ICE T65/Z80/6809

[RobC]

MY timing in getting an epizza board was pretty bad. Had hoglet's ICE boards made and the epizza and programmer ordered. Just for the vendor to (likely understandably) disappear.

I'm in the same boat - I have Dave's adapter board but, having ordered an EEPIZZA, it never arrived. (At least I got a refund.)

[bprosman]

This might be an alternative though its lacking the serial (FTD) interface.
https://www.ebay.com/itm/DC5V-XILINX-Sp ... 2912966752

Same for this :
https://www.ebay.com/itm/Xilinx-Spartan ... 4652356207

Or this one :
https://www.ebay.com/itm/Mojo-V3-FPGA-D ... 2969705764

[obiwanjacobi]

That last one should be usable, it has USB. Perhaps redefine the pin mappings...

BTW: what is the issue with the EEPROM? I found it (M25P80-VMW6) on ebay: 10 for 7 euros.
https://www.ebay.com/itm/10pcs-ST-M25P8 ... 2535498277

[hoglet]

That last one should be usable, it has USB. Perhaps redefine the pin mappings...

This also seems to have been discontinued by the original manufacturer.

The Eagle schematics and board design are here:
https://alchitry.com/products/mojo-v3

I see a few problems:
- The connectors use different pins for power, so it's more than just changing the .ucf file.
- The connector spacing is 2.1 inches, on the EEPIZZA board it's 1.7 inches.
- The EEPROM is a SST25VF040B which is 4Mbit, so only has space for one design (i.e. not multi-boot).
- It doesn't have the additional 6x2 connector

The only FPGA board I'm aware of with a similar main connecror footprint and power supply connections is the DE0-NANO.

Unfortunately, this costs ~£85 in the UK.

Dave

[obiwanjacobi]

Hmm, I guess you always run the risk of ebay/ali products not being available at any time.
If you want to provide consistency you have to make one yourself (can be a copied design) with still available parts for a reasonable price. Bonus is that you can integrate level shifting and make the EEPROM big enough to hold all the code you need.

@Dave: can you explain what the problem is with switching the brand of FPGA? Are you using IP that is bound to Xilinx tooling?

Edit: I see Farnell still carries the XC6SLX9TGQ144 for around 17 euros (without VAT)...

[hoglet]

@Dave: can you explain what the problem is with switching the brand of FPGA? Are you using IP that is bound to Xilinx tooling?

There's no real technical issue here, it's just rather mundane to do this for about 20 different projects. It's much more than just recompiling, because Altera DCMs are different, and the dual port RAMs are less flexible. I'd just rather spend my time on other things.

I'm still hoping that at some point eepizza will open his ebay shop again as China starts to recover. If this doesn't happen in the next few months, then I think cloning the existing board is the best way to go. Roland's had good experiences using PCBWAY for Atom2K18, and they do double-sided assembly and part sourcing.

In any event, all the projects are open source, so anyone is free to fork them and add support for different boards/FPGA. If you feel motivated to design a new all-in-one board, then you have my blessing.

Dave

[bprosman]

I made some HiRes pictures.

2 Questions I have :

@Dave do you have an idea (or educated guess) how many layers the EEPizza board has ?
My guess is at least 4,with (maybe?) a power and ground plane.

Looking at my board pictures it looks that the voltage regulators are already larger than SOT-223

[hoglet]

@Dave do you have an idea (or educated guess) how many layers the EEPizza board has ?
My guess is at least 4,with (maybe?) a power and ground plane.

I think it's 4-layers.

Looking at my board pictures it looks that the voltage regulators are already larger than SOT-223

Here's the datasheet for the AMS1117:
http://www.advanced-monolithic.com/pdf/ds1117.pdf

The SOT-223 package is 6.3mm x 3.3mm.

Looking at your pixtures, that seems to match what you have.

Do you have any calipers? You could try to measure them.

Dave

[Prime]

I think I've found a bug in the 6809 core on the Godil.

If you set a write to memory watch, it always lists the written data as $FF

With the following watches set :

Code: Select all

>> blist
0: FF90 mask FFFF: Mem Wr Watch (trigger: Always)
1: FF92 mask FFFF: Mem Wr Watch (trigger: Always)
2: FF93 mask FFFF: Mem Wr Watch (trigger: Always)

I get the follwowing results :

Code: Select all

CPU free running...
00.306531 : Mem Wr Watch hit at CE61 writing FF90:FF  .
00.306803 : Mem Wr Watch hit at CE75 writing FF90:FF  .
00.307152 : Mem Wr Watch hit at CE7B writing FF92:FF  .
00.307157 : Mem Wr Watch hit at CE7E writing FF93:FF  .

The first write to $FF90 was $80 from the A register
The second write to $FF90 was $C0 from the A register
The writes to $FF92 and $FF93 where $00 (both clr instructions).

This is on a CoCo 3, but I've used it on there before without problems. I can be certain about what
is being written as this is my code I am debugging

So there seems to be either a small bug or I'm doing something wrong (which is entirely possible!).

Cheers.

Phill.

[hoglet]

So there seems to be either a small bug or I'm doing something wrong (which is entirely possible!).

There is indeed a bug, which only affects the GODIL, which is why I've not noticed it before.

I've committed what I think is a fix to the dev branch:
https://github.com/hoglet67/AtomBusMon/commits/dev

I've also pushed a change that adds an optional address to the next command.

Give this a try and see if it helps; hopefully I haven't introduced any new bugs!

Dave

[zwarder]

Great work going on here, I have old systems vectrex spectrum etc. Was looking for a fluke 6809 pod but this setup would be ideal for fault finding.