20 Feb
2010
20 Feb
'10
2:16 p.m.
I ordered my board as soon as I got the announcement on Thursday. I received
the board today (Saturday) and took it for a spin - imaged an Apple
16-sector disk and an MSDOS 1.2M disk using a YD-380 as the source drive.
Both worked great - the Apple disk produced a .dsk file that opened and read
in CiderPress and the DOS disk produced a .img file that looked good in
WinImage.
I was tempted to hook it up to an 8" drive using my DBIT adapter but
hesitated based on the discussions here. Would the DBIT adapter provide the
necessary buffering to drive a Shugart 801? I really like the idea of an 8"
drive connected to a netbook.
Jack
20 Feb
20 Feb
2:22 p.m.
On 20 Feb 2010 at 16:16, Jack Rubin wrote:
I was tempted to hook it up to an 8" drive using
my DBIT adapter but
hesitated based on the discussions here. Would the DBIT adapter
provide the necessary buffering to drive a Shugart 801? I really like
the idea of an 8" drive connected to a netbook.
No--the DBIT adapter pretty much just employs a PIC to count the step
pulses to derive TG40. AFAIK, the other signals aren't touched.
--Chuck
9:47 p.m.
I ordered mine the same time, hopefully it'll show up on Monday, looking
really forward to trying it out.
Curt
Jack Rubin wrote:
I ordered my board as soon as I got the announcement
on Thursday. I received
the board today (Saturday) and took it for a spin - imaged an Apple
16-sector disk and an MSDOS 1.2M disk using a YD-380 as the source drive.
Both worked great - the Apple disk produced a .dsk file that opened and read
in CiderPress and the DOS disk produced a .img file that looked good in
WinImage.
I was tempted to hook it up to an 8" drive using my DBIT adapter but
hesitated based on the discussions here. Would the DBIT adapter provide the
necessary buffering to drive a Shugart 801? I really like the idea of an 8"
drive connected to a netbook.
Jack
21 Feb
21 Feb
7:01 a.m.
On Sat, 20 Feb 2010, Jack Rubin wrote:
I ordered my board as soon as I got the announcement
on Thursday. I received
the board today (Saturday) and took it for a spin - imaged an Apple
16-sector disk and an MSDOS 1.2M disk using a YD-380 as the source drive.
Both worked great - the Apple disk produced a .dsk file that opened and read
in CiderPress and the DOS disk produced a .img file that looked good in
WinImage.
I was tempted to hook it up to an 8" drive using my DBIT adapter but
hesitated based on the discussions here. Would the DBIT adapter provide the
necessary buffering to drive a Shugart 801? I really like the idea of an 8"
drive connected to a netbook.
Jack,
Why would a Shugart 801 represent a different load than a 5.25" drive?
Aren't they both 5V TTL level interfaces?
Steve
--
2:14 p.m.
On 21 Feb 2010 at 10:01, Steven Hirsch wrote:
Why would a Shugart 801 represent a different load
than a 5.25" drive?
Aren't they both 5V TTL level interfaces?
Pullups on late 5.25" (e.g. 1.2MB) floppies as well as 3.5" drives
tend to be in the 2K range. Pullups on an 8" drive are almost always
150 ohms. How much current can that microcontroller sink for
sustained periods?
You might get away with it, but I wouldn't recommend it. Add to the
mix that 8" drive cables are usually much longer than 5.25" or 3.5"
cables (read: higher inductive load) and you have an "interesting"
cocktail. I never could make my CW MK 3 work reliably with 8" drives
until I constructed a buffer board.
--Chuck
22 Feb
22 Feb
4:35 a.m.
On Sun, 21 Feb 2010, Chuck Guzis wrote:
On 21 Feb 2010 at 10:01, Steven Hirsch wrote:
Hm. Never realized that. However, since the pullups are generally DIP
resistor packs why not replace them with a higher value?
Why would a Shugart 801 represent a different
load than a 5.25" drive?
Aren't they both 5V TTL level interfaces?
Pullups on late 5.25" (e.g. 1.2MB) floppies as well as 3.5" drives
tend to be in the 2K range. Pullups on an 8" drive are almost always
150 ohms. How much current can that microcontroller sink for
sustained periods? You might get away with it, but I wouldn't
recommend it. Add to the
mix that 8" drive cables are usually much longer than 5.25" or 3.5"
cables (read: higher inductive load) and you have an "interesting"
cocktail. I never could make my CW MK 3 work reliably with 8" drives
until I constructed a buffer board.
Even more interesting. I've noticed that my CW MKIV+ is "fussy" about the
drives it works with. Now I have an explanation! Before going to the
trouble of building buffers, I'm more inclined to use a higher value
pullup and a shorter cable.
Steve
--
11:16 a.m.
Pullups on
late 5.25" (e.g. 1.2MB) floppies as well as 3.5" drives
tend to be in the 2K range. Pullups on an 8" drive are almost always
150 ohms. How much current can that microcontroller sink for
sustained periods?
Hm. Never realized that. However, since the pullups are generally DIP
resistor packs why not replace them with a higher value?
3:13 p.m.
On Tue, 23 Feb 2010, Tony Duell wrote:
Well, you could, but I wouldn't. The resistors are not only pull-ups,
they also terminate the cable (regarded as a transmission line). I am a
great beleiveer in properly terminated cables, reflections on said cables
cause all sorts of 'interesting' problems.
I suspect it would work with 1k pull-ups and a short cable, but
constructing a buffer board is not going to take long (it's only 2 or 3
14 pin TTL chips), and it means it _will_ work. If you have problems with
data corruption, etc, you then know it's not due to ringing on the cable.
Fair enough :-). The TTL chips of which you speak are bidi bus
transceivers, correct? Would you use the R/W line to switch directions,
or is there other magic involved?
Any construction tips? (Schematic would be nice - but hand-waving works,
too!)
Steve
--
Pullups on late 5.25" (e.g. 1.2MB) floppies as
well as 3.5" drives
tend to be in the 2K range. Pullups on an 8" drive are almost always
150 ohms. How much current can that microcontroller sink for
sustained periods?
Hm. Never realized that. However, since the pullups are generally DIP
resistor packs why not replace them with a higher value?
24 Feb
24 Feb
7:53 a.m.
On 23 Feb 2010 at 18:13, Steven Hirsch wrote:
Fair enough :-). The TTL chips of which you speak are
bidi bus
transceivers, correct? Would you use the R/W line to switch
directions, or is there other magic involved?
Any construction tips? (Schematic would be nice - but hand-waving
works, too!)
Not even that complicated. What Tony and I have been talking about
are bog-plain 7407 open-collector non-inverting buffer chips. The
floppy bus (if it can be called that) is unidirectional. So,
basically, for controller outputs, you feed said output into a 7407,
adding a small pullup (2K2 should be fine) and run the output of the
7407 to your floppies. Inputs to the controller reverse this and use
a 150 ohm pullup on the input to the 7407 from the floppy, with the
output feeding the controller (which should have its own pullup, so
none required on your part).
Add power and decoupling caps and you've got the design.
--Chuck
9:57 a.m.
On Wed, 24 Feb 2010, Chuck Guzis wrote:
On 23 Feb 2010 at 18:13, Steven Hirsch wrote:
Ok, so you are not conditioning the data lines? That's where I had the
notion that bidirectional buffers were needed.
--
Fair enough :-). The TTL chips of which you
speak are bidi bus
transceivers, correct? Would you use the R/W line to switch
directions, or is there other magic involved?
Any construction tips? (Schematic would be nice - but hand-waving
works, too!)
Not even that complicated. What Tony and I have been talking about
are bog-plain 7407 open-collector non-inverting buffer chips. The
floppy bus (if it can be called that) is unidirectional. So,
basically, for controller outputs, you feed said output into a 7407,
adding a small pullup (2K2 should be fine) and run the output of the
7407 to your floppies. Inputs to the controller reverse this and use
a 150 ohm pullup on the input to the 7407 from the floppy, with the
output feeding the controller (which should have its own pullup, so
none required on your part).
12:44 p.m.
On Wed, 24 Feb 2010, Steven Hirsch wrote:
On Wed, 24 Feb 2010, Chuck Guzis wrote:
Nevermind... I've been spending too much time with modern storage devices
(SCSI, ATA, etc.). Keep forgetting that data communication between FDC
and drive is serial data....
Steve
--
On 23 Feb 2010 at 18:13, Steven Hirsch wrote:
Ok, so you are not conditioning the data lines? That's where I had the
notion that bidirectional buffers were needed. Fair enough :-). The TTL chips of which you
speak are bidi bus
transceivers, correct? Would you use the R/W line to switch
directions, or is there other magic involved?
Any construction tips? (Schematic would be nice - but hand-waving
works, too!)
Not even that complicated. What Tony and I have been talking about
are bog-plain 7407 open-collector non-inverting buffer chips. The
floppy bus (if it can be called that) is unidirectional. So,
basically, for controller outputs, you feed said output into a 7407,
adding a small pullup (2K2 should be fine) and run the output of the
7407 to your floppies. Inputs to the controller reverse this and use
a 150 ohm pullup on the input to the 7407 from the floppy, with the
output feeding the controller (which should have its own pullup, so
none required on your part).
1:29 p.m.
Ok, so you are not conditioning the data lines?
That's where I had the
notion that bidirectional buffers were needed.
What data lines?
The floppy interface has a write data line, which is unidirectional
controller to drive and a read data line which is unidirectional drive to
controller. They are both bit-serial (and 'raw', in that they directly
relate to magnetic flux transitions at the head).
There is no bidirectional data bus on the normal floppy disk interface.
-tony
1:41 p.m.
From: snhirsch at gmail.com
---snip---
Ok, so you are not conditioning the data lines? That's where I had the
notion that bidirectional buffers were needed.
--
There are no bidirectional lines used from or to a floppy drive.
Dwight
_________________________________________________________________
Hotmail: Trusted email with Microsoft?s powerful SPAM protection.
http://clk.atdmt.com/GBL/go/201469226/direct/01/
1:42 p.m.
On 24 Feb 2010 at 12:57, Steven Hirsch wrote:
Ok, so you are not conditioning the data lines?
That's where I had
the notion that bidirectional buffers were needed.
No, not at all. All data lines are unidirectional (e.g. READ DATA,
WRITE DATA, WRITE GATE). We're just matching the drive specs of the
two devices. You could, if you wanted, insert a schmitt trigger
device on the read data line for some conditioning (e.g. 74LS14).
--Chuck
12:58 p.m.
Fair enough :-). The TTL chips of which you speak are
bidi bus
transceivers, correct? Would you use the R/W line to switch directions,
or is there other magic involved?
No, they're just palin unidirectional buffers, open collector outputs.
Look up the 7406 (inverting) and 7407 (non inverting) data sheets. I am
sure they're on the web somewhere (I've not checked, I've got several
paper TTL data books here).
The lines to a floppy dirve do not reverse direction. Eachg signal is
either controller to drive (for example, the drive select lines, write
data, write gate, step, direction, etc) or drive to controller (e.g. read
data, index, track 0, etc).
You need to buffer the signals from controller to drive, certainly. I'd
use a circuit like this for each signal -- and put it close to the
controller board. :
+5V
---
|
/ 1k
\
/
|
| |\
Controller output signal -----+----||>--------------- Drive input signal
|/
1/6 7407 buffer
Each 7407 will buffer 8 signala. The 1k resistor is a pull-up and is
necessary if teh controller has open-collector outputs (as it should do,
bit I bet some don't). ou need to fit the 150 Ohm terminating resistors
in the drive, which will act as pull-ups on the 7407 open-collector outputs.
You can probably get away with leaving the input signals alone. If you
want to do things properly, you might try putting a termination resistor
between each input signal and +5V at the controler end of the cable. If
the cotnroller has 1k pull-ups, then try a 180 Ohm resistor (180 // 1k
gives about 150 Ohms).
If you are a purist (as I guess I am :-)), you might try something like
:
+5V
---
|
\
/ 150 Ohm
\
/
| |\ |\
Drive output----+---|#>o---|#>o------Controller input
|/ |/
2/6 74LS14 Schmitt trigger inverter
Each 74LS14 contains 6 schmitt trigger inverters, you need 2 for each
signal (since you don't want ot invert the signal, and AFAIK there is no
non-onverting schmitt trigger in the TTL range). So each 74LS14 will
buffer 3 signals. Again, you want to put this as close to the cotnroller
as you can.
-tony
22 Feb
22 Feb
12:42 p.m.
You might get away with it, but I wouldn't
recommend it. Add to the
mix that 8" drive cables are usually much longer than 5.25" or 3.5"
cables (read: higher inductive load) and you have an "interesting"
Isn;'t it the extra capacitance (from the longer cable) that's the problem?
cocktail. I never could make my CW MK 3 work reliably
with 8" drives
until I constructed a buffer board.
I was going to ask that. What is the problem with construction an add-on
buffer board, say using 74x07 chips and pull-up resistors, to interface
between the microcontroller/USB interface thingy/... and thr 8" driv
-tony
2:03 p.m.
On 22 Feb 2010 at 20:42, Tony Duell wrote:
There's that, but the issue of using a long cable with totem-pole
drivers creates interesting issues with under- and over-shoot. ISTR
that there are situations where a large inductance driven with a CMOS
totem-pole can result in both the pull-up and -down transistors being
momentarily turned on. Clamping diodes could minimize that.
You might get
away with it, but I wouldn't recommend it. Add to the
mix that 8" drive cables are usually much longer than 5.25" or 3.5"
cables (read: higher inductive load) and you have an "interesting"
Isn;'t it the extra capacitance (from the longer cable) that's the
problem? I was going to ask that. What is the problem with
construction an
add-on buffer board, say using 74x07 chips and pull-up resistors, to
interface between the microcontroller/USB interface thingy/... and thr
8" driv
That's exactly what I did--and mounted the thing on a PC-type bracket
with a DC-37F connector on it. Two birds with one stone--an external
drive connector and ample drive capability. For good measure, I
added some schmitt trigger receivers with their own 150 ohm pullups.
--Chuck
12:33 p.m.
Why would a Shugart 801 represent a different load than a 5.25" drive?
Aren't they both 5V TTL level interfaces?
A Shigart 801 would be the same load as a 'classic' 5.25" drive, but not
the same as a modern-ish half-height ex-PC drive.
The difference is the termination/pull-up resistors. The original spec
(for 8" -- SUhgart SA800 series, etc, 5.25" -- Shugaert SA400 series, etc
and 3" (not 3.5") -- Hitachi, etc) was to have a removeable 150Ohm
termination resistor package. This was fitted on the last drive on the
interface cale (that is the drive furthest from the controller) only, and
acted both as a pull-up for the open-collector drivers used and as a
termination for the cable. So of course the driver chips have to be able
to sink 5V/150Ohms = 33mA
3.5" drives and later 5.25" ones have non-removable pull-up resistors (I
won't call them terminators, they don't even attempt to match the
characteristic impedance of the cable). These are normally about 1k (I've
seen various values), and are fitted on all drives. Of course if you have
2 drives on the cable, uou get the 2 pull-up resistors in parallel, and
so on.
Electrically, this is the wrong thing to do (the cable should be
terminated), but it's probably OK you have a shert interface cable (the
drives are in a PC cabinet along with the controller. It suprisies me,
actually, that even early 3.5" drives (like the full-height Sony units)
didn't ahve proepr termination resistors.
Another issue sprints to mind. The termination resisotr pack in the
drive terminated/pulled up the signals from the controller to the drive
only. The signals from drive to controller were terminated by 150 ohm
rsistors on the cotnroller PCB. So the drive outputs have to be able to
sink 33mA too. I know that the full-height Sony 3.5" drives used 7438s to
drive these lines, and coule sink that current. What about more modern
drives? Is there a problem with linking those to a 'classic' controller
board?
-tony
1:31 p.m.
On 22/02/2010 20:33, Tony Duell wrote:
Another issue sprints to mind. The termination
resisotr pack in the
drive terminated/pulled up the signals from the controller to the drive
only. The signals from drive to controller were terminated by 150 ohm
rsistors on the cotnroller PCB. So the drive outputs have to be able to
sink 33mA too. I know that the full-height Sony 3.5" drives used 7438s to
drive these lines, and coule sink that current. What about more modern
drives? Is there a problem with linking those to a 'classic' controller
board?
We used to see problems on Acorn Archimedes machines, which had internal
modern-style 3.5" drives and LS TTL drivers. Adding a normal 5.25"
drive often didn't work because there was not enough drive current.
However, changing the LS TTL for standard TTL fixed that, and the
internal 3.5" still worked fine.
--
Pete Peter Turnbull
Network Manager
University of York
5196
days inactive
5200
days old
19 comments
7 participants
participants (7)
-
ard＠p850ug1.demon.co.uk -
cclist＠sydex.com -
curt＠atarimuseum.com -
dkelvey＠hotmail.com -
jack.rubin＠ameritech.net -
pete＠dunnington.plus.com -
snhirsch＠gmail.com