3.5 inch floppy drive and disks

[j...@mdfs.net]

Has anybody any experience of plugging a 3.5" floppy drive into
PDP-11 equipment as a replacement for a 8" or 5.25" drive?

jgh

[glen herrmannsfeldt]

j...@mdfs.net wrote:
> Has anybody any experience of plugging a 3.5" floppy drive into
> PDP-11 equipment as a replacement for a 8" or 5.25" drive?

I have used 5.25in HD drives, which work fine.

The difference is that 3.5HD drives run at 300RPM, while
the 5.25HD run at 360RPM. The only thing that should affect
is the low-level format, which might time out while writing
the track.

If the drive has a speed control, try to get closer to 360RPM.

-- glen

[Christian Corti]

glen herrmannsfeldt <g...@ugcs.caltech.edu> wrote:
> The difference is that 3.5HD drives run at 300RPM, while

There are quite some 3.5 drives that can be jumpered to 360 rpm. The
jumper (or solder pads) may be just labelled "300/360" or "360".

Christian

[Lee Gleason]

>Has anybody any experience of plugging a 3.5" floppy drive into
>PDP-11 equipment as a replacement for a 8" or 5.25" drive?

This fellow wrote about doing it on some old Cromemco gear

http://majzel.blogspot.com/2009/04/converting-from-8-to-35-inch-floppy.html

I haven't tried it yet, been meaning to on some old PDT11/150s - the 8
inchers are way too noisy and the media is not all that easy to find.

--

Lee K. Gleason N5ZMR
lee.g...@comcast.net


---
This email is free from viruses and malware because avast! Antivirus protection is active.
http://www.avast.com

[Bill Gunshannon]

In article <52bd8105$0$29782$862e...@ngroups.net>,

"Lee Gleason" <lee.g...@comcast.net> writes:
>
>
>>Has anybody any experience of plugging a 3.5" floppy drive into
>>PDP-11 equipment as a replacement for a 8" or 5.25" drive?
>
>
> This fellow wrote about doing it on some old Cromemco gear
>
> http://majzel.blogspot.com/2009/04/converting-from-8-to-35-inch-floppy.html
>
> I haven't tried it yet, been meaning to on some old PDT11/150s - the 8
> inchers are way too noisy and the media is not all that easy to find.
>

Unless you have some kind of third-party 8" controller you are
using RX01/02's and they do not have any kind of interface that
even looks like other disks. And based on your comment about
noise, I am betting you do have an RX02. :-) I have been
meaning to look into the interface between the RXV21 and the
RX02 with a hope towards building some kind of emulator either
using a PC or maybe a RaspberryPi or Arduino. I would like to
do the saame for the RL01/02 but I suspect that is more complex.

Just out of curiosity....

I work with other old 8-bit systems made by Tandy. (OK, it's a
really wierd hobby!!!) yesterday I booted OS-9 on a Color Computer
that has no disk drives at all and, no, it it's not ROMed. I booted
it over the serial line and all disk access from then on is over the
serial line. It also does IP Networking over the serial line. All
of this thru a PC server. I wonder what the chances are that a client
couldbe written for the PDP-11? Performance would not be as good as
the COCO which can do 56K over the bit-banger serial port but it
would probably perform as well as an RX disk when you take out the
latency due to the mechanics of the 8" floppy. Just rambling, but
if anyone else is curious now, it's called DriveWire and you can
find inof and pointers to it at a company called Cloud9,

bill


--
Bill Gunshannon | de-moc-ra-cy (di mok' ra see) n. Three wolves
bill...@cs.scranton.edu | and a sheep voting on what's for dinner.
University of Scranton |
Scranton, Pennsylvania | #include <std.disclaimer.h>

[Lee Gleason]

>

>Unless you have some kind of third-party 8" controller you are
>using RX01/02's and they do not have any kind of interface that
>even looks like other disks. And based on your comment about
>noise, I am betting you do have an RX02. :-) I have been
>meaning to look into the interface between the RXV21 and the
>RX02 with a hope towards building some kind of emulator either
>using a PC or maybe a RaspberryPi or Arduino. I would like to
>do the saame for the RL01/02 but I suspect that is more complex.

These are on a PDT-11/150, so it doesn't really have a controller at all -
just an 8085 emulating one.

I have the 8085 code disassembled, so I'm pretty confident I can modify it
enough to see a 3.5 or 5.25 inch device as the 8 inch.

I anticipate it will take a little cable making and the odd chip or two as
well...

>I work with other old 8-bit systems made by Tandy. (OK, it's a
>really wierd hobby!!!) yesterday I booted OS-9 on a Color Computer
>that has no disk drives at all and, no, it it's not ROMed. I booted
>it over the serial line and all disk access from then on is over the
>serial line. It also does IP Networking over the serial line. All
>of this thru a PC server. I wonder what the chances are that a client
>couldbe written for the PDP-11?

Might work on a real 11, but not likely on an 11/150 - the serial ports on
it are seriously slow.

[Al Kossow]

On 12/27/13 7:28 AM, Bill Gunshannon wrote:
> I would like to
> do the saame for the RL01/02 but I suspect that is more complex.
>

http://www.pdp11gy.com/indexE.html#file:///E:/homepage/indexE.html

[Bill Gunshannon]

In article <52bda2bd$0$29570$862e...@ngroups.net>,

"Lee Gleason" <lee.g...@comcast.net> writes:
>
>
>>
>
>>Unless you have some kind of third-party 8" controller you are
>>using RX01/02's and they do not have any kind of interface that
>>even looks like other disks. And based on your comment about
>>noise, I am betting you do have an RX02. :-) I have been
>>meaning to look into the interface between the RXV21 and the
>>RX02 with a hope towards building some kind of emulator either
>>using a PC or maybe a RaspberryPi or Arduino. I would like to
>>do the saame for the RL01/02 but I suspect that is more complex.
>
> These are on a PDT-11/150, so it doesn't really have a controller at all -
> just an 8085 emulating one.
>

So I take it that it uses something like Shugarts? You might try looking
for Qume 8" drives like wre used on the Tandy Model 16 and 6000. Just as
quiet as any 5.25 drive.

> I have the 8085 code disassembled, so I'm pretty confident I can modify it
> enough to see a 3.5 or 5.25 inch device as the 8 inch.
>
> I anticipate it will take a little cable making and the odd chip or two as
> well...

You might try talking to John at Dbit. He does a little board that lets
you connect 8" drives to standard PC controllers so he probably already
has the differences documented.

>
>
>>I work with other old 8-bit systems made by Tandy. (OK, it's a
>>really wierd hobby!!!) yesterday I booted OS-9 on a Color Computer
>>that has no disk drives at all and, no, it it's not ROMed. I booted
>>it over the serial line and all disk access from then on is over the
>>serial line. It also does IP Networking over the serial line. All
>>of this thru a PC server. I wonder what the chances are that a client
>>couldbe written for the PDP-11?
>
> Might work on a real 11, but not likely on an 11/150 - the serial ports on
> it are seriously slow.

Most older real 11's have very slow Serial as well, but even at 9600
baud I suspect it would be better than your average RX01/02. And
I think there were some faster modules available. If a disk driver
needs to be written anyway, one could shoot at a more efficient way
considering the inout device is going tobe different from the average
typist. Another thing I laways wanted to try was a new UART. Something
like the 16550 which did more than one char per interupt. Given the
dearth of disk drives that will work on a PDP-11 and the desire to keep
using them, I am surprised others haven't looked at this. But maybe
hardware hacking is a dying activity, too. How about emulating MSCP
on an Arduino? :-) Hardest part would be slowing down the Arduino
and keeping it from getting bored waiting on the PDP-11. :-)

[Al Kossow]

On 12/27/13 8:48 AM, Bill Gunshannon wrote:

> So I take it that it uses something like Shugarts?

DEC liked single-sided Calcomp drives.
Docs are all up on bitsavers if you want to compare pinouts.

[Johnny Billquist]

On 2013-12-27 17:48, Bill Gunshannon wrote:
> In article <52bda2bd$0$29570$862e...@ngroups.net>,
> "Lee Gleason" <lee.g...@comcast.net> writes:
>>> I work with other old 8-bit systems made by Tandy. (OK, it's a
>>> really wierd hobby!!!) yesterday I booted OS-9 on a Color Computer
>>> that has no disk drives at all and, no, it it's not ROMed. I booted
>>> it over the serial line and all disk access from then on is over the
>>> serial line. It also does IP Networking over the serial line. All
>>> of this thru a PC server. I wonder what the chances are that a client
>>> couldbe written for the PDP-11?
>>
>> Might work on a real 11, but not likely on an 11/150 - the serial ports on
>> it are seriously slow.
>
> Most older real 11's have very slow Serial as well, but even at 9600
> baud I suspect it would be better than your average RX01/02.

Ugh! Did you ever consider doing some math on that?
A normal RX01/02 would read a block in something like 200ms. (probably a
bit faster, but conservative numbers are always good).
That is 256 bytes in 200 ms for an RX02. That gives a bit rate of 10240
bps, so it would beat 9600... (Of course, I might have numbers wrong in
here, feel free to correct my errors...)

> And
> I think there were some faster modules available. If a disk driver
> needs to be written anyway, one could shoot at a more efficient way
> considering the inout device is going tobe different from the average
> typist. Another thing I laways wanted to try was a new UART. Something
> like the 16550 which did more than one char per interupt.

Any serious serial interface on a PDP-11 already can do more than one
char per interrupt. About the only thing that stupid is the DL11.
The better ones do DMA for output. No interrupt until all done.
Interrupts on input, but definitely no need for one interrupt per char.

> Given the
> dearth of disk drives that will work on a PDP-11 and the desire to keep
> using them, I am surprised others haven't looked at this. But maybe
> hardware hacking is a dying activity, too. How about emulating MSCP
> on an Arduino? :-) Hardest part would be slowing down the Arduino
> and keeping it from getting bored waiting on the PDP-11. :-)

Who cares if an Arduino is bored? It don't have a soul anyway. :-)
MSCP is well enough figured out in emulators today that it should be
doable to get a hardware controller done as well. But yes, hardware
hacking is more dead than software...

Johnny

[Bill Gunshannon]

In article <l9kl7o$qa9$1...@iltempo.update.uu.se>,

Johnny Billquist <b...@softjar.se> writes:
> On 2013-12-27 17:48, Bill Gunshannon wrote:
>> In article <52bda2bd$0$29570$862e...@ngroups.net>,
>> "Lee Gleason" <lee.g...@comcast.net> writes:
>>>> I work with other old 8-bit systems made by Tandy. (OK, it's a
>>>> really wierd hobby!!!) yesterday I booted OS-9 on a Color Computer
>>>> that has no disk drives at all and, no, it it's not ROMed. I booted
>>>> it over the serial line and all disk access from then on is over the
>>>> serial line. It also does IP Networking over the serial line. All
>>>> of this thru a PC server. I wonder what the chances are that a client
>>>> couldbe written for the PDP-11?
>>>
>>> Might work on a real 11, but not likely on an 11/150 - the serial ports on
>>> it are seriously slow.
>>
>> Most older real 11's have very slow Serial as well, but even at 9600
>> baud I suspect it would be better than your average RX01/02.
>
> Ugh! Did you ever consider doing some math on that?

Well, I'm just rambling.... :-)

> A normal RX01/02 would read a block in something like 200ms. (probably a
> bit faster, but conservative numbers are always good).
> That is 256 bytes in 200 ms for an RX02. That gives a bit rate of 10240
> bps, so it would beat 9600... (Of course, I might have numbers wrong in
> here, feel free to correct my errors...)

But, is that theoretical speed or actually benchmarked. I may be wrong,
but it always seemd to me that my RX02 spent a lot of time clinking and
clunking before it delivered the first character. Thus the reason I
figured serial would be acceptable even if not particularly better.
And the alternative is to run out of 8" floppies (I still have at least
three boxes of 10 unopened!) and not be able to use it at all anyway.

>
>> And
>> I think there were some faster modules available. If a disk driver
>> needs to be written anyway, one could shoot at a more efficient way
>> considering the inout device is going tobe different from the average
>> typist. Another thing I laways wanted to try was a new UART. Something
>> like the 16550 which did more than one char per interupt.
>
> Any serious serial interface on a PDP-11 already can do more than one
> char per interrupt. About the only thing that stupid is the DL11.
> The better ones do DMA for output. No interrupt until all done.
> Interrupts on input, but definitely no need for one interrupt per char.

I guess I was never that lucky. The couple I have with built-in serial
ports seem to be faster, but most of mine use DLV-11's and were lucky
if I could get the console to work well. I have some Terak Serial
Modules that I frequently used for consoles, but although they actually
have a dip setting for 19200 the 11/02 it had was never able to do even
9600 reliably (and that was even with the interupt driven ring buffer
in the terminal program!!)

>
>> Given the
>> dearth of disk drives that will work on a PDP-11 and the desire to keep
>> using them, I am surprised others haven't looked at this. But maybe
>> hardware hacking is a dying activity, too. How about emulating MSCP
>> on an Arduino? :-) Hardest part would be slowing down the Arduino
>> and keeping it from getting bored waiting on the PDP-11. :-)
>
> Who cares if an Arduino is bored? It don't have a soul anyway. :-)

Well, now you have hurt my Arduino's feelings. Looks like it's the
naughty list for you next year. And it only took you two days to
get on it. :-)

> MSCP is well enough figured out in emulators today that it should be
> doable to get a hardware controller done as well.

Two reasons I wasn't even thinking of controllers. One is the lack
of availability of some of the special interface chips used by DEC.
Not sure if there are modern (or semi-modern) equivalents. And, two,
we already have lots of interface modules and the drivers to use them.
What we are lacking is media.

> But yes, hardware
> hacking is more dead than software...

Probably too late in my life to plan on accomplishing much of what I dream
of, but I am definitely getting back into hacking hardware. That's the
other reason my hobby involves so many older systems. :-)

[glen herrmannsfeldt]

Johnny Billquist <b...@softjar.se> wrote:
> On 2013-12-27 17:48, Bill Gunshannon wrote:

(snip)

>> Most older real 11's have very slow Serial as well, but even at 9600
>> baud I suspect it would be better than your average RX01/02.

> Ugh! Did you ever consider doing some math on that?
> A normal RX01/02 would read a block in something like 200ms. (probably a
> bit faster, but conservative numbers are always good).
> That is 256 bytes in 200 ms for an RX02. That gives a bit rate of 10240
> bps, so it would beat 9600... (Of course, I might have numbers wrong in
> here, feel free to correct my errors...)

The bit rate should be 250kHz/500kHz for SD/DD (RX01/RX02), but
you have to add seek and latency. For random sector access, it
will end up pretty slow, but if you can stream consecutive
sectors, not so bad.

-- glen

[glen herrmannsfeldt]

Bill Gunshannon <bi...@server2.cs.scranton.edu> wrote:

(snip)

> I work with other old 8-bit systems made by Tandy. (OK, it's a
> really wierd hobby!!!) yesterday I booted OS-9 on a Color Computer
> that has no disk drives at all and, no, it it's not ROMed. I booted
> it over the serial line and all disk access from then on is over the
> serial line.

The first computer I owned was a Color Computer, and also the
first one I had a disk drive for. I had OS/9 running on it,
with the original drive (35 track SS), an 80 track SS drive,
and two Teac FD55-F (double side, 80 track/side). I rewrote
the OS/9 disk driver to drive them. (Came with an assembler,
and I believe disassembler, though I might have done it by hand.)

At the time I was wondering about 8 inch drives, or the
Teac FD55G (I had the Teac manual, which included the specs.
for the 55G.) I figured once that to do the I/O fast enough the
loop would be unrolled. I didn't have a drive to try it on, though.

We moved not so long after I got a PC/AT clone and I gave away
the CC, but kept the disk controller and drives.

> It also does IP Networking over the serial line. All
> of this thru a PC server. I wonder what the chances are that a client
> couldbe written for the PDP-11? Performance would not be as good as
> the COCO which can do 56K over the bit-banger serial port but it
> would probably perform as well as an RX disk when you take out the
> latency due to the mechanics of the 8" floppy. Just rambling, but
> if anyone else is curious now, it's called DriveWire and you can
> find inof and pointers to it at a company called Cloud9,

If I get back to 8 bit processors, I will probably do it in FPGA
hardware emulation.

Actually, I have an Epson QX-10 that sometimes runs. I think I need
to clean the heads, though.

-- glen

[Al Kossow]

On 12/27/13 11:43 AM, Johnny Billquist wrote:
> hardware hacking is more dead than software...
>

Brad and I would disagree with you

http://www.heeltoe.com/retro/udisk/

[Antti Louko]

Nice!

Q-bus version would be nice, too.

[Bill Gunshannon]

In article <l9kqiu$6gt$1...@speranza.aioe.org>,

That was exactly my point. While actual bit transfer is faster an
emulated serial disk would not have nay of the mecahnincal latency.

Speaking of which, anybody with device driver experience ever looked
at the uDRIVE from 4D Systems? There's a serial disk drive already.

[Bill Gunshannon]

In article <l9kr5a$844$1...@speranza.aioe.org>,

glen herrmannsfeldt <g...@ugcs.caltech.edu> writes:
> Bill Gunshannon <bi...@server2.cs.scranton.edu> wrote:
>
> (snip)
>
>> I work with other old 8-bit systems made by Tandy. (OK, it's a
>> really wierd hobby!!!) yesterday I booted OS-9 on a Color Computer
>> that has no disk drives at all and, no, it it's not ROMed. I booted
>> it over the serial line and all disk access from then on is over the
>> serial line.
>
> The first computer I owned was a Color Computer,

Far from the first for me :-) but I still have a handful of them.
TRS-80 Model I, III, and 4 as well. Had a couple 16's but they
are gone.

> and also the
> first one I had a disk drive for. I had OS/9 running on it,
> with the original drive (35 track SS), an 80 track SS drive,
> and two Teac FD55-F (double side, 80 track/side). I rewrote
> the OS/9 disk driver to drive them. (Came with an assembler,
> and I believe disassembler, though I might have done it by hand.)

Should have kept in touch with the regular users. The Dmode command
(not from MicroWare) let you modify all the parameters for disks and
run pretty much anything the controller supported. I was running
lots of different 5.25 and eventually even 3.5". I am digging thru
boxes and boxes of disks now trying to find all the stuff I had then.

>
> At the time I was wondering about 8 inch drives,

I am hoping to try 8" disks on the COCO and the later Z80 boxes using
John Wilsons little widget. I have OSes for the Model III that know
about 8" drives so I am hoping it will work. Would make accessing old
CP/M disks easier.

> or the
> Teac FD55G (I had the Teac manual, which included the specs.
> for the 55G.) I figured once that to do the I/O fast enough the
> loop would be unrolled. I didn't have a drive to try it on, though.

You must mean in HD mode. I use FD55-GFR's all the time. It's the
best and most adaptable 5.25 ever made. Works well hanging off a
RQDX3 as well. :-)

>
> We moved not so long after I got a PC/AT clone and I gave away
> the CC,

To me, PC's were just what I did for work. Most of mine run BSD and
the nes running Windows do it because I have an application that dowsn't
work under BSD (Like my EPROM Programmer!)

> but kept the disk controller and drives.

Any time you need to clean house let me know, I'll send you my address. :-)

>
>> It also does IP Networking over the serial line. All
>> of this thru a PC server. I wonder what the chances are that a client
>> couldbe written for the PDP-11? Performance would not be as good as
>> the COCO which can do 56K over the bit-banger serial port but it
>> would probably perform as well as an RX disk when you take out the
>> latency due to the mechanics of the 8" floppy. Just rambling, but
>> if anyone else is curious now, it's called DriveWire and you can
>> find inof and pointers to it at a company called Cloud9,
>
> If I get back to 8 bit processors, I will probably do it in FPGA
> hardware emulation.
>
> Actually, I have an Epson QX-10 that sometimes runs. I think I need
> to clean the heads, though.
>

Sometimes it is just plain fun to go back to what I did earlier. But,
I doubt I will ever have a 1401 in the house (although I have had some
pretty strange medium sized iron in the house that probably used just
as much electricity!)

[j...@mdfs.net]

How do people nowadays get data "into" and "out" of hardware
PDP-11 systems? A serial connection to an adjoining PC would
be workable for local transfer, but not if you want to give
something to somebody else to use on their hardware PDP-11.

Last time I was in front of a hardware PDP-11 about a year
or so ago it had 5.25" drives. I have various computers that
can read and write 5.25" disks in various speeds and densities,
but I spent most of about twenty years ago migrating most of
my floppy disk needs onto 3.5" media.

jgh

[glen herrmannsfeldt]

Bill Gunshannon <bi...@server2.cs.scranton.edu> wrote:

(snip, I wrote)

>> or the
>> Teac FD55G (I had the Teac manual, which included the specs.
>> for the 55G.) I figured once that to do the I/O fast enough the
>> loop would be unrolled. I didn't have a drive to try it on, though.

> You must mean in HD mode. I use FD55-GFR's all the time. It's the
> best and most adaptable 5.25 ever made. Works well hanging off a
> RQDX3 as well. :-)

I believe before the GFR, there was just a G that only ran in HD mode.
Not like the GFR that you can select the write current to allow for
writing non-HD disks.

That was about 1983. Not that I saw the drive, but the Teac
documentation included it.

-- glen

[Jerome H. Fine]

>j...@mdfs.net wrote:

Has anybody any experience of plugging a 3.5" floppy drive into
PDP-11 equipment as a replacement for a 8" or 5.25" drive?

The DEC names of floppy drives are:

RX01 - DX  -  8" SSSD 488 blocks  (Identical to IBM SSSD)
RX02 - DY  -  8" SSDD 988 blocks
RX03 - DY  -  8" DSDD 1976 blocks  *
RX50 - DU  -  5 1/4" SS?? 800 blocks
RX33 - DU  -  5 1/4" HD 2400 blocks   (Identical to PC HD)
RX23 - DU  -  3.5" HD 2880 blocks  **  (Identical to PC HD)

*NEVER officially supported by DEC software, but was
  included with V04.00 of RT-11 in the DY.MAC file.
  DEC NEVER sold an RX03 for the PDP-11, but the
  DSD 880/30 single 8" floppy drive supported the
  RX03 as both DSSD and DSDD media whenever the
  alternate double sided index hole was sensed.  Note
  that both the DEC  RX02 and the DSD 880/30 supported
  ONLY 18 bit addresses.  I seem to remember that other
  third party controllers supported 22 bit addresses, but I
  can't remember the name.

**My math for the RX23 suggests 80 cylinders by 2 surfaces
    by 18 sectors per track (blocks of 512 bytes).  Can
    anyone confirm these numbers or provide others?


It is not possible to plug either a 5 1/4" or a 3.5" drive into
a DEC controller which supports 8" floppy drives.  I could
be wrong, but I don't think so.  Some non-DEC third party
controllers might be able to do so, but I can't remember
which one.  ALSO, many non-DEC controllers could perform
a Low Level Format (LLF) of floppy media.  For DEC controllers,
the only success with an LLF on a PDP-11 was the RX33
on an RQDX3.  I think that the Rainbow was able to perform
an LLF on the RX50.

I have used all of the above drives under RT-11 on DEC  PDP-11
Qbus hardware.  The DEC RQDX3 supports both RX50 and
RX33 floppy drives.  I have never attempted to plug in a standard
PC 3.5" floppy drive with the same connector as a 5 1/4" PC floppy
drive to the DEC RQDX3, so I do not know if that 3.5" drive is
recognized as an RX23.  Has anyone else attempted to do so?

I do have a customized CQD 220M host adapter which does
specifically support a 3.5" SCSI floppy drive (which RT-11
then recognizes as an RX23) and a standard SCSI hard drive.
It is possible that some standard Qbus host adapters support
SCSI 3.5" floppy drives as an RX23.  I do not remember ever
seeing a SCSI 5 1/4" floppy drive.  Have anyone else?  Are
there still both regular and SCSI 3.5" floppy drives available?

Note that Ersatz-11 supports emulation of all of the above
floppy drives.  In addition, Ersatz-11 supports using actual
hardware drives on your PC when the operating system
does not block direct hardware access.  I don't think that
Windows XP supports direct access, but Windows 98SE
DOES support them under which:

5 1/4" HD floppy drive supports use as either RX50 or RX33
3.5" HD floppy drive supports use as RX23.

You can use PUTR from John Wilson to transfer data between
5 1/4" floppy media and 3.5" floppy media.  The file structure
will be RT-11 if you transfer individual files.  PUTR may also be
able to transfer a complete device AS  IS with the caveat that
there will be extra or missing blocks due to differences in device
size.  If PUTR can't, then for SURE, Ersatz-11 can do so but
with the limitation that only the very old operating systems like
Windows 98 can see raw physical floppy drives.

---------------------------------------------------------------

If you have any more questions, please ask.  If anyone has
answers to my four questions, please reply!

Jerome Fine

[Roger Ivie]

On 2013-12-28, Jerome H. Fine <ever...@nospam.com> wrote:
> RX01 - DX - 8" SSSD 488 blocks (Identical to IBM SSSD)
> RX02 - DY - 8" SSDD 988 blocks
> RX03 - DY - 8" DSDD 1976 blocks *
> RX50 - DU - 5 1/4" SS?? 800 blocks
> RX33 - DU - 5 1/4" HD 2400 blocks (Identical to PC HD)
> RX23 - DU - 3.5" HD 2880 blocks ** (Identical to PC HD)

Also RX26, 3.5" XD (2.88MB).

> I do not remember ever
> seeing a SCSI 5 1/4" floppy drive. Have anyone else?

It's not well known, but the DEC SCSI floppy adapters can handle 5.25"
drives as well as 3.5" ones. The controller detects the drive type by
disk select; at reset, it walks through the disk selects zeroing the
drives and whichever select responds dictates whatever the controller
thinks it is.

RX50 support's a bit spotty. I decided (I wrote the floppy half of the
embedded code) that block interleave was the job of the operating
system. The operating systems disagreed. RX33s would work, but the
controller and the OS disagreed about the block numbers of RX50s.

The last version of the RX26-capable firmware did RX50 interleaving in
the controller, which allowed the controller to work with RX50
diskettes. DEC had a customer that required RX50 support, so they beat
me over the head until I fixed it.

Controllers that have a WD2793 floppy chip support LD and HD media. The
newer controllers sporting a ud750-compatible chip (we used Intel chips
for the prototype; I don't know what they did in production) also
support the RX26 XD 3.5" drives.
--
roger ivie
ri...@ridgenet.net

[glen herrmannsfeldt]

Jerome H. Fine <ever...@nospam.com> wrote:

> [-- text/plain, encoding 7bit, charset: us-ascii, 95 lines --]

> >j...@mdfs.net wrote:

>>Has anybody any experience of plugging a 3.5" floppy drive into
>>PDP-11 equipment as a replacement for a 8" or 5.25" drive?

> The DEC names of floppy drives are:

> RX01 - DX - 8" SSSD 488 blocks (Identical to IBM SSSD)
> RX02 - DY - 8" SSDD 988 blocks
> RX03 - DY - 8" DSDD 1976 blocks *
> RX50 - DU - 5 1/4" SS?? 800 blocks
> RX33 - DU - 5 1/4" HD 2400 blocks (Identical to PC HD)
> RX23 - DU - 3.5" HD 2880 blocks ** (Identical to PC HD)

> *NEVER officially supported by DEC software, but was
> included with V04.00 of RT-11 in the DY.MAC file.
> DEC NEVER sold an RX03 for the PDP-11, but the
> DSD 880/30 single 8" floppy drive supported the
> RX03 as both DSSD and DSDD media whenever the
> alternate double sided index hole was sensed. Note
> that both the DEC RX02 and the DSD 880/30 supported
> ONLY 18 bit addresses. I seem to remember that other
> third party controllers supported 22 bit addresses, but I
> can't remember the name.

Around 1988 I used an RT-11 system with a DZ driver, but
I think they just renamed the DY to DZ. As well as I remember,
that allowed one to read/write single side disks in double side
drives as DY, or double side as DZ.

> **My math for the RX23 suggests 80 cylinders by 2 surfaces
> by 18 sectors per track (blocks of 512 bytes). Can
> anyone confirm these numbers or provide others?

That is the usual current format for HD 3.5in disks, both on
DOS/Windows and Apple systems.

Normal for 8 inch disks is 77 cylinders. I don't know how many
will allow one to seek past 76. Also, 15 sectors/track at 360RPM
vs. 18 at 300RPM. (Different numbers for other block sizes.)

> It is not possible to plug either a 5 1/4" or a 3.5" drive into
> a DEC controller which supports 8" floppy drives. I could
> be wrong, but I don't think so.

I have a cable where I wired the appropriate pins from
a 50 pin cable onto a 34 pin connector. It can also be done
using a small PC board with 50 and 34 pin connectors on it.

> Some non-DEC third party
> controllers might be able to do so, but I can't remember
> which one. ALSO, many non-DEC controllers could perform
> a Low Level Format (LLF) of floppy media. For DEC controllers,
> the only success with an LLF on a PDP-11 was the RX33
> on an RQDX3. I think that the Rainbow was able to perform
> an LLF on the RX50.

I now have an actual RQDX3, but previously the only Q-bus
controller I had was the non-DEC MXV-21.

I just got out hte MXV-21 manual, and noticed some that I
had forgotten before. One is that it does support two-side
drives. Seems it also has an option to support two (virtual)
single sided disks instead, using DS drive and media.

But the main thing that the MXV-21 has is the ability to do
low-level formatting.

> I have used all of the above drives under RT-11 on DEC PDP-11
> Qbus hardware. The DEC RQDX3 supports both RX50 and
> RX33 floppy drives. I have never attempted to plug in a standard
> PC 3.5" floppy drive with the same connector as a 5 1/4" PC floppy
> drive to the DEC RQDX3, so I do not know if that 3.5" drive is
> recognized as an RX23. Has anyone else attempted to do so?

I don't know, either. It could be done by timing the disk
index pulse to sense 300RPM vs. 360RPM.

> I do have a customized CQD 220M host adapter which does
> specifically support a 3.5" SCSI floppy drive (which RT-11
> then recognizes as an RX23) and a standard SCSI hard drive.
> It is possible that some standard Qbus host adapters support
> SCSI 3.5" floppy drives as an RX23. I do not remember ever
> seeing a SCSI 5 1/4" floppy drive. Have anyone else? Are
> there still both regular and SCSI 3.5" floppy drives available?

Q-bus SCSI adapters are pretty hard to find.

I do have a SCSI 3.5in drive, though, I believe from an HP-UX
system. It has an adapter board and connects to a more or less
normal (maybe different jumper options) floppy.

(snip)

-- glen

[glen herrmannsfeldt]

Roger Ivie <ri...@ridgenet.net> wrote:
> On 2013-12-28, Jerome H. Fine <ever...@nospam.com> wrote:
>> RX01 - DX - 8" SSSD 488 blocks (Identical to IBM SSSD)
>> RX02 - DY - 8" SSDD 988 blocks
>> RX03 - DY - 8" DSDD 1976 blocks *
>> RX50 - DU - 5 1/4" SS?? 800 blocks
>> RX33 - DU - 5 1/4" HD 2400 blocks (Identical to PC HD)
>> RX23 - DU - 3.5" HD 2880 blocks ** (Identical to PC HD)

> Also RX26, 3.5" XD (2.88MB).

>> I do not remember ever
>> seeing a SCSI 5 1/4" floppy drive. Have anyone else?

(snip)

> The last version of the RX26-capable firmware did RX50 interleaving in
> the controller, which allowed the controller to work with RX50
> diskettes. DEC had a customer that required RX50 support, so they beat
> me over the head until I fixed it.

For 8 inch floppies, it was usual to do interleave as a look-up
table, usually in the driver. The physical sector numbers were
sequential, and often with the factory formatted numbers.

For many years, it was usual for 5.25in floppy disks to come
without a low level format. When you do the low-level format, it
is easy to number the sectors with the desired interleave.
No need for a look-up table. The system doesn't need to know what
the interleave is, either.

Some years after the IBM PC, the format became popular enough
that floppies, eventually both 5.25in and 3.5in, came preformatted
in that format.

As noted above, RX01 is IBM standard, SS/SD.

RX02 uses the same sector headers, but a different format for
that data, as RX01. That conveniently allows one to use RX01
low-level format disks.

The IBM standard DD uses a different, DD sector header format.
Soft sector disks (just about all by now) need some way to
uniquely identify the sector header. Both IBM standards use
a pattern of clock bits that can't occur in data.

-- glen

[Johnny Billquist]

On 2013-12-27 21:30, Bill Gunshannon wrote:
> In article <l9kl7o$qa9$1...@iltempo.update.uu.se>,
> Johnny Billquist <b...@softjar.se> writes:
>> On 2013-12-27 17:48, Bill Gunshannon wrote:
>>> In article <52bda2bd$0$29570$862e...@ngroups.net>,
>>> "Lee Gleason" <lee.g...@comcast.net> writes:
>>>>> I work with other old 8-bit systems made by Tandy. (OK, it's a
>>>>> really wierd hobby!!!) yesterday I booted OS-9 on a Color Computer
>>>>> that has no disk drives at all and, no, it it's not ROMed. I booted
>>>>> it over the serial line and all disk access from then on is over the
>>>>> serial line. It also does IP Networking over the serial line. All
>>>>> of this thru a PC server. I wonder what the chances are that a client
>>>>> couldbe written for the PDP-11?
>>>>
>>>> Might work on a real 11, but not likely on an 11/150 - the serial ports on
>>>> it are seriously slow.
>>>
>>> Most older real 11's have very slow Serial as well, but even at 9600
>>> baud I suspect it would be better than your average RX01/02.
>>
>> Ugh! Did you ever consider doing some math on that?
>
> Well, I'm just rambling.... :-)

Suspected as much. :-)

>> A normal RX01/02 would read a block in something like 200ms. (probably a
>> bit faster, but conservative numbers are always good).
>> That is 256 bytes in 200 ms for an RX02. That gives a bit rate of 10240
>> bps, so it would beat 9600... (Of course, I might have numbers wrong in
>> here, feel free to correct my errors...)
>
> But, is that theoretical speed or actually benchmarked. I may be wrong,
> but it always seemd to me that my RX02 spent a lot of time clinking and
> clunking before it delivered the first character. Thus the reason I
> figured serial would be acceptable even if not particularly better.
> And the alternative is to run out of 8" floppies (I still have at least
> three boxes of 10 unopened!) and not be able to use it at all anyway.

Like I said, I just grabbed some conservative numbers. But yes, I would
think that would with heads loaded.
The clickety-clack noise is when you load heads, and also when you do an
INIT to the drives. The INIT takes forever, but I hope drivers don't do
that all the time...

So it might be that for the first block, you might be able to beat the
floppy with a serial port, but I wouldn't be too sure. The seek times
are most of the time not really as high as those averages. And it's seek
times that is the majority of the time as usual.

In the end, it depends (as usual), but even though floppies are no speed
demons, a serial port is still not that hard to beat. Heck, even at 9600
bps, we barely gets 960 charactes per second, as there is some overhead
there too...

>>> And
>>> I think there were some faster modules available. If a disk driver
>>> needs to be written anyway, one could shoot at a more efficient way
>>> considering the inout device is going tobe different from the average
>>> typist. Another thing I laways wanted to try was a new UART. Something
>>> like the 16550 which did more than one char per interupt.
>>
>> Any serious serial interface on a PDP-11 already can do more than one
>> char per interrupt. About the only thing that stupid is the DL11.
>> The better ones do DMA for output. No interrupt until all done.
>> Interrupts on input, but definitely no need for one interrupt per char.
>
> I guess I was never that lucky. The couple I have with built-in serial
> ports seem to be faster, but most of mine use DLV-11's and were lucky
> if I could get the console to work well. I have some Terak Serial
> Modules that I frequently used for consoles, but although they actually
> have a dip setting for 19200 the 11/02 it had was never able to do even
> 9600 reliably (and that was even with the interupt driven ring buffer
> in the terminal program!!)

See, there you go. The DLV11 is just a DL11 for the Qbus. Horrible
interface.

>>> Given the
>>> dearth of disk drives that will work on a PDP-11 and the desire to keep
>>> using them, I am surprised others haven't looked at this. But maybe
>>> hardware hacking is a dying activity, too. How about emulating MSCP
>>> on an Arduino? :-) Hardest part would be slowing down the Arduino
>>> and keeping it from getting bored waiting on the PDP-11. :-)
>>
>> Who cares if an Arduino is bored? It don't have a soul anyway. :-)
>
> Well, now you have hurt my Arduino's feelings. Looks like it's the
> naughty list for you next year. And it only took you two days to
> get on it. :-)

Ouch. Poor me. :-)

>> MSCP is well enough figured out in emulators today that it should be
>> doable to get a hardware controller done as well.
>
> Two reasons I wasn't even thinking of controllers. One is the lack
> of availability of some of the special interface chips used by DEC.
> Not sure if there are modern (or semi-modern) equivalents. And, two,
> we already have lots of interface modules and the drivers to use them.
> What we are lacking is media.

I think that modern replacements that work for the common case can be
found. They don't meet the specs, but then again, you don't see ten
meter Unibuses anymore... (Nor Qbuses)

Media. Yes, well, we could do SD-cords, USB memory sticks, or
whatever... With MSCP that is all very realistic.

>> But yes, hardware
>> hacking is more dead than software...
>
> Probably too late in my life to plan on accomplishing much of what I dream
> of, but I am definitely getting back into hacking hardware. That's the
> other reason my hobby involves so many older systems. :-)

Enjoy! :-)

Johnny

[Johnny Billquist]

Right. Which is why I took something like block access time instead of
bitrates.
But your point is very valid, and more precise. :-)

Johnny

[Johnny Billquist]

Oh, I use RL02, DECnet or TCP/IP...
Once in a moon magtapes also comes in handy. TK50 occasionally.
I have also just used a plain SCSI disk a few times.

Johnny

[Johnny Billquist]

Looking at ebay, I found about 10 of them for sale right now. Price
somewhere between $200 and $800 or so...
So, not so hard to find, but not cheap either.

Johnny

[Roger Ivie]

On 2013-12-28, glen herrmannsfeldt <g...@ugcs.caltech.edu> wrote:
> For 8 inch floppies, it was usual to do interleave as a look-up
> table, usually in the driver.

Indeed, which is one of the things that went into my decision that
RX50 interleaving was the operating system's problem. Another was
that the documentation DEC gave me about extant RX50 formats
indicated that different operating systems were using different
interleaving factors. I didn't want to try to figure out which
OS was involved in order to set up the interleaving.
--
roger ivie
ri...@ridgenet.net

[Mark G Thomas]

Every now and then one shows up cheaper. I recently scored an Emulex
UC08 for $60, though I had to remove the S-box handles and
unsolder/replace the scsi connectors with right-angled headers to
fit my BA23 PDP11 box.

Mark

--
Mark G. Thomas (Ma...@Misty.com)
http://mail-cleaner.com/

[Dennis Boone]

> I work with other old 8-bit systems made by Tandy. (OK, it's a
> really wierd hobby!!!) yesterday I booted OS-9 on a Color Computer
> that has no disk drives at all and, no, it it's not ROMed. I booted
> it over the serial line and all disk access from then on is over the
> serial line. It also does IP Networking over the serial line. All
> of this thru a PC server. I wonder what the chances are that a client
> couldbe written for the PDP-11? Performance would not be as good as
> the COCO which can do 56K over the bit-banger serial port but it
> would probably perform as well as an RX disk when you take out the
> latency due to the mechanics of the 8" floppy. Just rambling, but
> if anyone else is curious now, it's called DriveWire and you can
> find inof and pointers to it at a company called Cloud9,

To stir in a few additional thoughts:

1. vtserver does something similar, so code might be borrowed.

2. Many -11s can take ethernet hardware, which would kick the pants
off anything serial.

De

[pra...@gmail.com]

Hi Glen!
Maybe you would be interesting in selling your SCSI 3,5" floppy? Or know a place where one could get them?
thanks ;)