EOMA68 Computing Devices

An Earth-friendly way to easily upgrade and fix your own computer

Jan 08, 2018

EOMA68-A20 2.7.5 Gerbers Off to Factory; Thank You to Everyone for the Sponsorship

by Luke Kenneth Casson Leighton

So, finally, the HDMI Review is complete: the gerber files have been sent to the factory for PCB manufacturing. Estimates can now be made (between five to seven months) until shipping. There’s been a great response from people with donations and sponsorship.

Thank you!

In total so far, the response has been amazing: you’ve sent nearly \$400 USD on PayPal, 0.207 in Bitcoin, \$1200 USD as a loan, \$1000 USD in sponsorship (for a separate project), $14 USD in BitcoinCash, and as a nice side-effect of receiving that I discovered that I happened to already have 0.4 BitcoinCash thanks to the bitcoin fork. There are also commitments of $96.75 EUR a week on LiberaPay which is really great.

Below, at last, now that the "unknown and unknowable" amount of time which was associated with the Revision 2.7.5 HDMI Review has passed, estimates can now be made until production (about six months), which means that I will still need sponsorship and donations to cover all that time. Please do bear that in mind! Here’s the list again if you’d like to help:

HDMI Review completed, Gerbers sent.

At last, Richard’s happy with the way things look on the HDMI tracks, and I spent several days - around 50 or 60 small sets of modifications re-generating the Gerber files approximately 30 times - as I’ve found it’s much cleaner to use gerbv to view the layers. The reason I don’t do it during actual development is because creating photo-quality masks takes a long time and that dramatically slows down the iterative process. However, during the part where I am mostly staring and zooming in and out for several minutes looking at whether tracks could be moved to tidy it up, or add some extra GND vias, or any power track that doesn’t look wide enough, generating Gerbers and staring at those in uniform colours is definitely worthwhile.

So those are off to the Factory: it’s uncomfortably close to Chinese New Year but we might be lucky and get them back in four weeks. There’s no rush (reasoning follows in a bit), and from experience "rushing" is known to cause problems. Here’s the thing: now that the "unknown and unknowable" time (HDMI Review) on the critical path is done, everything else is quite straightforward. It’ll go something like this:

give about 10 samples.

2.7.4… just without an HDMI connector.

500 for Micro Desktop 1.7 (six weeks?). Casework gets ordered, too.

So that’s about… five to seven months, and a lot to get done. This is where it would have been nice to have been, right when the campaign started, last year. C’est la vie: we can’t ship stuff that’s broken or had to be changed because a component went EOL, and we can’t fix or plan for what we don’t know we don’t know we don’t know.

Practical-wise I will almost certainly have to go over to Shenzhen to help them with the testing at the factory. Remember, that also has to be taken into account in the form of sponsorship! The plane flight to HK, cost of travel and hotels in Shenzhen and so on.

DDR3 RAM and Capacitors

This is worthwhile specifically writing about, as one of the "sub-projects" to solve: the out-of-control spiralling cost of DDR3 RAM. (I’ve already replaced all 10 µF 0805 capacitors with pairs of 4.7 µF 0603). What I suspect is happening is that the insatiable demand for Apple products has not only caused the demand and price of capacitors to sky-rocket, but whatever RAM Apple is using in its products is so overwhelming the capacity of foundries worldwide to produce it that all other RAM and eMMC products are falling behind capacity as well.

So the price of the chosen (previously tested) RAM IC is insane: it was a Hynix 1033 MHz 1.5 V part. It turns out that the 1800 MHz equivalent might be compatible, but it will be necessary to test it… oh, and prices on that are just sky-high, not completely insane. The issue is: the available budget is fixed and limited. Whatever RAM is available to buy within the available budget, that’s what has to be bought. I will have to do the calculations, and get some quotes: please do not be surprised if the cost of 2 Gigabytes of RAM is a whopping $20, making it completely unviable to consider, and the decision has to be made to go with 1 Gigabyte of RAM instead.

Many of you will have noticed that this was mentioned in a previous update (actually the past two). I am repeating this practical warning because it turns out that, despite repeating things many, many times it surprises the heck out of me to receive private messages from people saying, "what do you think you’re doing, you didn’t say anything about that decision!" and, um… actually I did… many months ago.

So, again, here, if anyone has any ideas or input on how to deal with the DDR3 RAM situation, such that the actual promise to deliver 2 GB of RAM can be met, now is the time to say so… on the mailing list. Why the mailing list? So that your ideas can be reviewed and discussed, transparently, by other people, as opposed to making me the man-in-the-middle "sole, exclusive person working on this". Nice as it is to hear from people sending me private messages, it doesn’t serve the nature of this project to have private discussions, as it is an open project, not a closed one. As always, you can join the mailing list here.

Sign up to receive future updates for EOMA68 Computing Devices.

Subscribe to the Crowd Supply newsletter, highlighting the latest creators and projects