• Please review our updated Terms and Rules here

Powertran Cortex

astharot said:
Hello friends, i am a TI99 user.. i looked last summer a real powertran cortex and looked that is more near at my computer. i start at use the emulator and i wont try at learn a basic.. now i have a problem.. i used emulator v2.0, and i load a dump disk image with a basic program.. exist a mode for print a listate or extract the file in ascii mode for read it and print? I no have a real powertran in Italy is more rarest :)
thanks Ermanno
Click to expand...

Hi Ermanno. You want to be able to list a BASIC program to a text file on the PC? It's possible using the serial terminal feature of the emulator, but takes a couple of steps:

(1) Load the program into the emulator.
(2) In the emulator, select Terminal > Serial Terminal from the menu. This should display a second window.
(3) Select Terminal > Start Terminal Log from the menu. This will display a Windows file dialog where you specify the file to list the program to. By default the file will have a .trm extension, but it's just a plain text file.
(4) At the Cortex prompt in BASIC, type UNIT 2<Return>. Everything that appears on the screen will now also be sent to the serial terminal window. And everything sent to the serial terminal window will be captured in the log file.
(5) At the Cortex prompt in BASIC, type LIST<Return>. You should be able to work out what's going to happen. ;-)
(6) Select Terminal > Stop Terminal Log from the menu. The plain text log file now contains the program listing.
(7) Typing UNIT -2<Return> at the BASIC prompt will stop output going to the serial terminal window.

I'm expecting someone to point out a far easier way to do it that I've missed!

Stuart.
 
Stuart PERFECT!!!!! Write your solution in software section that i have created GREAT BOSS!!! :)
 
On Sitopway, I've made something in the vicinity of 20 orders from them, and most of them have a few "spares" in the order. I've only twice had a board that was actually bad though (out of about 1,000 boards), so the spares are almost always that--a nice extra bit that just makes this board house that much better in my eyes.
 
Gentlemen,

Herewith revised schematics and PCB for the Mini Cortex.

I think I've reflected all feedback, from CRUCLK handling, to decouplers, to headers for other interface options, etc. I've also added a F18A to the design. Once again your constructive critique would be much appreciated.

The Mini Cortex should be able to run the following software:
- Cortex Basic
- TurboForth
- Monitor
- MDEX
- Unix (up to full version 6)

Possibly, it can run NOS (harddisk version of MDEX) and TXDS (the OS for the TI990/4 mini). It won't run DX10. Possibly, it could run some Geneve software with a modest porting effort.

As a long term project it would be interesting to port INTERACT to the Mini Cortex.
 
Hi

Do you have a Bill of Materials (BOM) list and project costs /costings for your wonderful Mini-cortex ??

Thanks
 
Paul asked me to clarify a few details about the F18A, so here are the questions and answers:

1. Can the F18A can be hooked up to a 9995 without wait states?
A: Short answer, yes. The F18A's host-interface state machine needs about 120ns (worse case) between reads and writes. As long as the host CPU does not have a memory cycles faster than that, it should work fine. After checking the 9995 datasheet, it looks like the F18A should be ok with a 12MHz 9995.

2. There seem to be various versions of the F18A regarding pin size and board orientations?
A: The F18A is wider than the original VDP 40-pin DIP package, and hangs off the socket to the left side (the side with pins 1..20). That is currently the only orientation. I offer two PCB pin styles, short and tall, because when I went to install an F18A into a ColecoVision I noticed that the CV used rather tall capacitors around the DRAM chips. The F18A with short pins required those capacitors to be seriously bent over or removed, thus I found some taller PCB pins that would raise the F18A higher and allow it to fit better in systems with components near the VDP socket. There is no other difference. The tall pins are about as tall as a typical SIP pin header used in hobby electronics and on boards like the Arduino and such. Either way, I highly recommend the F18A be installed in a socket, and leave room to the left of the socket for the F18A profile.

3. The F18A could be a nice prototyping board for hardware floating point. Not sure how to reprogram the F18A, didn't see a JTAG header on the board.
A: The F18A comes with a JTAG header soldered to the board, however the photo on my web store does not show the header (I need to update those photos). You can certainly use the F18A as an FPGA devboard and I was hoping some people would do just that. It only needs 5V and ground to operate, has a simple 8-bit bidirectional bus, 12-bit VGA color, 1MB flash, and several I/O pins. You can develop HDL for it using the free Xilinx ISE WebPack software.
 
ti99/4a said:
Do you have a Bill of Materials (BOM) list and project costs /costings for your wonderful Mini-cortex ??
Click to expand...

I'll work on a detailed BOM. In brief the key parts are:
U1 TMS9995
U2 74LS612
U3 74LS14
U4 AT28C256
U5 628512-55
U6 74HCT688
U7 Sparkfun slim CF Card Board
U8 TMS9902A
U9 Elektor FT232R Board
U10 74LS259
U11 GAL22V10D
U12 74LS74
U13 74LS112
U14 74LS138
U15 F18A Board

Of the above the breakout boards are the most expensive: an F18A is about 75, and the other two are about 25 each. All together the Mini Cortex might cost as much as 200. If you leave out the F18A and use the cable Jim suggested instead of the Elektor board, it might come out at about 100. Obviously it all also depends on what you already have in your parts drawer, whether you source locally or from China, whether you make 1 or 10 PCB's, etc.

Forum member "lezanderson" has a lot of experience with component kits, perhaps he is willing to chime in with estimated costing for a Mini Cortex kit.

Once again, please be aware that this design is unproven as yet. I strongly recommend that if you want to build this you wait until a prototype has been built and is confirmed working.
 
Since a 512K x 8 static ram chip is not much more expensive than a 64K x 8 chip, would you make the PCB able to take either chip, perhaps 2 chips giving either 128KB or 1MB maximum.
Even the Sinclair Spectrum had 128KB ram, and the Cortex was designed to be expandable to 1MB.
 
mnbvcxz said:
Since a 512K x 8 static ram chip is not much more expensive than a 64K x 8 chip, would you make the PCB able to take either chip, perhaps 2 chips giving either 128KB or 1MB maximum. Even the Sinclair Spectrum had 128KB ram, and the Cortex was designed to be expandable to 1MB.
Click to expand...

I forgot to provide a circuit description, sorry: the 628512 is a 512K x 8 static ram chip. There is no room on the PCB for a second ram chip. Even if there were, I'm using one line of the '612 mapper to be a flag to mark a memory page as read only, leaving only 19 address lines = 512KB. This is a deviation from the Cortex design which has a use for Unix (shared memory pages, making 'fork' operations much faster).

Are there applications that would benefit from having 1MB of ram?
 
matthew180 said:
The F18A is wider than the original VDP 40-pin DIP package, and hangs off the socket to the left side (the side with pins 1..20).
Click to expand...

Ah, somehow I convinced myself it was extending to the right. On the left it will have to clear the CF Card breakout board. How long are the long pins? About 10mm?

matthew180 said:
You can certainly use the F18A as an FPGA devboard and I was hoping some people would do just that. It only needs 5V and ground to operate, has a simple 8-bit bidirectional bus, 12-bit VGA color, 1MB flash, and several I/O pins. You can develop HDL for it using the free Xilinx ISE WebPack software.
Click to expand...

That is great news! Do you have a suggestion for a USB-JTAG cable to use?

One last question: I believe you build the boards yourself, do you have a good home technique for mounting SMD parts? I am considering to have the CF Card socket directly on the main PCB, but I do not look forward to soldering 0.025" spaced leads by hand...
 
Last edited:
Just emailed Funkward-tech they can do the ICs for $25 all in (with shipping etc)

U1 TMS9995
U2 74LS612
U3 74LS14
U4 AT28C256
U5 628512-55
U6 74HCT688
U8 TMS9902A
U10 74LS259
U11 GAL22V10D
U12 74LS74
U13 74LS112
U14 74LS138

** Not 512K SRAMs are normally 55ns Speed
AT28C256 are 120ns,150ns,200ns !!
Maybe using a W27C512-45 (45ns) 64K EEPROM may be better than the 32K AT28C256 ??


********

kit $25


B/Rgds,
Paul


Funkward Technology Co.,Ltd
Room 903,B Bldg,Ze Run Hua Ting,Baoli Road,
Buji Town,Longgang District,
ShenZhen, 518112,China
Tel:0086-755-83212558
Fax:0086-755-83212558
Email: info@funkward.com
Facebook: Paul Huye
Twitter: paulwooyee
Skype: paulwooyee


If enough people ask he may pop it onto his website (ebay Shop) !

Price is for one Kit.. you could get a discount if buying more than one !

Hope this is useful ?


*** Best to specifiy EXACT part numbers including speed rating for this like RAM,ROM, GALs etc otherwise you may get something which is too slow !
 
The most cost effective option would be :

128K x 8 Bit SRAM DIP32 (55ns) TC551001-5,KM681000-55 etc <$1.0 each
DS1250 512Kx8bit NVSRAM (70ns) DIP32 about $2.50 each
W27C512-45 64Kx8bit EEPROM (45ns) DIP28 <$0.80 each

Whereas a 512K SRAM 55ns are usually $3.50 each wholesale and the AT28C256 are rather slow >100ns or so !

mnbvcxz said:
Since a 512K x 8 static ram chip is not much more expensive than a 64K x 8 chip, would you make the PCB able to take either chip, perhaps 2 chips giving either 128KB or 1MB maximum.
Even the Sinclair Spectrum had 128KB ram, and the Cortex was designed to be expandable to 1MB.
Click to expand...
 
lezanderson said:
I don't think Sparkfun still stock this ??
Click to expand...

Correct, it is out of production. However, there is still some stock floating around in the retail channel.

The boards you link to are 'IDE mode' boards, the Sparkfun board is 'memory mode'. The two modes are similar but not identical, unfortunately. I have not found another memory mode board.

For the handful of people building the Mini Cortex the remaining stock is probably sufficient. In the unlikely event that there are suddenly tens of people building this, the best solution imho would be to have the CF Card socket directly on the main PCB and to find a PCB manufacturer that is willing to solder on the CF Card socket for a modest fee.

Paul
 
Mini Cortex EPROM AT28C256 ??

With reference to the Mini Cortex....any reason for using the rather slow 28C256 as these are normally 100,120,150ns or so whereas the W27C512-45 is 45ns and the same price (if not cheaper) ! The AT28C256 must clearly need wait states slowing down the system ?
 
lezanderson said:
Just emailed Funkward-tech they can do the ICs for $25 all in (with shipping etc). (...) If enough people ask he may pop it onto his website (ebay Shop) !
Click to expand...

That is very convenient. I hope that answers the question of "ti99/4a" about costing, thank you for looking into this.

lezanderson said:
With reference to the Mini Cortex....any reason for using the rather slow 28C256 as these are normally 100,120,150ns or so whereas the W27C512-45 is 45ns and the same price (if not cheaper) ! The AT28C256 must clearly need wait states slowing down the system ?
Click to expand...

With no wait states, access time for the EPROM must be 120ns or better (TMS9995 timing requires 115ns). The memory mapper takes 40ns, so the RAM must be 70ns or better. The previous breadboard circuit runs without wait states and after a lot of fiddling is running stable. I've added the wait state circuit so that it can run stable without fiddling (and so that I can try a 4Mhz 9995 if I want to -- hence the 55ns RAM). With the wait state (for ROM and I/O access only!) access time can be as slow 450ns.

As for system speed, it is only important that the RAM is fast: the ROM is only used at boot time, afterwards everything runs from RAM.

Why the 28C256? Main reason is that I had those in my parts drawer, any other pin compatible 32Kx8 (E)EPROM will do. A pin compatible 64Kx8 part is possible too, but only the bottom 32K would be used. I've thought about an in-circuit ROM update facility with the 28C256, but that is not in the current design.
 
Using CPLDs instead of Old Slow 74LS ICs ??

As old 74LS TTL ICs now start to look like dinosaurs , things like 74LS612 (40ns !!) perhaps we should consider using more modern CPLDs to do the job ? CPLDs like the Altera MAX 7000 series, namely EPM7128SLC84,EPM7064SLC44 typesd which are cheap and fast.. The EPM7128SLC are normally 10ns or 15ns the EPM7064SLC can be as fast as 5ns !

And as these are cheap ... the EPM7128SLC are $2.0 each, programming them to be a Memory Mappers, CRU decoders etc may be the future ? Especially if using CPUs like the TMS99105/TMS99110 !
 
lezanderson said:
Using CPLDs instead of Old Slow 74LS ICs ??

As old 74LS TTL ICs now start to look like dinosaurs , things like 74LS612 (40ns !!) perhaps we should consider using more modern CPLDs to do the job ? CPLDs like the Altera MAX 7000 series, namely EPM7128SLC84,EPM7064SLC44 typesd which are cheap and fast.. The EPM7128SLC are normally 10ns or 15ns the EPM7064SLC can be as fast as 5ns !

And as these are cheap ... the EPM7128SLC are $2.0 each, programming them to be a Memory Mappers, CRU decoders etc may be the future ? Especially if using CPUs like the TMS99105/TMS99110 !
Click to expand...

If you enjoy playing/designing/tinkering with the hardware, then reducing everything down to just one or two big, fast chips is often not the object of the exercise. Personally, I like a board to look as if its actually doing something, and a board containing just one great big CPLD holds no interest whatsoever. Might as well just run it in emulation on a laptop and not bother with the hardware at all.

Just my $0.02.

Stuart.
 
Stuart said:
If you enjoy playing/designing/tinkering with the hardware, then reducing everything down to just one or two big, fast chips is often not the object of the exercise. Personally, I like a board to look as if its actually doing something, and a board containing just one great big CPLD holds no interest whatsoever. Might as well just run it in emulation on a laptop and not bother with the hardware at all.

Just my $0.02.

Stuart.
Click to expand...

The point of CPLDs is to reduce costs and board size by putting a lot of very small and fiddly TTL ICs onto one chip... in the same way a CPU is all on one chip. The Cost and Speed advantages of using things like GALs and CPLDs can be great.

CPLDs are one of those things you either hate them or love them...but either way they are a VERY useful tool to both the professional and hobby designer ! And learning new electronics skills must be good ?
 
You must log in or register to reply here.
Back
Top