News from the meeting at GC


So, why didn't you order it 2 weeks ago?

giphy.gif
 
So, why didn't you order it 2 weeks ago?

The design spec for the latest revision with it's simulation testing simply did not exist 2 weeks ago. This is the speed of invention. It is not a part or spec that can be simply ordered off the shelf.

IMHO there is significant risk in @EvilDragon 's decision to go 'all in' on this SoC board revision. However, I do understand the logic. There are no further gains addressing the 1.2 Ghz limit issue to be had within the confines of the processor board. Either it gets us 'there' or 'there' must be redefined to what this new board can do. The risk, as I see it, is if the changes to fix the 1.2 Ghz limit issue result in a new issue. So, those of you inclined should proceed with crossing your fingers for the next several weeks while a big batch of bare SoC boards are produced. I hope this works out. The project could use a solid win.
 
ED, you really need to start putting your own health ahead of 'business'. Those hours cannot be good for you.

The CPU board has been improved as much as possible (thanks again to Juliano for the simulations!), so we're ready to start the production. As the circuit itself has not changed (we just added vias and a few capacitors), I am planning to order enough PCBs for the mass production already. A new prototype run is not really needed, as we now know we cannot further improve it and it should be good enough for 1,5GHz!
This would save us 6 weeks.

Is the main concern here the 6 week delay or is it the financial implications of another costly prototype phase? If the concern is only time based then surely it makes sense to have another prototype run just to be sure the 1.5Ghz fix has worked. Guess this will also depend on whether the pick and place issue can be solved.
 
Is the main concern here the 6 week delay or is it the financial implications of another costly prototype phase? If the concern is only time based then surely it makes sense to have another prototype run just to be sure the 1.5Ghz fix has worked. Guess this will also depend on whether the pick and place issue can be solved.

Whether or not it fixes the 1.5Ghz issue, this is as far as development can go to fix the issue. I.e. whatever this version of the SoC board does for processing, it's as good as it gets.

Now, don't panic. The material outlay for bare system boards isn't small, but it's a fraction of what a populated board costs. I'm sure he'll have a pre-production validation board population run with these new boards to keep from risking a pile of dead units.

For producing the boards, the up-front setup costs are probably the same whether he orders 20 or 1000+. The 1000+ may or may not be enough to get a break on volume pricing. But - the additional 980+ bare boards incremental costs probably aren't that much in the grand scheme.

The gamble isn't on whether or not the 1.5Ghz issue is solved. That one is as solved as it is going to get. The gamble is on whether or not the steps to remedy the 1.5Ghz issue results in something else having issues - which it shouldn't. IMHO @EvilDragon is making the right call.
 
So if the not happen Prototype run saves us 6 Weeks, this means this Massproduktion Run also takes 6 Weeks because they make the same steps on the factory expect its 1000 Boards..
so in 6 weeks, the boards will be shipped to global components, than populated whit the tiny electronics bits, and then testet, and assembled to shiny new Pyra??
So i quess, maybe the last 2 months
 
@Grench yeah, think you nailed it there, basically as you say its a no gamble, gamble, - frankly speed isn't everything I'd be quite happy with "only" 1.2ghz (but fairly confident most units will get to 1.5ghz!)
 
The gamble is on whether or not the steps to remedy the 1.5Ghz issue results in something else having issues - which it shouldn't. IMHO @EvilDragon is making the right call
I agree because it's not really a gamble. Worst case scenario, the 1000+ boards get delivered and absolutely none of them can be used: net loss is, what, $3000? Which is not good but realistically for a project like this it isn't bad. Doing the full run will not take longer than the prototype and may actually take less time, so in the grand scheme the gamble is really just paying extra for a "prototype" run (and having 1000 unusable boards) vs being ready to actually go in 6 weeks.
 
It was only a short reply.
That's ok. He could have written it while sitting on the toilet.
Don't judge people that fast :p
Toilet time means sending an email to the case company to make sure everything gets finished there. ;)
 
  • Like
Reactions: rSl
https://pyra-handheld.com/boards/threads/with-full-steam-into-the-new-year.79641/
"Jan 19, 2017"
"Time to produce a few more CPU boards with that new layout - which should definitely be the final one!"
Your point is?

He wanted to produce a few CPU boards for prototypes, there still a bit of questions if the simulations actually work. He then at later point decided to waive the risk and order all the boards. To me this seems a bit risky, but if it works it will save time. If it doesn't then either we're waiting longer until it's fixed or suffering with a processor that doesn't run at full speed. I hope the risk is worth it.
 
Wich is actually less than 2 weeks before.

But the point, and imho the reasons of the some harsh reactions to your postings is:

At the "full steam" - thread, wich was overall a very promising one, you stated it was "very, very bad" news, because another prototype run (which would be a completly logical step and not surprising at all) would be needed.
Then, in this very thread ED said that he can probably skip another intermediate run (wich is a bit risky, as others already stated) - and it is still not fast enough for you?

11 Days is not that much for this kind of descision, especially if you take a look on his schedule from the last week.

Don't get me wrong, waiting is not an easy time for us preorderers, and of course we would like to hold a pyra in our hands soon. But a few days more or less realy do not matter that much imo, especialy when it comes to such decisions.
 
Last edited:
ED, you really need to start putting your own health ahead of 'business'. Those hours cannot be good for you.

What hou... oh, I get it. Okay, off to bed with me :D

I agree because it's not really a gamble. Worst case scenario, the 1000+ boards get delivered and absolutely none of them can be used: net loss is, what, $3000? Which is not good but realistically for a project like this it isn't bad. Doing the full run will not take longer than the prototype and may actually take less time, so in the grand scheme the gamble is really just paying extra for a "prototype" run (and having 1000 unusable boards) vs being ready to actually go in 6 weeks.

Well, except that 1000 bare CPU boards do cost around 20k USD, they're pretty complex.

However, the simulation showed we're now within TIs specs for the most important values (VDD_MPU) and are NEARLY there for the resistance values - but we can't improve that with a full redesign.
Nothing else has changed, so the PCB shouldn't stop working, it'll be vastly improved compared to the previous version.

As all manufacturers have tolerances, the greatly improved design should easily be enough for 1,5GHz (some OMAP5 devboards can be overclocked to 1.7GHz). But as there's nothing we can improve anymore, it doesn't really matter :)

So yes, we're saving some time AND money here. Well, time depends on the arrival of the remaining stuff (cases, keymats and LCD Cables), but we can already populate PCBs without those, so it's a win in any way :)
 
What hou... oh, I get it. Okay, off to bed with me :D



Well, except that 1000 bare CPU boards do cost around 20k USD, they're pretty complex.

However, the simulation showed we're now within TIs specs for the most important values (VDD_MPU) and are NEARLY there for the resistance values - but we can't improve that with a full redesign.
Nothing else has changed, so the PCB shouldn't stop working, it'll be vastly improved compared to the previous version.

As all manufacturers have tolerances, the greatly improved design should easily be enough for 1,5GHz (some OMAP5 devboards can be overclocked to 1.7GHz). But as there's nothing we can improve anymore, it doesn't really matter :)

So yes, we're saving some time AND money here. Well, time depends on the arrival of the remaining stuff (cases, keymats and LCD Cables), but we can already populate PCBs without those, so it's a win in any way :)
Hopefully that means only a few more months then, give or take a few weeks!
 
Is there any disclosure for the public what does cost how much?
I think something about:

CPU board(2gb) 200€
Mainboard(standard ed.) 100€
Display+touch 50€
Case+keymat 30€
Battery 20€
Development(all the work, licences, prototype production) 100€
---
sum 500€

So you have to sell something about 3000-4000 units to get to the development costs funded and after this, you will get profit for the project?
I dont know if this is more or less secret, just asking.
 
Is there any disclosure for the public what does cost how much?
I think something about:

CPU board(2gb) 200€
Mainboard(standard ed.) 100€
Display+touch 50€
Case+keymat 30€
Battery 20€
Development(all the work, licences, prototype production) 100€
---
sum 500€

So you have to sell something about 3000-4000 units to get to the development costs funded and after this, you will get profit for the project?
I dont know if this is more or less secret, just asking.

There's an ancient news thread, but I don't think it's valid anymore.

EDIT: three years old, soon: https://pyra-handheld.com/boards/threads/how-to-calculate-a-price-for-the-pyra.73755/
 
Well, except that 1000 bare CPU boards do cost around 20k USD, they're pretty complex
Oh, oops. Wow those boards cost a lot more than anything I've ever had to deal with. I'm not used to PCBs costing more than a couple dollars.
 
Back
Top