Z-1 Observations

64 views
Skip to first unread message

Martin Eberhard

unread,
Aug 23, 2011, 12:39:28 PM8/23/11
to Cromemco
As I have been bringing up my Cromemco Z-1, I learned a few things
worth noting here:

1: The Z-1 has the Cromemco Blitz Bus. However, memory accesses become
unreliable if the 16KZ boards are too far away from the ZPU. Memory
accesses seem solid when the 16KZs are right next to the CPU, and
pretty bad when 8 slots away. Question: Did Cromemco ever provide an
active terminator board for the Blitz Bus? Any other experiences with
this kind of problem?

2: The 16KZ board (as manufactured) is not compatible with the 16FDC,
when the 16KZ is set for address C000h. One of my 16KZ boards came
modified to make it work. The modification is a jumper from IC3 pin 3
to IC4 pin 4. On this board, the bank select 0 switch should be OFF.
With this modification, the 16KZ will reset to "Not selected", but a
write to the Bank Select port (port 40h) that selects bank 0 (bit 0
set), will select the 16KZ.

This is necessary because the RDOS-2 PROM on the 16FDC is addressed at
C000h, so RAM at this address must be disabled after reset. But when
you execute (for example) the T command (which tests memory), then
RDOS-2 copies itself into low memory,then writes 01h to port 40h
(which disables the 16FDC PROM, selects bank 0, enabling RAM at
C000h), then copies itself into RAM at C000h, then tests memory.

This mod also allows 64K of RAM for CDOS...

2a: The same modified 16KZ also came with three 1.5K pullup resistors
on the back, pulling up 9 and 16 of IC34, pin 9 of IC51. It also has a
1.5K resistor on the back from +5V to ground directly behind C9. (I
suspect they had a problem with this capacitor discharging too slowly
after power-down.) I don't know what these are for, but they have the
look of factory mods. My other 3 16KZs do not have these resistors
added.

3: The 16FDC is not compatible with the front panel of the Z-1 because
it grounds pin 20, which is the S-100 PROT signal. With this signal
grounded, the Examine/Examine Next, Deposit/Deposit Next, and Single
Step functions of the front panel are disabled.

The solution is to cut the fat trace on the 16FDC that goes to S-100
pin 20.

I think pin 20 (and pin 70) is one of the signals that was changed
from the original Altair definitions when the IEEE-696 standard came
about. The Altair's front panel has a switch that generates PROT on
pin 20 and UNPROT on pin 70. The Imsai has no such switch, but it
still looks at the signal on pin 20 to enable/disable front panel
operations. (The Imsai front panel grounds pin 70.) The Cromemco Z-1
front panel is an Imsai front panel.

Cheers,
Martin

Michael Leon Peterson

unread,
Aug 23, 2011, 1:21:38 PM8/23/11
to Cromemco
Please see red below:
 
-----Original Message-----
From: Martin Eberhard
Sent: Tuesday, August 23, 2011 9:39 AM
To: Cromemco
Subject: Z-1 Observations
 
As I have been bringing up my Cromemco Z-1, I learned a few things
worth noting here:
 
1: The Z-1 has the Cromemco Blitz Bus. However, memory accesses become
unreliable if the 16KZ boards are too far away from the ZPU. Memory
accesses seem solid when the 16KZs are right next to the CPU, and
pretty bad when 8 slots away. Question: Did Cromemco ever provide an
active terminator board for the Blitz Bus? Any other experiences with
this kind of problem?
 
There were capacitors added I believe to the Bus in later versions of it.
 
2: The 16KZ board (as manufactured) is not compatible with the 16FDC,
when the 16KZ is set for address C000h. One of my 16KZ boards came
modified to make it work. The modification is a jumper from IC3 pin 3
to IC4 pin 4. On this board, the bank select 0 switch should be OFF.
With this modification, the 16KZ will reset to "Not selected", but a
write to the Bank Select port (port 40h) that selects bank 0 (bit 0
set), will select the 16KZ.
 
The 16KZ was never designed to run with all 4 boards, totally 64K, it was intended to be 48K.  I had a System 3 purchased with 32K, 2 (16KZ) and build from a kit the 2 memory boards (16KZ)  in 1978 to bring up to 64K and remember this issue.
 
This is necessary because the RDOS-2 PROM on the 16FDC is addressed at
C000h, so RAM at this address must be disabled after reset. But when
you execute (for example) the T command (which tests memory), then
RDOS-2 copies itself into low memory,then  writes 01h to port 40h
(which disables the 16FDC PROM, selects bank 0,  enabling RAM at
C000h), then copies itself into RAM at C000h, then tests memory.
 
This mod also allows 64K of RAM for CDOS...
 
2a: The same modified 16KZ also came with three 1.5K pullup resistors
on the back, pulling up 9 and 16 of IC34, pin 9 of IC51. It also has a
1.5K resistor on the back from +5V to ground directly behind C9. (I
suspect they had a problem with this capacitor discharging too slowly
after power-down.) I don't know what these are for, but they have the
look of factory mods. My other 3 16KZs do not have these resistors
added.
 
I do not have any notes on the 16KZ mods to refer to, maybe someone else does.
 
3: The 16FDC is not compatible with the front panel of the Z-1 because
it grounds pin 20, which is the S-100 PROT signal. With this signal
grounded, the Examine/Examine Next, Deposit/Deposit Next, and Single
Step functions of the front panel are disabled.
 
The solution is to cut the fat trace on the 16FDC that goes to S-100
pin 20.
 
I think pin 20 (and pin 70) is one of the signals that was changed
from the original Altair definitions when the IEEE-696 standard came
about. The Altair's front panel has a switch that generates PROT on
pin 20 and UNPROT on pin 70. The Imsai has no such switch, but it
still looks at the signal on pin 20 to enable/disable front panel
operations. (The Imsai front panel grounds pin 70.) The Cromemco Z-1
front panel is an Imsai front panel.
 
Cheers,
Martin
 
--
You received this message because you are subscribed to the Google Groups "Cromemco" group.
To post to this group, send email to crom...@googlegroups.com.
To unsubscribe from this group, send email to cromemco+u...@googlegroups.com.
For more options, visit this group at http://groups.google.com/group/cromemco?hl=en.
 

MikeS

unread,
Aug 23, 2011, 2:13:53 PM8/23/11
to crom...@googlegroups.com
Thank you for those useful tips, Martin!

If I may make one suggestion: it might avoid some confusion when discussing
this sort of thing if you mentioned the revision and mod levels of the cards
involved.

It sounds like your 16KZ is a Revision C with mod level 2 or 3:

The three 1.5K resistors around IC34 and 51 were added to speed up switching
times.

Could that other 1.5K resistor go from 12V (instead of 5V) to ground? I
don't have a modified 16KZ handy, but apparently Mod 2 added a resistor from
12V to ground in order to maintain sufficient load on the regulator during
WDI DMA operations.

Apparently Mod 3 replaced R24, 2.2K with 680 Ohms to maintain sufficient
current through a 1N5231 Zener on the -18V line.

I've never had any bus-related problems, but I've never been lucky enough to
have a Z-1; then again, I've also always made it a rule to keep the CPU and
memory cards adjacent.

Sounds like you're having fun; well done!

mike

Martin Eberhard

unread,
Aug 23, 2011, 4:26:23 PM8/23/11
to Cromemco

> If I may make one suggestion: it might avoid some confusion
> when discussing this sort of thing if you mentioned the revision
> and mod levels of the cards involved.
>
> It sounds like your 16KZ is a Revision C with mod level 2 or 3:

Will do! These 16KZs are indeed Rev C, unknown mod level. (How do I
tell? I think any Mod Level stickers have fallen off.) The 16KZ with
modifications is stamped "Cromemco Factory Assembled and Passed".

The other three do not have this stamp, though they look factory-
assembled. They appear to be older units, and all three appear to be
the same age. My bet is the older 3 came with the Z-1, and the last
one was added later to allow 64K operation for a later version of CDOS
or for Cromix - and that Cromemco made the modification to the 4th
16KZ to support the original customer.

The 16FDC is rev E. The Mod Level sticker is somewhat erased. Maybe
rev 2? But something smudgy had been stamped on the sticker as well.
It's a good bet that the 16FDC did not come with the Z-1, and a
reasonable bet that the Z-1 came with a 4FDC - because the PFD was
originally sold with the 4FDC. And I bet that 4FDC worked fine with
the front panel. If anyone has a 4FDC, take a look at S-100 bus pin 20
for me - does the 4FDC ground this signal?

The ZPU is Rev E, with a low serial number: 1290.

> Could that other 1.5K resistor go from 12V (instead of 5V) to ground?
> I don't have a modified 16KZ handy, but apparently Mod 2 added a
> resistor from 12V to ground in order to maintain sufficient load on the
> regulator during WDI DMA operations.

Yes indeed, it goes to +12V. Just testing to see if anyone was paying
attention ;-) Actually, I can't read the writing on the 3-terminal
regulators anymore without reading glasses :-(

Martin

Bill Sudbrink

unread,
Aug 23, 2011, 8:13:14 PM8/23/11
to crom...@googlegroups.com
Martin Eberhard wrote:
>
> 1: The Z-1 has the Cromemco Blitz Bus. However, memory accesses become
> unreliable if the 16KZ boards are too far away from the ZPU. Memory
> accesses seem solid when the 16KZs are right next to the CPU, and
> pretty bad when 8 slots away. Question: Did Cromemco ever provide an
> active terminator board for the Blitz Bus? Any other experiences with
> this kind of problem?

I got tired of messing with the cromemco DRAM and switched to a 64K SRAM
board. The IMSAI power supply has more than enough juice to handle it.
Faster for the Dazzler too.

> 2: The 16KZ board (as manufactured) is not compatible with the 16FDC,
> when the 16KZ is set for address C000h. One of my 16KZ boards came
> modified to make it work. The modification is a jumper from IC3 pin 3
> to IC4 pin 4. On this board, the bank select 0 switch should be OFF.
> With this modification, the 16KZ will reset to "Not selected", but a
> write to the Bank Select port (port 40h) that selects bank 0 (bit 0
> set), will select the 16KZ.

With the "standard" (big grin on that "standard" by the way) S-100 signals,
using the phantom signal was the way to go. Just tie the chip enable from
the RDOS ROM to the /PHANTOM line.

> This is necessary because the RDOS-2 PROM on the 16FDC is addressed at
> C000h, so RAM at this address must be disabled after reset. But when
> you execute (for example) the T command (which tests memory), then
> RDOS-2 copies itself into low memory,then writes 01h to port 40h
> (which disables the 16FDC PROM, selects bank 0, enabling RAM at
> C000h), then copies itself into RAM at C000h, then tests memory.
>
> This mod also allows 64K of RAM for CDOS...

Note that the standard CDOS boot disables the RDOS ROM on the 16FDC.
You don't have to do it manually with 'T'. I run 64K (or 56K or whatever
it comes out to) also.

> 3: The 16FDC is not compatible with the front panel of the Z-1 because
> it grounds pin 20, which is the S-100 PROT signal. With this signal
> grounded, the Examine/Examine Next, Deposit/Deposit Next, and Single
> Step functions of the front panel are disabled.
>
> The solution is to cut the fat trace on the 16FDC that goes to S-100
> pin 20.

My front panel already addressed this when I got it. The original
owner cut the trace on the front panel board.

Bill

Martin Eberhard

unread,
Aug 23, 2011, 8:59:20 PM8/23/11
to Cromemco
>With the "standard" (big grin on that "standard" by the way) S-100
> signals,using the phantom signal was the way to go. Just tie the
> chip enable from the RDOS ROM to the /PHANTOM line.

Yes, I completely agree - this is what I have done on my Sol20/Helios
machine and on one of my Altairs. Works perfectly, very low power. But
because this Z-1 is such a relic, even amongst Cromemco machines, I
thought it would be cool to keep it purely Cromemco. I might change my
mind as I fight these ^%#^! 16KZ boards - right now, they seem to
misbehave when hot - and they get plenty hot with use.

I have fought with the idea of switching in a 4FDC because it is the
same vintage as the Z-1. But I am afraid that if I do, the format of
the diskettes will be incompatible with my other machines, especially
the Dell machine that I use for Image Disk. My reading of the specs
for the WD1771 disk controller chip at the heart of the 4FDC leads me
to believe that it will have real incompatibility issues with
controllers/disks that are IBM 3740-compliant. The WD1793 at the heart
of the 16FDC, on the other hand, is 3740 compliant. Hmm...

Bill Sudbrink

unread,
Aug 24, 2011, 1:12:30 AM8/24/11
to crom...@googlegroups.com
Martin Eberhard wrote:
>
> I have fought with the idea of switching in a 4FDC because it is the
> same vintage as the Z-1. But I am afraid that if I do, the format of
> the diskettes will be incompatible with my other machines, especially
> the Dell machine that I use for Image Disk. My reading of the specs
> for the WD1771 disk controller chip at the heart of the 4FDC leads me
> to believe that it will have real incompatibility issues with
> controllers/disks that are IBM 3740-compliant. The WD1793 at the heart
> of the 16FDC, on the other hand, is 3740 compliant. Hmm...

I have several 4FDCs sitting in bags. They are all flakey as bakery
fresh croissants. My first IMSAI was fitted with one (same /CE->/PAHNTOM
modification (that's where I learned it)). I spent a lot of time trying
to get the system stable before I gave up and switched to the 16FDC.

Bill

Michael Leon Peterson

unread,
Aug 24, 2011, 9:46:21 AM8/24/11
to crom...@googlegroups.com
All,

The biggest problem with ALL the really old Cromemco boards was the socket
brand that they were using, they would cause intermittent connection issues
and heat related issues. In later boards Cromemco moved to a different
brand that was much much more reliable and very hard to remove chips they
were tight, but I cannot remember the brand name, latter they removed the
sockets and only had them on chips that where intended on removal or update.
Problem with this change it was hard to find the bad chip by the
remove/replace repair process. Also the later boards did NOT like to have
parts removed, you would end up removing the pads...... they than improve
the manufacturing of the boards. It was a learning process....

Michael

Bill

--

Bill Sudbrink

unread,
Aug 24, 2011, 9:56:06 AM8/24/11
to crom...@googlegroups.com
Michael Leon Peterson wrote:
> The biggest problem with ALL the really old Cromemco boards was the
> socket brand that they were using, they would cause intermittent
connection
> issues and heat related issues. In later boards Cromemco moved to a
different
> brand that was much much more reliable and very hard to remove chips
> they were tight, but I cannot remember the brand name, latter they
removed
> the sockets and only had them on chips that where intended on removal or
> update.
> Problem with this change it was hard to find the bad chip by the
> remove/replace repair process. Also the later boards did NOT like to
> have parts removed, you would end up removing the pads...... they than
> improve the manufacturing of the boards. It was a learning process....

The above is certainly true... (also note the early TI 7400 series chips
with
the pins that tarnish to a lovely black and become brittle) but in the case
of
the 4FDC, the floppy controller chip (part number escapes me at the moment)
was known to be a poor design, making 5 1/4 inch operation iffy and 8 inch
operation almost impossible.

Bill

Martin Eberhard

unread,
Aug 24, 2011, 10:32:00 AM8/24/11
to Cromemco
> The biggest problem with ALL the really old Cromemco boards was
> the socket brand that they were using, they would cause intermittent
> connection issues and heat related issues.

Yes I was afraid of that. The sockets on all my Cromemco boards look
like the cheap Texas Instruments sockets that were nfg the day they
were made. And the pins of many of the ICs are indeed tarnished black.
Has anybody tried any contact cleaners in these sockets? Removing and
replacing all those sockets looks like quite the chore :-(

Bill Sudbrink

unread,
Aug 24, 2011, 10:47:09 AM8/24/11
to crom...@googlegroups.com
Martin Eberhard wrote:
> Yes I was afraid of that. The sockets on all my Cromemco boards look
> like the cheap Texas Instruments sockets that were nfg the day they
> were made. And the pins of many of the ICs are indeed tarnished black.
> Has anybody tried any contact cleaners in these sockets? Removing and
> replacing all those sockets looks like quite the chore :-(

Pull chips. Inspect carefully for cracked/broken pins. If pins ok, dip
them in Caig DeoxIt and reinsert carefully (don't bend or break pins on
reinsertion). In my experience, many years of flawless operation follows.

Bill

Michael Leon Peterson

unread,
Aug 24, 2011, 12:05:52 PM8/24/11
to Cromemco
If you try the socket replace make sure you buy a professional de-solider
tool, do not use a pullet type, mechanical suction, needs to be gentle and
constant air, otherwise you will ruin the solider pads.....

-----Original Message-----
From: Martin Eberhard

--

Martin Eberhard

unread,
Aug 24, 2011, 3:28:10 PM8/24/11
to Cromemco

> If you try the socket replace make sure you buy a professional de-solider
> tool, do not use a pullet type, mechanical suction, needs to be gentle and
> constant air, otherwise you will ruin the solider pads.....

You bet - learned that decades go when I worked as a technician at a
computer store. these days I use a Hakko 808 desoldering tool, which
is pretty gentle and does not jam.

But I will try the Deoxit approach first, per Bill's suggestion.

By the way, I really appreciate the useful hints from this group!

Martin

Martin Eberhard

unread,
Aug 25, 2011, 1:51:59 AM8/25/11
to Cromemco
>With the "standard" (big grin on that "standard" by the way) S-100
> signals,using the phantom signal was the way to go. Just tie the
> chip enable from the RDOS ROM to the /PHANTOM line.

Thanks for the tip, Bill. I did more or less as you suggested: I have
a Tanner/DRC 64K SRAM board. This board has a "half-phantom" option,
where the phantom signal only hides 1/2 the board. Originally, this
hid the low 32K, but someone modified it so that phantom hid the high
32K. Perfect... I wired the 16FDC ROM enable signal to the S-100
phantom line, such that whenever the LED on the 16FDC was lit, the
phantom signal was active. Now, the system has 32K when running from
ROM, 64K otherwise.

And the flaky behavior of my Z-1 seems to be gone. This makes me
confident that the problem is on a 16KZ memory board - not on my
16FDC. Good news. Now I will wash the IC pins on every chip on the
16KZ boards with Deoxit and see if I can make them reliable - as soon
as my can of Deoxit arrives...

By the way: the mod to the 16KZ that allows 64K with the 16FDC (I
described this in an earlier post) is actually recommended in my old
version of the CDOS manual - it was a factory-designed mod.



.

Bill Sudbrink

unread,
Aug 26, 2011, 3:05:56 PM8/26/11
to crom...@googlegroups.com
Martin Eberhard wrote:
> Thanks for the tip, Bill. I did more or less as you suggested: I have
> a Tanner/DRC 64K SRAM board. This board has a "half-phantom" option,
> where the phantom signal only hides 1/2 the board. Originally, this
> hid the low 32K, but someone modified it so that phantom hid the high
> 32K. Perfect... I wired the 16FDC ROM enable signal to the S-100
> phantom line, such that whenever the LED on the 16FDC was lit, the
> phantom signal was active. Now, the system has 32K when running from
> ROM, 64K otherwise.

One thing to note... if you ever want to install another /PHANTOM
generating board (you can have multiples) you will need to replace
the wire with a diode so that the other /PHANTOM doesn't activate
the ROM on the 16FDC.

> And the flaky behavior of my Z-1 seems to be gone. This makes me
> confident that the problem is on a 16KZ memory board - not on my
> 16FDC. Good news. Now I will wash the IC pins on every chip on the
> 16KZ boards with Deoxit and see if I can make them reliable - as soon
> as my can of Deoxit arrives...

Another note... the DeoxIT dip will not make any appreciable difference
in the appearance of the black pins but it makes a world of difference
in the performance of the chips. You might just chalk this up to the
removal and resocket of the chips but I found that the "helpful" effect
of resocketing the tarnished chips wore off over a few weeks/power cycles.
The DeoxIT made the effect permanent (so far anyway).

Bill

Reply all
Reply to author
Forward
0 new messages