Shop OBEX P1 Docs P2 Docs Learn Events
what are the possible ways to use GPRS with the propeller? — Parallax Forums

what are the possible ways to use GPRS with the propeller?

Bobb FwedBobb Fwed Posts: 1,119
edited 2010-11-16 08:21 in Accessories
What are the possible ways to use GPRS with the propeller? It doesn't need to be GPRS specifically, but some type of cellular communication. What hardware is used to do this?

How is this setup with the wireless carriers? We want to setup a data-only (low quantity of data at that) system that conveys about 60 bytes of data per hour (maximum of 500KB per month). Is there some way to work with the carriers to get low cost accounts? We would likely add a few hundred accounts per year.

I can figure out the carrier stuff, that's just going to be a lot of talking with the companies, the hardware is the big thing I care about.

Comments

  • TubularTubular Posts: 4,646
    edited 2010-09-13 04:12
    Hi Bobb,

    GPRS is a good way to go. A couple of years ago i picked up a Telit GPRS modem, played with it for a few hours, and was surprised at what it could do. I wouldn't do it any other way.

    Telit easy-gprs scripts are built in for email, ftp, web page get and put, socket connections for sending data like for your application, and more. There are other brands such as Sagem, Siemens, Wavecom that have equivalent gprs modules.

    You're right about getting the SIM cards and plans being the hardest part. Its easier to get started now there are so many smartphones needing data plans. We pay something like $3 a month for a SIM that would do 500kB.

    regards
    tubular
  • Bobb FwedBobb Fwed Posts: 1,119
    edited 2010-09-13 08:41
    Thanks for the info, I will look into it. I am guessing (hoping) the Telit modem is designed as a module for a microcontroller? Or is there some way to inface it with the propeller?
  • TubularTubular Posts: 4,646
    edited 2010-09-13 16:10
    Yep, the gprs modems are absolutely designed to connect to a microcontroller. There are two serial wires TXD and RXD, and these are signalled at TTL levels for a microcontroller, rather than say RS232 for a PC. You need a few other wires to turn the module ON/OFF, and sometimes additional handshaking lines such as RTS/CTS and DCD can be useful but not essential.

    You have to be careful with the power supplies. The Telits are designed to run from Lithium batteries, about 3.4 to 4.2 volts. But the modems have sharp current spikes (about 2 amps) for short periods when transmitting, so you need big caps to try and smooth it out. The prop is only rated to 3.6v, so there are a couple of approaches:-

    1) Try and run everything at 3.6v. Your regulator will need big caps so the Telit voltage doesn't sag below 3.4 volts when transmitting (and drawing its 2A spike).

    2) Run the Telit from a LiPo battery (3.7..4.2v), Prop from a separate 3v3 LDO reg off that battery, and have small protective resistors in series on each of the interfacing lines (RXD TXD ON/OFF etc), to protect the Prop.

    Last week I rushed out a small Telit breakout board that would fit the prop proto board. I'm only half expecting it to work, it will definitely require a further revision or two. I'll let you know what happens with it
  • Bobb FwedBobb Fwed Posts: 1,119
    edited 2010-09-13 16:37
    I got a chance to look at the manual a bit. Doesn't look too complicated to interface. The project I am working on doesn't have super tight constraints on budget (always a good thing) so I figured for the power I would run my Prop(s) on my normal 3.3V reg and have a separate adjustable reg for the 3.8V (maybe I'd go 4V) supply.

    You say "big caps" that's a broad range. How short of a spike is the 2A draw? I was hoping a 1A linear regulator (don't want to mess with switching regs on an RF device) with a couple thousand micro-farads of caps would be enough. The problem will likely be the ESR required to pump out during the spikes. What kind of setup did you go with? I think at some point it would be cheaper to just get a 2A/3A regulator instead of loading the board up with caps which could produce questionable for reliability especially over the long haul (5/10+ years down the line is very important for us).

    Where did you order your Telit from? Directly from them? The only supplyer we use that has any is Newark, but they only have 1. I figured I could get the 1 and mess with it, then find a normal supplier.

    If you have any extra breakout boards, or want to share the design files, I would love to have them so I can start messing with one. Are they functional for standard breadboards? I don't have a Parallax proto board.
  • TubularTubular Posts: 4,646
    edited 2010-09-13 20:27
    Bobb, Linear PSU is certainly a good place to start. We attempted a TI switcher design but went back to a linear approach. We used large supercapacitors (50-150Farad) as the primary supply for longevity, so I couldn't tell you what the minimum that would work would be. The evaluation board that came from Telit had two big electo's on it. The spikes are at 216 Hz I think.

    We are just now having a second crack at a suitable switcher from Linear Tech. We'll see how it goes this time round.

    Regarding suppliers, we get ours from Glyn (Germany company with local offices here). Not sure if they are in the US. Sparkfun have a range of Telit modules and breakouts. Arrow electronics have some in stock according to FindChips.com. But if they aren't "readily" available to you, I would look at another brand. I know Newark/Farnell carry other brands too. I found the Easy-GPRS feature of the Telits very easy to use, but the others probably have equivalent scripts that are equally good. Its quite a competitive little area, GRPRS modems.
  • TubularTubular Posts: 4,646
    edited 2010-09-13 21:21
    Bobb regarding board designs etc, let me fire up my design when the board comes back shortly. It was done in a hurry, I'm not really expecting it to work without some modification, but if it does work out it might be an easy way of getting Propeller Proto Boards online...
  • Bobb FwedBobb Fwed Posts: 1,119
    edited 2010-09-14 08:57
    Tubular wrote: »
    The spikes are at 216 Hz I think.
    I saw that, but is that 50% duty? Or what? I didn't see any information on that.
  • TubularTubular Posts: 4,646
    edited 2010-09-14 13:59
    Bobb, it's a low duty cycle, and the duty will depend on how many "slots" the GPRS uses.

    Average current - I seem to remember a few hundred mA for a 2G module, and up to about 800mA for a 3G module. 1 amp with lots of smoothing should be fine at least for 2G.
  • Tracy AllenTracy Allen Posts: 6,658
    edited 2010-09-14 22:39
    I've used the LT1963 linear regulator, which is a one optimized for fast transient response. Enfora GPRS modems and also the MultiTech socket modems spike from 400mA up to about 1.4 Amps, but the regulator limits the supply transient to around 50 mV with a 100µF low-ESR output capacitor.

    I'm curious about those Telit modules. They look very small. At first glance, it appears that they have to be mounted on a PCB using a ball grid scheme. Is that right?
  • TubularTubular Posts: 4,646
    edited 2010-09-15 00:42
    The raw modem has ball grid, then they progressively add more pcbs and connectors until you have the solution you need. For instance

    GE864 is the small essential modem, cheapest form factor
    GC864 adds a pcb w/sim pads, and 80 pin molex connector (so its thicker and longer)
    GT864 is the terminal in plastic case with RS232, power supply etc.
  • Bobb FwedBobb Fwed Posts: 1,119
    edited 2010-09-15 17:10
    So I did the math, if the power drops back down to 19mA between the bursts of 2A, that gives us a duty cycle of about 30%. So with that, I SPICE'd it with the LT1963A regulator (1.5A output). And it will work fine up to about 19V input with 2000µF electrolytic + 200µF ceramic. Lucily I am running at 14V where it works better. The fluctuation goes from 4V (where I have the reg set) down to 3.6V during bursts.

    I also plan on trying to keep output bursts as low powered as possible (so 1X TX?), so hopefully that will bring the duty down to 14%, which will make it all better as well (@19V it goes down to 3.75V). I think it's worth a try with this setup....

    With both duty draws, at 14V it only drops 20mV or so. This all assumes I can dissipate the heat that will be produced. I will have to do the math on that.
  • Tracy AllenTracy Allen Posts: 6,658
    edited 2010-09-15 18:45
    Bobb, were you using the Linear Tech Spice model for the LT1963A? My jaw drops about the 2000µF electrolytic + 200µF ceramic. The data sheet suggests at least 10µF at the output, and some of the graphs go as high as 100µF tantalum at the output plus 10*1µF of ceramics distributed near chips. But I don't see a recommendation of 2000+. If that is hanging on the output of the regulator, it will present a heavy load for starting up from the shutdown state. There has to be a good bypass on the input, say another 10 to 100 µF for stability, and of course the input supply has to have enough punch for the spikes. I haven't actually measured the spikes, and it would be interesting to do so under operating conditions. I've been using the TPSC series low ESR solid electrolyte capacitors from AVX.

    Tubular, does your new PCB use the ball grid devices?

    I wonder why the manufacturers insist on unfriendly connection methods. I don't like the 80-pin Molex sockets either. So many connections, all for tx, rx, power and ground and maybe a few handshaking lines. The MultiTech socket modems have pins on 2mm spacing, but only a few of the 64 possible positions are implemented, and that is done in a way that leaves it mechanically unstable as it sits in a socket. Go figure.
  • Bobb FwedBobb Fwed Posts: 1,119
    edited 2010-09-15 19:00
    the reason such large caps are needed is because I am going over the current limiter, so the caps have to supply power by themselves for a little while. The limiter is 1.6A at 20V and scales up as the input voltage goes down. so at 14V it is close to 3A if I remember (maybe that was 10V), so I could probably drop some caps...
  • TubularTubular Posts: 4,646
    edited 2010-09-15 20:55
    Tracy, I am indeed using the BGA Telit GE865 in the new pcb. The pcb is 1.2" square, covering the core of the propeller proto board, Telit "down" toward the P8x32A, and sim card holder and other components "up" inside the 1" square.

    I didn't have time on that board to lay out the power supply, so its just brought out to a corner terminal. For test purposes I will just use a Max155x and 2000 mAH lipo battery. There is an Attiny85 onboard for initialisation if required.

    I need to preserve the reset button somehow on the next rev, since this board covers the central core of the proto board.
  • Bobb FwedBobb Fwed Posts: 1,119
    edited 2010-09-16 09:35
    Bobb Fwed wrote: »
    the reason such large caps are needed is because I am going over the current limiter, so the caps have to supply power by themselves for a little while. The limiter is 1.6A at 20V and scales up as the input voltage goes down. so at 14V it is close to 3A if I remember (maybe that was 10V), so I could probably drop some caps...

    I forgot to say, yes, I am using Linear Technology's SPICE.

    Current limit @ input voltage (for 4V output)
    1.05A @ 20V
    1.6A @ 18V
    2.15A @ 16V
    2.7A @ 14V
    3A @ <13V

    This is interesting, because they say 1.6V minimum current limit if input is only 1V above output. At 18V, that's a 12V difference. And I don't see in the datasheet anything about the 3A max.

    So, above 16.1V, the capacitors will be powering the device for a bit. My target input is 14V, but it could periodically be close to 16V.

    What I will likely do on my production design is switch the voltage down to 5V maybe. That way I could power my 5V devices, and use the linear regs for 3.3V and 4V, with very little heat. At 5V input, there is only a 10mV fluctuation.
  • Bobb FwedBobb Fwed Posts: 1,119
    edited 2010-09-16 10:50
    I found the current limit information, it wasn't in the tables, because it's more fluid than I though. View graphs:attachment.php?attachmentid=73270&stc=1&d=1284658850

    So if I keep it cool (below 50 or estimated 70C), at 5V (1V differential) the current limiter won't kick in. But now that makes be want to bump the switching pre-reg up to 8V maybe, so I get more into that hump above 2A, but still fairly low to keep the heat dissipation to a minimum. Ug. Maybe I should just get a 2A reg.

    Seeing as this device is supposed to last at least a decade (hoping for decades), I shouldn't skimp too much here, and it will put my mind more at ease for removing the electrolytic caps. 2A reg would allow me to keep the 5V pre-reg which would save me the money I lost by not needing another, separate, 5V reg.

    Any suggestions on a fast transient response linear technology 2A regulator? I'll look around.
  • TubularTubular Posts: 4,646
    edited 2010-09-16 12:20
    Bobb, we're trying a LT3680 (3.5A step down) design, as soon as the boards arrive in the coming days. http://www.linear.com/pc/productDetail.jsp?navId=H0,C1,C1003,C1042,C1032,C1082,P38541

    For a linear regulator solution, the LT1528 works (and is used on several designs including Sparkfun)

    tubular
  • Bobb FwedBobb Fwed Posts: 1,119
    edited 2010-09-16 12:26
    Ya, I found that one (LT1528) and LT1764. But I think I like NCP5663DSADJR4G (ON Semi) better. The price is almost half, and the performance may be slightly better. But if a switching solution works, I will be very tempted to use that instead.
  • Tracy AllenTracy Allen Posts: 6,658
    edited 2010-09-18 16:20
    The LT3680 is attractive for a number of reasons, but I'm concerned about the switching noise applied to the RF modem. It looks like about 10 mV of ripple at the switching frequency (~1MHz) and other frequency components if it goes into the low power burst mode. Maybe it will be fine, so please do report back what you find. I guess a lot depends on how well the RF modem rejects that noise, and PCB layout. That thick BGA package sounds like a challenge!

    The NCP5663 looks like a good find, attractive price for the BoM. It appears to be an NPN output regulator, with 1.5V dropout, as opposed to the LT1963 which has a PNP output, 300mV dropout. Same pinout, higher maximum current, and fast transient response. I'm tempted to use it, except there is the 30µA current in shutdown compared with <1µA for the LT1963. 30µA isn't bad, but my apps are solar powered and I obsessively count microAmps.


    There is a Linear Tech app note, an104, written by LT tech guru Jim Williams about how they do load transient testing. The example is the LT1963. His cartoon...
    attachment.php?attachmentid=73311&stc=1&d=1284851835
    630 x 440 - 24K
  • TubularTubular Posts: 4,646
    edited 2010-09-18 21:29
    Lol, if only datasheets were that entertaining.

    Since the D2PAK pinout is the same for the OnSemi and LT parts, makes sense to design the PCB around those, and load with the right part for the application.
  • Bobb FwedBobb Fwed Posts: 1,119
    edited 2010-09-20 10:11
    The NCP5663 looks like a good find, attractive price for the BoM. It appears to be an NPN output regulator, with 1.5V dropout, as opposed to the LT1963 which has a PNP output, 300mV dropout. Same pinout, higher maximum current, and fast transient response.

    1.3V dropout is the theoretical maximum dropout at 3.0A. According to the charts and graphs, nominal dropout is closer to 1.0V @ 3A or (more importantly) 0.92V at 2A. My Li-Ion battery shuts off at 5.5V, so 5.5 - 1.3 is still above 4V, so I am good, no matter what.

    The shutdown amperage is important to me as well. My project has to go into a standby mode and run on battery for weeks at a time. I will continue looking, but I think I can manage with a few extra µA that this will use.
  • Bobb FwedBobb Fwed Posts: 1,119
    edited 2010-10-20 09:44
    So what is the current status on your work with the GPRS module?
  • TubularTubular Posts: 4,646
    edited 2010-10-20 13:28
    Fwiw , this is the GE865 breakout I designed to work with the propeller proto boards.

    The telit goes face to face with the prop chip, exposing the SIM card uppermost. All the useful pins are broken out around the rim, and the essential ones are in a single bank of 8.

    The pads are for an AtTiny45v if required for initialization or standalone operation

    On the back on the board is the SIM holder, bulk caps, status Leds, antenna connector, etc. No reg on it at this stage (doing that externally)
    1024 x 771 - 291K
  • Bobb FwedBobb Fwed Posts: 1,119
    edited 2010-10-20 15:05
    So have you used it yet? How have you (or have you) setup the service/data plan?
  • TubularTubular Posts: 4,646
    edited 2010-10-20 15:47
    Hi Bobb,

    That one hasn't been connected just yet, but I have used '864s successfully in several projects over the past 3 years, and some Sony GR64s before that.

    When you purchase a data sim you need to find out what the APN is for the provider you're purchasing from. You will need that so the Telit knows which of the many networks to connect to.

    See the screenshot in this message for the commands to get started. Change the vfinternet.au to the APN given to you by your provider.
    http://forums.parallax.com/showpost.php?p=877355&postcount=26

    If you have any questions, I'll be happy to help where I can...
  • Bobb FwedBobb Fwed Posts: 1,119
    edited 2010-11-11 13:58
    I've just finished the prototype of a big project here, so while we are waiting for boards back, I'm going to jump into trying to use the Telit. You wouldn't happen to have a component for DipTrace for the Telit, would ya?
  • TubularTubular Posts: 4,646
    edited 2010-11-15 22:14
    Bobb,

    Just getting up to speed with a couple of threads. Do you still need the Diptrace footprint? I can probably organize this (extract from existing design) for the 80 pin molex (GC864 and/or UC864). I don't have anything in DipTrace for the GE865 yet, again could solve this if you still need it.
  • LeonLeon Posts: 7,620
    edited 2010-11-16 02:55
    The Telit modules are DCEs and transmit on RxD and receive on TxD, that caught me out once.
  • Bobb FwedBobb Fwed Posts: 1,119
    edited 2010-11-16 08:21
    Tubular wrote: »
    Bobb,

    Just getting up to speed with a couple of threads. Do you still need the Diptrace footprint? I can probably organize this (extract from existing design) for the 80 pin molex (GC864 and/or UC864). I don't have anything in DipTrace for the GE865 yet, again could solve this if you still need it.
    That would be great if you could do that. It would help me out a lot!
Sign In or Register to comment.