Atom 2k18 ?

discussion of games, software, hardware & emulators relating to the Acorn Atom and Acorn System machines.
User avatar
1024MAK
Posts: 8663
Joined: Mon Apr 18, 2011 4:46 pm
Location: Looking forward to summer in Somerset, UK...
Contact:

Re: Atom 2k18 ?

Post by 1024MAK » Fri Mar 15, 2019 9:43 pm

roland wrote:
Fri Mar 15, 2019 8:30 pm
I also added a bunch of capacitors that go the the + and - of the IC's (I don't know the exact English words for it but the description will do, at least I hope so).
Do you mean decoupling capacitors?

Mark

User avatar
hoglet
Posts: 8095
Joined: Sat Oct 13, 2012 6:21 pm
Location: Bristol
Contact:

Re: Atom 2k18 ?

Post by hoglet » Fri Mar 15, 2019 10:38 pm

A couple of possible errors in the audio section.

- C6 can be deleted (it was part of the low pass filter for the SID audio which is now gone).

- The order of the J14/J15 and C13/14 should be swapped so the capacitors are directly connected to the DAC. Then the connections to J11 should come from after the capacitor, but before the jumper. Otherwise there is a big DC offset on the audio signals on J11.

- I'm not sure that C5/C13/C14 need to be so large. The capacitor forms a high pass filter to block DC. The impedance it is driving is ~10Kohms. Which gives a cut-off frequency 0.7Hz. It doesn't need to be this low. I would use something like 0.47uF polyester film type. Electrolytic capacitors are best avoided this type of application .

- I'm not sure that C8 is needed. What's it doing?

(large values like 22uF make more sense when driving an 8R speaker directly - is there where the value came from?)

Dave
Last edited by hoglet on Fri Mar 15, 2019 10:46 pm, edited 3 times in total.

User avatar
roland
Posts: 3138
Joined: Thu Aug 29, 2013 8:29 pm
Location: Born (NL)
Contact:

Re: Atom 2k18 ?

Post by roland » Sat Mar 16, 2019 8:31 am

hoglet wrote:
Fri Mar 15, 2019 10:38 pm
- C6 can be deleted (it was part of the low pass filter for the SID audio which is now gone).
It is deleted.
hoglet wrote:
Fri Mar 15, 2019 10:38 pm
- The order of the J14/J15 and C13/14 should be swapped so the capacitors are directly connected to the DAC. Then the connections to J11 should come from after the capacitor, but before the jumper. Otherwise there is a big DC offset on the audio signals on J11.
I intended it this way. Both the DAC audio channels (stereo?) are available at J9. But if somebody wants to use the DAC for something else than audio then he can disconnect the audio connections and use J11. In that case a change of the voltage level does not appear on the speaker.
hoglet wrote:
Fri Mar 15, 2019 10:38 pm
- I'm not sure that C5/C13/C14 need to be so large. The capacitor forms a high pass filter to block DC. The impedance it is driving is ~10Kohms. Which gives a cut-off frequency 0.7Hz. It doesn't need to be this low. I would use something like 0.47uF polyester film type. Electrolytic capacitors are best avoided this type of application . (large values like 22uF make more sense when driving an 8R speaker directly - is there where the value came from?)
Yes, they come from the original Atom diagram. And I just copied and pasted them. I have changed them.
hoglet wrote:
Fri Mar 15, 2019 10:38 pm
- I'm not sure that C8 is needed. What's it doing?
I don't know where it came from. I think I had looked at the wrong place when I putted it there. In the cassette-in there is a similar capacitor. That might have confused me. In the Atom 2k14 diagram there is no C8 equivalent, so I removed it.

I still have to add a few extra pins near the PiTube interface. That's something for later this weekend, I'll have to go now...

Thanks for your feedback. It saves us a lot of troubleshooting afterwards :)
Last edited by roland on Sat Mar 16, 2019 8:32 am, edited 1 time in total.
256K + 6502 Inside
MAN WOMAN :shock:

User avatar
hoglet
Posts: 8095
Joined: Sat Oct 13, 2012 6:21 pm
Location: Bristol
Contact:

Re: Atom 2k18 ?

Post by hoglet » Sat Mar 16, 2019 9:11 am

roland wrote:
Sat Mar 16, 2019 8:31 am
Thanks for your feedback. It saves us a lot of troubleshooting afterwards :)
You are welcome.

One thing to pay attention to when routing the PCB is the ground wiring. The following paths are most critical:
- the FPGA moduile to the level shifters, and especially U8 and U9
- the FPGA moduile to the SRAM

On a board this large, with fast/modern technology, and only two layers, good ground routing is will make a bit difference. I would say this is much more important than with Atom 2014, as FPGAa are very sensitive to noise spikes.

It might take some time, but you can change the connections to the FPGA to simplify the routing, to allow better grounding.

Dave

User avatar
roland
Posts: 3138
Joined: Thu Aug 29, 2013 8:29 pm
Location: Born (NL)
Contact:

Re: Atom 2k18 ?

Post by roland » Mon Mar 18, 2019 4:56 pm

hoglet wrote:
Fri Mar 15, 2019 8:52 pm
You should add "No Connects" to pins you really intend to be unconnected. Then the ERC will not flag them as errors.
Is there a way to do this only for the current schema? I don't want to mark pins as unconnected in the symbol because that will give other issues, like nc pins connected (in case of a 74*244)....
256K + 6502 Inside
MAN WOMAN :shock:

User avatar
hoglet
Posts: 8095
Joined: Sat Oct 13, 2012 6:21 pm
Location: Bristol
Contact:

Re: Atom 2k18 ?

Post by hoglet » Mon Mar 18, 2019 5:14 pm

roland wrote:
Mon Mar 18, 2019 4:56 pm
Is there a way to do this only for the current schema? I don't want to mark pins as unconnected in the symbol because that will give other issues, like nc pins connected (in case of a 74*244)....
Yes, there is.

I don't have access to Kicad at the moment, but I think it's in the Place menu, and it's called the No Connect flag.

Place one on each pin; the result looks like an X.

Dave
Last edited by hoglet on Mon Mar 18, 2019 5:14 pm, edited 1 time in total.

User avatar
roland
Posts: 3138
Joined: Thu Aug 29, 2013 8:29 pm
Location: Born (NL)
Contact:

Re: Atom 2k18 ?

Post by roland » Mon Mar 18, 2019 10:02 pm

Thanks again, I have found it and reduced the ERC to 4 items. After that I changed the pins of the power connector (J4) to Power Outputs and that solved a recurring error that pins are connected but no pin to drive them. And I had to change the pins of the schottky diodes to Power Input and Power Output to get rid of the markers at the +5V of the FPGA board. Result: 0 warnings and 0 errors in CRC check \:D/

I did find an additional error which was not detected by ERC: the break key (J8 pin 20) was not connected to the RST signal on the bus.

Regarding the debugging, Hoglet asked to add a small header with RDY and SYNC, is that really necessary because those signals are already available at the I/O connectors. I don't mind in adding them, but there has to be a good reason. The board is complicated enough :wink:
256K + 6502 Inside
MAN WOMAN :shock:

User avatar
hoglet
Posts: 8095
Joined: Sat Oct 13, 2012 6:21 pm
Location: Bristol
Contact:

Re: Atom 2k18 ?

Post by hoglet » Mon Mar 18, 2019 10:19 pm

roland wrote:
Mon Mar 18, 2019 10:02 pm
Regarding the debugging, Hoglet asked to add a small header with RDY and SYNC, is that really necessary because those signals are already available at the I/O connectors. I don't mind in adding them, but there has to be a good reason. The board is complicated enough :wink:
It just makes it easier to connect the logic analyzer with the usual 15cm fake "Dupont" cables if all the connections are in close proximity to each other. But if these signals are difficult to route, the forget about it. Getting good ground routing with relatively thick traces is far far more important!

Dave

User avatar
roland
Posts: 3138
Joined: Thu Aug 29, 2013 8:29 pm
Location: Born (NL)
Contact:

Re: Atom 2k18 ?

Post by roland » Tue Mar 19, 2019 5:42 pm

A little while ago we discussed the type of RAM for the new Atom.
hoglet wrote:
Wed Mar 13, 2019 8:54 am
Could you make a small carrier PCB to adapt it to a 32-pin DIP footprint, then socket it?
Edit: There is a design here:
https://www.retrobrewcomputers.org/foru ... &#msg_4823
I had a look at the design and it is quite complicated.
RAM-adapter.png
In theory the adapter can be much simpler just by swapping address and data lines, for example:
A0 connects to A1 (both pin 1)
A1 connects to A16 (both pin 2)
and also some "easy" connections between the data lines.

In theory it doesn't matter how the address lines and data lines are connected as long as you write and read in the same order.

But are there any problems to expect in real life, especially when fast read and writes are used in combination with a FPGA?
256K + 6502 Inside
MAN WOMAN :shock:

User avatar
hoglet
Posts: 8095
Joined: Sat Oct 13, 2012 6:21 pm
Location: Bristol
Contact:

Re: Atom 2k18 ?

Post by hoglet » Tue Mar 19, 2019 6:20 pm

roland wrote:
Tue Mar 19, 2019 5:42 pm
In theory it doesn't matter how the address lines and data lines are connected as long as you write and read in the same order.

But are there any problems to expect in real life, especially when fast read and writes are used in combination with a FPGA?
I don't believe there are any issues with swapping address and/or data lines on fast SRAMs. Use whatever order makes the routing easier.

Dave

User avatar
roland
Posts: 3138
Joined: Thu Aug 29, 2013 8:29 pm
Location: Born (NL)
Contact:

Re: Atom 2k18 ?

Post by roland » Sat Mar 23, 2019 10:16 pm

It has been a bit quiet on this topic but still some progress:
  • I fixed an error in the diagram: the 7445 must be powered with 5V and not 3.3V
  • The +3.3V connections to the Pi Tube should not be connected to the board. The Pi Zero gets 5V from the Atom or an external power source
  • A3 on the Pi Tube connector is at pin 29 and not pin 27
  • I have added a footprint for a fast RAM in SOJ-36 housing (it will be located beneath the FPGA to keep the tracks short and routable.
Further, I tested this circuit for bi-directional level shifting:
LevelShifter.png
This simple circuit works nice. First I just build it and tested it with a multi meter. It worked as designed and intended. After that I connected my Atom 2k15 with 3.3V serial I/O to a Linux host with a 5V serial to USB device with this level shifter in place. I could send characters from the Atom to the Linux host. That went fine.

Unfortunately I don't have the equipment to generate a 3.3V or 5V square wave at different frequencies and a scope to watch the output signal. Well I can use my Atom to generate a square wave but the scope is still lacking. Maybe a nice experiment for one of the next meetings.

I am setting up the components on the PCB to see how it can be populated. And it shows design errors that are passing the ERC check (like connecting the 7445 to 3.3V). Another preliminary setup is in the attached pdf file; a lot of resistors are not placed on the board yet.
Attachments
FPGAtom-pcb-preview.pdf
(552.02 KiB) Downloaded 11 times
FPGAtom-schema-draft.pdf
(519.67 KiB) Downloaded 9 times
256K + 6502 Inside
MAN WOMAN :shock:

User avatar
roland
Posts: 3138
Joined: Thu Aug 29, 2013 8:29 pm
Location: Born (NL)
Contact:

Re: Atom 2k18 ?

Post by roland » Sun Mar 24, 2019 12:37 pm

Today I finalized the diagram. This should be it. The last changes are:
  • Added A16, A17 and A18 to the Atom Bus Expansion connector. It might be a nice to have in future projects.
  • Switched MMC_MOSI and MMC_MISO labels in the FPGA module.
  • Added GND to the VGA connector.
  • Changed pin numbers of J9A (audio connector) to match the footprint.
Now I can starting designing the PCB board :)
Attachments
FPGAtom-schema-final.pdf
(522.81 KiB) Downloaded 13 times
256K + 6502 Inside
MAN WOMAN :shock:

User avatar
oss003
Posts: 2939
Joined: Tue Jul 14, 2009 11:57 am
Location: Netherlands
Contact:

Re: Atom 2k18 ?

Post by oss003 » Sun Mar 24, 2019 1:13 pm

Hi Roland,

I just had a look at the schematic diagram but does the serial port only have TX/RX, no handshake?

Greetings
Kees

User avatar
hoglet
Posts: 8095
Joined: Sat Oct 13, 2012 6:21 pm
Location: Bristol
Contact:

Re: Atom 2k18 ?

Post by hoglet » Sun Mar 24, 2019 1:20 pm

roland wrote:
Sun Mar 24, 2019 12:37 pm
Now I can starting designing the PCB board :)
Be careful with the SOJ footprint that's in KiCad. I used the default one (Package_SO:SOJ-36_10.16x23.49mm_P1.27mm) on my 1MHZ Bus FPGA adapter board. I found yesterday it's not very suitable for hand soldering, because the pads don't extend at all beyond the edge of the device. I was able to solder it, using lots of flux, but it was not as easy as I had expected. Really a custom footprint should have been used here, with longer pads.

gob33
Posts: 25
Joined: Thu Jun 02, 2016 10:15 am
Contact:

Re: Atom 2k18 ?

Post by gob33 » Sun Mar 24, 2019 2:28 pm

Hem. I would like to point out that there is another (clever) way by using one Parallax Propeller chip with a 6502.
This simplify greatly the PCB. Then you program the Propeller in Spin language to manage your I/O
Do a search on Google to see many projects with 6502/65C816 + Propeller.

Examples:
L-Star Plus with Atom emulation mentioned somewhere.
The ProPeddle software defined 6502 computer kit
The Hive project. 3 Propellers (main host, graphic subsystem, general I/O) , extremely detailled documentation how it works

Book:
Programming and Customizing the Propeller (Official Guide)
Last edited by gob33 on Mon Mar 25, 2019 12:50 am, edited 12 times in total.

User avatar
roland
Posts: 3138
Joined: Thu Aug 29, 2013 8:29 pm
Location: Born (NL)
Contact:

Re: Atom 2k18 ?

Post by roland » Sun Mar 24, 2019 3:50 pm

oss003 wrote:
Sun Mar 24, 2019 1:13 pm
I just had a look at the schematic diagram but does the serial port only have TX/RX, no handshake?
That's correct. Just like the Godil this Atom will only have TX and RX, no handshake. I did run out of pins on the FPGA module because of other nice features such as six VGA colour outputs which gives this Atom more colours (although not simultaneously). But if you leave out the ADC then you have a few pins available for hardware handshake at the serial port. It's just a matter of adjusting the FPGA.
hoglet wrote:
Sun Mar 24, 2019 1:20 pm
Be careful with the SOJ footprint that's in KiCad. I used the default one (Package_SO:SOJ-36_10.16x23.49mm_P1.27mm) on my 1MHZ Bus FPGA adapter board. I found yesterday it's not very suitable for hand soldering, because the pads don't extend at all beyond the edge of the device. I was able to solder it, using lots of flux, but it was not as easy as I had expected. Really a custom footprint should have been used here, with longer pads.
Thanks for the warning. I looked at your design which footprint to use, so I have the same one. I will enlarge the pads. I moved the SOJ to the other side of the board because that makes soldering easier if it's ever necessary to replace the RAM. The headers of the FPGA module won't be in the way.
I'll order a memory chip before submitting the board for production (I always do) so I can check that the parts will fit on the board.
gob33 wrote:
Sun Mar 24, 2019 2:28 pm
I would like to point out that there is another (clever) way by using one Parallax Propeller chip with a 6502.
This simplify greatly the PCB. Then you program the Propeller in Spin language to emulate I/O
Thanks, that's also a nice project - such as all projects with 6502 processors :lol:
However, it can't replace this project because I want the new board to be a replacement of my Atom 2k15 design. It should fit in my Atom housing. And, for me that is also important, I want to learn working with FPGA's. The Atom is a good start because I know that computer very well. With my new board also other computers can be emulated. Hoglet mentioned BeebFPGA already but IMHO it should also be possible to implement a Dragon 64 or even an IBM PC (running @5 MHz or so - not suited for Windows 10 :wink: )

The creator of the Parallax Propeller has a Dutch name: Jac Goudsmit. I assume he lives in the Netherlands or perhaps in Belgium. I will try to contact him to see if he wants to visit our TechSat in June and give us a demonstration of his board.
256K + 6502 Inside
MAN WOMAN :shock:

User avatar
oss003
Posts: 2939
Joined: Tue Jul 14, 2009 11:57 am
Location: Netherlands
Contact:

Re: Atom 2k18 ?

Post by oss003 » Sun Mar 24, 2019 5:28 pm

Roland,

what about a switch for cassette/serial mode, then you will have 2 extra signals (cas-in/cas-out) for serial communication if they are freely programmable.

Greetings
Kees

User avatar
roland
Posts: 3138
Joined: Thu Aug 29, 2013 8:29 pm
Location: Born (NL)
Contact:

Re: Atom 2k18 ?

Post by roland » Sun Mar 24, 2019 6:36 pm

It is possible to add a set of jumpers that makes CAS-IN and CAS-OUT also DTR and RTS. In that case it might be better to route CAS-OUT through the last available buffer port. To this set of jumpers you can also connect an external switch.
256K + 6502 Inside
MAN WOMAN :shock:

User avatar
Multiwizard
Posts: 1525
Joined: Wed Jan 11, 2012 9:03 pm
Contact:

Re: Atom 2k18 ?

Post by Multiwizard » Sun Mar 24, 2019 7:06 pm

Hi,

as you know this goes all way above my head/knowledge, but,,,

while scrolling down I just saw something about a cassette port...

Is it possible (in this new Atom design) to load apart of SCOS/COS0 and FCOS/COS1 to load programs with MJCOS speed? :-k

I thought it would be handy, because some of us still have tapes with MJCOS programs... :D


Greetings, Wim... :-)

User avatar
roland
Posts: 3138
Joined: Thu Aug 29, 2013 8:29 pm
Location: Born (NL)
Contact:

Re: Atom 2k18 ?

Post by roland » Sun Mar 24, 2019 8:51 pm

Hi Wim,

The new Atom also has an 6522 so it is possible. I will have a look at the MJCOS design to see if it can be implemented into the FPGA. I doubt that there are many programs stored on MJCOS tape formats.
256K + 6502 Inside
MAN WOMAN :shock:

User avatar
Multiwizard
Posts: 1525
Joined: Wed Jan 11, 2012 9:03 pm
Contact:

Re: Atom 2k18 ?

Post by Multiwizard » Sun Mar 24, 2019 8:55 pm

roland wrote:
Sun Mar 24, 2019 8:51 pm
Hi Wim,

The new Atom also has an 6522 so it is possible. I will have a look at the MJCOS design to see if it can be implemented into the FPGA. I doubt that there are many programs stored on MJCOS tape formats.
Oh that would be nice... :-D

I still have several tapes with MJCOS programs on it... 8)

Even a (for me unknown) MARIO game/title... :shock:


Greetings, Wim... :-)

User avatar
roland
Posts: 3138
Joined: Thu Aug 29, 2013 8:29 pm
Location: Born (NL)
Contact:

Re: Atom 2k18 ?

Post by roland » Sun Mar 31, 2019 7:24 pm

I have now completed the location of the parts on the pcb. Let the routing begin .... I have made a small start :mrgreen:

The Pi Zero (optional) and the fast ram must be mounted on the solder side of the board. The two schottky diodes can be mounted on which side you prefer. I left both of them in the design but if you think that one is better or sufficient you can make a wire bridge.

The board is intended to have one ram chip operational but with some hardware mods in the future we can sacrifice a led of the MMC and use that for an extra select between one of the rams which gives the Atom 1 MB of (practically useless) ram memory =P~
Attachments
FPGAtom-pcb-setup.pdf
(345.91 KiB) Downloaded 11 times
256K + 6502 Inside
MAN WOMAN :shock:

User avatar
roland
Posts: 3138
Joined: Thu Aug 29, 2013 8:29 pm
Location: Born (NL)
Contact:

Re: Atom 2k18 ?

Post by roland » Sun Apr 07, 2019 6:54 pm

After a few evenings puzzling I finally have the first draw of the print layout \:D/
The design rules check indicates some errors that I have to fix but it is a nice start. I'm afraid there is not much room for useful ground planes :mrgreen:

Atom2019-board.png
Please do not make a pcb from this - it contains some design rule errors!
256K + 6502 Inside
MAN WOMAN :shock:

User avatar
hoglet
Posts: 8095
Joined: Sat Oct 13, 2012 6:21 pm
Location: Bristol
Contact:

Re: Atom 2k18 ?

Post by hoglet » Sun Apr 07, 2019 7:22 pm

roland wrote:
Sun Apr 07, 2019 6:54 pm
I'm afraid there is not much room for useful ground planes
I do worry that without good ground connections, noise will be a very big problem.

Have you considered going to a 4 layer board?

What's the size of the board, and your preferred PCB vendor?

User avatar
roland
Posts: 3138
Joined: Thu Aug 29, 2013 8:29 pm
Location: Born (NL)
Contact:

Re: Atom 2k18 ?

Post by roland » Sun Apr 07, 2019 7:39 pm

The size of the board is about 367 x 112mm. Most of my boards were made by MakePCB but because their website does not work correctly I ordered my last Yarrb boards at AllPCB.com. They are cheaper but I don't know their quality yet. The Yarrb board have been shipped today.

What would be the best practise if I want to add two layers? Or should I just take a chance and order five boards and we'll see what happens?

BTW: I have found and corrected the errors... now it's 100% error free.
Last edited by roland on Sun Apr 07, 2019 7:40 pm, edited 1 time in total.
256K + 6502 Inside
MAN WOMAN :shock:

User avatar
hoglet
Posts: 8095
Joined: Sat Oct 13, 2012 6:21 pm
Location: Bristol
Contact:

Re: Atom 2k18 ?

Post by hoglet » Sun Apr 07, 2019 7:53 pm

roland wrote:
Sun Apr 07, 2019 7:39 pm
The size of the board is about 367 x 112mm. Most of my boards were made by MakePCB but because their website does not work correctly I ordered my last Yarrb boards at AllPCB.com. They are cheaper but I don't know their quality yet. The Yarrb board have been shipped today.

What would be the best practise if I want to add two layers? Or should I just take a chance and order five boards and we'll see what happens?
It should be easy in KiCAD to add two new layers, one for ground, the other for 5V/3V3.

Looking at the prices for a board that size on JLCPCB:
- 367mm x 112mm x 2 layers x 10 boards = £27.25 + £23.11 2-3 day DHL shipping = £5.36 per board
- 367mm x 112mm x 4 layers x 10 boards = £57.22 + £23.11 2-3 day DHL shipping = £8.03 per board

(MakePCB seems much more expensive)

If it were me, I would go with the 4 layers, but at the end of the day it's your call.

Dave

User avatar
roland
Posts: 3138
Joined: Thu Aug 29, 2013 8:29 pm
Location: Born (NL)
Contact:

Re: Atom 2k18 ?

Post by roland » Sun Apr 07, 2019 9:42 pm

hoglet wrote:
Sun Apr 07, 2019 7:53 pm
It should be easy in KiCAD to add two new layers, one for ground, the other for 5V/3V3.
Should these power and ground layers be on the outer or inner layers? Are these only for eliminating noise from the environment and not for interference between two tracks?

I have no experience with high speed pcb designs, only low speed (max about 5MHz bus).

MakePCB is more expensive but their boards are of good quality. But if you recommend JLCPCB I will give it a try.
256K + 6502 Inside
MAN WOMAN :shock:

User avatar
1024MAK
Posts: 8663
Joined: Mon Apr 18, 2011 4:46 pm
Location: Looking forward to summer in Somerset, UK...
Contact:

Re: Atom 2k18 ?

Post by 1024MAK » Sun Apr 07, 2019 11:09 pm

The ground and power layers are often the two inner layers. This means if there is a problem with a signal track, you may still be able to get to it.

The objective is to minimise the noise on the signal lines caused by fast switching transistors in the chips. When the transistors that form a logic gate (especially a line driver output transistors) change state, a short duration high current is drawn through the chips power pins. It is worse if multiple gates or outputs change state at the same time (such as an address or data bus). This can result in the local supply voltage dropping or the local 0V/ground voltage rising slightly. Possibly enough to affect the same chip (or a nearby chip) from being able to correctly determine the logic level on an input. A ground plane and a power plane significantly reduces the resistance and impedance of the supply, hence the decoupling capacitors are more effective. Further, the inclusion of both a ground and a power plane allows the currents to take the shortest and most direct path between the source and the load. Combined this significantly reduces switching noise on the ground and supply rails and on the signal lines.

Mark
Last edited by 1024MAK on Sun Apr 07, 2019 11:11 pm, edited 1 time in total.

User avatar
myelin
Posts: 643
Joined: Tue Apr 26, 2016 9:17 pm
Location: Mountain View, CA, USA
Contact:

Re: Atom 2k18 ?

Post by myelin » Tue Apr 09, 2019 1:41 pm

Something I overlooked on my first 4 layer board: you want the ground plane to be the inner layer adjacent to your highest speed signals. So if your fast signals are on the top layer, put ground on the top inner layer and power on the bottom inner layer. If they’re on the bottom layer, make the bottom inner layer ground.

(Even if you don’t do that, your grounding will still be way better than on a 2 layer board, but it’s a good recommendation that I only recently heard!)
Last edited by myelin on Tue Apr 09, 2019 1:41 pm, edited 1 time in total.
SW/EE from New Zealand, now in Mountain View, CA, making BBC/Electron hardware projects for fun.
Most popular: fast serial port, FX2+PiTubeDirect Tube/Cartridge adapter, USB cart interface.

User avatar
roland
Posts: 3138
Joined: Thu Aug 29, 2013 8:29 pm
Location: Born (NL)
Contact:

Re: Atom 2k18 ?

Post by roland » Tue Apr 09, 2019 7:52 pm

The most critical with highest speed will be around the 10ns ram so I added a ground plane at layer 2 and a power plane at layer 1 (KiCad has F.Cu, In1, In2, B.Cu as layers). The ram chip is surface mounted on layer B.Cu.

These are my planes:
ground-plane.png
power-plane.png
Will this do? If so I'll start removing the power and ground traces on front and back layers.
256K + 6502 Inside
MAN WOMAN :shock:

Post Reply