Saturday, March 27, 2004
ConnectMe Life Support Board
An open-source schematic for a minimal ConnectMe life support board is posted on my website:
http://www.gluelogix.com/ConnectWiMeRS232.shtml
Scroll to the bottom.
Larry Martin
www.GlueLogix.com
Copyright (c) 2004 Larry Martin. All Rights reserved.
http://www.gluelogix.com/ConnectWiMeRS232.shtml
Scroll to the bottom.
Larry Martin
www.GlueLogix.com
Copyright (c) 2004 Larry Martin. All Rights reserved.
Monday, March 01, 2004
First Encounter
My First Encounter with the Digi ConnectMe came in Feb2004 when my employer found a customer who refused to consider our equipment unless it had an Ethernet jack. Bit of background: the last five years of my life have been spent with an RF startup. We got the baseband interfaces defined and implemented early on, giving our early customers something stable to work with. Then we concentrated on the RF for a few years. The problem was, the baseband world changed on us. In 1999, "internet devices" were hokey dialup things like the ATM at your local squishy-mart. At that time, it was still reasonable to field a device with an RS232 interface. Even in multidrop networks, it was easier at that time to get electricians to pull RS485 than Ethernet. No more. Now that 10BaseT and 100BaseT and God-knows-whats-next has taken over, you can't compete unless you have Ethernet too.
So we got Ethernet. It took around two weeks.
We had already noticed the ConnectMe, and were waiting for an excuse to use it in something (there was one other small Ethernet-to-serial filter device available at the time, but it was based on the 80186 and had a fraction of the ConnectMe's memory).
When the customer delivered their ultimatum, there was actually plenty of time to do a deliberate development. But by the time management took the ultimatum seriously (i.e. it became a crisis) we were around 3 weeks away from a make-or-break demo. As it turned out, we were able to put together a couple of prototypes with days to spare.
Since there was no longer time to respin our main circuit card with the ConnectMe on board, I designed a very simple and small circuit card that could be wedged into an underused corner of the enclosure. All it really needed was an LM-series voltage converter, the ConnectMe itself, and a place to land jumper wires. Our tech cut the square hole in the case and ran jumpers from the ConnectMe to the main board. Plug in, turn on, and fugedaboudit.
But even at that there were some problems.
Our first ConnectMe that was actually soldered into a circuit communicated very intermittently. It sometimes took minutes to boot up, sometimes never booted up at all. Its brother, on the JTAG development board, never had a bit of trouble.
The problem turned out to be my reset circuit. I had put a couple of bells on the interface board, like 3.3 to 5 volt translation and a nice RC reset circuit. Turned out, the ConnectMe has an internal RC network on the Reset pin. They expected people to implement Reset with a simple switch to ground, no debounce, etc., but as far as I can tell, that fact is undocumented in the Hardware Reference. In any case, when we removed my RC network, we fixed the problem. The Ethernet prototypes shipped and the customer checked off the Ethernet square on their acceptance test.
Two weeks calendar time, probably a man-month spread between me, another software guy and two techs, plus the cost of creating the interface board. One guy on the team said he told a friend what we'd pulled off. The friend, probably a big-company expert on these matters, said "That's impossible." Sorry.
Larry Martin
www.GlueLogix.com
Copyright (c) 2004 Larry Martin. All Rights reserved.
So we got Ethernet. It took around two weeks.
We had already noticed the ConnectMe, and were waiting for an excuse to use it in something (there was one other small Ethernet-to-serial filter device available at the time, but it was based on the 80186 and had a fraction of the ConnectMe's memory).
When the customer delivered their ultimatum, there was actually plenty of time to do a deliberate development. But by the time management took the ultimatum seriously (i.e. it became a crisis) we were around 3 weeks away from a make-or-break demo. As it turned out, we were able to put together a couple of prototypes with days to spare.
Since there was no longer time to respin our main circuit card with the ConnectMe on board, I designed a very simple and small circuit card that could be wedged into an underused corner of the enclosure. All it really needed was an LM-series voltage converter, the ConnectMe itself, and a place to land jumper wires. Our tech cut the square hole in the case and ran jumpers from the ConnectMe to the main board. Plug in, turn on, and fugedaboudit.
But even at that there were some problems.
Our first ConnectMe that was actually soldered into a circuit communicated very intermittently. It sometimes took minutes to boot up, sometimes never booted up at all. Its brother, on the JTAG development board, never had a bit of trouble.
The problem turned out to be my reset circuit. I had put a couple of bells on the interface board, like 3.3 to 5 volt translation and a nice RC reset circuit. Turned out, the ConnectMe has an internal RC network on the Reset pin. They expected people to implement Reset with a simple switch to ground, no debounce, etc., but as far as I can tell, that fact is undocumented in the Hardware Reference. In any case, when we removed my RC network, we fixed the problem. The Ethernet prototypes shipped and the customer checked off the Ethernet square on their acceptance test.
Two weeks calendar time, probably a man-month spread between me, another software guy and two techs, plus the cost of creating the interface board. One guy on the team said he told a friend what we'd pulled off. The friend, probably a big-company expert on these matters, said "That's impossible." Sorry.
Larry Martin
www.GlueLogix.com
Copyright (c) 2004 Larry Martin. All Rights reserved.
