![]() |
MAJOR IR PROBLEMS RESOLVED
Training method has been revised
I spent part of the afternoon at Microchip in Chandler with a couple super FAE's resurrecting our IR board. It was like being at the Mayo Clinic for uControllers. We had had very little sucess training our board. We thought the code may be corrupted and figured we needed to reprogram. We were likely wrong on this point. TEam 39 had tried the WinPIC freeware and the serial cable with resistors scheme and managed to erase the chip and then successfully program all but 3 memory locations correctly (many times). We basically rendered the chip totally non responsive. This serial cable bit banging programmer scheme may have worked for some teams with success but among other things, it depends a lot on the RS232 voltages from the PC being "just right". These voltages vary from 5 to 12V from Lap top to Desktops and everywhere in between. At Microchip We used MPLAB IDE and an IDC2 type programmer that is part of the PICkit II progrmming kit. Very easy to do. Nice because it uses the familiar MPlab. MPlab will show you the assembly languge code that correspondes to the hex code for the IR. You could tweak the code if you like!!! http://www.newark.com/jsp/search/pro...FC-GB100000001 After reprogramming, we found that the IR board perfomed just as unreliably as it did prior to the first reprogramming with the freeware. We could not get through the training of all 4 LED's without an error. After much trial and error we discovered that it trains real well if you hold down the key on the remote untill the 3rd flash of the error light. Wow , what a difference. I wish we knew this little detail at the start of all of this. REVISION: it turns out that this taining method is still in error. Our universal remote had lost it's identity. Yes it trained, but likely the frequency was incorrect. When universal remote was reprogrammed, it programmed without any error lights. Every key still needs to be pressed twice. If you are getting the error light to blink on power up but have not been able to train, I suggest that you try the "realase on 3rd blink" method above. The error light blinks at power up because the chip is working and the code is executing. FYI, we used an RCA RCR4558 universal remote ($10 from Walmart) and programmed with code 1001 which is a Sony remote. Don't throw away the paper with the code listings on it!!!!! Also note that this remote will loose its identity if you remove the batteries and press a key. The identity is held by a supercap (not FLASH). Not a big deal if you do not remove bateries. It is super easy to reprogram by holding down the TV key and putting in 4 digit code. http://www.ambientweather.com/ptrcarcr4258.html As some background, when we were still in the desparate mode, we had built an IR detector (phototransitor) and actually verified that the IR remote we were using was producing 38kHz pulses. With a box full of remotes collected from households, only a few were 38 kHz. Note that 36 and 40kHz remotes will train the unit but range will be limmited. Several web sites list the frequencies. Sony is mostly 38 kHz but there are exceptions. Universal remotes are your best bet. Good resource here: http://winlirc.sourceforge.net/ Alternate IR's are mostly DIY circuit diagrams. In my searching for over an hour, I found no USA source. Note that you can replace the PIC chip if you are careful. Use tweezers and a heat gun if you do not have a surface mount rework station. Have the heat gun dead center and directly over the part at ~ 1 inch. Use low fan speed. Use a directing nozzle if you have it. It will only take a few seconds. Kill the heat as soon as the chip is free. Let it cool for a couple minutes if you have made an usucessfull attempt and then retry. Use one person to hold the gun and another to pull the chip . Hold the board in a vise. If that heat gun is waving around while heating, you are not being precise enough and will likley enjoy the outcome. REPLACEMENT: It is not a fine pitch part so a thin tip soldering iron can be used to solder in the new part. Lay the flat part of the iron on the row of leads and slide it down the line. You will NOT likely need to add solder. I recommend just using the residue on the pads. Heat gun will work too .... but do so only if you have done this before. You could melt the part. It is a 2 person operation. Get the PIC16LS87 chip here for $5. Get one to go and one to blow (..or melt). http://parts.digikey.com/1/parts/561...lf87-i-so.html The L means that it can operate 2.5 to 5VDC. This makes it more brown out resistant...but not immune. We did turn on the brown out protection bit before reprogramming. This will cause the processor to halt and preven errant execution or a freeze if the voltage drops too much. It will reset/restart when voltage is back within spec. What a deal ... 5 Mips for $5 !!! Team 39 will bring a programmer to AZ and Vegas in case a team wants to reprogram their chip. We are in Gilbert if you want to do so before the regionals. This ends a frustrating chapter for Team 39. Bottom line is that if the error light still blinks once on power up, you have not likely fried or even corrupted the IR board. Each button needs to be pressed TWICE. You should not get the error light when training. Try multiple remotes if you do not have a universal remote. If you have a universal remote, set it to emulate a remote that is 38 kHz. |
Re: MAJOR IR PROBLEMS RESOLVED
Are you in violation of the Fix-it-window?
Just asking as this is somewhat of a grey area. |
Re: MAJOR IR PROBLEMS RESOLVED
We were just trying to figure out how to fix it.
We will do everything from scratch during the week1 fix it window again. |
Re: MAJOR IR PROBLEMS RESOLVED
Quote:
Is there a difference between learning for yourself and going open source (public) with that knowledge as opposed to finding out for your team and keeping it private (or even sharing it)??? . |
Re: MAJOR IR PROBLEMS RESOLVED
Quote:
Greyer yet, is that I figured I had at least 8:30 left to program autonomous modes. But that requires a lot of coffee, and I ran out of FloMax two days ago. I didn't know whether I could close the window each time, or let the time keep going and going too. :confused: |
| All times are GMT -5. The time now is 21:47. |
Powered by vBulletin® Version 3.6.4
Copyright ©2000 - 2017, Jelsoft Enterprises Ltd.
Copyright © Chief Delphi