Mazda-speed.com
July 04, 2020, 07:32:18 PM *
Welcome, Guest. Please login or register.

Login with username, password and session length
News: Mazda-speed.com is no longer open and is now a read-only site. For Mazdaspeed Miata discussions, please visit the Mazdaspeed section on Miata.net.
 
   Home   Help Login Register  
Pages: [1]   Go Down
  Print  
Author Topic: Knock Thresholds in Hydra 2.7 R84 and later  (Read 11691 times)
0 Members and 1 Guest are viewing this topic.
rotorglow
Toyotae non carborundum.
Fifth Gear
*****
Posts: 2699


« on: April 21, 2015, 06:15:23 PM »

As a follow up (kinda) to the discussions of how/if knock control works in Hydra 2.7 R84 (here and here).....

I've been digging into this with Jeremy and Hydra lately, as I've been getting some audible knock after installing the 3.63 gears a month or so ago.  Essentially, with the taller gearing, the engine is at peak torque and peak boost for longer than with the shorter gearing, and some of my timing in the 3500-4500 range at high boost is a little too aggressive.

It turned out that the Hydra wasn't doing knock retard like I was expecting (by which I mean, at all). The short version of "why" is that with the knock detection strategy introduced in R84 about a year ago, the knock threshold has to be much MUCH lower than we thought.  I'm currently experimenting with values in the .04V (yes, four hundredths of a volt) to .10V range underload, and am finally getting knock retard to occur, and around .14 to .18 near idle.  The old settings of ~.4V and up (tracking with increasing RPM) are way too high.  Even Jeremy's values of .20V (suggested in his January 2015 email) are much too high for my car.  

I've also bumped up my "knock retard per event" to 5 (from 1.5), and pulled down my excessive knock threshold to about .08V (from about half a volt).  Further testing will reveal if I end up on the backup spark map with those settings.  

As discussed before, the reason the knock signal is so low now is that "knock ramp" variable shows the processed voltage from the knock sensor, after the Hydra removes background noise and anything outside of a cylinder's knock window.  For reference, the raw knock sensor voltage, that we were all a little confused about no longer seeing, as discussed in one of those threads, is available on the "left knock sensor" variable.  That will look more familiar, but isn't what the Hydra pays attention to for knock retard/enrichment.

What hasn't been well-documented (as far as I can tell) is the specific impact on recommended knock thresholds the change makes.

There're a couple of things that are a little distressing about all this: the new values need to be anywhere from one half to one tenth of the old ones.  That's a pretty big adjustment about which Hydra hasn't been terribly forthcoming.  I'm working with Ken Hill on some remote adjustments to the spark map that he dialed in for me last summer (on the old gears) and he said he only found out about the magnitude of this change recently.

I'm also not really psyched that the Hydra now compresses the usable resolution of the knock detection system into a range from 0.0 to maybe 0.15 volts. That doesn't give much of a spread.  And it throws overboard the old way of a figuring out a noise floor for an individual engine, so you really have to hope that their signal processing algorithm is working properly and quickly enough.

I haven't seen any of this discussed in detail lately, and most people might still be on older firmware, so it might not apply.  But if you're on R84 or later, it'd probably be good to take a look at your knock thresholds. And if you've gotten new gears since getting tuned, double check to make sure the timing in some of the load points you now linger in isn't too aggressive.
« Last Edit: April 21, 2015, 06:17:52 PM by rotorglow » Logged

2004 VR Sport: FM Big Enchilada w/ Hydra 2.7 + DeatschWerks 700cc injectors + FM crossflow rad + 3.63 rear gears + ES poly diff bushings + comp motor mounts--Koni Sports + Ground Control 550/375 + Racing Beat 1.125" bar + blocks & bolts--Wilwood/AX6 brakes--15x8 6ULs + RE-71Rs--Speedhut + Clearwaters + Auxmod + other misc electrics
rotorglow
Toyotae non carborundum.
Fifth Gear
*****
Posts: 2699


« Reply #1 on: April 22, 2015, 06:17:52 PM »

Forgot to mention that at mid/high rpm, the knock signal spikes will happen so quickly that logging at lower than 48Hz probably won't pick them up.  I usually prefer to log at 12Hz so I can capture more drivetime, and the tiny voltage spikes that the Hydra reacts to weren't showing up at that rate.  Everything looked suspiciously quiet, even though near idle there was some noise filtering through.
Logged

2004 VR Sport: FM Big Enchilada w/ Hydra 2.7 + DeatschWerks 700cc injectors + FM crossflow rad + 3.63 rear gears + ES poly diff bushings + comp motor mounts--Koni Sports + Ground Control 550/375 + Racing Beat 1.125" bar + blocks & bolts--Wilwood/AX6 brakes--15x8 6ULs + RE-71Rs--Speedhut + Clearwaters + Auxmod + other misc electrics
barmato
Fifth Gear
*****
Posts: 743


« Reply #2 on: April 25, 2015, 03:40:01 AM »

I'm a Hydra newbie so forgive me if I sound clueless, I almost am.  I logged my knock ramp as well as a few other parameters and I do see some spikes on the knock voltage up to about 0.8V.  With the old knock sensor methodology you would figure out the noise background and set the threshold just a little higher.  With this new methodology are any voltage spikes necessarily knock?  I don't hear knock but that doesn't mean that there isn't any.  My knock threshold is set at 0.2V across the board so my little spikes weren't enough to retard the timing.  I'm not sure if I have a problem or not.  I agree, I wish there was more information about this out there.  I'm glad you're asking questions. 

http://img.photobucket.com/albums/v344/barmato/Miata%20Stuff/Hydra%20Log%204-24-15_1.jpg

Logged
rotorglow
Toyotae non carborundum.
Fifth Gear
*****
Posts: 2699


« Reply #3 on: May 14, 2015, 08:37:08 PM »

I'm a Hydra newbie so forgive me if I sound clueless, I almost am.  I logged my knock ramp as well as a few other parameters and I do see some spikes on the knock voltage up to about 0.8V.  With the old knock sensor methodology you would figure out the noise background and set the threshold just a little higher.  With this new methodology are any voltage spikes necessarily knock?  I don't hear knock but that doesn't mean that there isn't any.  My knock threshold is set at 0.2V across the board so my little spikes weren't enough to retard the timing.  I'm not sure if I have a problem or not.  I agree, I wish there was more information about this out there.  I'm glad you're asking questions. 


You mean 0.08V, right?

Not every spike is knock; I get a lot of them in vacuum, which just can't be knock.  And in boost/at higher rpm, the spikes can happen too quickly to show up on a log.  That makes it nearly impossible to catch inaudible knock and set the KT precisely, and it makes me suspicious of the detection strategy overall. 

Knock history has been useful.  Take a look at your ignition table while connected to the Hydra; the cells where knock supposedly happened will be highlighted orange.  You can clear the knock history via Tools --> Knock Detection --> Clear Knock History.

Ken, Jeremy and I are continuing to explore all of this.  A threshold of .2V is proving to be too high on my car.  I'm around .1 above idle, with the amplifier turned off.  This has solved a lot of false knock retard near idle, but the threshold is still too low, and I still get some KR in vacuum a little further up.  So occasionally there's a little stumble as the timing gets yanked back.

My current thinking is that because knock threshold is a 2D map (voltage vs rpm) the Hydra can't distinguish between false knock (in vacuum) and real knock (in boost) at the same engine speed with a reasonable knock threshold.  I had the thought that if knock threshold was a  3D map (voltage vs rpm vs MAP), the "false knock" in vacuum could be more accurately ignored while still keeping the knock threshold appropriately low to catch the real stuff (that can't always be heard).  Jeremy is asking Hydra to do this, or to allow for a MAP floor below which the box won't react to sound that can't be knock (because it's in vacuum).

That would help, but still feels like a workaround for a feature that doesn't work quite right.  And it could take a while to be implemented.

I'm going to play around with an old firmware this weekend.  If there's one that implements the chattering AC relay fix, AND the fan/AC clutch delays, but doesn't have the per-cylinder knock detection, I'm going to give it a good hard look.  In the meantime, I still have the engine turned way back while we figure things out.

Logged

2004 VR Sport: FM Big Enchilada w/ Hydra 2.7 + DeatschWerks 700cc injectors + FM crossflow rad + 3.63 rear gears + ES poly diff bushings + comp motor mounts--Koni Sports + Ground Control 550/375 + Racing Beat 1.125" bar + blocks & bolts--Wilwood/AX6 brakes--15x8 6ULs + RE-71Rs--Speedhut + Clearwaters + Auxmod + other misc electrics
davin
Third Gear
***
Posts: 212


« Reply #4 on: May 15, 2015, 01:20:49 PM »

I have almost the exact same data log with mine. I would have a 0.04v spike just after I push in the clutch after going full throttle in third or fourth gear. I had mild audible knock when I first installed the hydra and set the base timing to the wrong mark, about 10 degrees more than it should have been. It sounded like a plastic bag ruffling. The hydra never recognized it but at that time I had it set to 0.4v with the amplified signal. Since I have fixed the timing and set the knock threshold to 0.06v I have not herd any noise or read anything over 0.04v. 
Logged
barmato
Fifth Gear
*****
Posts: 743


« Reply #5 on: May 15, 2015, 10:55:50 PM »

Yes, .08V not .8V, sorry.
Logged
davin
Third Gear
***
Posts: 212


« Reply #6 on: May 16, 2015, 05:40:26 AM »

I don't know if it's worth mentioning, but when I had the wrong timing adjustment and I used an octane booster "31 points" in a 93 octane tank. I had absolutely no knock noise and the car was scary fast in third gear. It felt much faster than my friends M3 E46 I drove from Florida to California. I wish I would have recorded knock voltage then, but I do have other variable data logs of it, if anyone wants to look at them. I can not wait to get the FM2 turbo upgrade, to get that feeling again. Honestly it was put the car in third gear at 3,500 rpm cruising, quickly but smoothly push the gas to the floor, look down to see 10.5 psi on the boost gauge, then everything that was around me was completely gone. it was an amazing feeling.
Logged
rotorglow
Toyotae non carborundum.
Fifth Gear
*****
Posts: 2699


« Reply #7 on: May 16, 2015, 09:07:35 PM »

I don't know if it's worth mentioning, but when I had the wrong timing adjustment and I used an octane booster "31 points" in a 93 octane tank. I had absolutely no knock noise and the car was scary fast in third gear. It felt much faster than my friends M3 E46 I drove from Florida to California.

Well, sure.  You were accidentally tuned for higher octane fuel.   If you'd had the correct timing, it wouldn't have done anything.
Logged

2004 VR Sport: FM Big Enchilada w/ Hydra 2.7 + DeatschWerks 700cc injectors + FM crossflow rad + 3.63 rear gears + ES poly diff bushings + comp motor mounts--Koni Sports + Ground Control 550/375 + Racing Beat 1.125" bar + blocks & bolts--Wilwood/AX6 brakes--15x8 6ULs + RE-71Rs--Speedhut + Clearwaters + Auxmod + other misc electrics
Pages: [1]   Go Up
  Print  
 
Jump to:  

Powered by MySQL Powered by PHP Powered by SMF 1.1.21 | SMF © 2015, Simple Machines Valid XHTML 1.0! Valid CSS!

Copyright 2007-2016, Eunos Communications LLC