Tests, tests, and more tests.


EvilDragon

Administrator
Staff member
Joined
Mar 4, 2003
Messages
29,986
Age
46
Location
Ingolstadt
It's that time of the year again:
The time when the hockey season starts in Germany... which means: I have to prepare tons of video clips for the new season.

Which is also the reason I'm not often visiting the forums - it's simply a lack of time.

However, this doesn't mean I'm also lacking time to work on the Pyra. It means I'm having less time for sleep than usual...

Thankfully, the season has now started, all clips have been prepared, so I can finally give you updates about what happened in Pyra-Land!

As the keymats have a leadtime of 12 weeks, I'm trying to get everything else done within these 12 weeks as well!

Various tests of the PCBs

I've assembled another 8 Pyras now which will be used to do a burn-in test of the new PCBs next week.
In this test, all Pyras are booted off an SD Card. They will log the remaining battery capacity and voltage, the current CPU speed, temperatures as well as the uptime onto the SD-Card. So if they crash, freeze or whatever, I can see when that happened in the logfile.

Hopefully, they won't crash, but if they do, it might be related to the remaining battery capacity (as there's nothing that shuts down the OS if the battery gets too empty at the moment).

Afterwards, I will also make a test with full CPU load and use an infrared camera to check on the heating.

Two of the units have been built a bit different:
One has two heatsinks installed. According to Julianos simulation, this shouldn't make a big difference - but as such complex simulations are not always 100% correct, I'd like to test that anyways.

And another one has a copper plate instead of an aluminium plate installed, just to see, if there's any difference in heat distribution.

I also grabbed an incredibly useful small USB power tester.
This can be used to check the power the connected device is pulling.
It allowed a quick test of the 22 prototype PCBs I had here - I didn't even need to boot them up!

Here's how:

First, you need to know that the OMAP5 boots up right away once you connect it to a power supply (like the OMAP3).
It looks for a bootloader on various media: SD Card slot, eMMC, USB.

If it doesn't find one, it continues the cycle, until it finds a bootloader.

So connecting the USB power tester to a prototype that doesn't have any bootloader anywhere looks like this:

PyraWithoutBootLoader.jpg

Power consumption always cycles between 0.204A and 0.225A. This is because while looking for a bootloader, power varies a bit.

This quick test confirms within seconds, that there's no short circuit on the mainboard (at least none that increases the power consumption) and that the CPU board at least tries to boot, so the CPU is working.

And yes, one of the 22 PCBs actually had a slightly higher power consumption:
It was between 0.260A and 0.280A - something seems to be wrong here!

But is it the CPU board or the mainboard?
Easy way to test: Remove the CPU Board!

And this is what happened with that board without CPU-Board:

PyraBrokenM10.jpg

It's firing up the LEDs! That shouldn't happen!

This is what a different mainboard looks like without CPU board:

PyraMainboardOnly.jpg

Also take note of the power consumption, which is a LOT less than with that "broken" mainboard.

While I don't know what exactly is wrong with that mainboard (it even boots without issues, but we haven't tested all components), that quick check actually showed right away that something is wrong.

Well, but that means:
ALL CPU Boards that have been produced at least start the CPU.
And only one mainboard has an issue. That would be an AWESOME failure rate. But more tests will be needed.

With that power consumption tester, I could do some more quick tests.

This is a Pyra booted up with no LCD connected:

PyraWithoutLCD.jpg

The same one with an LCD connected and keyboard brightness set to maximum:

PyraWithFullKeybBL.jpg

With keyboard LEDs disabled:

PyraWithoutKeybBL.jpg

And with keyboard LEDs set to minimum brightness:

PyraWithMinimalKeybBL.jpg

Usually, minimum keyboard brightness is enough. You can easily see all keys in the dark with that one and you won't need the LEDs enabled when you're using it in a bright surrounding.

While the CPU is doing nothing spectacular at the moment, everything has been enabled:
Wifi, Bluetooth, USB-Ports, all LEDs, etc.
Things we could switch off to save some more power.

With normal usage, the battery should last around 9 hours right now - and this is also what Askarus experience in daily usage has been.

However, powersaving can be improved A LOT. You can also lower the maximum clockrate or disable a CPU if you want to have more battery time - it will still be faster than the Pandora is right now.

This is okay for a start I think (the Pandora had 8 hours battery time at the beginning), but there's still a lot of improvement possible.

Okay, so much for the current status of the tests, let's head to...

The cases

FormAction has reopened the factory after the summer break and should hopefully finish those cases soon, but right now I'm concentrating on Uniqueplast (the company where we'll move the production to).

I'm in good contact with the one of the junior CEOs (we also communicate quickly via messengers and I get almost instant replies to any questions - and he is also a hardcore gamer, so he knows how shoulder buttons should feel...).

There's an exhibition coming up in Germany, the Fakuma International.
It's an exhibition for industrial plastic processing.

Both CEOs from Uniqueplast will be there - and I'll go there as well, so we can look together for different plastic materials and hopefully find the perfect one for the Pyra.

This will be pretty interesting - and it's nice to see they actually are interested in what they're doing (and new developments), which is something that was sadly lacking from our current company...

The exhibition takes place in October. If we can get proper plastic there, we can produce the cases in November and have them in time with the keymats.

Please note the disclaimer: These are possible dates, but it's what I'm trying to achieve.


Other things...

Nikolaus is working hard on other software-based low-level stuff (for example, debugging why the Pyra hangs when you reboot it or why the VBUS detection of the charger chip doesn't work reliable yet) and is making good progress there.

I'll try to finish the tests as fast as possible, as these units would then be ready for our Pyra Prototype preorderers (yep, finally!)

As usual, feel free to throw any questions here at me and thanks again for your interest and support.
I'll keep you updated!
 
Wow! This is so exciting, I can't wait! 2.5W of power consumption is pretty impressive! I'll take 9 hours of battery life any day of the week! Thanks for the update ED.
 
As the keymats have a leadtime of 12 weeks, I'm trying to get everything else done within these 12 weeks as well!
What is the new leadtime on the boards? (Sorry, I forgot.) I remember that it got longer than it had previously been. If it's 8 weeks, then you're going to have a Herculean task to have the boards delivered no later than the keymats, if you do the burn in test next week and order the boards after that!

Usually, minimum keyboard brightness is enough. You can easily see all keys in the dark with that one and you won't need the LEDs enabled when you're using it in a bright surrounding.
Hmm... That sounds like the keyboard back light is going to be too bright for some people's preferences.

Is there a convenient spot on the board where a user could solder a series resistor in the power line to the LEDs? (Assuming of course, they have the tools and skills for such fine surface-mount work.)
 
Last edited:
Lets get real here

There is no chance for this device to be released (starts shipping?) in 2018.
Even H2 2019 looks far too optimistic at this point.

I am not very motivated especially with those "once in a month" updates with zero progress in them from the past few years.
 
Lets get real here

There is no chance for this device to be released (starts shipping?) in 2018.
Even H2 2019 looks far too optimistic at this point.

I am not very motivated especially with those "once in a month" updates with zero progress in them from the past few years.
I feel ya, I remember the Pandora being delayed. I'm not as worried about this one taking so long though, as the Pyra is up-gradable, I was worried the Pandora would be outdated before it even came out. I have to say, for emulation it is great though, I was playing mine today.

Living though the Pandora has taught me patience.

Not to alarm you but getting the Pandora and shipping was half the wait, after that they came out painfully slow. It was hard reading peoples experience with their Pandora while I waited. That was much much more difficult for waiting for it to start shipping so I hope that when it does start shipping they can come out as quickly as people want them.
 
What is the new leadtime on the boards? (Sorry, I forgot.) I remember that it got longer than it had previously been. If it's 8 weeks, then you're going to have a Herculean task to have the boards delivered no later than the keymats, if you do the burn in test next week and order the boards after that!


Hmm... That sounds like the keyboard back light is going to be too bright for some people's preferences.

Is there a convenient spot on the board where a user could solder a series resistor in the power line to the LEDs? (Assuming of course, they have the tools and skills for such fine surface-mount work.)
I think it just looks so bright because of the transparent case.
If there is no light bleeding, the light will only emit from the signs of the keys so it will be much less disturbing for the eyes.
But we will have to wait and see (no pun intended).
I am so hyped for the Pyra at the moment, seeing the prototypes etc.
Let's hope everything will come together in these 12 weeks.
Good job ED!
 
So no news was no news.Hopefully the prototypes will be ready soon, sounds like the new boards have created some new issues though.Still hopeful for a March 2019 release.
 
Lets get real here

There is no chance for this device to be released (starts shipping?) in 2018.
Even H2 2019 looks far too optimistic at this point.

I am not very motivated especially with those "once in a month" updates with zero progress in them from the past few years.
I know how you feel and I would be more than happy if all that would work faster.
Which it would, if more people with more time were involved.

Steps are small these days (and which is normal, as the last few things always take longest, simply because everything that doesn't take long to fix has already been fixed).

With the Pandora, notaz had been a huge help, as he's great with debugging and fixing software issues.
However, he doesn't really have time and there is no one else except for Nikolaus and working on these lowlevel things. And he's a hardware designer, not a software developer (though he knows a lot about that as well).

Without a person who does that, it's extremely hard to determine for some bugs whether they're hardware or software bugs - which is why it takes so long.

I wished I had the money to pay a lowlevel software development company to speed this up, but I don't have such funds.

I totally know how you feel, I would've never thought it would take that long... But I can also remember that there were a lot of Pandora newsposts which literally sounded like nothing is happening at all.

But that's the thing with the last development processes: you can sit for months over one single problem (which seems like nothing is happening at all) and then BOOM - it's solved and everything moved ahead.
 
If nothing else, the one takeaway from that (questionably) broken board that we can say for sure is that the backlights consume very little power: it's noticeable and you will gain quite a few extra minutes if you switch it off, but you won't need to worry about it if you don't.
 
3,5 Watts sounds pretty good, and compare to the Standbye Current of the XBOX OneX dos its sound much much better, as this needs 30 - 40 Watt in "Fast Boot" Mode,
And 9 Hour Batterielifetime at the Start of the Firmwaredevelopment sound also great, my GPD Win got much less Batterie Time..
The Rest of the News also sound pretty good..
 
With the Pandora, notaz had been a huge help, as he's great with debugging and fixing software issues.
However, he doesn't really have time and there is no one else except for Nikolaus and working on these lowlevel things. And he's a hardware designer, not a software developer (though he knows a lot about that as well).

Without a person who does that, it's extremely hard to determine for some bugs whether they're hardware or software bugs - which is why it takes so long.

My position still stands from May, https://pyra-handheld.com/boards/threads/moving-along.82982/page-2#post-1437537 , although I'm going on a short cruise at the end of October and at some point I'll need to look at getting a job again.

I'd take any sort of board that boots to try to help out identifying and potentially fixing problems. Given the latest problems that Nikolas has been discovering and having to compare them to the uevm, I'd probably have to start considering buying one of those too.

I've been making good progress on my own case problem, which I was hoping to complete before showing this group.
DnItvvfU0AAkCWZ.jpg

Once I get the dpad done I can get to my preferred tasks, software engineering.
 
Thanks so much for the news update! It really makes the wait easier, and I don't recall I'd seen the new CPU boards booting yet.

I guess those power ratings are measured with wifi rfkilled off, and most other periperhals powered down too. But modern wifi chips shouldn't be as noticable in terms of power drain compared to the old one in my Pandora; I now have a Nokia/HMD 8110 4G which I can leave the wifi on and connected for four or five days between charges, and that's only got a 1500mAh battery.
 
@levi ED said

While the CPU is doing nothing spectacular at the moment, everything has been enabled:
Wifi, Bluetooth, USB-Ports, all LEDs, etc.
Things we could switch off to save some more power.

With normal usage, the battery should last around 9 hours right now - and this is also what Askarus experience in daily usage has been.

;-)
 
Ah thanks. I must have been looking at the nice pictures, then skipped to the state of the case plastic.

In other good news, uniqueplast seem to at least have a web presence. I've never been able to find anything about form action via a web search. Now we know where to chase them up if they double cross us!
 
Back
Top