From: Kim H. <ki...@ki...> - 2004-08-04 23:33:33
|
I'm having some problems with my gumstix that I'm not sure are related to the ram refresh rate issues. Maybe someone has some ideas. Here's the sequence of events that I've followed: * First I blew up one gumstick. Completely and utterly. I had the voltages mixed up on the pinout. * Then I took my waysmall apart and followed the traces to get the layout correct. I made a small breakout board that had an SMT pressure connector on it to connect to my little circuit that contains some MAX 3233 chips on it. I took the stick out of the waysmall at that stage to use with my new breakout board. That worked fine. It would talk to both rs232 ports. * It was time to do some work with the thing, so I put it back into the waysmall so I could use both serial ports and work via a USB network connection. At that stage, I started receiving double echos via the terminal emulator. I figured this was transmit and receive lines cross on the pressure connector. Jiggling it didn't stop this and after a little while it wouldn't take any input at all anymore on the original waysmall. The stick still worked on my little chassis however. * Time to upgrade the uboot. First I wanted to simply upgrade the linux as it seemed less ominous. I was able to flash the linux, but the waysmall didn't boot. Craig informed me that this was due to changed environment variables in the eeprom. I changed these and it booted successfully. However, I only booted it about twice, then it failed to come up to even a uboot stage. Now, I still haven't upgraded the uboot yet, so it's conceivable that it is the ram refresh rate bug kicking in I suppose. I guess it needs this to be correct to even boot the uboot, surely. It's not using non-volatile ram as part of the uboot after all right? * Time to make a jtag cable. I made a circuit based on this one: http://home.comcast.net/~staskh/projects/tools/ and soldered this onto the waysmall's jtag connector. But no go. Nothing. The ocd commander software stated that the cable was disconnected. Well, the last time I used the waysmall, I had some problems with the pressure connector, so maybe this was the cause. * So I made another chassis computer, with some max3233's for the rs232 ports, an ldo regulator so make sure that the 74HC244's would provide 3.3V outputs. Another new pressure connector and a 14 pin connector that would take a ribbon cable connection to the jtag board (But providing the same signals to the same places as the 20 pin connections). Still no go. * I then received a new (Old-style) gumstick. I plugged this into my own original computer board and I saw it booting linux, so the stick works. I turned off the power and on again, this time it only booted to the uboot prompt. Off and on again and nothing. I never saw it boot again, no matter how much pressure connector jiggling I did. * I tried this new gumstix in the waysmall chassis. Also no go. * At this point I'm thinking that the pressure connectors must get out of alignment after a certain amount of use or abuse and it's not possible to see the connections properly on the inside when the board is in place. So I took the second gumstix that died during the flash experments and soldered a pressure connector onto the stix with a hot air rework iron. I then soldered little wires onto this and connected soldered them onto the similar pressure connector on my new harness that also contain the jtag connectors. Nothing... So right now I'm at a loss. What are the possibilities: 1. Maybe I'm blowing the stixs up with ESD? But I doubt this. I don't have an ESD clean room. There is some carpet next to me. But I'm not wearing sneakers (In which case, I would generate ESD) and I've never actually blown anything up with ESD before. 2. Maybe when the pressure connector jiggles and cross adjacent pins it damages the stix? I think this one is a definate possibility, because apart from the stix that died during the flash experiments, all of the others do experience a little bit of pressure connector shorting. However, I can't see how it would cause problems, as the VCC and GND are on opposite sides of the connection and I don't see the other lines having a problem with some voltage on them. But maybe they do? A fully charged li-poly is about 4.2V and one chap earlier stated that the jtag pins on the xscale chip can take up to 4V inputs. If vcc slipped, it would also not have any current limiting there either. However, in a connector slips, vcc would go to rts and gnd would go to nreset. The rest would be signal shorts. I doubt that that is a problem. What do you think Gordon? 3. Maybe my stix are not broken but are simply suffering from the ram refresh rate problem. Because I don't have a genuine mcgraigor cable I can't eliminate this as a possibility yet. Though I thought that the circuit that I was using was essentially the same circuit as Benny is using and he *is* able to talk to his stick. Because no amount of pressure sensor jiggling can get a new gumstix working again, I think that they are breaking. I don't want to break any more if that's the case. Does anyone have any ideas? Does it appear to others also that my stix are probably fried, or does the above still leave the possibility of a jammed, but otherwise still functioning gumstix? PS. I do have one new module gumstix available here. But I have only two connectors for them and hence two shots to get them soldered right. I only have a rework hot air gun to do this with. I think there will be two problems. The biggest I think is getting the solder paste at exactly the right thickness to solder correctly. This will be very hard because the pins are so close together that the chance of two much paste is quite high. The second problem is simply seeing the correct placing of the connector before blasting it with air. Once the solder paste is in place, you can't see the connectors clearly anymore. I'm currently also making my own boards. I can make tracks down to 0.1mm no problems, and I do chemically tin place them. But they don't have solder mask on them nor to I have a solder paste template. Does anyone have any tips or suggestions of how to get this right first time or at worst, the second time :) ? - Kim |
From: Kim H. <ki...@ki...> - 2004-08-04 23:54:03
|
Slight correction on "some" of the last mail. My latest new (Old style) gumstix is working now after all. I noticed a wire became unsoldered from the battery :) But... with much of the previous checking, I tested all the voltages carefully. This must have happened after I did all that. I guess that means that I do have the means now to test whether or not a particular home made jtag cable works or not. Except that in the meantime I've trashed my pressure connector on the new jig by soldering to it :( And I lost the tiny screw from the waysmall and kind find any around here of the same size. I could drill out the gumstix board to fit a close match that I have here, but it worries me quite a bit doing this, as it's the only working stix I have at the moment. - Kim |