New Interface - Yet Again
8 messages · 2004-07-12 → 2004-07-19 · Yahoo Group era · View archive on archive.org
Participants: Jeff, aralbrec, Jack Boatwright, Philip Kendall, Jarek Adamski, joe schmoe
Preserved from the Timex/Sinclair 2068 Yahoo Group (2001–2019), which is no longer online. Text reproduced from the archive.org archive; email addresses masked.
Messages
1. New Interface - Yet Again
Jeff · Mon, 12 Jul 2004 13:56
Jack;
You are probably right about making this interface compatible with
the Spectrum series computers. I've been half-heartedly looking at
the Speccy specs (I've ALWAYS wanted to say that :)) but I guess I'll
need to be more intentional about it. This is one reason I am
seriously considering making the thing reprogrammable to accomodate
the Commodore series computers as well.
One problem with a project like this is balancing the feechur set
against the final price of the product. I don't expect to make any
money on this but I can't afford to lose money on each unit either.
Building these things in low volumes will make them fairly expensive -
my best guess is $100 to $150 per. A four layer PCB (anything less
with today's high performance FPGAs is just not practical) of the
size needed would be approximately $50 alone. While not too bad for
those of us who are still working and just a little crazy, it might
be too expensive for those on fixed incomes or possessing a bit more
sanity. I guess the question is how much do you think the Sinclair
market would pay for a memory expansion/IDE/printer/serial
interface? I expect the other goodies that some of us want
(ethernet, USB, enhanced multi-tasking, etc.) could be implemented on
expansion boards or an FPGA upgrade for a nominal cost.
As an aside - what are your thoughts on designing a 640X480 256
color VGA display card for the Sinclairs. My thoughts would be to
offload scrolling and BitBLTing to the display card hardware so the
Sinclair would actually be able to process some data. I bring this
up because people keep complaining about the quality of the 2068
display and, with the ubiquity of VGA monitors, this may be a way to
address that issue.
As for bringing others up to speed on this: I taught electronics
for eight years, so writing tutorials and such would actually be fun.
2. Re: [ts2068] New Interface - Yet Again
Jack Boatwright · Tue, 13 Jul 2004 10:27
Jeff,
I'm all for VGA display, even though I have pretty decent RGB
display going right now.
As for the cost of the unit. I suppose it's really a concern,
however for the feechur set discussed and the long wait we've had
from Timex, I am favorable for moving ahead and will offer seed
money if you need it. We've had a few folks already state that
they'd be interested in one and after it has been produced,
debugged, etc., I'm sure more folks would step forward. Maybe I'm
only dreaming, but I think so.
I suppose there might be some folks in Europe that would be
interested as well. I know of a few who are not completely
Spectrum oriented and I think the Timex (& UK2086) computers are
gaining more respect. Possibly announcing on css would answer
that. (I know that Phil is part of this group, maybe he could
shed light on this)
The thought of having something compatible with both the Commodore
& Spectrum is an interesting concept. I can't wait for the flame
wars over that. :-D BTW - please make it compatible with the
Plus/4 ... that's the only Commodore I have.
In making Spectrum compatible ... I know that a lot of folks in
this group probably only have the TS machine, so making for the
Speccy may leave some out unless it is also twistor-compatible.
Jack
----- Original Message -----
From: "Jeff" <[email]>
To: <[email]>
Sent: Monday, July 12, 2004 6:56 AM
Subject: [ts2068] New Interface - Yet Again
> Jack;
> You are probably right about making this interface compatible
with
> the Spectrum series computers. I've been half-heartedly looking
at
> the Speccy specs (I've ALWAYS wanted to say that :)) but I guess
I'll
> need to be more intentional about it. This is one reason I am
> seriously considering making the thing reprogrammable to
accomodate
> the Commodore series computers as well.
>
> One problem with a project like this is balancing the feechur
set
> against the final price of the product. I don't expect to make
any
> money on this but I can't afford to lose money on each unit
either.
> Building these things in low volumes will make them fairly
expensive -
> my best guess is $100 to $150 per. A four layer PCB (anything
less
> with today's high performance FPGAs is just not practical) of
the
> size needed would be approximately $50 alone. While not too bad
for
> those of us who are still working and just a little crazy, it
might
> be too expensive for those on fixed incomes or possessing a bit
more
> sanity. I guess the question is how much do you think the
Sinclair
> market would pay for a memory expansion/IDE/printer/serial
> interface? I expect the other goodies that some of us want
> (ethernet, USB, enhanced multi-tasking, etc.) could be
implemented on
> expansion boards or an FPGA upgrade for a nominal cost.
>
> As an aside - what are your thoughts on designing a 640X480
256
> color VGA display card for the Sinclairs. My thoughts would be
to
> offload scrolling and BitBLTing to the display card hardware so
the
> Sinclair would actually be able to process some data. I bring
this
> up because people keep complaining about the quality of the 2068
> display and, with the ubiquity of VGA monitors, this may be a
way to
> address that issue.
>
> As for bringing others up to speed on this: I taught
electronics
> for eight years, so writing tutorials and such would actually be
fun.
>
>
>
> Yahoo! Groups Sponsor
> ADVERTISEMENT
>
>
>
>
>
> ----------------------------------------------------------------
----------------
> Yahoo! Groups Links
>
> a.. To visit your group on the web, go to:
> http://groups.yahoo.com/group/ts2068/
>
> b.. To unsubscribe from this group, send an email to:
> [email]
>
> c.. Your use of Yahoo! Groups is subject to the Yahoo! Terms
of Service.
>
>
3. Re: New Interface - Yet Again
aralbrec · Tue, 13 Jul 2004 19:15
--- In [email], "Jeff" <jburrell7@y...> wrote:
> You are probably right about making this interface compatible
with
> the Spectrum series computers. I've been half-heartedly looking at
> the Speccy specs (I've ALWAYS wanted to say that :)) but I guess
I'll
Specifying Speccy specs specifically for spectrophiles
is spectacular for the speccy scene.
Sorry, I'm not a poet :-)
Jeff, regarding a previous post which I may or may
not have time to answer this afternoon, I do believe
that using the DOCK to map in the hardware would be a
mistake for a number of reasons. One is it would
introduce incompatibility with anything plugged into
the cart slot. I know you've mentioned adding a cart
slot to the board to mitigate this, but IMO that's
an ugly and unnecessary solution.
Another has to do with compatibility with other systems.
The 48K Spectrum has a /ROMCS signal which is very similar
to the TS2068's /BE signal but only acts on the bottom 16K
of memory. This is the only route to add memory to the
original Spectrum without hardware mods, which Jarek has
suggested in another post. Having a simple signal that
says "disable all internal memory" is exactly what the
doctor ordered if you want to make this interface "easily"
portable to other systems. Simulating the DOCK bank in an
FPGA is easy enough, but it's like it's adding a second
unnecessary level to the mix. I am less familiar with
the 128K spectrums, but such a signal would also be
useful there.
Finally, it kind of violates the esthetics of the machine.
The TS2068 has the logic sitting on its board to support
bank switching and DMA using the /BE method. I would
like to support the distinction between internal and
external that Timex made in its hardware as it just makes
sense :-). Ok, this one is a little weak.
> need to be more intentional about it. This is one reason I am
> seriously considering making the thing reprogrammable to accomodate
> the Commodore series computers as well.
A thought here. I would love to have a 200K
gate FPGA at my disposal and at $30, I believe
the price is right. The trouble is, as you
pointed out, you can't get anything near that
size in 5V. The closest thing to TTL is the
LVTTL or CMOS LVTTL output configurations
of the chip (3V basically). I don't know if
you can drive the 2068/Spectrum/C64 signals
using CMOS LVTTL -- going rail to rail is
helpful but the noise margins are kind of
skinny. Perhaps the solution is a small
cheap CPLD of some sort sitting between the
FPGA and the computer bus that can perform
level conversion and at the same time unscramble
signals to the computer system attached -- kind of
a programmable twistor board? Just an idea
I'll throw out there.
Another thing: the power demands of such a
large FPGA cannot be met by either the
TS2068 or Spectrum power supply. There would
have to be an external power supply. Another
pain in the ass to consider.
Alvin
4. Re: Re: New Interface - Yet Again
Jeff · Tue, 13 Jul 2004 23:06
Alvin;
I have no problem using the /BE signal to manage the external
memory. I have looked at the Spectrum schematics and you are correct
that the EPROM bank is the only way to expand their memory. If this
is to be done it will probably need 4K banks and will be, to a
greater or lesser degree, a kludge. Kinda makes you appreciate the
2068 and how advanced it was for its time.
Theoretically, the 3.3V LVTTL (5V tolerant) should give you almost
1V of noise margin, but I have been bitten by that in our present
equipment design. One of the FPGA output signals that drove across
our backplane to other FPGA inputs rose smartly to 2.5V and then did
an exponential rise to +5Vish. Several FPGAs used this to start
their FSMs to acquire some data and they reflected/loaded the signal
enough during the squishy rise that it dropped below the switching
threshold. OOPS!!! This caused one of the boards to restart its FSM
and generate error codes. This depended on time, temperature,
voltage levels, etc. and was a mess to find. At any rate, after this
I take the gauranteed output levels on the data sheets only as hints
as to what might be expected under ideal conditons and after offering
a CPLD as a sacrifice to the logic gods.
Having said all of the above, the SN74LVC4245ADW is a nice 5V/3.3V
version of the '245. I think that anything connected to the computer
bus should use the SN74LVC4245ADW and anything on the expansion board
(even the 5V stuff) could probably use the FPGA ouputs as long as we
are careful in how we implement it.
The Spartan II devices are available from Digikey in 30K, 100K, and
200K versions. I'll check but I think they are all footprint
compatible.
I was fully expecting that we would need to provide some sort of
external power supply. I've been trolling the surplus sites on the
web (www.allelectronics.com is one) for power supplies. There are a
number of them that will give +5/+3.3/+12 at the current levels we
would need. I would post regulate the +3.3 to get the +2.5V FPGA
core voltage. The biggest pain is carving the power planes for all
of these voltages.
I think that it is feasible to make this puppy compatible with the
C64/C128. They both have memory banking schemes similar to the
Spectrum (C64) and 2068 (C128 in 128 mode). This might give us an
audience that would make the project more popular.
One thing you should know about is a project called the C-One. It
is an FPGA based reconfigurable computer that is being developed by
some people in the commodore community. The interesting thing about
this is that it will have VGA output and ATAPI hard drive
interfaces. Everything is defined by the FPGA configuration. It may
be possible to drop a Z80 core with the necessary peripherals to make
a 2068 work-alike. It probably wouldn't be completely compatible
with the 2068, but may be an interesting upgrade.
5. Re: [ts2068] New Interface - Yet Again
Philip Kendall · Wed, 14 Jul 2004 11:32
On Tue, Jul 13, 2004 at 10:27:23AM -0700, Jack Boatwright wrote:
>
> I suppose there might be some folks in Europe that would be
> interested as well. I know of a few who are not completely
> Spectrum oriented and I think the Timex (& UK2086) computers are
> gaining more respect. Possibly announcing on css would answer
> that. (I know that Phil is part of this group, maybe he could
> shed light on this)
Personally, I suspect there wouldn't be that much interest; the 'techie'
density on css is fairly low. Obviously, the interface being Spectrum
compatible would help significantly (I've lost track slightly of what
the current plans are on that front).
On the other hand, you're not going to lose anything to posting to css.
The other place which may generate some interest (perhaps more than css)
are the WoS forums (although probably not until the WoS web server
recovers).
Cheers,
Phil
--
The Babylon 5 Mantra: "Ivanova is always right. I will listen to
Ivanova. I will not ignore Ivanova's recommendations. Ivanova is
God. And if this ever happens again, Ivanova will personally rip your
lungs out." Ivanova: Babylon 5: A Voice In The Wilderness, Part 1
6. Re: New Interface - Yet Again
Jarek Adamski · Wed, 14 Jul 2004 14:57
Jeff,
--- In [email], "Jeff" <jburrell7@y...> wrote:
> I have looked at the Spectrum schematics and you are
> correct that the EPROM bank is the only way to expand
> their memory.
There's enough interfaces that disable the only 16kB
ROM only.
The solution I've talked about is external Z80CPU that
disables the internal one. So the Spectrum is used only
as:
1. First 16kB video RAM in #4000..#7FFF (include
TC2048).
2. Second 16kB video RAM in #C000..#FFFF, available
only after OUT #7FFD,#07 (OUT %01XXXXX11111101,%001XX111
- where "X" means don't care).
3. The first 16kB video RAM in #C000..#FFFF, available
also after OUT #7FFD,#05 (OUT %01XXXXX11111101,%001XX101).
4. OUT %01XXXXX11111101,%SXXX changes the displayed
video RAM depending on S bit.
5. Input/output ports with disabling option from the
build-in interfaces and YABUS slots.
So it works like a "video and keyboard card". Perhaps
some interfaces could be connected also in the middle.
A Z80CPU is ~2USD, an external board for it - ~4USD,
case, some simple logic - gives total 10USD and you
have the /BE for ZX Spectrum, TC2048, etc.
Feel free to make project of the interface for 64kB
address area using /BE and a duplication of #F4 and
#FF ports, then I will tell you what should be
changed to make it available for the Spectrum.
Jarek Adamski.
7. Re: New Interface - Yet Again
aralbrec · Fri, 16 Jul 2004 19:19
--- In [email], "Jeff" <jburrell7@y...> wrote:
> Theoretically, the 3.3V LVTTL (5V tolerant) should give you
almost
> 1V of noise margin, but I have been bitten by that in our present
> equipment design. One of the FPGA output signals that drove across
> our backplane to other FPGA inputs rose smartly to 2.5V and then
did
> an exponential rise to +5Vish. Several FPGAs used this to start
> their FSMs to acquire some data and they reflected/loaded the
signal
> enough during the squishy rise that it dropped below the switching
> threshold. OOPS!!! This caused one of the boards to restart its
FSM
> and generate error codes. This depended on time, temperature,
> voltage levels, etc. and was a mess to find. At any rate, after
this
> I take the gauranteed output levels on the data sheets only as
hints
> as to what might be expected under ideal conditons and after
offering
> a CPLD as a sacrifice to the logic gods.
I've never been involved in pcb design but it is
something that makes me nervous that a first
board won't work :-) Especially if you plan
to connect this to multiple different computers
that will place different TTL loads on the
signals, but I will leave that to you.
> Having said all of the above, the SN74LVC4245ADW is a nice
5V/3.3V
> version of the '245. I think that anything connected to the
computer
> bus should use the SN74LVC4245ADW and anything on the expansion
I will have to have a look at a data sheet. A couple
of things I would be concerned about are rise/fall
times (which can be programmed to 'slow' on the FPGA,
though certainly not on a '245 part) and the granularity
of such a chip. Eg: you may want to tristate individual
signals rather than groups of 8.
The CPLD solution would solve this, allow 'twistoring'
for attaching the same pcb to multiple computer buses
and might help in FPGA configuration.
>board
> (even the 5V stuff) could probably use the FPGA ouputs as long as
we
> are careful in how we implement it.
Okay.
> The Spartan II devices are available from Digikey in 30K, 100K,
and
> 200K versions. I'll check but I think they are all footprint
> compatible.
One thing I can guarantee right now -- I will
blow mine up. Do these parts come in packages
that can be fitted into a socket? Would help
with upgrading as well.
Should also check if Xilinx's free tools
can target all these parts.
> I was fully expecting that we would need to provide some sort of
> external power supply. I've been trolling the surplus sites on the
> web (www.allelectronics.com is one) for power supplies. There are
a
> number of them that will give +5/+3.3/+12 at the current levels we
> would need. I would post regulate the +3.3 to get the +2.5V FPGA
> core voltage. The biggest pain is carving the power planes for all
> of these voltages.
So we need space for a power supply and 2 ide drive bays
in the beu box, though maybe the power supply can be
external.
> I think that it is feasible to make this puppy compatible with
the
> C64/C128. They both have memory banking schemes similar to the
> Spectrum (C64) and 2068 (C128 in 128 mode). This might give us an
> audience that would make the project more popular.
It probably can be done.
> One thing you should know about is a project called the C-One.
Yes, I am aware of it. There is a similar project in the
Spectrum compatible Sprinter, but less general than the C1.
The other difference is that it was complete several years
ago but unfortunately is no longer being sold. The C1 is
more a standalone computer in my mind and wouldn't fill
my ts2068 fix :-)
Alvin
8. Re: [ts2068] New Interface - Yet Again
joe schmoe · Mon, 19 Jul 2004 09:27
try using the rgb interface as described in the 2068
technical manual and an rgb monitor and the display
clears right up...
__________________________________
Do you Yahoo!?
Vote for the stars of Yahoo!'s next ad campaign!
http://advision.webevents.yahoo.com/yahoo/votelifeengine/