Another build

Use this forum to describe how you are using, or planning to use, the parts kits generously donated to a number of members by Lez Anderson
Martin A
Posts: 799
Joined: 09 Nov 2013 21:03

Re: Another build

Post by Martin A »

74AHC is pretty quick 9ns for the 08 and 11.5 for the 138.

The real holdup is the time it takes the VDG to release the address bus to the video memory.

Maybe if I were doing a re-design I'd use dual port memory there. Vram access time would them be the same as the rom. And allow room to run a couple of Mhz faster.

Still even as it is, it 4-6 time faster than the original Atom, that's still a useful improvement!
lezanderson
Posts: 186
Joined: 14 Aug 2012 15:31

Re: Another build

Post by lezanderson »

Martin A wrote:74AHC is pretty quick 9ns for the 08 and 11.5 for the 138.

The real holdup is the time it takes the VDG to release the address bus to the video memory.

Maybe if I were doing a re-design I'd use dual port memory there. Vram access time would them be the same as the rom. And allow room to run a couple of Mhz faster.

Still even as it is, it 4-6 time faster than the original Atom, that's still a useful improvement!
**********************************************************************************************************

Being a Penny-Pincher....It's the cost of the 74AHC ICs I was thinking of !

And as EPM7032SLC44-10 (10ns) are only $2.50 each it would reduce PCB size and costs.

IDT7005, 35ns (8Kx8bit Dual Port SRAM) cost only a couple of dollars each too..

But as you say, perhaps it's something for the future.
Last edited by lezanderson on 02 Dec 2017 12:19, edited 1 time in total.
lezanderson
Posts: 186
Joined: 14 Aug 2012 15:31

Re: Another build

Post by lezanderson »

For anyone interested in building their own 'Super' Atom

As designed by Martin A.. >>.I've done some rough costing




Prices sourced from eBay.com

W65C02P-10 (10Mhz)..................... $2.50 each
82C55 PIO.................................. $1.05 each
6847 VDP....................................$3.50 each
R6522 VIA...................................$1.60 each (2pcs needed)
UM61256 32K 15ns SRAM.................$2.20 each
HM62256-7 70ns SRAM ...................$1.50 each
27C101-70 70ns 128K EEPROM..........$1.30 each
RTC DS12887 Clock.......................$1.50 each

ICs would cost about $15.0 (about £12.00)


74HC00..................................3pcs $0.99 >>> $0.33 each
74HC08..................................2pcs $0.99 >>> $0.50 each
74HC74..................................10pcs $0.99 >>> $0.10 each
74HC138.................................5pcs $0.99 >>>>$0.20 each
74xx145.................................Probably have to source Farnell or RS
74HC244................................2pcs $0.99 >>>>$0.50 each
74HC245................................5pcs $0.99 >>>>$0.20 each

74HC ICs will cost a couple of Dollars or Pounds, if AHC are used this will increase quite a bit.

Oscillallators, resistors capacitors, plus switches, connectors etc will cost ?? (probably $15 to $20 dollars, about £15.00).

If a PCB is designed /made then it's cost will depend on size and the number of layer ??
lezanderson
Posts: 186
Joined: 14 Aug 2012 15:31

Re: Another build

Post by lezanderson »

Information Only ;

Acorn Atom kit

https://www.ebay.com/itm/Acorn-Atom-Hom ... 2817148043

Cost :$14 about £11 which is reasoable. Obviously you'll need to buy all the smaller ICs and bits and bobs elsewhere...but it's a start.


And :

Dragon 32 /Tandy Color Computer

https://www.ebay.com/itm/Dragon-32-Tand ... Sw~RVaJ6gM

Useful if you want to design and build your own Dragon 32 machine !
Martin A
Posts: 799
Joined: 09 Nov 2013 21:03

Re: Another build

Post by Martin A »

An Update.

Lez has been kind enough to send (via Dave) some different speed CPUs and other parts for testing.

Test 1 was to replace all the AHC logic parts with the more widely available and usually cheaper HC, at that point the Intersil 82C55 PPI that controls the keyboard and VDG refused to take any keyboard input. With the clock reduced to 1mhz it still refused to play ball and was replaced with the supplied NEC 8255AC-2 which is rated to 8mhz and the keyboard sprang into life.

With the system apparently working, I replaced the modern 14mhz CPU with the older 8mhz part.

Test 2 was then to run some simple printing and scrolling tests on that CPU and then increase the clock speed to see where things started to go wrong.1,2 and 4mhz tests all ran perfectly. Running at 6mhz there were problems with a couple of crashes and some display errors when running the test program

Test 3 The 8mhz CPU was replaced with a 10mhz part, and the tests re-run. The system successfully ran all the tests at 6,8 and 10mhz. At 12mhz the overclocked CPU still booted but there was no response to the keyboard.

Test 4 was the original 14mhz part, that showed the same results as the 10. Indicating that the problem running at 12 was probably not due to over-clocking the 10mhz but something else - most likely the PPI.

The final test so far was to put the 10mhz CPU back in, with the 10mhz clock and run the old PCW benchmarks. Benchmark 1 was so quick I had to have an outer loop of 100 to get something long enough I could time - the others benchmarks had th eouter loop reduced to 10. So the adjusted times were:
BM1 0.048s
BM2 0.54s
BM3 1.00s
BM4 1.17s
BM5 1.46s
BM6 1.92s
BM7 3.16s

The board has Atom integer basic and doesn't have the floating point rom, so benchmark 8 couldn't be run. Still to be tested are some slower memory chips.

However some of Lez's earlier questions has been aswered:
HC parts seem to be working nicely
The 8mhz CPU works fine at 4mhz
The 10mhz CPU seems to be working up to and possibly beyond spec.
The NEC PPI seems to be a better match for HC parts than the Intersil one was.
lezanderson
Posts: 186
Joined: 14 Aug 2012 15:31

Re: Another build

Post by lezanderson »

Free Acorn Atom ICs

For anyone wanting to design or build their own homebrew Acorn Atom I’ve got some surplus ICs for the job. Anyone interested should email me (email me at lezanderson@gmail.com) . I can only handout stuff to those that are serious about building a DIY Acorn Atom. First come first served.

65C02 DIP40
6522
82C55
6847
Various SRAMs 8K,32K,64k, etc
Various EPROMs 8K, 16K, 32k etc
EPM7032SLC44-xx CPLD
DS12887 RTC
AY-3-8910
SN76489
7805 Regulator


74HCxx ICs 74HC00, 74HC08, 74HC138, 74HC245 etc

Various Other ICS
Z80 CPUs
HD63C09 CPUs
W65S816-xx CPU
R65C02P4 IC MPU DIP
R6522 IC MPU DIP X 2PCS
MB8877 IC FDC DIP
6845/68B45 DIP
6821/68B21 DIP
68B50 DIP
68B54 DIP
146818 DIP
SAA5050 DIP
D7002C DIP
SAA1099 Sound Chips
7805 REGULATORs


***********************************************************



If you’re repairing Acorn Atom/BBC computers then :

* Sadly I can’t supply ICs to repair old Atoms.. Best to go to eBay.com

eBay.com prices (for repairing)

R65C02P4 …….………... $1.90
R6522………………...…….$1.50
82C55………………...…….$1.05
6847………..………..…….$3.50
6845.....................$3.00 to $8.00 depending on type
Martin A
Posts: 799
Joined: 09 Nov 2013 21:03

Re: Another build

Post by Martin A »

Further testing done on the video memory.

The video ran is pushed harder that the main ram. It's shared between the VDG and the CPU with the VDG being forced off when the CPU wants to access it.

The Atom's OS takes care to avoid contention by waiting for the vertical blank to write to the screen, which avoids snow on the display too. However poking directly into screen memory bypasses the OS.

Using the modern 55ns ram and a 10mhz cpu and clock poking a full 6k into the screen memory produces the expected snow, but also produced other memory related errors and crashes. Slowing down to 8mhz and the snow still appeared but the crashes went away.
Switching to a slower, older 70ns rated ram from Lez, and the errors/crashes returned, there was also actual video memory corruption on certain parts of the display. Switching down to 6 then 4mhz and the problems continued. It looks like the older ram isn't an option for the video memory on the current setup. Though I'm not quite sure why the last test failed, it's more likely to be a particular timing contraint not being met, as over all it ought to have been fast enough.
User avatar
JonB
Posts: 146
Joined: 08 Nov 2013 11:16

Re: Another build

Post by JonB »

Gosh Martin you have too much spare time! :lol:
Martin A
Posts: 799
Joined: 09 Nov 2013 21:03

Re: Another build

Post by Martin A »

JonB wrote:Gosh Martin you have too much spare time! :lol:
So much spare time, that I've "finished" this incarnation of the video board.

The 3x 74 series chips on the right are there to isolate the 8k video ram when the CPU isn't accessing it. Next to the memory is the VDG, flip flop and 14.4mhx 4x colour burst frequency Oscillator.

The 3 phono sockets provide a colour component video output. Yellow plug in Luminance, red in Pr, white in Pb. Since the TV here is happy with NTSC input like most modern sets, this design is a lot simpler than the old Atom colour board. As there is no 60hz to 50hz correction, and no PAL clock. The output from the VDG can more or less be used as it is.

Under the ram is the 82C55 PPI and the 74145 (or 7445) keyboard driver.

Above the keyboard header is a 2nd oscillator, which is 1.2288mhz and a 74HC4040 chip which divides that by 512 to get the 2400hz reference frequency for the tape interface.

The lower 2 connectors are tape in and tape out. The 8 pin device is the op-amp for tape input.

The remaining connection is a 2 pin header for the speaker, very much in the same style as the spectrum, as it's driven directly from one of the PPI pins.
Colour video board
Colour video board
colour output.jpg (57.55 KiB) Viewed 10359 times
Martin A
Posts: 799
Joined: 09 Nov 2013 21:03

Re: Another build

Post by Martin A »

The board containing the rest of the I/O is now ready for testing.

Its got a 6522VIA connected to a parallel printer port, and a BBC Micro style User port on a 20 way header. The atom rom contains code to drive the Printer, on channel A. Channel B is unused and free for the programmer and the user port seemed to be the best option for connecting it up.

The other 40 pin chip is the 2nd 82C55 PPI connected to the IDE with the same layout as CFX. The PPI is mapped in to the free I/O block at BC00 in the Atom memory map.

All it needs is some software to drive it.
Wiring side - not overly complex
Wiring side - not overly complex
Printer-IDE wiring.jpg (116.33 KiB) Viewed 10315 times
Top view of the VIA and PIA
Top view of the VIA and PIA
Printer-IDE top.jpg (84.83 KiB) Viewed 10315 times
Post Reply