Minimum required for data logging?

[Beauvais Motorsport]

I want to log 2 sensors already in the car and add one extra then display all three together to analyse.

Can I do this with a high resolution without breaking the bank or with major wiring changes?

 

[Steven103]

Absolutely, probably, depends.

Need to decide what sample rate & ideally what electrical format the sensor outputs. Analogue & digital square wave/PWM are doable. If it’s a digital message you’re screwed, fit a dedicated analogue/digital sensor in parallel.

You can get some EPAS related values by connecting the diagnostic pin on the DCI controller to the back of the OBD connector then access with a decent reader. Thread on CS somewhere. Very easy but all the usual drawbacks of OBD.

Loads of handheld multichannel USB oscilloscope type devices. Something like a picoscope or cheap copy. This is where knowing your required sample rate comes in.

Some of the usual brands do stand-alone analogue/digital loggers with basically one button and a memory card. I was looking at Race Tech DL1 as IIRC they were analogue/digital ~4 channel >25Hz and occasionally came up on eBay circa £200.

Exact opposite of plug & play, but this is completely doable for peanuts with an Arduino & bit of learning. Similar sort of thing to a raspberry pi. Can send live data from an Arduino straight to a laptop screen via USB. Or with (a lot) more effort & an expansion board you could log CSV files to an SD card for a more permanent install exactly like the DL1. Pros: endless possibilities for adding complexity. Cons: learn to code & do a bit of breadboard electronics yourself. Much like doing complex stuff in excel, someone will have posted something 90% similar online that you can copy/paste/modify.

 

[Beauvais Motorsport]

Thanks for the info! To be specific, its the yaw rate sensor and steering angle sensor, the added sensor will be a linear potentiometer. I dont know the sample rates.

From what you're saying I honestly dont think I'll be able to go custom with this and am looking at the ££ option (aim pdm with their own sensors) I want to start with the 3 sensors as they are the most important but then being able to add another 4 potentiometers and others down the line without too much fuss kinda makes up for the initial cost of 2K+.

 

[Steven103]

I went down the route of separate sensors for those two. Accelerometer (3 axis so just lat acc, no yaw) in front of gear stick and a string potentiometer around the steering column. I'm logging at a higher frequency and resampling to 5Hz to get rid of noise.

It meant no disassembly, no risk of DTCs or upsetting the standard ABS/EPAS/ECU, no hassle of working out sensor type or pin outs, no hassle of working out the conversion from volts/bits to real units, no hassle of integrating OBD (all of the signals from the sensors you mentioned are available from the ABS/ESP ECU via diagnostic connector) or risking insufficient & variable sample rate.

Annoying to add weight/clutter by duplicating existing sensors/wiring, and you need to do your own signal filtering/cleaning.


If you do decide to investigate reusing existing sensors, you need to probe each pin of the connector with a multimeter whilst moving the sensor. If it's analogue the voltage will change somewhat consistently and proportionally when you move it, if it's digital or on a bus you'll probably need an oscilloscope to determine which.

 

[Beauvais Motorsport]

I went down the route of separate sensors for those two. Accelerometer (3 axis so just lat acc, no yaw) in front of gear stick and a string potentiometer around the steering column. I'm logging at a higher frequency and resampling to 5Hz to get rid of noise.

It meant no disassembly, no risk of DTCs or upsetting the standard ABS/EPAS/ECU, no hassle of working out sensor type or pin outs, no hassle of working out the conversion from volts/bits to real units, no hassle of integrating OBD (all of the signals from the sensors you mentioned are available from the ABS/ESP ECU via diagnostic connector) or risking insufficient & variable sample rate.

Annoying to add weight/clutter by duplicating existing sensors/wiring, and you need to do your own signal filtering/cleaning.


If you do decide to investigate reusing existing sensors, you need to probe each pin of the connector with a multimeter whilst moving the sensor. If it's analogue the voltage will change somewhat consistently and proportionally when you move it, if it's digital or on a bus you'll probably need an oscilloscope to determine which.

Thanks again! The aim pdm actually has a yaw rate sensor inside so theres a start, if mounting/space is possible by the original sensor that is, its an e90 and want to keep everything as untouched as possible. I could use their rotary steering sensor too but I wouldn't consider this if I didnt have custom column internals that get rid of the play the standard columns have, as I want to know exactly how much the rack is moving compared to the steering wheel. I would like to be able have the option of fitting the original column back in to test exactly how much mine improves the steering but I'll think about that at a later time.

"Annoying to add weight/clutter by duplicating existing sensors/wiring, and you need to do your own signal filtering/cleaning."

Yes I dont want to have to splice or change the original wiring if I can help it, a 'stand alone' system would make things much easier for me and allow me to change the setup or remove/move to anther car if I needed to without much fuss.

 

[Robbie Corbett]

If this isn't a permanent fixture, so just for testing it would be relatively straight forward to do this with an arduino/esp32 dev board. You could quite happily log 3 sensors at >100Hz and store data onto an SD card (or stream live).

Over the years I have made several dataloggers for non real time stuff and almost always do signal processing on the computer, so other than simple rc filtering on the logger it doesn't do anything smart, it simply measures a voltage and adds that reading to a txt file.

If the txt file it makes is sensible you could just open it in excel and do any filtering/data display you need there.

However devils advocate... if you are wanting to understand the relationship between steering angle and rack travel..... you could just do this with some tape stuck the the steering wheel and a pencil stuck to the dash (ref zero) to give you wheel angle and a ruler to measure rack travel - it would take you 30-40mins to take maybe 100 readings which would be more than enough to make a nice excel graph...

 

[Beauvais Motorsport]

If this isn't a permanent fixture, so just for testing it would be relatively straight forward to do this with an arduino/esp32 dev board. You could quite happily log 3 sensors at >100Hz and store data onto an SD card (or stream live).

Over the years I have made several dataloggers for non real time stuff and almost always do signal processing on the computer, so other than simple rc filtering on the logger it doesn't do anything smart, it simply measures a voltage and adds that reading to a txt file.

If the txt file it makes is sensible you could just open it in excel and do any filtering/data display you need there.

However devils advocate... if you are wanting to understand the relationship between steering angle and rack travel..... you could just do this with some tape stuck the the steering wheel and a pencil stuck to the dash (ref zero) to give you wheel angle and a ruler to measure rack travel - it would take you 30-40mins to take maybe 100 readings which would be more than enough to make a nice excel graph...

Its so I know the angle of the wheel compared to the actual rack travel, how much play or roughly how much the torsion bar is twisting in certain driving conditions. If I can take the signal of the original steering angle sensor then I can put the original column back and compare the 2 and have an actual figure of how much my column improves the steering by.

See you guys are smart, but I'm.. Smart in my own way 😅 So I think im stuck with the option that does more for me but for more money.

 

[Robbie Corbett]

Ahh, thats why you want the yaw?

 

[Beauvais Motorsport]

Ahh, thats why you want the yaw?

Another data point to compare too! E.g. While turning lateral accel stays linear yet steering wheel to rack difference decreases slightly (rear end turning more), compared to lateral accel stays linear yet steering wheel to rack difference increases slightly (loading up the front end). You could say well just look at steering wheel angle only and throttle position, but this gives more accuracy and more data is always better, probably.. 😅