WiFi mysteries resolved once and for all — Power saving is clearly the culprit!


(oh, just a side note maybe unrelated, but now that I think of it, both yesterday and today I reached that maximum speed after I connected the power cable...)
 
WRT the whole "it's fast" "no it ISN'T!", folks should also remember that many users have noticed that your performance experience is strongly influenced by the access point you're using.  Not only its model, but possibly its firmware load and channel.  My OpenWRT WRT54GL's at home don't work well at all with my Pandora, whereas a Dlink at a friend's house seems to work very well.

What's clear is the Pandora Wifi is way more cranky and idiosyncratic than pretty much any other consumer Wifi device you could buy these days.  It would be awesome to have its problems addressed by a software upgrade, and I think everybody suspects that it *could* be fixed by a software change.  But can our community lay hands on the necessary software bits?  TBD.
 
I don't know if this source is the same of our driver, but at line 92 you can see a clear definition... now, where/how these flags can be used ? Maybe in "/etc/init.d/wl1251-init" where the driver is loaded ?

Or the driver must be recompiled with this value already set ?

http://lxr.free-electrons.com/source/drivers/net/wireless/ti/wl1251/acx.h

EDIT: in init.c this code seems using a default value of power save off... 

Anyway seems like it's time for me to learn how to use this git thing.... at least I will be able to look at the real code

EDIT2: :eek:   even the code in git seems using a power off value by default... now I'm lost

EDIT3: mmm... seems like the mac80211 is more related to the power mode than the other driver...

EDIT4: "ieee80211_set_power_mgmt" in "mac80211/cfg.c" seems like the function to fix to me... it's just a guess, anyway for now I don't know exactly how can be changed, C is not exactly my first language...
 
Last edited by a moderator:
Ok, so now I'm calm and collected, and can chime in with my observations slightly expanded. Hopefully we'll steer clear of the "NO! You're WRONG!!! I'm RIGHT!!! Shut UP!!!" arguments this time :/

Now, let's go back to the beginning. I have never had these kind of problems. When this subject was first opened, I did a run similar to Porgs, to verify my subjective experience. I have cleaned out the data set since then, but that is really not necessary, because it does not take measured data to observe that the kind of thing Porg is subjectively describing (that the pages never load, or only after several attempts at reloading, and so on) quite simply is not the case for me. Apparently not for Askarus, either. Now, I haven't made a controlled study over verified identities of access points either, but I have used my pandora in several quite different enviroments, with no reason to believe that the routers are all the same. And yes, I do note that my network speed gets better when I turn off power saving (as is to be expected) - The point being that even with powersaving on, it is quite usable. On my unit.

This does not in any way invalidate your experiences here. Nobody is saying that the Wifi on your pandoras is working better than you claim, or that it does so on all pandoras. Quite the opposite - I try to bring focus to the fact that the wifi reception and powersaving aggressiveness and so on seems to differ between different units, sometimes quite widely. Which, in the end, means that this is quite likely not primarily a problem in the firmware/OS, since we all run more or less the same OS and the same wifi drivers and so on. 

If the same software performs differently on similar hardware, it means the hardware (or something lower than the aforementioned software, at least) differs.

Now, someone (Notaz?) talked about the N900 calibration procedure, and the wifi chip calibration data file, and that different units had rather different calibration data. Here comes my guess: Our wifi chips are calibrated before they are put into the units, and this calibration is not as exact and unit-specific as the one Nokia did with their N900s. Which means that, basically, our wifi reception is more or less out of whack depending on how close our approximate calibration data is to the ideal. If that is correct, then that would mean that the solution is a more precise calibration rig in which to run complete units - Not a hack in the driver. Which would be an unfortunate conclusion, since that is far more difficult to fix (I guess), but I can't really logically come to another conclusion. Well, not as long as I don't assume that everyone who has different experiences than myself is either careless or lying, of course :p
 
I have to say that my own experiences much more closely match what Moxie describes than what Porg describes. I personally did not see enough improvement from turning off power saving on my units to bother with it, and I can max out my bandwidth with several concurrent connections in PND Manager even without touching it if I'm not far from the router.

To my mind there are far too many factors influencing the speed at which a web page loads to make it a useful measure for this in any case. For example I've found that on my pandora my browsing speed in Firefox on many pages was much improved by installing a flash blocking extension.

- Neelix
 
I am having WIFI issues too. I am not using it a lot, only for downloading from the PND store. Sometimes it just isn't possible to download a pnd due to the bad connection. Then I will usually download if from my desktop and copy it to my sd. 

I assume most people will not be as 'flexible' as I am in this situation.  Have the same problem on 2 classic Pandora's. It might be a good idea to add the sudo iwconfig wlan0 power <on|off> script prominent available as option in the new firmware (e.g. ask before using Wifi). There might be others like me, who don't really care that much for WIFI but the bad Wifi does limit the usability of the Pandora. 

I am not sure if the script helps a lot in my case though, tried it yesterday when I had connection issues with compo4all mame but it didn't work for that problem. 
 
I wasn't meaning to "broke" things for the ones that doesn't have problems, but I think too that it should be included the possibility to choose what is the default state of power save, so anyone can be able to change it as he want...
 
That is a good idea - Put the magic incantation on a button.
 
This may be slightly off topic, but I personally think it adds to the discussion.

My father, around 1994 or so, purchased a SyQuest drive. It's like a hard drive with removable platters. It didn't work right, so he exchanged it for a new one. That one didn't work right either. In the end he had to exchange it about 9 or 10 times 'til he got one that worked perfectly.

My point is, I think porg was extremely unlucky, and ended up with three Pandoras that just happened to have really bad WiFi. I have had three Pandoras as well. Two first batch, and one 1 gigahertz, which I am using now. They all had decent WiFi, and I never had press reload to get a page to load. I have some trouble downloading large files, but that's the opposite of the problem porg was having. I'm not saying porg was lying, I'm saying just because one person has bad WiFi, it doesn't mean we all do, or even a majority of us do.

Maybe someone should make a sitewide poll.
 
Porg is definitely not the only one with Wifi issues. 
 
No, and it is enough of an obvious problem to a large enough number of people that it needs looking into. Just don't fall in the trap of assuming that the problem is the same for everyone, cause it as obviously isn't. 
 
Porg is definitely not the only one with Wifi issues. 
I didn't say he was, I said I'm unsure how many people have them, it seems to me to be somewhere around half. And the whole thing is TI's fault for not releasing documentation. There's no reason for them not to, you have to buy the chip from them anyway, so it's not like they'd lose money if they gave out the documentation.

No, and it is enough of an obvious problem to a large enough number of people that it needs looking into. Just don't fall in the trap of assuming that the problem is the same for everyone, cause it as obviously isn't. 
I don't assume that at all.
 
@PowerGod: Thanks for your WiFi driver code exploration attempt!

 

@Moxie: Glad, that I could deliver my apology & that we are fine now!

I agree 100% with your latest analysis: The problem most likely is the per-chip-calibration (@notaz) and to some degree maybe within the driver (how it considers calibration).

 

@Neelix & @Moxie: Thanks for your positive reports. Detailed enough, because if stuff works, why further analyze it?! Should you know, that you only use it in a very particular router environment, identifying it could help gather some experience values though! Thanks.

@b_o_b: Thanks for another negative report.

@vandyswa: Thanks for a mixed report (working & failing routers).

@Silent-Hunter: Thanks for your report about extremely bad luck with consumer products.

 

 

@EvilDragon & @Askarus: As I told you privately via email, I had wanted my Pandora tested particularly scrutenely, before sending the third Pandora to me. WiFi was found to be working at OpenPandora Gmbh. When I got it, it obviously didn't (in various environments under high scrutiny). Therefore I doubted your testing method/environment. I cannot stress it enough: Please test with a default dumb consumer router, not one very particular one, which just luckily works fine with the "cranky and idiosyncratic" (@vandyswa) Pandora WiFi chip.

 

 

Regarding poll: Danger of being a) biased (who rather reports: proud or disappointed owners?) B)  again too superficial (WiFi: good/bad not significant enough, more choices difficult as c) the poller must know all possible issues prior polling and d) possibly too hard to answer for many users as this requires much monitoring/testing). Therefore this qualitative discussion is the better approach I hope.

 

Last but not least: I still hope for input from the hardware engineers / firmware devs on this, as the problem is quite probably already identified somewhere between driver and per-chip-calibration.
 
Last edited by a moderator:
@PowerGod: Thanks for your WiFi driver code exploration attempt!

 

@Moxie: Glad, that I could deliver my apology & that we are fine now!

I agree 100% with your latest analysis: The problem most likely is the per-chip-calibration (@notaz) and to some degree maybe within the driver (how it considers calibration).

 

@Neelix & @Moxie: Thanks for your positive reports. Detailed enough, because if stuff works, why further analyze it?! Should you know, that you only use it in a very particular router environment, identifying it could help gather some experience values though! Thanks.

@b_o_b: Thanks for another negative report.

@vandyswa: Thanks for a mixed report (working & failing routers).

@Silent-Hunter: Thanks for your report about extremely bad luck with consumer products.

 

 

@EvilDragon & @Askarus: As I told you privately via email, I had wanted my Pandora tested particularly scrutenely, before sending the third Pandora to me. WiFi was found to be working at OpenPandora Gmbh. When I got it, it obviously didn't (in various environments under high scrutiny). Therefore I doubted your testing method/environment. I cannot stress it enough: Please test with a default dumb consumer router, not one very particular one, which just luckily works fine with the "cranky and idiosyncratic" (@vandyswa) Pandora WiFi chip.

 

 

Regarding poll: Danger of being a) biased (who rather reports: proud or disappointed owners?) B)  again too superficial (WiFi: good/bad not significant enough, more choices difficult as c) the poller must know all possible issues prior polling and d) possibly too hard to answer for many users as this requires much monitoring/testing). Therefore this qualitative discussion is the better approach I hope.

 

Last but not least: I still hope for input from the hardware engineers / firmware devs on this, as the problem is quite probably already identified somewhere between driver and per-chip-calibration.
You're welcome. Adding to that, the disk drive in question still works perfectly after almost 20 years.

And you've got a point regarding the poll.
 
Am I right that calibration is done in the factory (before the chips are shipped to PCB population facilities)?

If so, is there any way to fix dodgy calibration (if that is actually causing the problem)?

My pandora (summer 2012 rebirth), metal tape removed from antenna

works consistently well with

my router (a Draytek vigor ap700)

where is is normally the only client device. Adding one or two others does not make a noticeable difference.

There are between one and three other networks within range (depending on what the neighbours are doing)
 
Now, I might be talking completely rubbish here, since I do not know anything at all about this, really, but my thought would be that with Notaz special kernel module, that can dump calibration data to the file system and read it back again, there might be a possibility to invent some sort of self-calibration tool? As in, something that

* reads the dumped-out data

* does some measurements

* writes back a new version of the calibration data with some adjustments

* restarts the module/reloads calibration data

...and then loops.

Now, this would hinge on firstly, having a good understanding of the calibration data format and what part of the data controls what; secondly, a possibility to do meaningful measurements in some way on the pandora itself; thirdly, to be able to map perceived problems in the measurements to the correct calibration data. That is not at all trivial, I guess. But, still - An idea.

Notaz, feel free to tell me I'm daft :)
 
From memory, you need to have access to the manufacturer's documentation to understand the calibration file.

The files are not obviously similar from one chip to  the next.
 
From memory, you need to have access to the manufacturer's documentation to understand the calibration file.

The files are not obviously similar from one chip to  the next.
Why, if OP is a customer of TI, is TI allowed to withhold vital information about the product?
 
TI has moneys

TI has lawers

TI has customers

OP doesn't
I should think there's some law against them doing it though.

For example, let's say I go into an electronics store, and purchase an MP3 decoder chip (just an examle, it could be anything). Then I get home and open the box, only to find that there's no pinout chart, or any documentation whatsoever. So I call the people who made the chip, and they refuse to give it to me. That makes the chip a faulty product, and by refusing to help me, they are breaking the law.

At least, that's the way it is where I come from. I think.
 
Back
Top