LinkMeter v2 Homebrew BBQ Controller - Part 1


 
Status
Not open for further replies.
Originally posted by E. Fox:
To add to my predicament; I am unable to set a root password so I am attempting to tftp the new bin and reflash the router. However no matter how many times I try (using a switch between computer and router) I cannot get the router on 192.168.1.1. I have tried catching it during the bootwait and also setting the router in failsafe mode and no luck. The UI loads and I can see probe temps etc. but I cannot set the password or access any of the linkmeter configuration items.
Click to expand...
You can get the router into failsafe but not connect to it? There's no TFTP in the failsafe mode so I'm not sure if you're tying to upload an image there. Also remember the failsafe IP is 192.168.200.1. If you can get in you can erase the root password or the whole configuration area or flash a TRX using the sysupgrade command.

The LCD garbled issue is something that happens sometimes. Some builds are more susceptible than others, and even with an oscilloscope I can't see any reason why it is happening. It happens so rarely (once every month maybe in continuous running) to me that I can't debug it. The v3.2 board changes the connection method to hopefully that makes it better, not that that helps you. Maybe check your connections to the decoupling capacitor C2, swap out the AND diode D2 for another 1N4001 or a 1N914/1N4148.
 
Originally posted by Bryan Mayland:
It looks fine to me, other than C2 (the 3.3V power cap) not having a value. Something you may want to mull over because I'm not sure about it, does the ATmega need a decoupling capacitor? I've heard that it has an internal 0.1uF capacitor on AVCC but I've never seen anything on the datasheet about it. The v3.1/v3.2 boards don't really have one but they do have a "spare" 0.1uF capacitor by the power section which may be acting somewhat as a decoupling capacitor.

I don't know what else you'd do for A0. I mean fixing it in code means custom compiled firmware, which becomes a pain to maintain. You could do a solder jumper to allow the trace to be cut more easily but really I don't think it matters because these boards are so easy to make, it isn't terribly hard to just make something custom.

I don't know what the lid toggle would hang. That would just mean a new status update has never been received after you click on it? It could take up to 10 seconds to reflect the new state if your browser is polling.
Click to expand...

Okay so I am done with the board design on the MiniLink v1.1. I think she is ready for the board house. I estimate that the cost per board will be about $4.50 which is less than half of the cost of the original Linkmeter v3.1. You do lose the LCD and button but the software that Bryan has designed is so good you don't really need it.

Bryan, I read about decoupling capacitors and it seems it was a good idea to add one so I did. I matched all part numbers with your part numbers so it should be easy to assemble using the existing parts list.

Mind checking it one more time? MiniLink v1.1 (2)
 
Originally posted by Dave S (GeoDave):
Mind checking it one more time? MiniLink v1.1 (2)
Click to expand...
Sweet! I like the new placement of the resonator closer to the destination pins. I never know how far away one of those can be.

Since you've now routed it, I'll say that 6mil traces with 6mil spacing is the absolute minimum for dorkbot and I'd say you'd want to be larger than that by a good margin because the boards aren't always perfect. I use 12mil traces, 15mil for 5V/3.3V, and 20mil for GND/12V with I think 10mil spacing. I know your board is pretty tight though so it might be impossible with those dimensions.
 
I ended up JTAGing the router and TFTPing the latest snapshot. Everything is working perfectly now. Thanks for all the help.
 
Brian, I've got the garbled LCD issue as well. To test the LCD, I loaded the sketch you posted previously. It works except for one slight deviation, it's displaying a "D" instead of an "A" on the LCD. How's that possible?...

#include <ShiftRegLCD.h>
ShiftRegLCD lcd(8, 4, TWO_WIRE, 2);

const unsigned char lcdBacklight = 5;

void setup() {
pinMode(lcdBacklight, OUTPUT);
analogWrite(lcdBacklight, 255);
lcd.clear();
}

void loop() {
lcd.print("A");
delay(5000);
}
 
Originally posted by Bryan Mayland:
Sweet! I like the new placement of the resonator closer to the destination pins. I never know how far away one of those can be.

Since you've now routed it, I'll say that 6mil traces with 6mil spacing is the absolute minimum for dorkbot and I'd say you'd want to be larger than that by a good margin because the boards aren't always perfect. I use 12mil traces, 15mil for 5V/3.3V, and 20mil for GND/12V with I think 10mil spacing. I know your board is pretty tight though so it might be impossible with those dimensions.
Click to expand...

I made a couple more changes and I read more about routing, nets, and the like. I ended up using 10mil traces, 13mil for 5V/3.3V, and 18mil for GND/12V with 10mil spacing.

Here she is: MiniLink v1.1 (3)

We are still at less than $5 per board if you order through DorkbotPDX. Should be call her good and send her to Dorkbot?
 
If anyone hasn't noticed yet, the wiki has been updated with the v3.2 PCB parts list, schematics, and the rest. There's also a 1"x1" button board, which is designed to be easy to use more than tiny and looks something like this


Except the corners are rounded and it says UP on the top in the actual version. The photo shows a round button cap instead of the square I keep pushing because I wanted to try it out and it stuck on and didn't want to come off. The round button has the benefit that you don't have to cut the stem out of the cap and it is probably easier to cut a hole for.
 
Originally posted by Joe Loyd:
Brian, I've got the garbled LCD issue as well. To test the LCD, I loaded the sketch you posted previously. It works except for one slight deviation, it's displaying a "D" instead of an "A" on the LCD. How's that possible?...
Click to expand...
Is it possible you have your D4 and D6 reversed? I'm fairly certain that it wouldn't display anything at all if you did, but 'A' is 100 001 and 'D' is 100 100. As with all these LCD issues I just don't know what to do because it is so random. It is unusual that yours prints D consistently instead of A though. Are your D0-D3 connected? Have you tried with them disconnected?
 
Originally posted by Dave S (GeoDave):
We are still at less than $5 per board if you order through DorkbotPDX. Should be call her good and send her to Dorkbot?
Click to expand...
The new routing looks great. Really good job there. The only issues I see are silkscreen. Only major one which really isn't is that the A0 label on the "extra" header got rotated (probably by mistake) so it is on the pad. Minor silkscreen issues, a majority of the pin labels are going to be cut off, like D6, and 5V and RST on the FTDI connector.

The magical word that EAGLE calls soldermask is "Stop". Turn on tStop and bStop and you can see where there won't be soldermask around the pads. If there's no soldermask the silk doesn't stick. I'd just set your grid to 0.01 and pull them all away from the pads where you have space.

I mean those are super-minor things, it can go as-is because it is just some cut off labels. I always get bit in the *** by the silk.
 
Is it worth upgrading to the latest schematic over the last? I got the previous version working , however if there are major advantages I think I will rebuild.

Happen to have any extra button board PCBs?

Thanks
 
Originally posted by Bryan Mayland:
The new routing looks great. Really good job there. The only issues I see are silkscreen. Only major one which really isn't is that the A0 label on the "extra" header got rotated (probably by mistake) so it is on the pad. Minor silkscreen issues, a majority of the pin labels are going to be cut off, like D6, and 5V and RST on the FTDI connector.

The magical word that EAGLE calls soldermask is "Stop". Turn on tStop and bStop and you can see where there won't be soldermask around the pads. If there's no soldermask the silk doesn't stick. I'd just set your grid to 0.01 and pull them all away from the pads where you have space.

I mean those are super-minor things, it can go as-is because it is just some cut off labels. I always get bit in the *** by the silk.
Click to expand...

Thanks for the help Bryan...it was a lot of fun to learn Eagle! I noticed the A0 silkscreen issue this morning so I did fix it before I sent it to Dorkbot.

I will fix the other silkscreen issues which I agree will probably come up when I get the boards. The silkscreen is a pain!

I have been testing the MiniLink v1.0 today and it works perfectly. I think v1.0 was all 6mil on those so they did a pretty good job with the boards. I am glad you told me the traces for certain nets should be larger. I feel more comfortable with the larger traces after reading a bit online.

The Linkmeter interface has gotten better yet again! I am glad that you switched to a popup window when you commit changes to the config.
 
Originally posted by Bryan Mayland:
...
Is it possible you have your D4 and D6 reversed? I'm fairly certain that it wouldn't display anything at all if you did, but 'A' is 100 001 and 'D' is 100 100. As with all these LCD issues I just don't know what to do because it is so random. It is unusual that yours prints D consistently instead of A though. Are your D0-D3 connected? Have you tried with them disconnected?
Click to expand...

No, D4 and D6 are correct. I disconnected D0-3 with no change, sill all "D"s....
 
When I had my LCD issues I was getting D instead of A as well. I had to swap out to the other diode to get it working.

After that I didn't have any issues with that one. The other two I built still have garbled LCD issues periodically and I have to power toggle them to get it cleaned up. I usually get the issue when I plug in a probe, this action causes the LCD to garble sometimes.

dave

Originally posted by Joe Loyd:
<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by Bryan Mayland:
...
Is it possible you have your D4 and D6 reversed? I'm fairly certain that it wouldn't display anything at all if you did, but 'A' is 100 001 and 'D' is 100 100. As with all these LCD issues I just don't know what to do because it is so random. It is unusual that yours prints D consistently instead of A though. Are your D0-D3 connected? Have you tried with them disconnected?
Click to expand...

No, D4 and D6 are correct. I disconnected D0-3 with no change, sill all "D"s.... </div></BLOCKQUOTE>
 
Originally posted by E. Fox:
Is it worth upgrading to the latest schematic over the last? I got the previous version working , however if there are major advantages I think I will rebuild.

Happen to have any extra button board PCBs?
Click to expand...
The only difference between the V3.1 and V3.2 boards is that the LCD is being driven over an SPI latched shift register instead of the previous 2-wire version. I'm hoping folks will find this a lot more reliable than the original version.

EMail me your physical address to bmayland leoninedev.com and I'll send you my spare button board PCB.
 
Originally posted by Dave S (GeoDave):
Not having much luck with these IKEA probes. They are way off using the presets...anyone else having issues?
Click to expand...

They're the only probes I have and till now it worked o.k.
Do you use a 50K fixed resistor with them?
 
Originally posted by Dave S (GeoDave):
I have been testing the MiniLink v1.0 today and it works perfectly. I think v1.0 was all 6mil on those so they did a pretty good job with the boards. I am glad you told me the traces for certain nets should be larger. I feel more comfortable with the larger traces after reading a bit online.

The Linkmeter interface has gotten better yet again! I am glad that you switched to a popup window when you commit changes to the config.
Click to expand...
Yeah EAGLE is a giant *** when you first start using it and it feels like the least user friendly program in the universe. Eventually you figure out what all their paradigms are, like how the heck grouping items to move them works and what an alt grid is and where the hotkeys are, and you can do things pretty quickly. It doesn't mean it is a very well designed app at all from modern perspective, but it does do a good job once you learn it.

I'm going to be switching all the pages that take you to a blank text page to the popups at some point. I didn't want to drag in yet another javascript include to do such simple dialogs, but I may end up going that direction. I also want the config dialog to close faster, but it has to stay open for at least 2 seconds now because it is hoping that by the time it closes the config will have been updated (in linkmeterd so it can be sent back to the config page). So many things to work on!
 
Originally posted by Dave S (GeoDave):
Not having much luck with these IKEA probes. They are way off using the presets...anyone else having issues?
Click to expand...
My IKEAs are working OK at room temperature where they read right on for ~75F but I haven't tested them at grill temperature yet. I kinda want to write a sketch to have it do the "LUA implementation of Levenberg–Marquardt algorithm for a 2-pass (rough estimate, refine) multi-iteration determination off coeffs" from the TODO list.

EDIT: I use the standard 10k resistors.
 
Bryan: can you explain me the difference between using 10k and 50k for the probes? As far as I understood, the fixed resistor should be around the same as the resistor of the probe at 25 degrees C. so that's why I used the 50k, but actually I don't know the influence..
 
Originally posted by Bryan Mayland:
<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by E. Fox:
Is it worth upgrading to the latest schematic over the last? I got the previous version working , however if there are major advantages I think I will rebuild.

Happen to have any extra button board PCBs?
Click to expand...
Email sent, really appreciate it.
The only difference between the V3.1 and V3.2 boards is that the LCD is being driven over an SPI latched shift register instead of the previous 2-wire version. I'm hoping folks will find this a lot more reliable than the original version.

EMail me your physical address to bmayland leoninedev.com and I'll send you my spare button board PCB. </div></BLOCKQUOTE>
 
Status
Not open for further replies.

 

Back
Top