November 2021
29Nov
New sensor
The hall-effect sensor turned up from Active Sensors. Just waiting now on the 3way DTM connector to arrive and I can fit it.
I've redesigned the lower front suspension bracket for the Mygale. This has additonal webbing and is 5mm longer overall than the Radical bracket, so shouldnt flex or crack under braking.
22Nov
Gear position voltages
I've sourced a replacement sensor, just waiting for it to arrive in the post. The sensor I'm replacing is a Novitechnik SP2846-345-065-101, which is a mechanical sensor. I'm replacing it with a hall-effect device, to eliminate any future issues with noise. The gear voltages from all the drives in 2021 are shown below. Again I used the data in SQL to extract the min, max and average. The lines either side of each average gear voltage, are the min and max voltages seen on all the runs at that event.The standard deviation does seem to increase during the year, but then you'd expect that to some extent with a mechanical device. The reason there are gaps in the 6th gear trace is because I didnt reach 6th gear on some of the shorter tracks. This is another example of using the data to determine potential issues. It's just a shame I didnt spot it before the false neutral at Castle Combe.
18Nov
Gear position sensor
Looking at the data from 2021, there looks to be an issue with the signal from the gear position sensor. This resulted in the first and only false neutral all year, at Castle Combe. With the trace from April compared to October, there is significantly more noise on the gear voltage (mV) seen at the last event, so I'll replace the sensor with a new one for 2022. The gearbox is coming out of the car anyway, as it needs a service, the diff needs shimming, and two gear ratios need changing, so I'll replace it when it's returned from MBR.
Looking at the data taken from the car, I wrote a SQL query to count the number of gear changes, and the figure is 2538 during 2021, all of which were trouble free, apart from one false neutral. Thats not bad at all is it?
17Nov
British Sprint Awards
I collected my 4th place award on Saturday. An odd introduction from Paul Parker, I was expecting at least a Well Done, or such like, but no, I got "I wish the exhaust sounded nicer, 4th place, Graham Blackwell". I wonder what the intro will be next year should I win it :D
Left to Right: Graham Blackwell, Rob Tonge, Terry Holmes, Pete Goulding, Graham Porrett, John Loudon, Matt Hillam, Mark Anson, Simon Bainbridge, and champion Steve Miles. Absent were Steve Broughton and Simon Wallis.
Congratulations to all the drivers on a great season.
13Nov
2021 review
What modifications worked and what didn't
Changing to Yaw based traction control - Worked
This was the single biggest change for 2021, and one that made the largest impact, and took the longest amount of time to tune. Lots of experimentation with the settings, lead to it really starting to work mid-season, where I could finally build the confidence to apply the throttle earlier and earlier out of the slow corners. There is still some tweaking to do, particularly in the wet, during launch, however, its really paid dividends. I've not been working alone either, I've had the invaluable assistance of a leading Syvecs expert, and during the year we've honed the settings to make it better and better.Launch control adjustments - Worked
During 2021 I made 83 launch control starts, with zero drive shaft failures. My quickest launch was 2.16s 64ft at Castle Combe in October. There's definitely a sub 2.0s launch in the car, and with constant improvements during the year, I'll continue analysing the data and adjusting the settings in 2022 until I get down to the times I'm happy with.Redesigned rear wheel speed sensor pickups - Worked
I've run for the whole of 2021 without a single issue with the rear wheel speed sensor pickups. I designed the reluctance rings, and the 3D printed plastic GT101 mounts, and the system has worked perfectly all year. The rings and brackets are available in my shop, and can be seen fitted on several other cars in the paddock.Dual fuel pump setup - Worked
My twin tank and twin pump setup has worked 100% all year, with not a single instance of fuel surge. Noticeable at Knockhill and Lydden before I changed the setup, this year I've been able to knock seconds off my PB's and the class records, at both circuits, where the car was previously prone to fuel surge. The only issue I have got, is with the fuel level sender unit inside the fuel tank. Its not giving consistent readings, however, that could be caused by the digital gauge I've sourced, so I may wire the sender unit in to a spare input on the ECU, so the ECU can report the fuel level via CAN on the dashboard instead.Race Technology DASH4PRO CAN conversion - Worked
The CAN version of the dash replaced the RS232 version, and after some programming of the CAN stream and the dashboard layout, I've got a fully featured rich display of all the parameters that matter, and the dashboard never missed a beat all year.VVT timing - Worked
I wired in the exhaust VVTi cam solenoid to the ECU, which had never been connected previously, and with the output enabled, the ECU has reported not a single VVT Failure all year. This means when I go back to the Dyno we can work on tuning the Exhaust camshaft timing, to see if we can get even more power from my Area-Six EcoBoost engine.Powerlite battery upgrade - Worked
The PS-09 battery really struggled to start the engine during 2020, so I contacted Powerlite and they arranged for a free PS-20 replacement. I must say that the performance of the PS-20 has been brilliant. It starts the engine from cold, every time, and of the 83 runs this year, the battery never missed a beat.Cold air intake - Worked
The CAI definitely drove the ambient air temperatures down. It does need insulating from heat soak though. I'll do that over the winter break.Funk Motorsport Turbo blanket - Worked
The heat released from the turbo was kept under control by the turbo blanket, which in turn reduced temperatures under the engine cover.Fast acting air charge temperature sensor - Worked
This has had quite an affect on the performance this year. On the one hand the ECU is able to react more quickly to temperature changes. On the other, had I not fitted the fast acting sensor, then the ECU may not have tripped at Snetterton when the ACT rose above 80C. But I'd rather the ECU trip, than the engine suffer damage because a sensor wasnt reporting temperatures quickly enough.Alternator under ECU control - Worked
All year I've had the ECU turn the alternator output on and off depending on the throttle position, and battery voltage, and its worked really well, reducing the load on the engine on full throttle, when the conditions allowed.Rewiring the X10 - Worked
I rewired the X10 expander unit to allow me to connect damper potentiometers, and to move the wheel speed sensors off the ECU and on to the X10, and its all worked perfectly all year. I did have a noisey front left sensor, which was a recurring problem, until I fitted a replacement at Kirkistown and that seems to have finally cured the issue.Rewiring the chassis - Worked
At the start of the year, I removed the battery terminal posts from the cockpit floor, relocating the Lithium battery to the intercooler side pod, and this allowed me to remove the heavy duty copper cables that ran inside the car, and use shorter cables to provide power to the starter motor. This freed up space inside the cockpit, and I had no issues at all with battery power during the year.Barge board removal - Worked
For Snetterton I raised the front wing angle to a new higher angle than I'd ever run before, and on the first timed run the ACT temperature exceeded 80C and tripped the ECU. After some head scratching, rather than lower the wing angle I tried removing both barge boards, and this transformed the ACT temperatures. The front end grip was phenomenal too, so since that weekend, the barge boards have been consigned to the garage roof, and the car has been incredible to drive. I had to trim the nose cone to allow the steeper wing angle, but by doing that I have a new Max Aero setting for the front, and this has been used to good effect at several circuits, allowing me to make huge improvements in my times. I'll continue to use the new settings in 2022.Diffuser floor attachment - Worked
After several years of struggling to get the leading edge of the diffuser floor to remain attached to the underside of the chassis floor, I found a way of securing it properly, and this has made a big difference to the balance of the car.All in all, its been a great year, with just a single ECU trip at Snetterton, and the exhaust pipe fracture at Castle Combe on the final run of the year, which given the abuse the car has had, and compared to other less fortunate drivers, I think is a fair return on the time and effort invested in preparing and racing the car.
7Nov
Numbers on
I've replaced the old number 8's with the new number 4's. And I've replaced the IGN switch with a new one, again, it's IP67 spec so its weather proof. Everything tests out ok. The next job I'll be starting is to check the chassis over. So all joints and suspension components will be inspected for cracks/wear etc.I'm still getting to grips with analysing the engine data. I've tried a few different tools to enable Machine Learning, and I'm now looking at Orange, which is very easy to use, and is providing some promising results so far.
4Nov
Exhaust repaired
The exhaust pipe is now fitted back on the turbo. It had to be Tig welded, and Nigel's done a great job. I'll make a support bracket for the pipe, so hopefully it wont happen again. Having said that, the weld was 9 years old, and the pipe had been through a very large number of heat cycles, so I guess it was just one of those things, and I should have checked it every now and again. I've also replaced all the plastic pipes providing air to the pneumatic shift system.
3Nov
Data analysis
With the dust settled on the 2021 championship, I've started looking through the data captured from the car, and I've discovered a potential issue with the wiring. Let me explain.Over the weekend, I've taken every LRD log file catpured from the Life ECU during 2021, and exported them one at a time, in to CSV format. Basically, every time the engine is started and stopped, the ECU saves the data during that run, in the ECU's internal memory, for downloading. If you dont download the data, its eventually overwritten, so its best to download the data after every run at the circuit, so you have a record of the whole year.
When you export the data to CSV, you're allowed to specify the frequency that the data is exported at. It suggests 1000Hz, as that the highest sampling frequency used when the engine is running. However each log file would be 1000s of megabytes in size, and thats too much detail to wade through. So instead, I've decided on a 10Hz export rate, which results in files up to 10MB in size, which is far more manageable.
I've installed Microsoft SQL Server Express on my Desktop PC, and I've created a database called EHM. In the database, I've a table called mygale_data, in to which I have imported all of the data from the eighty three exported CSV files. Prior to importing the files, they were all modified with an extra column, which I've called eventID, and each file has been given a unique number which the eventID column is populated with. This allows me to group the imported rows in to seperate events, using the eventID. I've created another table called event_details, which holds the filenames, the corresponding eventid's, and the run order for that event.
That task took me around 6 hours to process and import all of the CSV files, and I've now got a table with ~320,000 rows of data in it. So what I can do now, is create SQL reports which interrogate all of the information from the car. The target is to use Machine Learning, so the computer can be taught what a healthy engine looks like, and as the data is analysed it can potentially pinpoint issues which could indicate impending component failure.
One thing that became apparent, was the engine died on me at Castle Combe on my T2 run off run, albiet briefly, and the cause was the ENGINE ENABLE signal went from OK to OFF to OK, at which point the power returned and I carried on around the circuit. So, had this happened before? At Curborough the week before I did experience the same symptons on one run, but at the time I didnt find the cause. Would a report in SQL find otherwise?
Yes. The results show that the ENGINE ENABLE signal also changed during the P2 practice run at Curborough on the figure of 8 course.
I've included the lat and long G forces in the results, the vehicle speed, and the throttle position, as well as the onTime (the time the engine had been running prior to the enable switch glitching)
So I've checked the toggle switch that provides the IGN on/off signal (ENGINE ENABLE) to the ECU. I cant find anything wrong with it. The wiring looks ok, again, all seems to be working correctly. However, to be on the safe side, I've ordered a replacement switch, as it is 8 years old, and I'll swap it out. I dont know if the G forces present during the glitches caused the switch to fail briefly. I cant find anything else in the data that would pinpoint the cause, other than potentially a faulty switch/wiring. Anyway, with the switch replaced I'll have to wait until I next go testing to see if the problem has gone away.
I'm now looking through the other channels in the database, to see if there's anything else I can learn from the information I've captured. I've installed Pytorch and I'm learning how to use Python to provide the machine learning element, which is going to take some time to master. What else could we learn from analysing the data? What other areas can we benefit from examining? Driver performance, maybe?
