normalized knock mode

[k fuku]

 

トリムノックモードから正規化ノックモードに移行しています。

ノック周波数などの基本的な部分は問題ないのですが、その下のトランジェント領域の設定の微調整に苦労しています。

ログを見てアドバイスをいただければ幸いです。

 

 

https://drive.google.com/file/d/1kV6uyL66ylAs2UdcXfdtVCzXW_GkKvy-/view?usp=drivesdk

 

 

 

 

[Adamw]

I think there is too much filtering on the engine speed and TPS, can you set engine speed to 10Hz and TPS to none filtering and do another log.  Also set RPM delta learning rate to Normal so I can see how that looks.

[k fuku]

2021/10/9の午後3時48分に、アダムは次のように述べました。

エンジン速度とTPSのフィルタリングが多すぎると思います。エンジン速度を10Hzに設定し、TPSをフィルタリングなしに設定して、別のログを作成できますか。また、RPMデルタ学習率を標準に設定して、それがどのように見えるかを確認します。

設定を書き直しました。明日ログを取ります。

 

[k fuku]

On 10/9/2021 at 3:48 PM, Adamw said:

I think there is too much filtering on the engine speed and TPS, can you set engine speed to 10Hz and TPS to none filtering and do another log.  Also set RPM delta learning rate to Normal so I can see how that looks.

Thank you very much for your patience.
We have acquired four types of data.

(1) Learning standard RPM filter 10
(2) Learning fast RPM filter 10
(3) Learning standard RPM filter 0
(4) High speed learning RPM filter 10
TPS filter 10

Thank you very much for your cooperation.

 

https://drive.google.com/folderview?id=1V8ZIQZNdK8ZhAzHqu5QfWl561guXW30V

 

[Adamw]

To me the log with learning rate normal and RPM filter at 10Hz looks the best.  But the RPM ROC is still very unstable and noot very useful, I suspect this is your low cylinder count combined with the low tooth count cam trigger not giving accurate engine position.  Can you do another log with learning rate normal and RPM ROC filter at 5Hz.  

You can see below, engine speed is quite flat, but RPM ROC is very unstable.  In this area RPM delta is keeping the threshold high but since engine speed is flat it should really be the MAP and TPS delta triggering the threshold gain.

[k fuku]

5 hours ago, Adamw said:

To me the log with learning rate normal and RPM filter at 10Hz looks the best.  But the RPM ROC is still very unstable and noot very useful, I suspect this is your low cylinder count combined with the low tooth count cam trigger not giving accurate engine position.  Can you do another log with learning rate normal and RPM ROC filter at 5Hz.  

You can see below, engine speed is quite flat, but RPM ROC is very unstable.  In this area RPM delta is keeping the threshold high but since engine speed is flat it should really be the MAP and TPS delta triggering the threshold gain.

At standard rotation speed learning speed
I've got 5k logs.
It seemed to be oversensitive.

I also tried logging at 2.5khz, 1khz, 0khz with the same learning speed.

https://drive.google.com/folderview?id=1qKBUtcyuTYEwwYzHyi6MbVEUMO2d1nm0