[Tinkerphones] Problems seem to repeat...

H. Nikolaus Schaller hns at goldelico.com
Sat Dec 7 14:05:45 CET 2019


> Am 07.12.2019 um 13:35 schrieb Paul Boddie <paul at boddie.org.uk>:
> 
> On Saturday 7. December 2019 08.37.23 H. Nikolaus Schaller wrote:
>> I just came across:
>> 
>> https://www.crowdsupply.com/eoma68/micro-desktop/updates/measurements-and-a-> hypothesis
>> 
>> So they also ran into unexpected solder bridges which must be X-Rayed. This
>> sounds very very familiar to me...
>> 
>> With our experiences collected in the GTA04 project many years ago, I think
>> it is easy to fix - if you have enough money or can borrow money to make
>> some costly experiments with the production line setup.
> 
> I mentioned on the arm-netbook mailing list that similar problems were 
> experienced before, albeit with the RAM chips:
> 
> http://rhombus-tech.net/ingenic/jz4775/news/EOMA68-jz4775_XRay_Photos/
> 
> So, I thought that the manufacturer might have built up some experience since 
> then. From the latest update, it sounds like this is a production process 
> problem, with the prototypes having been made without issues, so I guess that 
> the matter of soldering the chips is not a fundamental problem but just 
> something that needs to be done correctly in the production environment.

Yes, exactly the same issue we had with the GTA04. We knew that Nokia had
built millions of phones incl. the N900 using the OMAP3 + PoP memory but
failed.

Then we got a manufacturer who adapted and learned how to properly produce
them on their line.

Then, OpenPandora switched from omap3530 to dm3730 (like we aready had) and
had only failures. Finally their production line learned how to produce them.

And then we did go to the same production line as the 1GHz OpenPandora
with the GTA04A5 - and they failed. Almost. Well approx. 30% yield which
was too low. We tried to optimize, but the GTA04A5 project did run out of
money.

> 
> (I wouldn't know what the difference between the pre-production and production 
> environments are in this case. There were also some other production-related 
> issues around board-edge component assembly, if I remember correctly.)
> 
>> But borrowing money requires a business model which promises future
>> revenues.
>> 
>> Without money it needs a lot of creativity and time to replace the missing
>> money.
> 
> I think that these experiments rely on a helpful and rather generous factory 
> owner who is looking to gain experience in the processes involved.

Indeed. Or a big enough budget to pay for his experiments.

I think I now quite precisely know what did go wrong with the GTA04A5 but it
would have cost several 1000€ to run another experiment. And it could have
failed again.

> 
>> Which leads to delay and delay and finally the product is working but
>> obsolete and only the brave initial sponsors remain. So there is nobody
>> following up the initial funding round.
> 
> Certainly, things have become a lot quieter around the EOMA68 initiative, and 
> there are people who have probably lost interest and do not expect to see 
> anything come out of it.
> 
> I have probably made a point about technological projects many times to anyone 
> who might listen, reflecting on my own observations and experiences as well as 
> what is already common knowledge, that the longer an uncompleted project 
> takes, the less likely it will be realised.

> Factors include the finances (as you mention above) plus technological 
> progress (making a product less attractive over time) and obsolescence 
> (components becoming unavailable and technologies becoming unsupportable). 
> Keeping a project short reduces the exposure to such risks.

Exactly. And that's the big challenge to keep the project duration short and
to do something very complex. All the projects we're talking about here are
at the level of the highest industry league, but are run by enthusiasts with
far too little budget. So we can only admire everyone who takes the challenge
and still tries to win (contrary to me - everyone is getting older).

BR,
Nikolaus



More information about the Community mailing list