GTA04b7/Neo900

Issue 526: Choice of GPS receivers

Reported by Nikolaus Schaller, Dec 7, 2013

To fulfill the FSF "freedom" requirement:

"* The microphone and the GPS should be connected to the main 
computer,
  not to the radio modem."

1. built into UMTS/LTE module (does not fulfill requirement but is 
the cheapest)
2. W2SG0084
3. uBlox EVA 7m 
http://www.u-blox.com/de/press-and-events/press-release-archive/1556-
u-blox-launches-eva-7m-industrys-smallest-standalone-gnss-module.html


The idea is not really to choose between them but allow any option. 
I.e. have the fottprint of all and maybe just solder a different 0R 
to connect the LNA output to either one. Maybe 2/3 are alternative 
(because they also have to share an UART).

Problem is that 2.&3. may be expensive (2. is ca. 15 EUR in BOM, 
3. may be 50 EUR) while 1. comes for free.

Comment 1 by Joerg Reisenweber, Dec 7, 2013

FSF evidently has no ideas about hw alternatives, so their 
"requirements" are just the "standard" way to 
go. Thinking about the rationale behind the GPS requirements you 
find that same purpose can get accomplished by reliably disabling 
the GPS antenna.
Easy to implement, no RF routing nightmare with 0R (DNP) 
"jumpers" and no wasting of PCB real estate for footprints 
of chips (DNP).

btw FSF has more "requirements" for hw, some of them 
totally insane. I don't plan to comply with Mr Stallman's ideas, 
rather carry on the spirit and do better.
According to USB authorities we mustn't offer hostmode, according to 
FSF we mustn't use WLAN with firmware blob etc pp. I need no help to 
define a free and useful good hw design.

Comment 2 by Nikolaus Schaller, Dec 7, 2013

There are good reasons to think about having a separate GPS receiver 
(independently of FSF's argumentations):

* we can offer GPS on a device that has no modem at all (PDA style)
* we can get a highly precise Pulse Per Second interrupt from a 
separate receiver
* we can be sure that the UMTS modem can't control or read out the 
GPS receiver (this is what FSF wants to protect us against, not from 
using GPS at all)
* the u-blox can provide 10 position updates per second
* it may (has to be checked) need less power

Therefore my proposal to think about it and if we have the space on 
the Lower board we can add the footprints. And people can either get 
the cheapest variant or pay more for one of the benefits described 
above.

Comment 3 by Joerg Reisenweber, Dec 7, 2013

we can test it on one of the protos, see what it gives us: options 
or flaws. Footprints are cheap, cost real estate only

Comment 4 by ben deering, Mar 25, 2014

I often think of my GTA04 as a GPS that can make telephone calls.  
If neo900 has no internal GPS that is isolated from the UTMS chip 
that would be a dissapointment and would appear to run contrary to 
the design goals of neo900 + untrusted UTMS chip; the UTMS chip 
would have access to location data any time the user wanted location 
data.

A GPS that can be trusted not to broadcast position seems like one 
of the best uses of the limited real-estate in the neo900 board, or 
possibly a daughter board that would let users move the antenna coax 
from a UFL connector connected to the UTMS internal GPS to the 
(optional?) daugherboard.  I don't know enough about the space 
created by the spacer ring, or about RF circuit design to know if a 
daugherboard is feasible at all.

Comment 5 by Joerg Reisenweber, Mar 25, 2014

thanks to tight monitoring of the modem you can still trust in it 
not sending anything to the air, even when you activate it for GPS 
usage. Of course you couldn't stop it fast enough when it ever would 
decide to actually send something despite you instructed it not to 
do, but then OTOH this is still better than a GPS you don't know if 
it has a transmitter built in and nobody told us about it. At least 
you'd know the very second it tries to cheat on you and you could 
decide what to do about it.
If that's still too high a risk, you better choose a GPS device that 
can't "make telephone calls"

Comment 6 by Joerg Reisenweber, Mar 25, 2014

ps: nota bene that using WWAN and still hoping to hide your position 
from network while user _can_ readout the position from an isolated 
GPS chip is a useless dream since WWAN gives away your position even 
without any GPS, when active. Just a bit more effort needed by 
carrier to find out about your position, and precision might be in 
the 20..50m range instead of 2..5m with GPS

Comment 7 by ben deering, Mar 26, 2014

I don't want to clutter the issues discussion page, but I think 
there is value in separating these functions.  Where I live there is 
often only 1 visible tower and where I play there is often none so 
base-station location tracking gets me within ~10km.  The modem 
would also have the possibility of logging location history when no 
basestation is available for hours/days.  

Several products now contain GPS-related 'anti-features' such as not 
allowing geotagging while in China, imposing a maximum speed on 
sports-cars unless the GPS indicates they are at a race-track, etc.  


The neo900 discussions on the forum and the mailing list have had to 
answer the 'open device with a closed modem' question dozens of 
times.  I am satisfied with the answer 'modem will only have access 
to what it needs, can be turned off, and can be monitored', but GPS 
location is not something the modem needs.

Comment 8 by Joerg Reisenweber, Mar 26, 2014

We'll see what initial tests will yield.
See Nikolaus' first post that states a 100EUR for BOM, if we go for 
a separate GPS which really is worth it. Development and testing 
costs (prototypes, field tests, PV) go on top of that, so depending 
on how many would want to buy this option, we have to increase the 
price by *significantly* >100EUR per device for that option. When 
you think a 100 users would be willing to pay 200EUR on top 
(ballpark figures) then it's no problem.

Comment 9 by Joerg Reisenweber, Mar 26, 2014

OOPS, sorry. Usual doble-normalization-mistake. BOM: 50EUR, sales 
price 100EUR (though the rule of thumb usually is "3 times the 
chip cost is sales price"). However the needed prototypes 
particularly for that separate GPS chip solution and the development 
and testing work still is the same, and we can't make all devices 
more expensive to finance those expenses for a few users who want 
the very special option to  use GPS while WWAN active and yet WWAN 
doesn't know about GPS position.
The logging inside WWAN would be theoretically possible *only as 
long as user enables GPS (otherwise it's blocked by shorting the 
antenna)* but that would mean the modem firmware would need a 
malware function supporting this, since readout of such log isn't a 
part of standard GSM specs. Such malware function would be useful 
for Neo900 only since for all other devices using the Cinterion 
modem there are way simpler methods of acquiring same data realtime 
and without firmware patch.

I honestly wonder if a GPS module attached to the expansion slot 
wouldn't be a better option for those interested in such 
"modem-less" GPS

Comment 10 by Kuper, Mar 27, 2014

If it is true that a primary reason for considering buying a Neo900 
is that it does not compromise any more than absolutely necessary 
w.r.t. privacy and freedom, then surely the default option should be 
to have the microphone and the GPS connected to the main computer, 
not to the radio modem.

I suppose you could offer a compromised phone at a lower price, but 
the more compromises, the more you'd be putting it in direct 
competition with the Replicant-compatible mainstream devices, the 
Fairphone, etc.

Comment 11 by Joerg Reisenweber, Mar 27, 2014

please don't abuse this as forum! there's 
http://talk.maemo.org/showthread.php?t=91142 for that
Thanks!

Comment 12 by ben deering, Apr 15, 2014

http://www.digikey.com/product-search/en?K=943-1019-1-ND;site=us;lang
=en&WT.z_lt=943-1019-1-ND&WT.z_supplier_id=943&WT.z_ref_p
age_id=14&WT.z_tz_page_sec=Modules&WT.z_TechZone_Name=Wireles
s 
Jupiter SE880 is in 4.7 x 4.7 x 1.4 mm, 34 pins LGA package

Very small, a little power hungry, capable of 5hz updates.

Comment 13 by ben deering, Jul 7, 2014

Note:  I was wrong about this device being power hungry.  The ma 
looked high compared to others, but it was because the operating 
voltage is lower.

Created: 6 years 4 months ago by Nikolaus Schaller

Updated: 5 years 9 months ago

Status: New

Followed by: 2 persons

Labels:
Type:Enhancement
Priority:Critical