20 March 2012

ks14 status update

The 14 Channel Rx-Only kismet server (I'm calling it ks14) is sort of in a holding pattern looking for a enclosure to house the radio/splitter/lna bundle. I also have others things taking up my time lately.

The Twitter feed shown to the right of this post is where I've been putting the latest pictures of the project. 

I tried a Hoffman NEMA 1 enclosure, 12x12x4, but it was way way too heavy. It felt like a firesafe, I returned it.

I need about 11" x 9.25" x 2.5"  preferably aluminum but will settle for plastic. I could rearrange things and resize the mounting board to 11x8"x3.5" or so, maybe.... but I'd rather not. I've looked around some stores for a plano tackle box or parts type organizer with no luck. Nothing seems quite right, or what seems good doesn't quite fit.

I want something relatively slim that will let the entire assembly slip into a backpack easily. I've considered going with something bigger to house the radio/splitter/lna plus a mini ITX board etc.... but that gets fairly large and I may want to connect to a desktop computer sometimes.

Anyway, I have several rack mount type enclosures that would work but they're larger than necessary. I may try to rig up something from scratch that will protect all the cables and connectors.

Oh yeah, I ordered (and received) a replacement for the Atom 330 that died. It's an ECS AMD E-350 mini-ITX motherboard. I swapped it into an an old Mobile Computing Solutions ITX case that had an ancient Via Epia-M Series motherboard in it.

From what I found, the E-350 should be around the performance of the Atom 330... maybe more.

I need to get the USB hubs mounted and setup on the splitter/radio/lna mounting board and then I can test things out while still searching for a enclosure. I was thinking I should mount more than 2 hubs to help spread the radio traffic load across more than 2 USB ports.... we'll see how many I can reasonably fit on the mounting board.

win7, not winning


Just spent a day trying to repair or recover a win7 install that suddenly (after using windows to shrink the C: volume) declared itself "Not Genuine" among other issues. The start of it all seems to be error logs during boot that show LoadUserProfile failed with "...the specified file is not in a registry file format". But multiple new accounts I created all had the same error.

Also, Firefox and IE both started up slow and could not save any downloads... but gave no error either. I finally fixed that problem by copying over inetcpl.cpl from another win7 machine. Initially after that inetcpl.cpl fix IE and Firefox were back to quick startup, but eventually they were back to about a 2 minute startup time..... although Firefox could restart itself in a couple seconds (after disabling/enabling add-ons and telling it to restart itself).

What a mess. I performed countless reboots to recovery console on cd or whatever was installed on disk and accessed via F8. Nothing would ever let me successfully run sfc (system file checker). I replace all .dll's in windows/system32 which seemed to make things slightly better. chkdsk, system repairs etc all failed to improve anything.

Anyway, I'm copying off all the data and reinstall win7 and will not waste anymore time trying to repair it. Just a few minutes to vent on this blog entry here.

The "interesting" thing is that the linux install still works fine even after massive upheaval/overhaul to add 1 gig of space to it from the ~67G windows partition. I needed the 1G to install KDE,Gnome, and LXDE to the original XFCE-only install.

sda 80GB drive - Original layout
sda1 100meg windows "system reserved" partition
sda2 ~67G windows C: drive (windows disk management shrunk this to ~66GB)
sda3 200M /boot partition for fedora
sda4 Extended partition
     sda5 ~8G LVM pv/vg for / and swap logical volumes for fedora

  1. dd sda3 to the free space created when windows shrank its own volume(sda2),
  2. delete original sda3, recreate it around the dd'ed sda3 image
  3. deleting the extended partition(sda4) and single logical partition(sda5)
  4. recreate larger ext partition (sda4) to include the remaining free space from the shrunken windows volume and free space from the original deleted sda3.
  5. Recreate the previous sda5 as sda6 on it's original sectors and a new sda5 (consisting of the rest of the new extended partition).
  6. lvm up the new sda5, extend the volume group, and online resize2fs / to add ~1GB
  7. fixup grub since /boot has been physically relocated on disk.
  8. Reboot 
(Actually I rebooted once before adding the new sda5 because neither partprobe nor kpartx worked with the new sda5 created, adding only sda6 they worked. After the reboot and creating sda5,partprobe worked. Probably something i was doing wrong?)

sda 80GB drive  - Final layout

sda1 100meg windows "system reserved" partition
sda2 ~65G windows C: drive
sda3 200M /boot partition for fedora install
sda4 Extended partition
sda5 ~1G LVM pv/vg for /  (all new freed space here)
sda6 ~8G LVM pv/vg for / and swap logical volumes (was sda5, untouched, recreated as sda6).


All that stuff went off without a hitch but windows still won't work properly or give clear error logs after a full day of effort. I'm no windows expert but I'm able to follow clues and use Google, there's just no help for it. It was the windows volume shrinker that broke itself I guess. I didn't touch the windows partitions using fdisk.

I'm not trying to make any OS #1 is better than OS #n arguments, just whining.
 
I feel slightly better now.

14 March 2012

backups mlocate.db and PRUNEPATHS on CentOS 5.x


While carving out some free space on a full backup server, I found a 1.6GB sized file /var/lib/mlocate/mlocate.db that was present in all backups of the backup server itself (yeah I know, don't backup a machine to itself). My backup scheme uses rsync with hardlinks and takes no space for copies of files that don't change... this file was changing.

My first thought was to add it to the excludes for that backup job but after looking into what mlocate.db is, I changed my mind.

mlocate.db is the index/database for the locate command. It's generated nightly by a cron job. updatedb has a config file at /etc/updatedb.conf. There's a setting for PRUNEPATHS in there which by default includes some preset dirs to exclude from indexing.

All the hosts that are backed up to this server have their backups kept in subdirs of /backup which was not being excluded from the nightly updatedb indexing. So, the executables for all backed up hosts were being indexed by the nightly updatedb run, resulting in a 1.6GB mlocate.db file

I have /var/log, /var/cache, /backup and some others excluded from the nightly "backup of the backup server" itself, but not /var/lib/. So, I was backing up the /var/lib/mlocate/mlocate.db file daily and since it was changing daily, my hardlink scheme was ineffective at saving any space, chewing up 1.6GB/day.

So, the proper fix is to put "/backup" into PRUNEPATHS in /etc/updatedb.conf which will prevent updatedb from indexing your executables in /backup.

I then ran the updatedb cron job manually (sh /etc/cron.daily/mlocate.cron on my system) to see how things looked. Now the mlocate.db file is >99% smaller:
 -rw-r----- 1 root slocate 1.4M Mar 14 08:51 /var/lib/mlocate/mlocate.db

That should save at least 20Gb in 2 weeks of daily backups, more with weeklys kept for a year.

While that may not seem like alot of space with modern SATA disks, it is for servers with SCSI arrays using disks from the 36 and 72 GB era.

09 March 2012

splitter math and hands on

Hands on first with some pics, then the long wordy text after the jump.

I did some quick 2.4Ghz band checks with splitter inline, without, and then with the splitter+LNA inline and switched on.

These were done in the same manner as the previous traces for my LNA Functionality blog post.

A quick description is:
Cable an Andrew DL-2402 antenna to the spectrum analyzer input and adjust so the 2.4Ghz band is roughly full screen. Then run the spectrum analyzer for a couple minutes in Max Hold for trace A, then freeze the trace, leaving it onscreen. Then run Max Hold for trace B with another set of connections, eventually freezing that one after a couple minutes.

(See the LNA Functionality post for a more detailed explanation of how the traces are acquired.)


Here's the Splitter vs NoSplitter test setup:

NoSplitter setup shown, already did Splitter trace



Closeup of the screen from picture directly above. I must have had the video bandwidth turned up more here than in the next test setup so these traces look less fuzzy/more smooth. But it's still valid to compare the two within this picture to each other:

2.4Ghz band Lower trace is splitter inline, upper trace, no splitter (10dB/div)



This is the full setup, antenna into LNA, into splitter, into spectrum analyzer:
Test setup for splitter+LNA vs splitter only(shown in splitter only setup)


Closeup of screen from picture directly above.

2.4Ghz band. Upper trace is splitter+LNA inline, lower trace is splitter only (10dB/div)


Basically that last picture there is the "proof" that the LNA into the splitter can work pretty much as well as a single antenna into a single radio for an RX only setup. I don't have a comparison of splitter+LNA to Direct-to-Antenna. I don't' know how i forgot that but... you can look at the LNA Functionality post and see a direct to antenna trace for comparison. (Don't compare absolute position on screen, I'm not sure things were all referenced the same between those tests and these ones here.)

I'll put the wordy rambling about splitters after the jump where I'll talk about
  • Perfect or Ideal Splitter Math
  • Splitter Reality



tgif

It's Friday, hoping to make some progress on this 14 channel monitor project. I thought I'd post an update first.

Thanks to dragorn for giving me a shout-out (or tweetout) yesterday. Now I have more motivation to actually finish this thing so I don't look like a quitter.

I'm going full steam ahead with this internet stuff and have set myself up on twitter. A few clicks and a twitter feed now shows up over on the right side of this page. Actually it took a bunch of clicks but most of them were to disable Ghostery and NoScript blocking so I could click through the twitter gadget addition process and allow the "twitter badge" to actually show up in my browser. I'm going to slow down before I go and do something totally crazy like get on facebook.

I broke out my old samsung moment the other day and installed the latest UberMoment rom by TheTehk17. The thing runs *much* better. Anyway, I got "the twitter" on it so it should be easier to post progress pictures that way. So check out the pictures in the tweets listed on the right.

Here's a few pics of what I'm calling "the mounting" progress (it's not much progress :) )
The plan is to get the radios and the LNA mounted to a small sheet of aluminum that is attached to the splitter, wire up a power harness (for ext. USB and LNA ) and get the whole thing inside some sort of box/case etc. with bulkhead connections for RF, usb, and power. We'll see.
Holes drilled for pigtail exit to splitter (on extra hole visible due to mistake, more mistakes on the other side of the radios!)

$5.75 wifi sticks have no built-in RF shields. Also, pigtail in new exit hole/groove

Pigtail exit, finished look

SMA to RP-SMA adapters weren't present when I took this mock up of how things will be mounted

08 March 2012

Last parts arrived

Finally got in the sma to rp-sma adapters i needed to connect the usb radios to the splitter. I guess they were on the slow boat from China. That's the last part that I was waiting for.

I decided to build RF shielding wrappers for each radio. The radios were the cheapest I could find and they have no shielding at all on the circuit board. There's solder pad and holes for shield tabs to be installed, but no shields.

I tried a few tests with shielding a 2.4gh antenna using anti-static bags, anti- static foam, and some other stuff I found. Aluminum foil actually worked the best, blocked out all signals I could see with the spectrum analyzer.

I decided to use aluminum flashing I already had around to make covers or wrappers for each usb radio. I did some cutting and bending and I have the dimensions all worked out, I just need to make 14 of them and figure out some way to mount everything together.

The usb radios have a short pigtail inside going tot he RP-SMA connecter mounted in the end of the stick. I drilled tiny holes in edge of the usb sticks and moved the pigtails so they hang out the edge and and can connect to the splitter. All the usb radio sticks can sit on edge on top of the splitter.

I'll have pictures later.

I haven't figured out what to replace my blown atom330 motherboard with yet so the first mobile test run will probably be using a core2 duo laptop. Assuming I can get it all together this weekend before I get a new atom box ordered in.

I've been busy and haven't had much time to work on it this week. Hopefully I'll have time Friday afternoon

05 March 2012

LNA Functionality

The LNA (Low Noise Amplifier), marked as "Low Noise Preamp" in the picture below, is from an old Avantek DR2 radio. I don't remember the specifics (and Avantek was a little before my time) but it was probably a 4-16 T1 radio from the 1980's. It operated in the old common carrier 2Ghz band(something like 1.9 to 2.1 Ghz but don't quote me) which was auctioned off to create  "PCS" services, which I guess are no longer differentiated from the old 800Mhz cell spectrum. Now 800Mhz and 1.9Ghz is all just "cell phone" service, I never hear "PCS" referred to anymore(but I'm not cell tech).

Without looking it up in the DR2 manual(if I could find one somewhere), it seems to be happy with about 12-15VDC for power. Quick checks detailed later in this post show it provides about 20db of gain, give or take 5 or 6dB. Those numbers are in line with fuzzy memories of other brands of LNA's on some microwave radios I've worked with.

Keep in mind we're talking about 20dB gain on a small signal... say a -95dBm signal amplified to -75dBm. It's not going to add 20dB to a +24dBm output of a wifi card to give you +44dBm. The thing has a point on it's input beyond which distortion ramps up. Keep going and you're just going to burn the LNA itself out.


I often see or hear wifi stories about "putting an amp on it" for better performance in some manner. I think alot of times that is a Tim the Toolman Taylor "MORE POWER!! ARRRRRGGH!!!" type of statement. In other words, "I cranked it up to 11 !!!!".

Granted there are occasions to use the kind of power amp being referred to in those stories, but in most setups it should only be because their radio has a weak output or they have large cable losses to the antenna. For normal setups(my 14 channel RX only project here is non-normal) wifi amps need to be bi-directional because the radios have a common RX+TX input.

That is not what a unidirectional LNA is about....

Read on for some "in-action" pictures of the LNA and my long rambling attempt to convey my understanding of LNAs. You've been warned.

04 March 2012

XPS13

Just saw a commercial for XPS13.....
http://www.dell.com/xps13

$999 1.6Ghz Core i5, 128M SSD,  4G, carbon fiber and aluminum, backlit keyboard, glass gesture capable touchpad,

Looks comparable to the mac air, of course it can't be "more cooler" though. Might make a nice used laptop purchase on ebay one day, assuming they dont' hold their value like a Mac.


02 March 2012

14 Stick Shootout

After much fiddling with USB hubs, ext power supplies, and a blown Atom 330 motherboard I managed to get 14 radios wired into 2 x 7port hubs and into a working computer.


lsusb with 14 radios


I only had to edit the NAME field in the auto-generated udev rules to have them come up as specific wlan1-14 numbers based on their MACs(this is on Fedora 15).
wlan1 through wlan14

 The next thing to do was run kismet_shootout.rb(from kismet source ruby/ dir).
kismet_shootout.rb running with 14 radios

The idea was to verify that all the radios seem to perform more or less the same.

Next I'm going to run some shootouts with the splitter, LNA, and plain jane usb stick etc. The numbers I got above should rule out having to swap radios between the different antennas and antenna coupling setups to prove any differences are not the radios themselves (well, informal seat-of-the-pants proof anyway).

As for the blown motherboard mentioned above... when I hooked all the radios to the MSI Wind Atom box it booted slow, the BIOS beep was a different pitch and longer, but it did boot. I plugged in the external power supplies to the USB hubs and things were ok, but when i power cycled the machine(using it's 12V power plug)... it no longer came up. I checked the board out as well as I could considering there was no visible smoked parts... the little surface mount fuses I could find on it were ok... power supply's fine.

So I suppose I'll never know if a Atom 330 can handle the task. I'll replace it because my truck has a DCDC-USB wired in and I need a low power type box, but I'm assuming I'll end up with a D525 or whatever the latest mainstream equivalent is.

first post

I've sometimes thought it would be cool to have one of those fancy blog things and document some "awesome" project or another I'm messing with. I usually decide that my project is not really that interesting to anyone(or even myself) or I only take time to snap a few pictures and never do anything with them. If anything I only jam them up on some web server email a few direct links to friends.

Well, apparently I'm pretty excited about this latest project since I've actually created a blog.... it only took until 2012 but here it is.

Whether this project gets completed or not, I might find the blog useful for documenting other projects in the future.

Ok, on to the current project and impetus for creating this blog:

Portable 12V powered 11 (or 14) channel kismet_server running from a single antenna
(no fancy name yet.... maybe NFNY14.
 I actually sort of like "ks14" )


Some goals:
  • Prove out the single antenna to multi-radio via LNA idea
  • Determine horsepower needed for 14 concurrent channel kismet_server
  • Document it along the way

Some hardware particulars(red stuff in shipping, green is in my possession now):

Some software/config particulars:
  • Openwrt trunk
  • Kismet-2011-03-R2 
  • gpsd talking to an Axiom Sandpiper OEM GPS board
  • RX ONLY, (no VAPs, interfaces in monitor mode only)
Most of the materials on hand now
That's the basic gist here. Further posts will address the splitter losses, how well the LNA overcomes them, USB power consumption, mounting, and whatever else crops up on the way to success.