legacy/normal" 48 tpi or 96 tpi floppies, does the MCU attempt to adjust the head position? In other words, does it try to adjust for the strongest signal?
 

Lwall

New Member
Joined
Dec 17, 2013
Messages
7
No, we just did two steps or 4 steps (for 3.3 MB) per track. Assumption was we should be in the middle of where the track was supposed to be, and since they had trim erase, there should be minimal adjacent track interference.
 

Chuck(G)

25k Member
Joined
Jan 11, 2007
Messages
39,428
Location
Pacific Northwest, USA
Thanks much, Lwall. This could turn into an interesting project.

I was thinking about the ability with fine adjustment to handle 100 tpi floppies, for example.
 

Lwall

New Member
Joined
Dec 17, 2013
Messages
7
That might work. The fine step was 200 uinches so two coarse steps, minus two fine steps would give you 100 tpi, at least for the first track. I did a quick spreadsheet to see if that would work without requiring too many fine steps. Max was 13 and I'm pretty sure there was that much range on the fine stepper.
 

Marmotta

Member
Joined
Feb 3, 2022
Messages
10
Sorry for resurrecting this very old tread, but I also have the Kodak 6.6mb floppy drive plus controller... anyone have the software/driver? Without it is useless :-(
 
Top