Page 2 of 2

Re: Re Cumlulus Build

Posted: Thu Mar 05, 2015 3:58 pm
by Godzil
jdavis6809 wrote:Hello,
I think the next batch planed by others is waiting for 100.
No, I said that I could launch a real production batch (not a hand one) with a number about 100, never say making lower count is not possible but the price in this cas would be higher
There is two cumulusbus boards (modified) not one, One IDE cable and SD card (70 euro price by others is based on qty 100 and sd card extra cost).
70€ without shipping is the price metadata sold the cumulus he build, and he wasnt calculating his price on large volume, but on really low volume (i'm sure his price was based on 10 build board including the 3D printed plastic)

I give the price of 70€ as a rough calculation, with a real 100 pieces batch the real component+pcb price would be lower, maybe about 50€ excluding all fees, shipping etc.. of course.
It also does not include the SD card price

Re: Re Cumlulus Build

Posted: Fri Mar 06, 2015 6:49 pm
by jdavis6809
Hello,

What does the survey sheet mean when it shows 50%

So you are only interested in doing a run of say 100 with company with pick and place machine.
with present gerbers for main pcb and cumulusbus so the buffer problem will remain.

If you do get to do a run of say 100 units, then I foresee that were may never be a another run.

I do understand why you are doing this so you are not tied to one person doing the build as before.

But It could be a long wait between production runs.

I have ask the question when is the next batch going to run, not received an answer yet.

I consider my price reasonable for a batch of 10 with 2 cumulusbus boards or one cumulusbus board and one amplibus v1 board with 3 ports (2 spare 34way and 1 cumulusbus 40way) also I would be using ide box headers so the cable cannot be fitted the wrong way round.

Also I have stated price would be lower for 25.

Regards

John Davis flag_uk Electronic Eng

Re: Re Cumlulus Build

Posted: Fri Mar 06, 2015 10:53 pm
by Symoon
Hi all,
Good to read any offer to build hardware.
But being a non-electronics person, I don't really understand what is being offered here.

The goal is to replace the existing Cumulusbus, hoping to have one that is more reliable on any machine, am I right? But why are there two boards then? Twice the same just in case one fails?
Just asking because I need to understand what I'm buying before saying if I'm interested ;)

Re: Re Cumlulus Build

Posted: Sat Mar 07, 2015 1:48 am
by Godzil
I never say I would directly use the gerber from Retromaster, I plan to rebuild the schematics and layout as the current version is not modifiable, and the used nokia screen is not available to buy on normal markets, so it can't be use for anything else than for low volume.

But I really wonder how you will buffer the D bus with your 74lvc8t245, it wont be as easy as putting a bidir buffer on the data lines, and use read/write to switch the direction...

Re: Re Cumlulus Build

Posted: Sat Mar 07, 2015 7:00 am
by jdavis6809
Hello,

Standard practice to use r/w on dir pin ( microtan 65 buffer all i/o and r/w is used for dir on 74ls245 buffer)

r/w is use to drive dir on the 74ls245 on the microdisc circuit as well. so I do not know want you on about?

Also there is 4 i/o not buffered at all at the cumulus end 3v3 logic pulled to 5v rail via resistors, that is a no no (R1,R2,R4 and R5)

I would never do that, you would translate the logic with dual voltage buffers as done with the data bus.

You can get away with 5v logic out from oric to 3v3 logic into the cpld because cpld is 5v logic tolerant

But to have 3v3 logic o/p from the cpld to drive 5 v logic i/p on the oric bus end via a cable is asking for trouble ( yes I seen it done on eg: coco sd and others but the bus connection is very short.

If you think you just need to buffer the cumulus end go right ahead, will end up with the same problems as as before (if the cable is short you may be ok as Retromaster stated "works with short cable but not long cable"

I have hi-lighted what tests should be done to iron out the problems ( see previous posts)

Bad idea to tell customers that it works on my machine but it may or not work on yours as it is the case now

Sort out the buffer at cumulus end ie 3v3 logic to 5v logic then test with different length cables ( without buffer at oric end)

Then proceed to buffer the oric end then test with different length cables.

Do get me wrong I want this project to work.

My mods were based on present main board design not a new design you propose.

If you do use buffers on new layout you need to translate all 3v3 logic to 5v and 5v to 3v3 logic and remove R1,R2,R4 and R5.

The reason I choose one bi-directional buffer instead of a mix of bi-directional and single directional buffers for all i/o is it cuts the price down when you buy 100's and also cannot find dual voltage single directional buffers

The buffer I would use allows all logic combinations and bi-directional making it truly universal.

There may well be a problem with the map signal being generated from 02 signal and 32Mhz crystal oscillator (oversampling)

The Oric Atmos is very fussy with the timing and jitter on that signal.

If you are going to resign the project please let me know as there no point in me doing so ( I am considering it)

Also let people know that the project is going to be resigned and when likely production run would start.

At the moment I think people believe the design is current and does not change and when is production run who knows?


Regards

John Davis flag_uk Electronic Eng

Re: Re Cumlulus Build

Posted: Sat Mar 07, 2015 9:52 am
by jdavis6809
Hello,

A question for you software eng's out there

I am good at hardware, but I can write C embedded AVR.

Why is the coco sd project so simple? (forget that it needs buffers)

Uses the same floppy controller (actually supports 3) with one avr clocking 10Mhz and two xc9572 cpld's ( one used to bank switch different os roms) which is not needed here. Also does not need a LCD display. Controlled from the computer with disk commands.
CoCoSDC_Schema_RevC.pdf
(43.6 KiB) Downloaded 1745 times
Coco SDC (rev A) (Darren Atkinson).pdf
(1.89 MiB) Downloaded 4226 times
Regards

John Davis flag_uk Electronic Eng

Re: Re Cumlulus Build

Posted: Fri Mar 13, 2015 12:51 am
by tingo
the CoCo SDC is an elegant solution. The question is, could we make a solution this elegant for the Oric?

Re: Re Cumlulus Build

Posted: Fri Mar 13, 2015 11:55 am
by Godzil
tingo wrote:the CoCo SDC is an elegant solution. The question is, could we make a solution this elegant for the Oric?
You will need large changes in Sedoric, or create a new OS, is it worth doing this? not sure

Re: Re Cumlulus Build

Posted: Fri Mar 13, 2015 10:37 pm
by Dbug
One drawback of these offerings has been that they aren’t compatible with software that was written to interact directly with a floppy disk controller. This isn’t so much a problem if you are primarily us- ing the CoCo for Basic programming or running OS9 software. There are however a number of titles (mostly commercial games) that fail to work with those other systems.
Basically the usual issue: If you want to create new software, fine.

If you feel like running the existing software, you need to patch it.

Re: Re Cumlulus Build

Posted: Fri Jun 19, 2015 4:39 pm
by Steve M
Maybe the amplibus should stay as separate as they are useful for people with a Microdisc?

If you made the new improved version then that could be of interest to existing Cumulus users and then if it works with fewer or no problems than it can be incorporated into a new Cumulus ?