[Ovmsdev] Can buses stop after some time
Mark Webb-Johnson
mark at webb-johnson.net
Thu May 24 11:08:55 HKT 2018
I don’t think we need to support other speeds. As you say, if necessary it is trivial as an optional parameter after bus name. Probably no harm adding that. But, even today, there is a workaround:
OVMS# obdii ecu start can3
OBDII ECU has been started
OVMS# can can3 status
CAN: can3
Mode: Active
Speed: 500000
OVMS# can can3 start active 250000
Can bus can3 started in mode active at speed 250000bps
OVMS# can can3 status
CAN: can3
Mode: Active
Speed: 250000
When the client (HUD, whatever) is trying to connect to the ECU, it can try 500K, 250K. Or it can try 250K, 500K. I suspect yours tries the first descending sequence, and hence doesn’t have any issues as it finds the match at 500K first.
Anyway, this MCP2515 can bus lockup is something we have to fix. The fact it is reproducible on obd2ecu is good and helpful for that.
Regards, Mark.
> On 24 May 2018, at 10:46 AM, Greg D. <gregd2350 at gmail.com> wrote:
>
> Hi Mark,
>
> Thinking about this more, I believe what I have done is correct. The OVMS OBD2ECU task is representing itself to the HUD / Dongle as an ECU, and ECUs run at one speed. My speed is 500kbps. I should never be connected to a car's ECU, as there could be a conflict in answering polls. Fortunately, the connectors won't match up (both female). I also never transmit unless requested by the device to do so. Don't mind just sitting there, minding my own business (no timeouts for inactivity).
>
> The HUD I have does 500k, so we're good. It actually tries 250k, probably among others, before finding me on 500k. It will also try extended framing if standard doesn't evoke a response (I support both types). My question was whether 500k is a universally supported speed, and that seems to be the case. The one ICE car that I own (2013 Honda CRV) also runs 500k for its ECU.
>
> As you note, since any particular car's ECU is always fixed, it's the HUD's job to adapt, not mine. They connect to me, not the other way around. I don't think I need to emulate every ECU, just the OVMS ECU. Also, if I were to try different speeds, while the device was doing the same, we might never meet up. Best to just sit where I'm known to be, and let them find me. Are there any devices that we need to support that can't adapt to 500k?
>
> When I use an OBDII Dongle (OBDWiz in my case), it also scans through the space of ECU speeds and protocols, starting at the low end, to find and connect to me. That has worked perfectly every time (no CAN bus stoppage). 'can can3 status' shows no errors logged after it has connected (Rx error & overflow counters are zero), same with the HUD. So I'm puzzled why your testing has shown different results. I have seen some occasions where after physically connecting and reconnecting things the communication stops, but that could also be a flaky cable / connectors. It's been through a lot of mechanical use, but haven't found a repeatable way to reproduce it.
>
> Hopefully we'll get some feedback from users (all positive!) when the OBDII cables hit FastTech. If there's a need to support a different speed, we can add that as an optional parameter to the start command line, after the bus name.
>
> Greg
>
>
> Mark Webb-Johnson wrote:
>> The OBDII client should determine the bus speed and the ECU side only needs to support one speed.
>>
>> Clients typically do this in one of two ways:
>>
>> Set CAN controller to ‘listen’ mode, then loop through supported bus rates, listening for a correctly formatted CAN message to let you know you found the right rate. If found, exit the loop.
>>
>> Set CAN controller to ‘active’ mode, then loop through supported bus rates, polling for OBDII data and ignoring errors. If you get a valid response, exit the loop.
>>
>> Obviously #1 is the least noisy approach, and the least likely to interfere with other vehicle systems, but won’t work on purely active polling can buses with nothing else active on the bus. I guess some clients may employ both approaches.
>>
>> I suspect that your HUD is working cleanly because it tries 500K rate first, so doesn’t generate errors.
>>
>> It would probably be worth us transmitting a CAN bus heart beat every few seconds when obd2ecu is started, 12v external power is on, but we haven’t heard from the HUD in a while. That would probably help clients lock onto us quicker.
>>
>> Regards, Mark.
>>
>>> On 24 May 2018, at 4:54 AM, Greg D. <gregd2350 at gmail.com <mailto:gregd2350 at gmail.com>> wrote:
>>>
>>> Hi Mark,
>>>
>>> The OBD2ECU task assumes that "all" HUDs and such devices operate at 500k. Are you aware of any that don't (can't) operate at that speed? I was hoping I wouldn't have to support multiple speeds, especially autosensing them.
>>>
>>> BTW, I have not seen any problems connecting an OBDII Dongle to the OVMS and letting it do its default scan through the various rates in order to connect. It just takes longer than it would if (as I usually do) tell it what rate and frame size to use. The various frames and speeds tried before figuring out the right one don't seem to bother it.
>>>
>>> Greg
>>>
>>>
>>> Mark Webb-Johnson wrote:
>>>> I am also now seeing this.
>>>>
>>>> Trying out the OBD2ECU HUD cables, I was having problems getting it to work. Those HUDs try to transmit at different baud rates, to probe for what is correct, and that is causing errors at our end. Once we get those errors, seemingly we can’t recover. A ‘can can3 start active 500000’ fixes the issue and the HUD connects.
>>>>
>>>> It looks something like this:
>>>>
>>>> OVMS# can can3 status
>>>> CAN: can3
>>>> Mode: Active
>>>> Speed: 500000
>>>> Interrupts: 1
>>>> Rx pkt: 0
>>>> Rx err: 105
>>>> Rx ovrflw: 0
>>>> Tx pkt: 0
>>>> Tx delays: 0
>>>> Tx err: 0
>>>> Tx ovrflw: 0
>>>> Err flags: 0x8000
>>>>
>>>> Or this:
>>>>
>>>> OVMS# can can3 status
>>>> CAN: can3
>>>> Mode: Active
>>>> Speed: 250000
>>>> Interrupts: 146
>>>> Rx pkt: 0
>>>> Rx err: 128
>>>> Rx ovrflw: 0
>>>> Tx pkt: 0
>>>> Tx delays: 0
>>>> Tx err: 0
>>>> Tx ovrflw: 0
>>>> Err flags: 0x800b
>>>> E (713021) canlog: Error can3 intr=1 rxpkt=0 txpkt=0 errflags=0x8000 rxerr=1 txerr=0 rxovr=0 txovr=0 txdelay=0
>>>> E (713021) canlog: Error can3 intr=2 rxpkt=0 txpkt=0 errflags=0x8000 rxerr=2 txerr=0 rxovr=0 txovr=0 txdelay=0
>>>> E (713021) canlog: Error can3 intr=3 rxpkt=0 txpkt=0 errflags=0x8000 rxerr=4 txerr=0 rxovr=0 txovr=0 txdelay=0
>>>> ...
>>>> OVMS# can can3 status
>>>> CAN: can3
>>>> Mode: Active
>>>> Speed: 250000
>>>> Interrupts: 3757
>>>> Rx pkt: 1
>>>> Rx err: 128
>>>> Rx ovrflw: 0
>>>> Tx pkt: 1
>>>> Tx delays: 0
>>>> Tx err: 0
>>>> Tx ovrflw: 0
>>>> Err flags: 0x800b
>>>> ...
>>>> OVMS# can can3 status
>>>> CAN: can3
>>>> Mode: Active
>>>> Speed: 250000
>>>> Interrupts: 3775
>>>> Rx pkt: 10
>>>> Rx err: 128
>>>> Rx ovrflw: 0
>>>> Tx pkt: 10
>>>> Tx delays: 0
>>>> Tx err: 0
>>>> Tx ovrflw: 0
>>>> Err flags: 0x800b
>>>>
>>>> And then the can bus dead (until ‘can start …’ to reset it).
>>>>
>>>> Good news is that with those HUDs, it is very easy to recreate the fault condition. I’ll see what I can do to find out what is going on. My guess is we are not clearing the MCP2515 error condition correctly. I will try to find out what is going on...
>>>>
>>>> Regards, Mark.
>>>>
>>>>> On 14 May 2018, at 6:17 PM, Tom Parker <tom at carrott.org <mailto:tom at carrott.org>> wrote:
>>>>>
>>>>> On 14/05/18 20:36, Stein Arne Sordal wrote:
>>>>>
>>>>>> I also tried to raise the stack size to 6144.
>>>>>> It seems like it got worse…Can buses (RX) stops working more often. TX is fine.
>>>>>
>>>>> I don't see an improvement either. I wrote the new firmware with updated sdkconfig at about 3pm yesterday and it rebooted and lost the state of charge metric at 8:45pm, the car woke up at midnight and started charging, providing data for the SOC metric, during the charge there were a couple of gaps in the telemetry, charging finished at 3:10am and the OVMS rebooted at 3:45. The OVMS then stopped sending telemetry completely at 7:20 am when the car was switched back on. I had a chance to plug a laptop in and the module was unresponsive on the serial port.
>>>>>
>>>>> I'm not sending the monotonic metric so it's only possible to see the first reboot after the car is switched off (when it forgets the SOC).
>>>>>
>>>>> I've built a version of the firmware with most things turned off (and found vehicle depends on webserver and webserver depends on OTA) I'll see how that goes tonight.
>>>>>
>>>>> Otherwise I'll get the datalogger out and/or try the sdcard logger again.
>>>>>
>>>>>
>>>>>
>>>>> _______________________________________________
>>>>> OvmsDev mailing list
>>>>> OvmsDev at lists.openvehicles.com <mailto:OvmsDev at lists.openvehicles.com>
>>>>> http://lists.openvehicles.com/mailman/listinfo/ovmsdev <http://lists.openvehicles.com/mailman/listinfo/ovmsdev>
>>>>
>>>>
>>>>
>>>> _______________________________________________
>>>> OvmsDev mailing list
>>>> OvmsDev at lists.openvehicles.com <mailto:OvmsDev at lists.openvehicles.com>
>>>> http://lists.openvehicles.com/mailman/listinfo/ovmsdev <http://lists.openvehicles.com/mailman/listinfo/ovmsdev>
>>>
>>> _______________________________________________
>>> OvmsDev mailing list
>>> OvmsDev at lists.openvehicles.com <mailto:OvmsDev at lists.openvehicles.com>
>>> http://lists.openvehicles.com/mailman/listinfo/ovmsdev <http://lists.openvehicles.com/mailman/listinfo/ovmsdev>
>>
>>
>>
>> _______________________________________________
>> OvmsDev mailing list
>> OvmsDev at lists.openvehicles.com <mailto:OvmsDev at lists.openvehicles.com>
>> http://lists.openvehicles.com/mailman/listinfo/ovmsdev <http://lists.openvehicles.com/mailman/listinfo/ovmsdev>
>
> _______________________________________________
> OvmsDev mailing list
> OvmsDev at lists.openvehicles.com
> http://lists.openvehicles.com/mailman/listinfo/ovmsdev
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openvehicles.com/pipermail/ovmsdev/attachments/20180524/5338b77d/attachment.htm>
More information about the OvmsDev
mailing list