1
General discussion / Re: New Microchip PIC32 chips in DIP packages are coming
TI already offers CortexM3 in DIP packages (originally made by Luminary).[/quote]
Which ones? The closest I see is 28SOIC for some of the X00 series
Messages
This section allows you to view all Messages made by this member. Note that you can only see Messages made in areas you currently have access to.
Messages - Entropy
TI already offers CortexM3 in DIP packages (originally made by Luminary).[/quote]
Which ones? The closest I see is 28SOIC for some of the X00 series
2
General discussion / Re: Eridani: ARM USB OTG/HOST/DEV development board (LM3S365
All of the wireless carriers out there are currently offering total-transfer-capped service.
If you want to do something like this with a cable or DSL provider, you're going to need something with multiple Ethernet ports. This particular unit isn't going to do the trick for the kind of bandwidth you are talking about, and you're more likely to have success doing channel bonding with something like a hacked wireless router supported by DD-WRT. (Play with the VLANs and you can do multi-WAN).
However, I think in nearly all cases, pricing tiers are such that it will be cheaper to bump yourself up a pricing tier or two within a single account than obtain multiple accounts. I already did this analysis for AT&T, but I'm fairly certain it holds true with most of not all cable/DSL providers.
Note that what you're trying to do has been done, however it was not done for cost savings, but to obtain performance greater than what was available from the service provider in the case of a service provider (Iridium Satellite) that only offered circuit-switched data at 2400 bps. (A research institution channel bonded four Iridium modems to get 9600 bps throughput at an insane cost to support a research station in Greenland. Modem co-interference was a challenge for them.)
3
General discussion / Re: Eridani: ARM USB OTG/HOST/DEV development board (LM3S365
Yup. There's also the fact that even without local co-interference, three users simultaneously in the same pattern might do weird things.
One thing to keep in mind regarding infrastructure upgrades is that to add more cells to a network is VERY difficult - in some areas there simply might not be any sites for new cells. (NIMBYism - it is a common thing for residents of a neighborhood to vehemently oppose construction of a new cell tower and then complain that their cellular service sucks.)
Spectrum availability is gated by the FCC, that is another limitation in terms of upgrades. Improved modulation schemes are improving spectrum efficiency, but it isn't the providers that are slow here, it is the standards organizations. LTE Advanced is not finalized yet...
Also, how exactly is are the costs the same? You need three separate wireless accounts in this situation. For example, AT&T's 200MB plan is $15/month, the 2GB plan is $25/month (less than double the price for 10x the monthly transfer), the 4GB tethering plan is $45/month (slightly less than double the price for double the monthly transfer and significantly reduced overage charges). Oh yeah, technically this approach is violating the terms of service anyway regarding tethering for the lower plans.
4
General discussion / Re: Eridani: ARM USB OTG/HOST/DEV development board (LM3S365
I appreciate your feedback and opinion.
The reason it is male is so that you can use a ribbon cable and have a daughter board really. For breadboard I understand that is not ideal. The main advantage of using ribbon cables is that you preserve signal quality and speed.
I thought of doing two rows, on each side, but I wanted all SMD and so I couldn't. I could have left it unsoldered. Perhaps if there is a REV 3 I will do that if people don't mind soldering. There are some layout issues with doing it also... If you scrap the second JTAG interface you can get enough edge space.
I use this sort of thing: http://www.sparkfun.com/products/9385
For prototyping. Yes they aren't super cheap but a good thing to have around for breadboarding. Lots of things have male headers. :-)[/quote]
A DIY version of those is described at http://forums.adafruit.com/viewtopic.ph ... ser#p77776 and http://forums.adafruit.com/viewtopic.ph ... ser#p77777 - It's a bit of a pain in the butt, but potentially cheaper than the SFE jumpers.
5
General discussion / Re: Eridani: ARM USB OTG/HOST/DEV development board (LM3S365
Is the GPIO header prepopulated? It appears that it is prepopulated with a male header. Given that it is a two-row header (cannot be plugged into a breadboard in any way), why not leave it unpopulated (a la the wing headers on Gadget Factory boards, or the TI/Luminary EKx-LM3S6965 board) permitting a female header to be placed there? (easier to jumper to a breadboard for prototyping than a male header).
Or, since you don't have many GPIOs, why not split them across the sides of the board in two single-row male headers? This would permit the board to "straddle" between two breadboards (similar to how the EKx-LM3S811 board can be used if populated with downward-facing single-row headers).
I remember you've made some negative comments regarding the way Arduino I/Os are arranged (And I agree with you here - it is an incredibly stupid design that makes it impossible to even achieve the "straddle" approach described above even when the female headers are replaced with downward-facing male headers. The fact that they rushed it out the door instead of just slipping a week is simply inexcusable for such a high-impact mistake.), but honestly your approach here doesn't seem much better.
If you had as many IOs as the EKx-LM3S6965 or STM32-H103 boards, this would be understandable since 26-pin single-row headers on each side is getting a bit too long, but in this case you only would need two 20-pin single-row headers on each side to support the "straddle" approach.
Edit: Correction, EKx-LM3S6965 and STM32-H103 would only need two 26-pin single-row headers, current config is dual 2x13, one on each side.
6
Open Bench Logic Sniffer / Re: Si570 clock generator
In terms of things that increase board cost, I think moving to an XC3S500E for increased sample depth would provide much greater "bang for the buck".
7
General discussion / Re: Getting started in the ARM world
I am a big linux noob, so if you can explain it to me I would love to year, learning is always a good thing.[/quote]
You don't. The tutorials on building your own toolchain are for those who do not wish to use the Codesourcery releases for whatever reason (unsupported distro, wanting to learn more about the toolchain, concerns about possible limitations in the Codesourcery tools.
For example, the CodeSourcery page claims that the free "lite" edition does not include JTAG debug support - many of the tools above cover building OpenOCD and GDB. CodeSourcery Lite has all you need to compile, but apparently not to debug.
8
General discussion / Re: Getting started in the ARM world
In this script:
http://www.fun-tech.se/stm32/gcc/index.php
Cam I change the gcc and Newlib to the most recent versions just changing its file names/download places and it will work or there is something more under the hood that will break?[/quote]
There are two things I did differently from that script (other than execute each step manually):
1) Update to the latest versions of everything. This eliminates the need for the patch against binutils. However I believe there might be a small bug in the latest version of binutils that affects you if you're trying to do particularly fancy stuff (Specifically something to do with an instruction needed by FreeRTOS, which is the only example I have failed to compile with that toolchain)
2) Replace all instances of STM32/STM with ARM - the toolchain works just as well on other ARM CPUs like the Luminary Stellaris series
Note: There is the possibility something released between late December when I did my toolchain build and today that might break something. However many of my packages were newer versions than those listed in the fun-tech script.
9
Open Bench Logic Sniffer / Re: American distributer of Open Logic sniffer is needed...
I'm no expert on marketing, but retailers hate to start out with a price that is higher than existing retailers. That's why you see musical artists who are contractually obligated to sell their own CD for the same price as retailers, or online retailers who list prices that are "too low to advertise." Retailers like to know, from the get-go, that they won't be undercut by some easy-to-find competitor.
[/quote]
I think the DP products might be an exception, due to the general stigma of ordering from a foreign vendor.
If the "cheap vendor" were a US domestic company, no one would want to sell at a higher price. However, any US retailer that focuses even remotely on customer service (and most of the people who would carry DP products such as Adafruit and SparkFun do) knows that they can compete with a foreign vendor on more than just price.
10
Open Bench Logic Sniffer / Re: American distributer of Open Logic sniffer is needed...
Domestic support
Rapid domestic shipping
Those two would be worth, to me, on the order of $10-15 for the OLS.
e.g. I would probably happily pay Sparkfun or Adafruit $60 for an OLS rather than <$45 to Seeed (assuming you were compensated in the same manner Ian.)
It's the same reason Saelig could sell Rigol DS1052Es for $500+ when the eBay sellers were well under $400. (The price delta on that particular item has gone down, and I think the eBay price has gone up.)
Semi-related: For those of us who are willing to pay for an XC3S500E instead of a 250E (more blockram for sample depth), is there a possibility of that option and matching bitstreams?
11
Project development, ideas, and suggestions / Re: Seeed Studio Adjustable Power Supply
I'm actually doing something like that myself, although in my case with the goal of replacing the 7805 on my Boarduino with something more efficient. I gutted a cheap Walmart USB car charger that had a bad USB connector (I accidentally let the cable dangle in a puddle on my car's floor mat during road salt season...) and am waiting for some Molex 16-02-0115 ( http://www.mouser.com/Search/ProductDet ... 16-02-0115 ) so I can add pins that allow it to be plugged right into a breadboard.
12
Open Bench Logic Sniffer / Re: Ubuntu Installation Progress
I've got the client working on 10.04 (Linux laptop-danny 2.6.32-22-generic #36-Ubuntu SMP Thu Jun 3 19:31:57 UTC 2010 x86_64 GNU/Linux).
steps i took:
Code: [Select]
apt-get install libftdi-dev default-jreThe make gave a bunch of warnings which i did not investigate (yet) and running the jar file directly does not work. Starting the Loader class directly does work however. I'll try some sampling later on but with the device recognized i'm sure that should be ok.
apt-get install librxtx-java
cvs -z3 -d:pserver:anonymous@jlac.cvs.sourceforge.net:2401/cvsroot/jlac co -P client
cd client
export CLASSPATH=$CLASSPATH:/usr/share/java/RXTXcomm.jar
make
java org/sump/analyzer/Loader
Regards, Danny.
[/quote]
Thanks Danny!
I didn't get the warnings, but was able to get an analyzer to run using your instructions.
However, it seems this version doesn't have some of the features of the newer SUMP clients that come with the OLS distributions. (Why the fork?) Unfortunately, both the 2.04 and 2.1 distributions give the following when I try to execute:
java -jar analyzer.jar
"Error while invoking application: null"
Not too informative... "Write once, run anywhere" my ***...
13
Open Bench Logic Sniffer / Re: Other uses for the SUMP pump
You could use it as a general development board. It will run any BIN you upload to the ROM chip over the USB connection.
Don;t hold you breath for the SRAM board. v2 with SRAM is a big design commitment that will probably take multiple revisions, and we're not ready with the first version yet. I hope the first board will be available next month (Feb 10), another revision... 6months, a year? Depends on how it goes. No idea of the price range, but our goal is always cheap.
[/quote]
As an interim, what about a version $6-10 more using the XC3S500E instead of the 250 for extra BRAM?
It appears the biggest problem with this is that 500Es in the VQ100 package seem very difficult to find...
14
Open Bench Logic Sniffer / Re: Digital Sampling Oscilloscope attachment (DSO wing)
Lets try for a more precise requirements list.
100 MSPS A/D converter ( I am not sure that sample rate is achievable without problems )
20MHz Analog frontend ( This is to high for the sample rate above, but you can at least check if the usual microcontroller oscillators swing).
1 to 20V input stage ( in 5 or more steps )
As Luke mentioned the buffer depth is not great but I can remember fancy Tektronix scopes with less buffer memory. I also think in Luke's
calculation is off. I get 120us at 50MSPS until the buffer is full.
Klaus Leiss
[/quote]
The general rule of thumb for scopes is samplerate = 10x analog bandwidth, but people have gotten by with 5x. You just have to be more careful about filtering and interpolation.
IF it is possible to drive the buffered inputs of the new LA design with an AD9288 it's possible to achieve 200 MSPS (100 per DAC in the 9288 with clocks 180 degrees out of phase, two 8-bit parallel outputs) - some of the remaining 16 GPIOs would have to be used for control and clock.
Problem is that from what I see, to maintain the standard Gadget Factory "wing" form factor, you are limited to a 16-bit wing - The 32 bits of I/O (actually 16 of I and 16 of I/O) are broken out to two separate wing headers, somewhat different from the original gadgetfactory design that permitted a single 32-bit wing.
In addition to the 16 bits for the data, you'd need more lines for control of the analog frontend and ADC clocks, so at least something along the line of 20-24 signal lines would be needed. (Two ADC clocks, and you might be able to control the frontend using I2C, allowing you to work with 20 signal lines. However once you go past 16, the next logical steps are 24 or 32.)
It appears that most of the faster ADCs (some can do 250 in a single ADC) need more I/O per single ADC - One of the 250 MSPS units from ADI interleaves data between two 8-bit outputs, and the other has 8-bit LVDS (16 lines).
A Spartan-3E in a QFP144 or QFP208 would have provided more flexibility for a DSO - either would have allowed for at least four interleaved AD9288s, coming up to 800 MSPS.
15
Open Bench Logic Sniffer / Re: Digital Sampling Oscilloscope attachment (DSO wing)
However, with a 16-bit wing, you'll be able to at best achieve 200 MSPS on a single channel with the -100 parts.
Unfortunately, since the current "sump pump" design has the voltage buffer for the other 16 channels permanently installed, you may not be able to use that 16-bit path for another ADC. Two AD9288s would get you to 400 MSPS.