Assembly Team: Work Slower, Please :)


EvilDragon

Administrator
Staff member
Joined
Mar 4, 2003
Messages
29,716
Age
43
Location
Ingolstadt
BackAssward said:
Someone did a board level hardware modification to the wifi, and it was a learning experience that you should test it afterwards? I recall asking X68000 last week if there was a post mod test being done for the wifi.

I mean, that seems a no brainer to me, not something that needed to be learned.

Michael tested it - he is measuring the new resistor value. So he knows that the mod is successful, no problem here.
However, it could also be that the antenna is bad.
And that can be tested in the fully assembled units, and that's what we'll do from now on.
 
Last edited:

BackAssward

Active Member
Joined
Apr 3, 2008
Messages
985
EvilDragon said:
Michael tested it - he is measuring the new resistor value. So he knows that the mod is successful, no problem here.
However, it could also be that the antenna is bad.
And that can be tested in the fully assembled units, and that's what we'll do from now on.
No offense intended, but of course MWeston was board level testing after the mod (I had no doubt of that). I meant that it was pretty obvious that actual WiFi connectivity should be tested after the mod. That, that wasn't something to be learned, but clearly a step that should be done without question. It is an obvious thing that should have been done right from the start.
 
Last edited by a moderator:

EvilDragon

Administrator
Staff member
Joined
Mar 4, 2003
Messages
29,716
Age
43
Location
Ingolstadt
BackAssward said:
No offense intended, but of course MWeston was board level testing after the mod (I had no doubt of that). I meant that it was pretty obvious that actual WiFi connectivity should be tested after the mod. That, that wasn't something to be learned, but clearly a step that should be done without question. It is an obvious thing that should have been done right from the start.

Well, the company that produces the PCBs did a test with the chip and Michael did test his mod here, so everything should be fine.
We just started to test WiFi yesterday (it connects to the net and downloads a file) and had one failed unit out of 60 so far - and that was a non-working WiFi chip, something the company should have noticed. But we're doing that now, and except for that one unit, everything was fine.

Interesting though that the Pandora actually gets a WiFi connection in the assembly hall - none of our netbooks nor mobile phones get anything there, the signal is too weak.
So I guess the Pandora is pretty fine there :)
 
Last edited:

Bosbeetle

Terminally lost
Joined
Sep 7, 2008
Messages
4,005
Age
39
Location
The Netherlands
Website
Visit site
EvilDragon said:
Interesting though that the Pandora actually gets a WiFi connection in the assembly hall - none of our netbooks nor mobile phones get anything there, the signal is too weak.
So I guess the Pandora is pretty fine there :)


static.gif
 
Last edited by a moderator:

marshal

Member
Joined
Jan 5, 2009
Messages
414
Location
Switzerland
EvilDragon said:
Interesting though that the Pandora actually gets a WiFi connection in the assembly hall - none of our netbooks nor mobile phones get anything there, the signal is too weak.
So I guess the Pandora is pretty fine there :)

OMG !!

I'll have to compare the Pandora to my N800. If the Pandora actually exeed the range of the N800 I would be freaking impressed !!

I am arround 2000 though so I am patiently wanting for now ;)
 
Last edited by a moderator:

cosurgi

http://janek.kozicki.pl/
Joined
Jul 30, 2008
Messages
1,098
Location
Gdansk, Poland
Website
janek.kozicki.pl
The best way to test the hardware is to compile some code. You get heavy RAM usage, and heavy CPU usage. linux kernel is too much of course, try to compile something lighter. Usually that is how I was discovering problems with hardware on my PCs - suddenly a very stupid compilation problem, seemingly impossible to fix. While everything else in the OS works fine.

Just add to your "downloading" script, another call that compiles something. And see of the compile was successful. 99.5% times it will be OK. But once or twice it could detect hardware problems.

Also there's and ARM version of memtest, called memtester, check it out. You don't want returns due to faulty RAM or CPU.
 

Jumpman

Advanced Member
Joined
May 20, 2006
Messages
3,095
Location
Sin City
WOW, that's great news Ed. I have that same problem at my home, some devices get a bunch of signals from all around the community, and then some devices barely pick up the router from inside my house. That means the Pandora has a great antenna! Whoo Hoo!

Hey Ed, can you please let me know right about how many units have went off for shipping? I'm 250-350 in Craigs "Q", and the wait is killing me. I was hoping to get mine shipped today, but several members lower than me have not heard anything yet. I was guessing you guys should be around 300 so far.

Thanks,
Chris
 

may88

Well-Known Member
Joined
Dec 17, 2007
Messages
1,179
Location
Bury St. Edmunds, UK
EvilDragon said:
BackAssward said:
No offense intended, but of course MWeston was board level testing after the mod (I had no doubt of that). I meant that it was pretty obvious that actual WiFi connectivity should be tested after the mod. That, that wasn't something to be learned, but clearly a step that should be done without question. It is an obvious thing that should have been done right from the start.

Well, the company that produces the PCBs did a test with the chip and Michael did test his mod here, so everything should be fine.
We just started to test WiFi yesterday (it connects to the net and downloads a file) and had one failed unit out of 60 so far - and that was a non-working WiFi chip, something the company should have noticed. But we're doing that now, and except for that one unit, everything was fine.

Interesting though that the Pandora actually gets a WiFi connection in the assembly hall - none of our netbooks nor mobile phones get anything there, the signal is too weak.
So I guess the Pandora is pretty fine there :)

Can you not use a Pandora to stream the Web cam feed then? :p
 
Last edited by a moderator:

jottt

Member
Joined
Oct 28, 2009
Messages
321
Location
Middle of Germany
cosurgi said:
The best way to test the hardware is to compile some code. You get heavy RAM usage, and heavy CPU usage. linux kernel is too much of course, try to compile something lighter. Usually that is how I was discovering problems with hardware on my PCs - suddenly a very stupid compilation problem, seemingly impossible to fix. While everything else in the OS works fine.

Hey nice ;) Testing like this myself ;) I used to overclock my old system once. Everything seemed stable. Everything? Well...to be sure-sure I let my Gentoo out of the box. It crashed soon. Sooo...whenever I overclock the final test is some nice compilation of openoffice (or kdelibs). Gives you a good impression on the performance, too ;)

Now we talk about it...I feel that urge to try that on my Pandora...Probably takes a week or something. I think my old Athlon XP 3200+ needed 20 hours for OpenOffice 2...
 
Last edited by a moderator:

Nuron_V

Member
Joined
Oct 11, 2005
Messages
127
may88 said:
Can you not use a Pandora to stream the Web cam feed then? :p

Just what I was thinking. Although they probably best not they would have to spent a lot of time faffing around to get it working when they could be doing other things.
 
Last edited by a moderator:

teppic

Still Fresh
Joined
May 17, 2007
Messages
36
Nuron_V said:
may88 said:
Can you not use a Pandora to stream the Web cam feed then? :p

Just what I was thinking. Although they probably best not they would have to spent a lot of time faffing around to get it working when they could be doing other things.

They could use the pandora as a router, which would probably save some configuration time.

wifi -> pandora -> usb network -> ED's netbook

Having a pandora serve the live stream would rock though :)
 
Last edited by a moderator:

bce23w

Member
Joined
Feb 21, 2010
Messages
273
Folks like you guys are the reason Big Brother had a hojillion seasons. What do we need six to eight hours of silent, grainy, repetetive assemblage footage for? It's beyond me.

I personally would be pleased to see any of

  • Liveblogging (optionally, from a Pandora)
  • Daily or semi-daily vlogging
  • A nice shot and edited video mini-documentary about the production first 500 (or 1,000 or whatever)

But their absence would not bother me. After all, they would all be tasks that take time away from making and fixing things.
 

Grench

Forum Addict!
Joined
Oct 3, 2008
Messages
6,623
cosurgi said:
Please don't haste through this assembly like crazy. Take your time. Be slow, careful and relaxed, and everything will go fine. You've been waiting for this moment two years. Don't spoil it now through haste. Be careful.

2x - we all want -ours- fast, but we also want them all to be -right- to keep the 'bad unit' press minimized. No point in feeding the trolls.


EvilDragon said:
BackAssward said:
No offense intended, but of course MWeston was board level testing after the mod (I had no doubt of that). I meant that it was pretty obvious that actual WiFi connectivity should be tested after the mod. That, that wasn't something to be learned, but clearly a step that should be done without question. It is an obvious thing that should have been done right from the start.

Well, the company that produces the PCBs did a test with the chip and Michael did test his mod here, so everything should be fine.
We just started to test WiFi yesterday (it connects to the net and downloads a file) and had one failed unit out of 60 so far - and that was a non-working WiFi chip, something the company should have noticed. But we're doing that now, and except for that one unit, everything was fine.

Interesting though that the Pandora actually gets a WiFi connection in the assembly hall - none of our netbooks nor mobile phones get anything there, the signal is too weak.
So I guess the Pandora is pretty fine there :)
VERY cool news. I can't wait to test this out! Of Course I think I'm somewhere in the 1500-2000 range so I'll be waiting for batch 1.3 or 1.4.


cosurgi said:
The best way to test the hardware is to compile some code. You get heavy RAM usage, and heavy CPU usage. linux kernel is too much of course, try to compile something lighter. Usually that is how I was discovering problems with hardware on my PCs - suddenly a very stupid compilation problem, seemingly impossible to fix. While everything else in the OS works fine.

Just add to your "downloading" script, another call that compiles something. And see of the compile was successful. 99.5% times it will be OK. But once or twice it could detect hardware problems.

Also there's and ARM version of memtest, called memtester, check it out. You don't want returns due to faulty RAM or CPU.
Great ideas - Having a script series run on the back bench for a bit would do wonders for QC. Heck, have it play a mp3 of "ITS...ALIVE!" when it's done and flash the screen through test pattern series so the builder who picks it up can see quick if the thing has anything wrong with it. Just ideas.
 
Last edited by a moderator:

EvilDragon

Administrator
Staff member
Joined
Mar 4, 2003
Messages
29,716
Age
43
Location
Ingolstadt
carlgeorge said:
svs-57 said:
EvilDragon
Why OPT ignore customers questions about order rest 3000 cases?
It's almost cruel isn't it :)

Because we're busy putting them together?
I have no idea, Fatih is the one that handles this. He just left the assembly place and is getting married this month (yeah, good time, eh?)

So please give him and us a break.
At the moment I'm assembling and sometimes visiting the boards here to help with issues from 10am till 4am without any breaks except for some food.

We are no gods and we are no company with 1000 employees who can take care about all of that.
 
Last edited:

cosurgi

http://janek.kozicki.pl/
Joined
Jul 30, 2008
Messages
1,098
Location
Gdansk, Poland
Website
janek.kozicki.pl
EvilDragon said:
At the moment I'm assembling and sometimes visiting the boards here to help with issues from 10am till 4am without any breaks except for some food.
That's 6 hours of sleep. How familiar to me ;)

But don't overstress yourselves. Try to get a longer sleep sometimes. It works wonders. Assembling when being very tired isn't a receipt for success. Though sometimes when I'm in coding frenzy I can work 30h straight.
 
Last edited by a moderator:
Top