Potential changes on a future board.
Posted: 05 Dec 2019 13:24
I have mentioned a lot of this previously but not had much feedback about any of it yet..
CAPACITORS
I am still in two minds about this..Would we stick to the leaded ones, or go with SMT next time ?
It is actually possible the large 100uF SMT caps may not even be needed.
RAM
Would we keep the Falcon RAM board on the next revision ?
I'm also thinking about a smaller plug in RAM board as mentioned here https://www.exxosforum.co.uk/forum/viewt ... =13&t=2356 Where the RAM board could actually double up for the Acorn computers. Mostly the 72pinn simm is probably the best option to keep.. or not ?
I am not wanting to solder any RAM to the motherboard, As people should know by now, if the RAM goes faulty and is soldered down, they can turn into a mini nightmare, in particular if that DRAM chip is no longer obtainable.
I have seen some large DRAM (again used on the Acorn memory boards) chips where they can be socketed..
But again, if these chips on a longer obtainable, and that is the only memory solution on the board... This is why I think some sort of pluggable solution is always the best option, the problem being I don't want to really create a custom board like the Acorn unless they are worthwhile to do so, hence a "second market" for the boards.
In relation to the SIMM socket, because these sockets are expensive, it will be better to try and not use them, but then again, creating a custom board to replace it is going to be more expensive than a custom RAM solution anyway.
It may be better for me to look into producing 4MB 72pin simms instead. if I went down that route, I could design my own simm and even have a simple row of header pins on the top of the board where it could be soldered directly into the motherboard without the socket. And it could still be used as a regular simm anyway.
I was talking to Icky about this a few days ago, where we was thinking about using more modern socket if we were to create a new sim. With more pins we could even add alt-ram onto a "super simm". But this would have course make this simm rather expensive. And it is still possible whatever socket we decide on, may not be on the market for long anyway. It is why I think just using header pins as much is possible is the way to go because headers don't generally vanish off the market and they are low-cost as well.
JUMPERS
We have a hell of a lot on this current board. But this is because it is a develop machine for any possible future FPGA boards. but technically if a revised motherboard was done, it would not necessarily need all the jumpers as it would be a more "final release board" than a "development board". We could go back to the solder blob jumpers for the blitter etc as well if people think that would be better.
OTHER STUFF
I don't know if anyone else has any suggestions for minor tweaks ? The only thing which I noticed was maybe moving the main OSC away from under the PSU may be beneficial? or any suggestions people may have in order to reduce parts costs for assembly time etc.
Also as mentioned in this thread.. https://www.exxosforum.co.uk/forum/viewt ... =73&t=2231 About potential low-cost versions of assembled boards.
I will probably keep the SMT GLUE & MMU for the final H4 release board. I would really want to remove these and even rearrange where the chips are located on the motherboard as well. Though this would mean I would need to reroute a lot of the board again which I am not really sure its worth it.
LONG TERM
As to if a final H4 board is released or not depends on demand from the community. I do still have several H4E boards to sell yet, these will likely appear in my store after Christmas.
Over the coming year work will continue and hopefully finish on all the projects myself and Icky have been working on lately, like IDE & flashy etc. Once all these designs are finished and tested, I would look to integrate them on a new series of motherboard.
We also have the FPGA stuff to do something with, so likely this you will keep us busy for a long time. But basically as mentioned in the "Future of the project" link above, if we go FPGA, we are actually wandering away from a legacy machine which is what the "H series" of boards are. So likely once all kits have been integrated onto that series, we can finally call that series of motherboard completed and future development would be towards a more FPGA based machine.
CAPACITORS
I am still in two minds about this..Would we stick to the leaded ones, or go with SMT next time ?
It is actually possible the large 100uF SMT caps may not even be needed.
RAM
Would we keep the Falcon RAM board on the next revision ?
I'm also thinking about a smaller plug in RAM board as mentioned here https://www.exxosforum.co.uk/forum/viewt ... =13&t=2356 Where the RAM board could actually double up for the Acorn computers. Mostly the 72pinn simm is probably the best option to keep.. or not ?
I am not wanting to solder any RAM to the motherboard, As people should know by now, if the RAM goes faulty and is soldered down, they can turn into a mini nightmare, in particular if that DRAM chip is no longer obtainable.
I have seen some large DRAM (again used on the Acorn memory boards) chips where they can be socketed..
But again, if these chips on a longer obtainable, and that is the only memory solution on the board... This is why I think some sort of pluggable solution is always the best option, the problem being I don't want to really create a custom board like the Acorn unless they are worthwhile to do so, hence a "second market" for the boards.
In relation to the SIMM socket, because these sockets are expensive, it will be better to try and not use them, but then again, creating a custom board to replace it is going to be more expensive than a custom RAM solution anyway.
It may be better for me to look into producing 4MB 72pin simms instead. if I went down that route, I could design my own simm and even have a simple row of header pins on the top of the board where it could be soldered directly into the motherboard without the socket. And it could still be used as a regular simm anyway.
I was talking to Icky about this a few days ago, where we was thinking about using more modern socket if we were to create a new sim. With more pins we could even add alt-ram onto a "super simm". But this would have course make this simm rather expensive. And it is still possible whatever socket we decide on, may not be on the market for long anyway. It is why I think just using header pins as much is possible is the way to go because headers don't generally vanish off the market and they are low-cost as well.
JUMPERS
We have a hell of a lot on this current board. But this is because it is a develop machine for any possible future FPGA boards. but technically if a revised motherboard was done, it would not necessarily need all the jumpers as it would be a more "final release board" than a "development board". We could go back to the solder blob jumpers for the blitter etc as well if people think that would be better.
OTHER STUFF
I don't know if anyone else has any suggestions for minor tweaks ? The only thing which I noticed was maybe moving the main OSC away from under the PSU may be beneficial? or any suggestions people may have in order to reduce parts costs for assembly time etc.
Also as mentioned in this thread.. https://www.exxosforum.co.uk/forum/viewt ... =73&t=2231 About potential low-cost versions of assembled boards.
I will probably keep the SMT GLUE & MMU for the final H4 release board. I would really want to remove these and even rearrange where the chips are located on the motherboard as well. Though this would mean I would need to reroute a lot of the board again which I am not really sure its worth it.
LONG TERM
As to if a final H4 board is released or not depends on demand from the community. I do still have several H4E boards to sell yet, these will likely appear in my store after Christmas.
Over the coming year work will continue and hopefully finish on all the projects myself and Icky have been working on lately, like IDE & flashy etc. Once all these designs are finished and tested, I would look to integrate them on a new series of motherboard.
We also have the FPGA stuff to do something with, so likely this you will keep us busy for a long time. But basically as mentioned in the "Future of the project" link above, if we go FPGA, we are actually wandering away from a legacy machine which is what the "H series" of boards are. So likely once all kits have been integrated onto that series, we can finally call that series of motherboard completed and future development would be towards a more FPGA based machine.