[Ovmsdev] v2.5.4 call for testing

Mark Webb-Johnson mark at webb-johnson.net
Wed Oct 16 16:16:28 HKT 2013


Wiro,

I see the problem. I will fix in github tonight and commit.

Can you build from source? If not, you will need to wait for 2.5.5, or I can build you a private firmware?

Regards, Mark.

On 16 Oct, 2013, at 2:54 pm, Wiro Zijlmans <wirozijlmans at gmail.com> wrote:

> Mark,
> Other commands seem to work. E.g. Stat, version, see attached picture.
> What I mean with message from app is a notification that shows up at my phone as notification of message receipt. There is no change of parameters in the app itself.<image.png>
> 
> Groet / Regards
> Wiro Zijlmans
> 
> Op 16 okt. 2013 om 03:34 heeft Mark Webb-Johnson <mark at webb-johnson.net> het volgende geschreven:
> 
>> Wiro,
>> 
>> Do you get any SMS responses? Such as a response to "STAT"...
>> 
>> Try "ACC STAT".
>> 
>> I don't understand about 'confirmation from the app it is received' - do you mean you see the parameter updated?
>> 
>> Regards, Mark.
>> 
>> On 16 Oct, 2013, at 1:41 am, Wiro Zijlmans <wirozijlmans at gmail.com> wrote:
>> 
>>> Hi Mark,
>>> I am running the v1 HW on my TR and upgraded to 2.5.4.
>>> ACC seems not work at all. When sending ACC HERE command I get confirmation from app it is received but no feedback by SMS.
>>> 
>>> Groet / Regards
>>> Wiro Zijlmans
>>> 
>>> Op 11 okt. 2013 om 15:12 heeft Mark Webb-Johnson <mark at webb-johnson.net> het volgende geschreven:
>>> 
>>>> 
>>>> OK. I've just committed v2.5.4 and pushed to github. I've also tagged it (as v2.5.4) and upload pre-built .hex files in the vehicle/firmware directory.
>>>> 
>>>> Short story: v2.5.4 seems stable to me, and has most of the bugs that I know about fixed. I’m still having trouble tracking down a random charge-time-estimate bug that is affecting CHARGEBY ACC profiles, but other than that is seems fine to me.
>>>> 
>>>> Big caveat: this is testing code, please don't use ACC if you _really_ need that charge to get to work / school / the beach / wherever. Don't blame me if your car doesn't start the charge as scheduled.
>>>> 
>>>> Warning: this is a long mail.
>>>> 
>>>> Firstly, the v2.5.x change log:
>>>> 
>>>> 2013-10-11 2.5.4       Firmware 2.5.4
>>>>                        # Make MINSOC drive NET_NOTIFY_CHARGE (rather than NET_NOTIFY_STAT)
>>>>                        # Input GPIO control for input RA2..RA5, and output RC0..RC3
>>>>                        # ThinkCity: Updates to Think City support
>>>>                        # iMiev: Updates to Mitsubishi vehicle support
>>>>                        # Think City: Added support for switching on/off external heater/auxilary via Valet Mode button
>>>>                        # ACC: Aggressively stop charge if cooldown is done
>>>>                        # ACC: If we are waiting for a charge, but the charge is started, then stop it
>>>>                        # Range and SOC limits stop exactly ON the limit
>>>>                        # Think City: car_parktime added
>>>>                        # Fix for command handler with net_msg_cmd_msg
>>>>                        # Tesla Roadster: fix for mins remain estimate if SOC limit but no RANGE limit
>>>>                        # Use <=0, not ==-1, as indication range limit is not set
>>>>                        # ACC: Move old STAT command to PARAMS?, and add a new STAT command to show true status
>>>>                        # ACC: If charge is done, in acc location, and unplugged, transition to parked-in state
>>>>                        # ACC: Repeat tx charge mode and limit commands, during cooldown and acc charge, if necessary
>>>>                        # Tesla Roadster: Disable cooldown on boot
>>>>                        # ACC: Race condition for charge finishes before cooldown flag is cleared
>>>>                        # ACC: Setting params now returns a summary
>>>>                        # ACC: Reset acc state, if enable/disable or setting parameters
>>>>                        # ACC: Integer/Byte overflow workarounds
>>>>                        # ACC: Workaround race condition where charge may take a few seconds to actually start (when commanded)
>>>>                        # Tesla Roadster: Provide defaults for range, soc and cac in charge estimates
>>>>                        # Support car_chargeestimate
>>>>                        # ACC: rework to use new car_chargeestimate
>>>> 
>>>> 2013-09-18 2.5.3       Firmware 2.5.3
>>>>                        ## Updates to thinkcity
>>>>                        ## Remove DIAG and add ACC to V1_Production build config
>>>>                        ## Remove InternalGPS from V2_TR_Production build config, as not required
>>>>                        ## Cooldown: safety check (only recycle if charging), and revise cooldown current to 13A
>>>>                        ## Support definition of location in SMS commands
>>>>                        ## Tesla Roadster: HVAC#1 should be 0x8F not 0x87
>>>>                        ## Logging: Don't reserve a log slot if the respective logging option is disabled
>>>>                        ## Improve cooldown status reporting on SMS STAT
>>>>                        ## Tidy-up of minSOC
>>>>                        ## Move chargelimit charge stop to common vehicle.c function (not acc)
>>>>                        ## Tesla Roadster: Updates to Tom's charge-time-predictor, with thanks
>>>>                        ## Log cooldown charges as mode=5, and fix excessive-logging-bug related to cooldown cycles
>>>>                        ## Logging: Send log messages one-by-one, to avoid buffer overflows
>>>>                        ## ACC: Number homelinks from 1..3
>>>>                        ## Tesla Roadster: support for chargelimits
>>>>                        ## Make SMS "RESET" reset the module completely
>>>>                        ## Complete rework and expansion of checkpoint numbers
>>>>                        ## Stop charge after cooldown, if no subsequent charge requested
>>>>                        ## Report CAC as part of charge log record
>>>>                        ## Typos in acc for net_state_enter
>>>>                        ## Show debug checkpoint, for DIAG SMS with debugcrash
>>>>                        ## Support vehicle_fn_minutestocharge
>>>>                        ## Add a timezone parameter (-)HH:MM
>>>>                        ## ACC: Support timed charges
>>>>                        ## Move car_cooldown_wascharging to global, and make use of it in ACC
>>>>                        ## Introduce some delays to allow car to wake up
>>>> 
>>>> 2013-08-27 2.5.2       Firmware 2.5.2
>>>>                        ## Updates to vehicle_thinkcity.c (getting to a pretty usable vehicle module)
>>>>                        ## Make car_tbattery signed integer, and integrate thinkcity changes
>>>>                        ## Only report range and soc limits if >0
>>>>                        ## Merge thinkcity changes:   Think City AC-line voltage/current in SMS STAT   SMS migrated from standard handler to vehicle_thinkcity.c
>>>>                        ## Tesla Roadster: Don't stop charge after cooldown, if previously the car was already charging
>>>>                        ## Log cooldown charges that become normal charges as two separate charges
>>>>                        ## Tidy up diag, by removing temp test code T1 T2 T3
>>>>                        ## Create individual configs for TeslaRoadster and RenaultTwizy (to allow all features for these cars)
>>>>                        ## Add TRACK (type XX) vehicle that just tracks GPS
>>>>                        ## string_to_mode utility function
>>>>                        ## Rework vehicle inclusions. Tesla Roadster, Renault Twizy and Volt/Ampera are production. Everything else is experimental
>>>>                        ## Add experimental vehicle Kyburz DXP
>>>>                        ## Add HVAC message details
>>>>                        ## Support HVAC bit in car_doors5, and cooldown cycles
>>>>                        ## Range and charge limits handled by ACC
>>>>                        ## Basic ACC implementation, but without timed charges
>>>> 
>>>> 2013-08-16 2.5.1       Firmware 2.5.1
>>>>                        ## Framework to build logging module in experimental modes
>>>>                        ## FIsLatLongClose utiliy function (courtesy of Tom Saxton)
>>>>                        ## Twizy: Mi/Km conversions updated to new functions
>>>>                        ## Switch to use macros, for clarity and maintainability
>>>>                        ## ACC integration to build environment
>>>>                        ## Parameter support for base64 encoded parameters
>>>>                        ## utils support for mode display
>>>>                        ## ACC support for net_sms
>>>>                        ## Switch logging system to use new "h" historical data submission, with acknowledgement
>>>>                        ## Go from 4->6 log records, based on free space
>>>>                        ## Tesla Roadster: Notify server if charge limit is changed
>>>>                        ## Tesla Roadster: Notify server if charge mode is changed
>>>>                        ## v2.5.1 protocol guide
>>>>                        ## Add core support for cooldown
>>>>                        ## Tesla Roadster support for cooldown
>>>>                        ## Stub implementation, with basic functionality, for ACC
>>>>                        ## Zero logging records on init, plus other safety checks
>>>>                        ## Only initialise logging on a normal power up
>>>>                        ## Standardise 3 diag tests: T1, T2 and T3
>>>>                        ## Only send logging msgs if link is connected
>>>>                        ## Fix logging prefix and some logging tests
>>>>                        ## Revisions to CAR_IS_CHARGING logic
>>>>                        ## Misc bug-fixes for logging
>>>>                        ## Report distance in miles (rather than 10ths of miles)
>>>>                        ## Remove dr++ and cr++ sequence numbers, as not required
>>>>                        ## Make ACC commands case insensitive
>>>>                        ## Fix bug acknowledging log record #0
>>>>                        ## Move digital speedo feature (Tesla Roadster) from experimental to fully supported, and implement opt-in feature bits
>>>>                        ## Honour opt-in flags for logging
>>>>                        ## Only report debug crash reason if not normal power up, and only report once
>>>> 
>>>> As you can see, a lot has changed. The key points are:
>>>> 
>>>> Some restructuring of the build targets:
>>>> V1_production.hex               V1 Hardware Module, production firmware
>>>> V2_experimental.hex             V2 Hardware Module, experimental firmware
>>>> V2_production.hex               V2 Hardware Module, production firmware
>>>> V2_RT_production.hex            V2 Hardware Module, Renault Twizy full firmware
>>>> V2_TR_production.hex            V2 Hardware Module, Tesla Roadster full firmware
>>>> Support for LOGGING (charge and drive)
>>>> Support for ACC (Advanced Charge Control) - Tesla Roadster only at the moment
>>>> Lots of other little nice fixes and enhancements
>>>> 
>>>> Apart from overall reliability testing (making sure it is stable in the cars), the two things I need help with are the LOGGING and ACC functions.
>>>> 
>>>> [1] Logging
>>>> 
>>>> The logging code is enabled by setting opt-in feature (#13) bit 1 (to log drives) and/or bit 2 (to log charges). Or, you can just set feature #13 to 255 and opt-in to everything :-)
>>>> 
>>>> Once you've opted in, whenever you complete a charge or drive, the car will submit a summary record to be stored on the server. You can then use the prototype HTTP API, or perl client, to retrieve those records for your own purposes.
>>>> 
>>>> Note that there is a privacy concern here, as both record types (drive and charge) store GPS locations - this is the reason why it is opt-in.
>>>> 
>>>> The testing I need done is just to ensure that this works for you, and that the resulting logged drives/charges match what you are expecting. It would also be useful to confirm if this works on all supported cars (which it should).
>>>> 
>>>> [2] Advanced Charge Control
>>>> 
>>>> The Advanced Charge Control system is designed to supplement the (aka take over from) vehicle's own charge scheduling. It should support any vehicle with charge control, but at the moment that means only Tesla Roadster. It has some pretty sophisticated features. For the moment, it is setup via SMS, but in future we will allow smartphone Apps to do this. Once setup, it runs autonomously, and doesn't even require a cellular connection.
>>>> 
>>>> ACC works from the concept of a 'charge location'. This is a geofenced location that is used for charging, and you can configure up to 4 of these (numbered #1 through #4). To define the car's current location as a 'charge location', sms "ACC HERE" to the car, and it will allocate a free slot and reply to you with it. You can use "ACC NOTHERE" to clear the current location.
>>>> 
>>>> If you want to clear a particular ACC location, you can sms "ACC CLEAR n" (to clear one location), or "ACC CLEAR" (to clear all locations).
>>>> 
>>>> You can show the current status off ACC with “ACC STAT”.
>>>> 
>>>> You can show the ACC details for the current location by sms "ACC PARAMS?" (or for a particular location by "ACC PARAMS? n" - which is very useful if you have crappy cellular connectivity like I do).
>>>> 
>>>> The ACC parameters for a particular location are configured with the "ACC PARAMS" SMS. You can list a location number (1, 2, 3 or 4) as the first parameter - or don't specify location if you just want to configure the current location the car is at. After 'ACC PARAMS", you can set the parameters you would like, from the following:
>>>> 
>>>> COOLDOWN - perform a cooldown charge (13A range mode, with cooling cycles) when the car is plugged in at this ACC location
>>>> NOCOOLDOWN - don't do a cooldown charge (default)
>>>> HOMELINK n - activate homelink when the vehicle drives to within 100m of an ACC location (the idea is to open our garage door / gate automatically as you approach)
>>>> NOHOMELINK - don't activate homelink (default)
>>>> CHARGEPLUGIN - charge the car on plugin (but after cooldown, if enabled)
>>>> CHARGEAT HH:MM - schedule charge to start at the specified time (hours:minutes, 24hour clock)
>>>> CHARGEBY HH:MM - schedule charge to complete by the specified time (hours:minutes 24hour clock)
>>>> NOCHARGE - don't charge (default)
>>>> LIMIT n - limit charge current to n Amps - must be specified for all CHARGE* actions
>>>> MODE m - set charge mode (where "m" is STA(NDARD), STO(RAGE), RAN(GE) or PER(FORMANCE)) - must be specified for all CHARGE* actions
>>>> STOPRANGE r - stop charge when range "r" is reached (specified in vehicle units, set to 0 (default) to not limit).
>>>> STOPSOC s - stop charge when SOC "s" is reached (specified as a percentage, set to 0 (default) to not limit).
>>>> 
>>>> The actions to take at a particular ACC location must be enabled with "ACC ENABLE" (or "ACC ENABLE n") before they will work. You can also disable with "ACC DISABLE" (or "ACC DISABLE n").
>>>> 
>>>> If you are using CHARGEAT or CHARGEBY, you need to specify the timezone of the vehicle. This is parameter #23 and is specified as HH:MM offset from GmT (with an optional leading "-" if west of Greenwich).
>>>> 
>>>> The default temperature limit for cooldown is 31Celcius, and time limit is 60 minutes. If you want to change these, you can set parameter #15 to templimit:timelimit (e.g.; "31:60").
>>>> 
>>>> Note that you can also cooldown manually, while charging, with the SMS command "COOLDOWN", and see the status with SMS "STAT".
>>>> 
>>>> I've spent quite some time testing CHARGEPLUGIN, CHARGEAT, COOLDOWN, LIMIT, MODE, STOPRANGE and STOPSOC - hopefully those are ok now, but the control logic does need wider testing. The CHARGEBY option is still giving me random problems, related to the charge time estimated - I am still working on that. I've also been unable to test HOMELINK and would be very interested to see if that works.
>>>> 
>>>> Some notes on cooldown on Tesla Roadster
>>>> 
>>>> The cooldown algorithm is to start a range mode charge at 13Amps, and to monitor the HVAC system of the car. Once the HVAC starts, runs for some time, then stops, we call it a 'cooling cycle'. We keep a record of how many such cycles have occurred.
>>>> 
>>>> Once we detect that the HVAC has stopped, we try to start it again. The logic at the moment is to switch to performance mode for ten seconds, then back to range mode, once every minute until the HVAC starts again. The _best_ way of starting the HVAC is to stop and start the charge, but that is painful on the contactors, so we don't do it. Switching Range->Performance->Range seems to be the second best, but I remain unconvinced that it makes much difference. In Hong Kong summer weather, I get one cooldown cycle every ten minutes or so. I can drop 6 to 8 celcius in one hour.
>>>> 
>>>> The cooldown will stop after either the desired temperature is hit (based on the temperature of the hottest brick in the battery) or the cooldown has been ongoing for the time limit.
>>>> 
>>>> Conclusions
>>>> 
>>>> I'm happy that the code is reasonably stable now, but really need help testing this in a larger number of cars. If you do see a problem, please get me (by eMail to OVMS developers mailing list, or my personal address):
>>>> 
>>>> Output of "SMS STAT" for the specified location
>>>> Your vehicle id
>>>> Output of base64 parameter string (from the cellphone app) - copy-and-paste
>>>> Date/Time you entered the ACC location (with timezone)
>>>> Date/Time you plugged in at the ACC location (with timezone)
>>>> In the smartphone app, call up the Features and Parameters screens (so the server logs those, and I can check)
>>>> Description of the problem you have
>>>> 
>>>> Please also share success stories. It is good to know what works.
>>>> 
>>>> Regards, Mark.
>>>> _______________________________________________
>>>> OvmsDev mailing list
>>>> OvmsDev at lists.teslaclub.hk
>>>> http://lists.teslaclub.hk/mailman/listinfo/ovmsdev
>>> _______________________________________________
>>> OvmsDev mailing list
>>> OvmsDev at lists.teslaclub.hk
>>> http://lists.teslaclub.hk/mailman/listinfo/ovmsdev
>> 
>> _______________________________________________
>> OvmsDev mailing list
>> OvmsDev at lists.teslaclub.hk
>> http://lists.teslaclub.hk/mailman/listinfo/ovmsdev
> _______________________________________________
> OvmsDev mailing list
> OvmsDev at lists.teslaclub.hk
> http://lists.teslaclub.hk/mailman/listinfo/ovmsdev

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openvehicles.com/pipermail/ovmsdev/attachments/20131016/4ca4f907/attachment.htm>


More information about the OvmsDev mailing list