|
|
Post by jimkernsjr on Aug 25, 2016 5:01:20 GMT
Hello Everyone... I have a very strange problem. Recently, my gateway which has been working very reliably for about 6 months nonstop now started to lock up. MQTT light stays lit, no radio activity light or data through MQTT. Generally unplug it, gave it a little wiggle on the stack of shields, and rebooted and it comes back, This started getting more and more frequent until it's daily. For some reason, I suspected a bad solder joint, because when my heat pump died I also started having this problem! (maybe just a coincidence.) A quick note of the setup: A Buono Uno R3 switched to 3v3, Wiznet 5100 ethernet shield, and a protoshield, with the RMF69 mounted on top (which I suspected I did a bad solder on). The RFM is also on a small OSHpark breakout pressed on the shield. I have identical spares of all of these parts. So, after a few lockups, I remove it from its aesthetically pleasing case  and let it run exposed until it locks up and start probing with my voltmeter or oscilloscope, which didnt take long anymore. I checked continuity from the radio all the way down through to the uno - I find no problems. I decide to check voltages just for so, and low and behold, I find 4.8 volts on the 3v3 lines! Figuring surely this is pissing off the radio, I start to figure where this is coming from. -Removed my RFM Protoshield - still a problem with UNO+Ethernet together only -UNO itself - no problem. So I notice the Ethernet shield has an 3v3 regulator...I assume it's to make it's own 3v3 voltage. So I decided to get crafty - I bent the 3v3 pin of the ethernet shield in 45 degrees so it does not drop down in the header when together. Put it all togther, started it up, and it works, good. So I check voltages again. -Ethernet shield - 3.3 on the 3v3 line -UNO - up to 5V now on the 3v3 line! Why is this happening? I tried it with two different ethernet shields, same voltage problems. New tests were done on the new UNO, not the locking up one. Hard to blame the ethernet shield - both have the same problem. Ok, maybe the first UNO has problems now, because now no more lockups with the new one. However, with 5v on the 3v3 line, I suspect I'm going to have problems again soon with the replacement Buono UNO. I also decided to throw an MCP1703 3v3 regulator on my proto shield to take the radio loose from the 3v3 line. (Not just for a troubleshooting step, because of a preference anyway - the 5v line has more power and also Sainsmart makes a switchable UNO, without a beefed up 3v3 line, so I could put it on this as well). So right now I am running it again, no lockups, my RFM is on it's own voltage regulator, and the ethernet 3v3 pin is seperated from the Buono. But like I said, I dont thing it's going to be good for the UNO long term. Any ideas? Thanks to all! |
|
|
|
Post by jimkernsjr on Aug 25, 2016 5:12:46 GMT
|
|
|
|
Post by jimkernsjr on Aug 26, 2016 1:03:23 GMT
Hi All...
I have an update. Upon looking tonight, The IOREF is not carried up on my ethernet shields, either one. I also came home to find it locked up again. (New UNO, Old Ethernet, Original Radio). I switched the ethernet controller this time, bending in 3v3 to protect the UNO. I'll see what happens now.
Any one has any ideas I would appreciate. If you have the same setup, can you check your 3v3 pin and see what up? I cant believe no one else has experienced this. If anyone else doesn't have this problem with the same setup, what ethernet controller are you using? Mine has no marking, but I only find one type controller with the Wiznet 5100....
Thanks...
|
|
|
|
Post by paulzy on Aug 26, 2016 5:20:24 GMT
I had this same issue and after nothing working i went over my board thoroughly and found the issue, turns out a design fail caused the filtering caps on the regulator got connected but was not clearly visible without removing the caps, once i done so and reworked the board i have not had any issues.  If your board looks the same in the image, the two caps circled in black are the ones and the red arrow is where the short was on my board. ![]() Hope this helps |
|
|
|
Post by jimkernsjr on Aug 29, 2016 1:48:28 GMT
Hi Paulzy... Thanks alot for your reply and suggestion. However, my 2 shields look different(attached). I see the line going between the two in yours, but I dont see such a line on mine and I'm not so sure what to check. Both caps seem to be sitting on a plastic insulating plate. I check the continuity and came up with the following results: -left cap seems to be input, ground on the pin closes reset, 5v closest ethernet jack. -regulator - left pin ground, center pin output 3v3, right pin input 5v, and 3v3 is the same on that large metallic shield on the front side. -right cap seems to be output, with ground on pin closest reset, and 3v3 on the output. I'm really not sure what to check other than that, but I suspected maybe the regulator went bad or always was bad? But my second controller I just opened, new out of the sealed package and it has the same problems. Troubleshooting things like that is not my strong suit, so if you have any other suggestions I'd be open. Meanwhile I ordered 2 new Ethernet controllers and for the meantime I'll leave the 3v3 pin bent out of the way. The problem with the lockup seems to be that in time it destroyed the radio. A new radio seems to have fixed it. But I'd love to hear your suggestion on how to fix these 2 ethernet controllers...and I'm sure everyone else as well because I'm sure I'm not the only one with this problem as this seems to be a common setup on this board! Best regards, Jim Attachments:
|
|
|
|
Post by papa on Sept 20, 2016 19:23:59 GMT
In a message to me, jimkernsjr wrote:
|
|
|
|
Post by papa on Sept 20, 2016 19:35:18 GMT
Jim, I'll respond to your message here on your thread. Perhaps, we can generate more discussion & wisdom here that would benefit more people.
I had seen your thread, "Gateway Lockups and strange voltage problem on 3v3 line".
Other than read your thread & scratch my head, I didn't do much at the time because ...
1) I've had a rough summer: 5 weeks helping my sisters while our mother was dying & then terrible Gastro-intestinal problems leading up to the hospital & eventually gall bladder surgery 6 weeks ago. I'm a lot better, but still getting my focus & motivation back & trying to catch up on things set aside for the above. Oh, we also flew to Seattle for 2 weeks to see family.
2) I have 3 Ethernet shields on 3 gateways & so far I have not had any problems with them. One was used for 2 or 3 months & is now unused. Another has been running for almost 3 months. (<< of course, these 2 Gateways have run less than 6 months). The 3rd gateway has been running for over a year with no problems.
You have me wondering about my Gateways. When I get a chance, I'll compare my Ethernet Shields to yours & paulzy's & see what voltage readings I get.
Others of you out there in DIY Home Automation land, ?? especially those with Ethernet Shields on top of Buono Unos? are you having Gateway problems like Jim's ?? or NOT?? How long have your gateways been running ??
|
|
|
|
Post by papa on Sept 21, 2016 19:43:26 GMT
OK, jimkernsjr, I've taken a closer look at my Gateways. AND I confirmed the voltage concerns you found, other than gateway lockups (yet). I have had 2 gateways running (without problems so far), but I have disconnected them for the time being. As I said above, I have 3 Ethernet shields on 3 gateways. One was used for 2 or 3 months & has net been used for a while. Its Ethernet Shield is labeled as being from Seeed Studios. Here are the voltage readings I got: Buono Uno, switched to 3.3 volts, + Seeed Studio Ethernet Shield, powered by usb connector (5 volts), running a gateway sketch 3.3 volt socket & ground = 4.64 volts (supposedly too high voltage for RFM69 radio !!) 5 volt socket & ground = 4.64 volts Buono Uno, switched to 5 volts, + Seeed Studio Ethernet Shield, powered by 9 volt adapter in the barrel connector, running a gateway sketch 3.3 volt socket & ground = 3.33 volts 5 volt socket & ground = 4.89 volts ^^ As Jim said, when switched to 5 volt operation, the 3.3 volt socket is truly that. Buono uno alone, switched to 3.3 volts, powered by usb connector 3.3 volt socket & ground = 3.33 volts (As Jim says, Buono alone at 3.3 volt operation gives true reading at 3.33 socket) 5 volt socket & ground = 4.83 volts Buono uno alone, switched to 5 volts, powered by usb connector 3.3 volt socket & ground = 3.33 volts 5 volt socket & ground = 4.81 volts ^^ As far as the 3.33 volt & 5 volt sockets go, there is little difference between switching to 3.3 volt operation or 5 volt operation =============================================================================================== Jim, my other two gateways (the ones in operation until today) use an Ethernet shield very similar to yours (not paulzy's): Concerning these gateways, I did not take voltage readings on the underlying Buono Unos alone, but with the Ethernet Shields inserted into them, the Shields' 3.3 volt & 5 volt sockets gave voltage readings similar to the gateway built with a Seeed Studio Ethernet Shield & to Jim's. That is, with the Buono switched to 3.3 volt mode, the Ethernet Shield's 3.3 volt socket showed 4.8+ volts which should be out of the RFM69 radio's safe range of 1.8 - 3.6 volts. My other concern (that I did not see you mention) is "When a Buono Uno is switched to 3.3 volt mode & an Ethernet Shield is inserted on top, are the Ethernet Shield data ports (D2, etc) operating at 3.3 volts or are they like the Shield's 3.3 volt socket closer to 5 volts?" Again the latter should not be safe for a RFM69 radio. I'll see if I can test those voltages. Does anyone else have similar results? Can others please check? With gateways built on top of switchable Buono Unos? With 3.3 volt Arduino pro minis? You might want to check, especially if you had baffling gateway problems or want to avoid them. |
|
|
|
Post by papa on Sept 21, 2016 20:37:56 GMT
Above, I wrote:
I connected a mulitmeter's COM port to the GND near the Ethernet Shield's D13 socket. I connected the meter's mAVohm port to various Ethernet Shield digital sockets while the gateway sketch was running. The following are the results I got. Am I doing this correctly?
Buono Uno, switched to 3.3 volts, + Seeed Studio Ethernet Shield, powered by usb connector,
running a gateway sketch:
GND near Digital sockets plus
D2 = 0 volts (data line not used until radio connected & communicating?)
D10 = 4.5 volts constant (does not seem safe for RFM69 radio)
D11 = 4.5 volts constant (does not seem safe for RFM69 radio)
D12 = 0 volts & sometimes 3.22 volts
D13 = 2.61 volts on reset
Buono Uno, switched to 5 volts, + Seeed Studio Ethernet Shield, powered by usb connector,
running a gateway sketch:
GND near Digital sockets plus
D2 = 0 volts (data line not used until radio connected & communicating?)
D10 = 4.64 volts constant
D11 = 4.64 volts constant
D12 = 0 volts & sometimes 3.22 volts
D13 = 2.3 volts on reset
^^ !! does not seem much difference between switching Buono to 3.3 volts or 5 volts.
With Buono Uno ALONE, switched to 3.3 volts,
I saw several voltages, but none above 3.3 volts so a NODE made of an RFM69 radio on a Buono Uno should be safe for the radio.
However, concerning a gateway built on a Buono Uno, my tests (if done correctly?) leave me with even more concerns than Jim reported.
By itself, the Buono Uno's voltages read as labeled. However, when the Buono is switched to 3.3 volt & an Ethernet Shield is inserted on top, the 3.3 volt socket & some of the data lines needed by RFM69 are nearer to 5 volts.
So far, I've had no gateway problems even with a Buono / Ethernet Shield gateway that's operated for about a year. However, jimkernsjr reports his Buono gateways went bad & my tests indicate he may be on to something.
Again, could others please check their gateways' voltages. Do you get similar results?
If this is a problem, what can we do about it?
|
|
|
|
Post by papa on Sept 22, 2016 15:00:34 GMT
My key confirmations & findings:
1) With NO Ethernet Shield, a Buono Uno switched to 3.3 volt operation has a 3.3 volt socket that is truly that & no digital sockets show more than 3.3 volts.
Thus an RFM69 radio should be safe when installed on a Buono Uno ALONE that is switched to 3.3 volt operation.
2) When an Ethernet shield is installed on a Buono Uno, the voltages seen at various Ethernet shield sockets are always as if the Buono is switched to 5 volt operation:
That is, even switched to 3.3 volt operation, the Ethernet Shield's 3.3 volt socket & key RFM69 sockets like D8, D10, & D11 are nearer to 5 volts.
Note: According to the data sheet, a RFM69 radio's safe range is 1.8 - 3.6 volts
jimkernsjr wrote:
The Ethernet Shield is used to working with 5 volts so I don't believe it needs protecting.
With an Ethernet Shield (ES) installed on a Buono Uno switched to 3.3 volt operation, running a gateway sketch ...
as jimkernsjr did, I tried bending the Ethernet Shield's 3v3 pin so it does not connect to the Arduino 3v3 socket below.
Hopeful result: The ES's 3.3 volt socket read 3.24 volts & should be safe to connect to the voltage input pin of the RFM69 radio.
That should give the radio SOME protection
However, I got other not so good results: The ES's D10 & D11 sockets still showed around 4.5 volts which seems NOT SAFE for an RFM69 radio.
While a creative attempt, using a voltage regulator on the 5v line to power the RFM69 does not seem necessary because bending the pin already gives a safe 3.24 volts.
Moreover, neither bending the pin nor using the voltage regulator prevent the ES's D10 & D11 sending about 4.5 volts to the RFM69.
|
|
|
|
Post by papa on Sept 22, 2016 16:35:47 GMT
OK, I believe I have solved this so one can build a gateway of a Buono Uno Arduino compatible (switched to 3.3 volt operation) + Ethernet Shield + RFM69 radio that truly operates around 3.3 volts that is safe for the RFM69 radio. jimkernsjr was on the right track to try bending an Ethernet Shield (ES) power pin so it did not connect to the Buono below. He just did not pick quite the right pin. Bending out the ES's 3.3 volt pin does put the ES's 3.3 volt socket at 3.3 volts. However, D8, D10, & D11 still send about 4.5 volts to the RFM69. Rising up from the Buono are a group of 6 pins called ICSP, one pin is VCC (power) Holding the Buono as it's oriented in the picture below, VCC is the upper right pin (& is labeled in the pic) The ICSP pin labeled "1" is MISO, & VCC is pin 2, the pin paired with pin 1.  << click on pic for larger view. On the Buono, somewhat bend out ONLY that ICSP VCC pin so it does not connect with the Ethernet Shield socket above. You can leave the Ethernet Shield's 3.3 volt pin UNbent & connected to the Buono Uno 3.3 volt socket below. You should still be able to sufficiently connect all the other Ethernet Shield pins to the Arduino below. Again, I bent out only the Buono's ICSP VCC pin so it did not connect to the matching Ethernet Shield (ES) socket. I sufficiently inserted all the other ES pins into the Buono Uno switched to 3.3 volt operation. The Buono already had a gateway sketch installed. Power up the Gateway. ES sockets' voltage results: 3.3 volt socket = 3.3 volts 5 volt socket = 5 volts D8, D10, & D11 sockets never go higher than 3.3 volts. Hooray. Success at least with voltage readings. Next step is seeing if the so assembled gateway works. |
|
|
|
Post by papa on Sept 22, 2016 19:15:14 GMT
The following applies only to a Gateway built from a Buono Uno Arduino compatible (switched to 3.3 volt operation) plus Ethernet Shield plus RFM69 radio:
As stated in the last post,
I bent out only the Buono's ICSP VCC pin so it did not connect to the matching Ethernet Shield (ES) socket.
(See the pic in the last post which shows the correct pin to bend.)
I sufficiently inserted all the other ES pins into the Buono Uno which was switched to 3.3 volt operation.
The Buono already had a gateway sketch installed.
Power up the Gateway. ES sockets' voltage results:
3.3 volt socket = 3.3 volts
5 volt socket = 5 volts
D8, D10, & D11 sockets never go higher than 3.3 volts. Hooray. Success with voltage readings.
Final test: Does the so assembled gateway work ?? ...
Yes, Hooray, the Gateway links to MQTT & node data is received & passed on to OpenHAB.
Yet to be seen: 1) Does the bent pin hack cause other problems that negate this fix?
2) Will RFM69 radios develop premature problems & failures if they had been previously used on a Gateway without this hack.
If others find this bent pin hack successful, jimkernsjr, perhaps you can mark this thread "SOLVED."
|
|
|
|
Post by jimkernsjr on Sept 24, 2016 5:16:09 GMT
Hi Papa...
Wow, thanks so much for that. I bent back my 3v3 pin and tried your hack, and it seems ok, according to my voltmeter.
I kind of agree with you, that the multimeter is a pretty good indicator of the digital pin voltage- normally I put a test sketch on and just raise the pins high to check it (also the way I check to see if unmarked Promini's are 3.3 or 5). The way you did, I'm surprised that worked as well as it did-I would have expected the multimeter to do more of an average. My next step was to throw an oscilloscope on it and see what was up, but I didnt get that far, and franky I was surprised there was not much mention of this problem until now. At any rate, you found the voltage out of spec on the digital pins before the fix, and ok afterwards, so to me the net result is fixed. You are also correct - I didn't even check that point, because I never could get past even the supply pin problem.
To your point about it affecting other people, I admit my setup is a bit different than most in that I power it through the barrel jack instead of the USB. (I have it on an end table in the living room, and use a POE injector just to minimize wiring and up the WAF!) The power is 12v with a pretty beefy brick - potentially everyone else that powered it with a cell phone charger type and just an average supply, it was able to dip the voltage down and not cause damage as quickly as my setup. But again, just a guess, I am far from an electrical engineer as you can tell.
As far as damage...I now have 2 Buono Unos, 3 ethernet, and 2 radios that only work together in strange combinations. One radio only starts up attached to one of the buonos, but not with ethernet card 1 and 3... Trying to sort it out is a mess...most likely the radios will end up in the trash, the Buono's for testing and Guini pigs, and I'll just start over. My hope is that everyone else that tries this configuration reads this, and I saved them from the mess I have!
Maybe to be thoughful to others, we should contact Inhaos about the problem with the board also...You take care of, or you want me to?
I will close the thread...let me know via message if you want to contact Inhaos or me.
Thanks again!
|
|
|
|
Post by papa on Sept 24, 2016 14:43:51 GMT
You're welcome, Jim. Thanks for noticing & persisting in asking about the Buono + Ethernet Shield problem. I have added the hack to my Gateway build instructions on the forum.
You wrote, "I was surprised there was not much mention of this problem until now." Apparently the RFM69 can tolerate (for a while) voltage out of the rated range. One of my Gateway RFM69s handled it for over a year.
You wrote, "I power it through the barrel jack instead of the USB." While programming & testing Gateways & nodes, I power them via the computer's USB. When I put them into service, more often than not, I power my Gateways & nodes via the barrel connector. Though some are powered by USB & cell phone chargers. The latter is what computourist recommends.
You wrote, "barrel jack ... The power is 12v with a pretty beefy brick." I believe 12v may be a bit high for the Buono's voltage regulator. From eBay, I get cheap 9 volt power adapters with barrel connectors.
You wrote, "Maybe to be thoughtful to others, we should contact Inhaos about the problem with the board also." I have mixed thoughts about such a notification. When used by itself, the Buono's 3.3v_5v switch truly changes turns the device to (mostly) 3.3 volt operation. & I assume (without testing) that an Ethernet Shield operates has proper voltages when installed on a FIVE volt Arduino compatible. The problem (without the hack) is using the Ethernet Shield with a Buono switched to 3.3 volt mode.
I guess I'll email Inhaos & try to be specific about how they might caution buyers.
Yes, you can close the thread as "SOLVED." We can revisit it if someone finds a serious problem with the hack.
|
|
|
|
Post by jimkernsjr on Sept 25, 2016 2:05:16 GMT
Hi Papa...
Again, thanks for all, especially having a look at it with all you have going on with the family.
However, on your 12v point on the barrel jack, the Data sheet for Buono UNO R3 says:
"6V to 23V width range input"
Standard UNO is max 12v.
Best wishes to you and the family. Hope things are looking up!
Best regards
Jim
|
|
|
|
Post by papa on Sept 25, 2016 13:01:09 GMT
Jim, I stand corrected on the range for the power barrel jack.
Thanks for your good wishes. We are working thru settling our mother's estate.
I am healing & feeling a lot better, more able to concentrate on things like OpenHAB.
|
|
and let it run exposed until it locks up and start probing with my voltmeter or oscilloscope, which didnt take long anymore. I checked continuity from the radio all the way down through to the uno - I find no problems. I decide to check voltages just for so, and low and behold, I find 4.8 volts on the 3v3 lines! Figuring surely this is pissing off the radio, I start to figure where this is coming from.
