HSFE (or how to use 3.5" drives with your H17 controller)

[Les Bird]

Greetings,

 

Chris Elmquist mentioned in another post something he developed and that we've been testing the last couple of weeks. This fantastic little device is called the HSFE (Hard Sector Floppy Emulator). It's a little gadget that sits between the H17 (or H88-1 for the 89) controller card and a couple 3.5" floppy drives. You can use 3.5" media just like your 5.25" hard sector disks and the 3.5" floppies are a LOT cheaper (50 pack for $25 or less), much more plentiful and more reliable then the hard sector floppies. If you run BIOS80 in CP/M or the HUG SY driver in HDOS you can store up to 400K per disk.

 

The HSFE does this by simulating the index pulses for a 10 sector disk so that the controller "thinks" it's working with a 5.25" hard sector disk.

 

I've been banging on this thing for a couple weeks now and it works great.

 

CP/M and BIOS80 are a bit glitchy. I'll occasionally get BDOS Select errors but I'm not entirely sure it's the HSFE that's at fault here. If we have time we'll look through the BIOS80 source and see if there's something that can be done to improve this but for now I just live with it. Once the drives get past the select error, reading and writing to them is flawless.

 

Under HDOS the thing is bullet proof! I've not had one problem with it under HDOS.

 

I purchased four NEC 3.5" drives from Newegg for $5.50/ea. and a pack of 50 floppies from Office Depot for $25. Cheap, reliable storage for our vintage computers.

 

Attached here is the silk screen layout of the HSFE which has been submitted to my PCB fabricator. We should receive the PCBs in about 2 weeks. I'll post another message with more details when the PCBs arrive.

 

- Les

 

 

[Mark Garlanger]

That sounds great. Were you able to use standard 1.44M floppy drives? Is it possible to use standard 1.44M drives with the soft-sectored controller? I was able to find one 720k 3.5" drive for the soft-sectored controller and have been looking for a second one, but they have just been going for way too much. If I could use a standard new drive, that would really help.

Mark

[Les Bird]

That sounds great. Were you able to use standard 1.44M floppy drives? Is it possible to use standard 1.44M drives with the soft-sectored controller? I was able to find one 720k 3.5" drive for the soft-sectored controller and have been looking for a second one, but they have just been going for way too much. If I could use a standard new drive, that would really help.

Mark

-

 

Mark,

 

You can use standard 1.44M floppy drives but they must be able to support 720K floppies in order to work with the HSFE. What we've been doing is taking the normal 1.44M HD floppy disks and putting black tape over the high density hole so that it looks like a 720K floppy to the drive. The drive must be able to support this to work with the HSFE. Same applies for the H37. I've been able to use standard 3.5" floppy drives with the H37 but only after taping over the high density hole on the 3.5" floppy disk. The high density hole looks like the write protect slider tab but it's on the other side of the disk.

 

My bet is you could get one of those 3.5" floppies for $5.50 from Newegg and it'll work with your Z89-37 controller.

 

- Les

 

[Chris Elmquist]

On Tuesday (07/07/2009 at 08:43PM -0500), Mark Garlanger wrote:
> That sounds great. Were you able to use standard 1.44M floppy drives? Is it
> possible to use standard 1.44M drives with the soft-sectored controller? I
> was able to find one 720k 3.5" drive for the soft-sectored controller and
> have been looking for a second one, but they have just been going for way
> too much. If I could use a standard new drive, that would really help.

yes... you sure can. You take a piece of black tape and cover up the
density select hole on the 1.44 MB media. This causes the drive to run
in 720K (250Kbps) mode which is compatible with the H17 format.

You do have to be a little careful choosing the 3.5" drive though-- some
have been made so cheap that they leave out the density select switch/opto
sensor and are hardwired for 1.44MB only. Samsung drives are an example
of this. Between Les and I, we have used Sony, Teac and NEC drives without
issue. We'll try to pull together a list of successful candidates when we
release the doc package for this little board.

Note too that all of this work has been done with 3.5" drives and not
5.25". I believe that the speed tolerance on 5.25" drives is too sloppy
for this to be a success. 3.5" hold their speed much more accurately
and therefore the pseudo pulses stay aligned as they should.

Chris

--
Chris Elmquist

[Mark Garlanger]

I actually picked up a box of 100 DD DS 3.5" disks a few years ago when I got the drive. So with the right drive, I'll be all set.

Thanks,
  Mark

[dwight elvey]

________________________________
> From: les...@bellsouth.net

Hi
I believe the problem is a timeout waiting for the index
pulse. Unlike the older drives that had a drive-ready signal,
the newer drives use this pin for disk change.
This means that a controller that is waiting for a fixed
time after starting the motor before seeing an index pulse
may be timing out.
The newer drives hold the index pulse until the drive feels
the disk is up to speed. Rather than being a fixed time,
it is some number of revolutions before the index pulse
is seen. This is often after the software has timed out.
At least this is my thinking. You might try starting
the disk in different positions to see if that makes a
difference in if it times out.
Dwight

_________________________________________________________________
Lauren found her dream laptop. Find the PC that’s right for you.
http://www.microsoft.com/windows/choosepc/?ocid=ftp_val_wl_290

[Chris Elmquist]

On Tuesday (07/07/2009 at 10:11PM -0700), dwight elvey wrote:
>
> Hi
> I believe the problem is a timeout waiting for the index
> pulse. Unlike the older drives that had a drive-ready signal,
> the newer drives use this pin for disk change.
> This means that a controller that is waiting for a fixed
> time after starting the motor before seeing an index pulse
> may be timing out.
> The newer drives hold the index pulse until the drive feels
> the disk is up to speed. Rather than being a fixed time,
> it is some number of revolutions before the index pulse
> is seen. This is often after the software has timed out.
> At least this is my thinking. You might try starting
> the disk in different positions to see if that makes a
> difference in if it times out.

Hi Dwight...

I tried to simulate this-- in all sorts of different ways-- generating
pulses immediately when the drive was selected, keeping the index active
for a entire revolution, numerous other tricks and just could not really
improve the situation.

I still get these BDOS errors even with an original 5.25" drive and
10-hole media in use so I am real tempted to blame BIOS-80.

Will dig into that code here sometime soon and see if there are any
tweaks we can make to it to improve the robustness.

Chris

--
Chris Elmquist

[Herbert Johnson]

On Jul 7, 10:11 pm, "Les Bird" <lesb...@bellsouth.net> wrote:

>
> Mark,
>
> You can use standard 1.44M floppy drives but they must be able to support 720K floppies in order to work with the HSFE. What we've been doing is taking the normal 1.44M HD floppy disks and putting black tape over the high density hole so that it looks like a 720K floppy to the drive. The drive must be able to support this to work with the HSFE. Same applies for the H37. I've been able to use standard 3.5" floppy drives with the H37 but only after taping over the high density hole on the 3.5" floppy disk. The high density hole looks like the write protect slider tab but it's on the other side of the disk.
>
> My bet is you could get one of those 3.5" floppies for $5.50 from Newegg and it'll work with your Z89-37 controller.
>
> - Les

There's two kinds of issues, in using 3.5" drives to replace 5.25-inch
drives. I identify these on my Web site page about floppy drives, I'll
provide a link below. These are technical issues which can be
confusing; they confuse me because I only look at them every few
years.

Very briefly, one issue is how IBM changed the rotation rates on 3.5-
inch drives from old "standard" rates to a new speed - but they also
changed the rate data bits were written to compensate. It's confusing
to discuss without a lot of detail, but watch out for running the 3.5"
drives at the "wrong" rotation speed. Some drives will run at the
"right" speed for 5.25" DD work (300 RPM), some will not. Some can be
set up properly by changing jumpers.

Very briefly, the other issue is using HD (high density) diskettes at
DD (double density). These disks are made with DIFFERENT magnetic
coatings, they expect to be "used" at the appropriate data rates. You
certainly cannot use DD diskettes at high density - they won't even
complete formatting! But running HD disks at double density, is
running them at too "low" a density (speed), and also writing to them
at the "wrong" level of magnetic "signal". They may not be reliable
over the longer term. This is harder to verify because they "work" at
least initially.

I discuss this on my Web page at:

http://www.retrotechnology.com/herbs_stuff/drive.html

in various sections as seperate topics. But I don't have a
comprehensive section like "using 3.5" HD floppy drives and diskettes
to replace 5.25" double density drives and diskettes". It was not an
issue until stores stopped selling double-density diskettes. I'm sorry
I don't have it as a tidy document.


Herb Johnson
retrotechnology.com

[Herbert Johnson]

On Jul 13, 9:35 pm, Herbert Johnson <herbrjohn...@gmail.com> wrote:

> Very briefly, the other issue is using HD (high density) diskettes at
> DD (double density). These disks are made with DIFFERENT magnetic
> coatings, they expect to be "used" at the appropriate data rates. You
> certainly cannot use DD diskettes at high density - they won't even
> complete formatting! But running HD disks at double density, is
> running them at too "low" a density (speed), and also writing to them
> at the "wrong" level of magnetic "signal". They may not be reliable
> over the longer term. This is harder to verify because they "work" at
> least initially.

I'm in discussion now, with someone who has done data recovery and
media conversion for many years. I hope to have a comprehensive
technical response soon. I'll post a Web link to my site with that
discussion. If I or they can identify ways to "use" HD diskettes in DD
format, I'll do so. But my recommendation is to find DD diskettes,
preferably name-brand ones, and use those at DD - and make sure your
3.5" drive is set up properly for DD use as a 5.25-inch drive
replacement.

P.S. I use my "gmail" email address for write-only, when I make public
posts. That directs any spam to gmail, which I hardly ever read. If
anyone wants to contact me privately, use my "retrotechnology" email
address, on my Web site or as listed below.

Herb Johnson

Herbert R. Johnson, New Jersey USA
http://www.retrotechnology.com/ retro-technology home pages
-- S-100, CP/M history by "Dr. S-100"
-- other old tech in iron, glass, rock
domain mirror: retrotechnology.net
email: hjohnson AAT retrotechnology DOTT com

[Herbert Johnson]

There's been no comments to my post here, on cautions about using 3.5"
HD disks as DD diskettes. If no one is interested, I'm sorry if I've
annoyed anyone about a practice no one is troubled about. The bottom
line seems to be, you can get away with it - but what happens when
those disks get old is not clear, however.

My technical discussions with Chuck Guzis of Sydex Inc are at the Web
page linked below:

http://www.retrotechnology.com/herbs_stuff/guzis.html

Chuck originally developed Anadisk and 22disk, which were used on MS-
DOS systems to copy and analyze CP/M and other diskettes. His company
does data recovery from diskettes and other media.

He says that HD diskettes are normally written at higher currents and
faster data rates. Also, the magnetic signals from HD diskettes are
weaker than the signals from DD diskettes. Consequently, using HD
diskettes as DD diskettes may result in disks that may be more likely
to lose data over time or be less readable on other drives. He also
cautions that you should use unformatted or degaussed (bulk erased) HD
diskettes.

(No one asked, but he mentioned that the differences with 5.25" DD and
HD disks are much greater. Don't mix those two types up.)

But, his practices are to just use old-stock, name brand, 3.5" DD
diskettes for DD use - he has plenty of them. And, my experience is
that one can find DD old-stock diskettes, and even recent-made no-name
DD diskettes. Chuck did say, he thought today's 3.5" HD diskettes were
inferior to name-brand disks of the past. I"ve found the same to be
true for 8-inch diskettes. Bottom line: old name-brand are better than
old no-brand or new lesser-name brand.

[Les Bird]

Herb,

Your long detailed posts are very informative. Don't stop. You are not
annoying anyone.

Currently Chris and I are the only 2 using the HSFE at this time. I read
your previous post and although I didn't reply it was very good advice, as
is this post. This is something I would've never thought about before.

Before I get into heavy use with the HSFE I'll see if I can find a good
source for DD disks.

- Les


--------------------------------------------------
From: "Herbert Johnson" <herbrj...@gmail.com>
Sent: Tuesday, July 21, 2009 1:27 PM
To: "SEBHC" <se...@googlegroups.com>
Subject: [sebhc] Re: HSFE (or how to use 3.5" drives with your H17
controller)

[Les Bird]

After a quick search here's what I found for some sources for 3.5" DD disks:

http://www.oldsoftware.com/floppy_disks.html
http://www.wdatamedia.com/dd.htm (minimum order 500 pcs but at .18/ea that's
$90)
http://www.floppydisks.com/items.php?itm=5&pgnum=1

- Les

[Gene Buckle]

Athana (http://www.athana.com) also sells 3.5" DSDD media, but they're
not NOS or chinese imports, they're new-manufacture media. Great
quality too. (Yes, I'm a customer. :) )

g.

[Chris Elmquist]

Herb,

I appreciate your research too. I really didn't have any intentions
on using the 3.5" media as an archival media. My goal was to just get
something to use in the machine that was more reliable than the 10-hole,
hard sector media over the short term. The 10-hole stuff has become
nearly unobtainium now and also doesn't allow us to do too much with
only 100K storage.

With the hard sector emulator, a 3.5" drive and BIOS-80 under CP/M,
we are getting 400K on a disk and that's pretty nice for hacking around.

I had hundreds and hundreds of HD 3.5" on hand and so those became the
target media. I would have had to find and buy DD ones and didn't really
see the need for this particular exercise.

I would hope that the long term archiving of these disks and software
happens on modern media... hard drives, DVDs, and the Internet... and we
just transfer from those archives to the floppies when we want to use the
stuff. Granted, you will need a robust boot disk to get the game started.

Chris

--
Chris Elmquist

[Les Bird]

>
> Herb,
>
> I appreciate your research too. I really didn't have any intentions
> on using the 3.5" media as an archival media. My goal was to just get
> something to use in the machine that was more reliable than the 10-hole,
> hard sector media over the short term. The 10-hole stuff has become
> nearly unobtainium now and also doesn't allow us to do too much with
> only 100K storage.
>
> With the hard sector emulator, a 3.5" drive and BIOS-80 under CP/M,
> we are getting 400K on a disk and that's pretty nice for hacking around.
>
> I had hundreds and hundreds of HD 3.5" on hand and so those became the
> target media. I would have had to find and buy DD ones and didn't really
> see the need for this particular exercise.
>
> I would hope that the long term archiving of these disks and software
> happens on modern media... hard drives, DVDs, and the Internet... and we
> just transfer from those archives to the floppies when we want to use the
> stuff. Granted, you will need a robust boot disk to get the game started.
>
> Chris
>

Unobtainium? I think you've watched The Core too many times... ;)

- Les

[dke...@hotmail.com]

On Jul 21, 12:41 pm, Chris Elmquist <chris.elmqu...@gmail.com> wrote:
> Herb,
>
> I appreciate your research too. I really didn't have any intentions
> on using the 3.5" media as an archival media. My goal was to just get
> something to use in the machine that was more reliable than the 10-hole,
> hard sector media over the short term. The 10-hole stuff has become
> nearly unobtainium now and also doesn't allow us to do too much with
> only 100K storage.
>

Hi
I punch my own 10 hard sectored disk.
Dwight

[Chris Elmquist]

On Tuesday (07/21/2009 at 05:29PM -0700), dke...@hotmail.com wrote:
>
> Hi
> I punch my own 10 hard sectored disk.

Did you build a punch/jig of some sort then?

Still have to come up with 5.25" soft-sector disks to punch which isn't
so easy anymore either...

I could use a punch for 16-hole hard sector :-)

Chris
--
Chris Elmquist

[dke...@hotmail.com]

On Jul 21, 6:03 pm, Chris Elmquist <chris.elmqu...@gmail.com> wrote:

> On Tuesday (07/21/2009 at 05:29PM -0700), dkel...@hotmail.com wrote:
>
>
>
> > Hi
> >  I punch my own 10 hard sectored disk.
>
> Did you build a punch/jig of some sort then?
>
> Still have to come up with 5.25" soft-sector disks to punch which isn't
> so easy anymore either...
>
> I could use a punch for 16-hole hard sector  :-)
>
> Chris
> --
> Chris Elmquist

Hi Chris
Yes, I had an old SA400 frame that someone had removed the
electronics from. I maked some blocks of aluminum to replace
the LED and senser for the index hole. I then drilled then through
with a #93 drill ( as I recall the number ). I sharpened the shank
of the drill bit to be a two fluke punch.
I use a grear to index the rotation of the flywheel and a center
drill to mark the flywheel with the 11 index positions.
To punch, I'd put the index hole in the window and as I clamped
the disk into the drive, I'd slip the drill bit through the index hole
and the punch blocks. I'd also align a pointer to the index mark
on the flywheel. Then I'd move the flywheel to each of the ten
locations and punch with the shank end of the drill bit.
I've used these disk on both my Poly 8813 and the H89/8.
I expect to use it on my N* as well.
I'd need another SA400 frame to do 16 hole on but the same
method should work.
I've been intending to make a mechanincal detent with a ball
and spring for the holes in the flywheel but so far, I've just
been using a pointer and eye ( and magnifying glass ).
I'd suspect that for 16 holes you'd need a detent. By eye
works well enough for 10 hole. One could connect a stepper
motor to the flywheel ( I wonder what the steps were on the
stepper from the drive ).
Dwight

[West, Ronald S.]

I just tried to log onto the www.sebhc.org/archive page and the old
account/password combo didn't work for me. Matter of fact after my
initial try it wouldn't prompt me for a password anymore. I exited all
copies of IE but still the same issue.

Does anyone know the status of that site?

Ron

[Les Bird]

Ron,

As far as I know the website is in limbo right now. If you're looking for a
file check my website which is a mirror of your old FTP but with new stuff
added.

- Les
http://www.lesbird.com/sebhc

[Chris Elmquist]

Very cool Dwight. If we put a stepper to precisely rotate the media then
we should put a solenoid on the punch too and the whole thing could be
automated :-)

Gee, maybe there is a market for these!? oh, wait...

In seriousness though, might also consider a "turns counter" attached
to the flywheel with some gearing and then you just dial in the position
by number. The turns counter I have in mind is often found attached to
potentiometers and variable capacitors on the front panel of radio and
test equipment...

Chris

--
Chris Elmquist

[West, Ronald S.]

Les,

Thanks. Was just on your site a few minutes ago.

Ron

[Les Bird]

>
> Herb,
>
> I appreciate your research too. I really didn't have any intentions
> on using the 3.5" media as an archival media. My goal was to just get
> something to use in the machine that was more reliable than the 10-hole,
> hard sector media over the short term. The 10-hole stuff has become
> nearly unobtainium now and also doesn't allow us to do too much with
> only 100K storage.
>
> With the hard sector emulator, a 3.5" drive and BIOS-80 under CP/M,
> we are getting 400K on a disk and that's pretty nice for hacking around.
>
> I had hundreds and hundreds of HD 3.5" on hand and so those became the
> target media. I would have had to find and buy DD ones and didn't really
> see the need for this particular exercise.
>
> I would hope that the long term archiving of these disks and software
> happens on modern media... hard drives, DVDs, and the Internet... and we
> just transfer from those archives to the floppies when we want to use the
> stuff. Granted, you will need a robust boot disk to get the game started.
>
> Chris
>

Just to let everyone know I did receive the PCBs for the HSFE on Friday and
was able to build and test several units. They work exactly as designed. For
those who want to move their collection over to a more reliable form of
media, the HSFE is a perfect solution until something better comes along.
Chris' explanation above pretty much sums it up.

To recap, it's a device that plugs in between the hard sector controller and
a pair of 3.5" disk drives. It then simulates the hard sector pulses on the
3.5" drives so that the controller thinks it's talking to a pair of 5.25"
hard sector drives. The disk drives use the standard PC-style floppy cable
with a twist. The drives are addressed as SY1 and SY2 (B and C in CP/M) but
the dip switches allow you to change this. We found the best combination to
be a primary 5.25" drive as SY0 and you can either boot from the 5.25" or
from the 3.5" drives. With this setup you can copy files from your 5.25"
drive to the 3.5" media and then just use the 3.5" drives. Under CP/M you
can use BIOS-80 and store up to 400K per disk. In HDOS use the HUG SY device
driver. BIOS-80 and the HUG SY device driver can be downloaded from my
website.

Like Chris, I've decided to stick with the pile of HD disks I had lying
around but if you want to use DD you can follow one of the links from one of
my previous posts.

Brand new NEC 3.5" drives can be purchased from Newegg for about $5.50/ea.

PCBs with a programmed chip cost $15/ea and fully assembled units are $25.
Add $5 for Priority Mail shipping.

Paypal me at les...@bellsouth.net if you want to order.

- Les

[Herbert Johnson]

Thanks for the several posts of support on my efforts. I got busy and
did not check the group until today. Thanks also to Dwight, who as
usual offers hard-core support for hard-sectored disks! It's rare to
see vintage computer support describe use of aluminum and re-
sharpening drill bits! But that's what it takes, to bring the best of
the 1970's into the 21st century!

Also, thanks Les for researching diskette prices. A buck or less per
diskette seems about par. I'm suspecious of buying disks in hundreds
for less, unless you want to test each one. Better to buy from a
dealer who will stand behind their products. Or, buy old BIG-name-
quality unused stock from individuals. Again, my colleague stated (and
I concur) that those disks from Verbatum, 3M, etc. will be of better
quality than the cheapest new-made disks today. The 3.5" diskettes are
not yet too old to suffer much from age.

But I've bought bad small-name 3.5" old-stock diskettes myself. Brand
called "Certron" give write errors half the time, even the paper
labels won't stick! Now I have a hundred or more of 'em, I can't
resell them, bleah!

Herb Johnson
retrotechnology.com

[Herbert Johnson]

Les, since you have several of the HSFE's and will distribute them,
I'd like to request the buyers to send you and each other some
physical sample diskettes. See if you can all read each other's disks.
The worst case will be reading the "high number" tracks, so fill the
disks with files.

Ideally, some kind of "checksum" program would be a way to verify
files. But since sectors and tracks have checksums (I think in the
hard sectored formats), just reading successfully is a good test. In
the CP/M world there are such checksum programs. But the simple modulo
255 checksum is simple to program up. Just read all bytes in the file
and add them up; as the sum exceeds 255, subtract 255. then what you
get is an 8-bit "checksum" per file. Just avoid reading past the EOF
or reading the EOF character as part of the file (depending on your
OS). CP/M has such checksum programs, don't know about HDOS.

Herb Johnson
keeping check
retrotechnology.com