Thread: Nanocom data spikes.

Originally Posted by Aussie_Oz_Born

​Hi All, I'm wondering if anybody else is getting data spikes in their CSV files when recording data using a Nanocom?
I thought it might be a loose connection somewhere at the diagnostics port, but not all the values jump and dip at the same time. I'm hoping my ECU isn't on the way out!
I have attached the .xlxs files (can't upload .csv files) if you'd like to see the spikes.
I'm using a Nanocom Evolution with my 1999 TD5 defender. -Cheers.

I wouldn't worry about it too much mine is exactly the same and has been like it for years
Was never really sure if it was the dodginess of the Nanocom or the ECU but they seem to do this

Ok, cheers. You don't notice any engine stutter around the time the spikes occur? It's probably only my imagination. You have to wonder if it is the ECU whether it reacts to the spikes or not.

If it is all running smoothly I wouldn't worry too much.

i've had the same problem in times past and it's pretty simple to write a formula in excel to filter the spikes out and then plot the graphs at a meaningful scale.

cheers

Steve

Originally Posted by Rok_Dr

If it is all running smoothly I wouldn't worry too much.

i've had the same problem in times past and it's pretty simple to write a formula in excel to filter the spikes out and then plot the graphs at a meaningful scale.

cheers

Steve

Ahh, great idea with using a formula to filter the spikes out! Thanks very much.

-Aaron.

It is a sad fact that the TD5 ECU does indeed very occasionaly reply to requests for dynamic data, as used in the instrument mode, with values that are clearly wrong and so in logging, causes occasional spikes etc.
This may be due to a lack of processing power in the ECU or it being overloaded with doing its normal job while also having to process repeated Diagnostic requests.

We did think about trying to automatically correct this, however that would mean putting pre determined limits on each value and distorting / changing the real results which in some cases might mask real problems.
We also cannot just remove an errant value without distorting the Time line.

In the end we decided to provide and log the values the ECU provides, exactly as is, warts and all, for all to see. After all it is hard enough to provide equipment without our having to also find ways to overcome problems and errors that are not of our making, although you would be surprised as to how often we do have to do this.

In this case, we have left recognising and if desired, fixing these errant values supplied by the TD5 ECU by editing the CSV data, down to the user.

Further proof that this is caused by the TD5 ECU rather than the nanocom is that the half a dozen other EMS ECU's the nanocom covers and has Instrument mode & recording of values, do not have these anomolies / spikes.

Originally Posted by BBS Guy

It is a sad fact that the TD5 ECU does indeed very occasionaly reply to requests for dynamic data, as used in the instrument mode, with values that are clearly wrong and so in logging, causes occasional spikes etc.
This may be due to a lack of processing power in the ECU or it being overloaded with doing its normal job while also having to process repeated Diagnostic requests.

We did think about trying to automatically correct this, however that would mean putting pre determined limits on each value and distorting / changing the real results which in some cases might mask real problems.
We also cannot just remove an errant value without distorting the Time line.

In the end we decided to provide and log the values the ECU provides, exactly as is, warts and all, for all to see. After all it is hard enough to provide equipment without our having to also find ways to overcome problems and errors that are not of our making, although you would be surprised as to how often we do have to do this.

In this case, we have left recognising and if desired, fixing these errant values supplied by the TD5 ECU by editing the CSV data, down to the user.

Further proof that this is caused by the TD5 ECU rather than the nanocom is that the half a dozen other EMS ECU's the nanocom covers and has Instrument mode & recording of values, do not have these anomolies / spikes.

Thanks Colin, I thought I may of had an excuse to swap out my MSB ECU for a NNN type..... Love the Nanocom by the way.

Thanks also Colin

It’s good to get confirmation of what I have felt for a long while, that this is unfortunately normal behaviour, which does not effect the operation of the vehicle.

I also totally agree with your position that the instrument only presents the raw data to the user, warts and all and it is the users responsibility to further process/clean the data so that they can interpret and derive meaning from it.

As to the removal of outliers (aka spikes) it’s a relatively simple exercise, computationally speaking, to remove the spike, while leaving the readings either side untouched and preserving the row integrity. It’s something I commonly do, to clean up instrument data feeds and large data sets of multi element chemical analyses, before undertaking analysis of the data set. Audio engineers have been doing it to multi-channel data for years (google median filter) and a formula can be easily created in Excel.

Cheers

Steve