General Tech Good at troubleshooting? Have a non specific issue? Discuss general tech topics here.
Sponsored by:
Sponsored by:

2005 CBR 600F Track version

Thread Tools
 
Search this Thread
 
  #21  
Old 09-05-2022, 03:46 AM
Mi_ka's Avatar
Member
Join Date: Aug 2022
Location: Thessaly, Greece
Posts: 38
Received 11 Likes on 10 Posts
Default

Originally Posted by dannoxyz
MSD or J&S unit with more advanced knock-detection algorithm: J&S Electronics SafeGuard Indivdual Cylinder Knock Control . These are based on Porsche KLR ignition-piggyback used on their Turbo models.
hmm, 1991?
This must be earlier than the SAAB knock sensing system based on monitoring the plugs before sending spark. My former 2003 Daihatsu Sirion (Storia) 1.3 has the same system.
So Porsche and not SAAB developed it originally?
 
  #22  
Old 09-05-2022, 06:31 AM
dannoxyz's Avatar
Senior Member
Join Date: May 2011
Location: Mesa, AZ
Posts: 1,987
Received 432 Likes on 391 Posts
Default

Porsche first used it on their 1985 944 Turbo (951) model. KLR computer was both electronic boost controller and ignition piggyback on top of Motronic 3.1 DME. Second generation 1989 911 Turbo (965) also used similar system. KLR was fed ignition signal from main DME computer and relayed it back to tell DME when to dump ignition coil to fire spark. Under most normal operation, ignition-trigger was reflected back immediately and spark fired at pre-programmed time in DME map.

KLR monitored TPS position and RPM to look up 3D table of pre-programmed boost levels. When there was positive manifold pressure from boost (using its own MAP sensor, while DME used AFM), it would hang onto ignition-trigger to add retard amount of time. Then relay back to DME to fire spark. This has effect of overlaying retarded correction map on top of DME's native ignition map.

KLR also monitored knock-sensor and incremented internal knock-countre. It also tickled knock-detected output line if you wanted to have external alert device such as light or buzzer. When knock count severity threshold was reached (combination of amplitude and frequency), it would retard ignition 3.1 degrees. If knock count remained high on next engine-rotation, it would retard ignition additional 3.1-degrees. Cycle would continue up to maximum of 33-degrees total ignition retard. Some overzealous programmer probably set this amount as no more than 10-12 degrees is usually needed to stop detonation within 4 rpms.

At same time KLR was retarding ignition due to knock, it would also reduce boost levels. This combination of boost-related ignition-retard, knock-based retard and boost-reduction kept engine safe and reliable.*

All this code was cleverly locked inside KLR's MCS48 CPU ROM with read-only bit set. It will not write any data to output lines when queried. All code executed and data used stayed internal to CPU. It was mysterious black-box for many years, no one really knew how it worked. Then some clever guys in Germany managed to unlock CPU's memory by mechanically shaving off chip's casing and exposing the silicone wafer inside. Using electron-microscope, they decoded memory-locations and read-only bit on wafer. Then they zapped read-only bit location with laser to flip its state!!! Now data-lines on CPU can export all its secrets!!!

It was increasing boost-levels by re-programming KLR's boost-map (along with matching fuel & ignition maps in DME) that I got the extra 100bhp! All modern hot-rodding's done in software!



BTW - Fun fact, boost-control solenoid used in Porsche 951 was same unit used in Ferrari F-40. Even though that car used Weber-Marelli EFI system. Also same Brembo front-brakes were used on both as well.

* Electronic boost control also tamed wild on/off boost behavior compared to earlier 1977 911 Turbo (930) which used mechanical boost controller. Crash from on/off powerband caused a death where Porsche was found at fault and got 930 banned from U.S. Only here can you successfully claim in court that Porsche built car so diabolical and so eeevil, that it would intentionally kill its owner! Medical malpractice lawsuits also started getting out of control around this time. Leading to doctors having to pay 1/2 of their income for malpractice insurance today.
 

Last edited by dannoxyz; 09-05-2022 at 07:18 PM.
The following users liked this post:
Mi_ka (09-05-2022)
  #23  
Old 09-05-2022, 09:05 AM
IDoDirt's Avatar
Super Moderator
Join Date: Aug 2006
Location: South Florida, USA
Posts: 5,302
Received 498 Likes on 450 Posts
Default

Originally Posted by dannoxyz
Then some clever guys in Germany managed to unlock CPU's memory by mechanically shaving off chip's casing and exposing the silicone wafer inside. Using electron-microscope, they decoded memory-locations and read-only bit on wafer. Then they zapped read-only bit location with laser to flip its state!!! Now data-lines on CPU can export all its secrets!!!
Yea, that's a fun story, but I think I'm going to call BS on that one.
 
The following 2 users liked this post by IDoDirt:
dannoxyz (09-06-2022), Mi_ka (09-05-2022)
  #24  
Old 09-05-2022, 12:34 PM
dannoxyz's Avatar
Senior Member
Join Date: May 2011
Location: Mesa, AZ
Posts: 1,987
Received 432 Likes on 391 Posts
Default

Originally Posted by IDoDirt
Yea, that's a fun story, but I think I'm going to call BS on that one.
heh, heh... That's story they told me at time and I'm sticking to it!!! 🤣👍
 
The following 2 users liked this post by dannoxyz:
IDoDirt (09-05-2022), Mi_ka (09-05-2022)
  #25  
Old 09-05-2022, 03:49 PM
Mi_ka's Avatar
Member
Join Date: Aug 2022
Location: Thessaly, Greece
Posts: 38
Received 11 Likes on 10 Posts
Default

Hey, who knows, maybe some director at the Max Planck institute wanted to hot rod his car and really wanted to see if that multi million Deutsche Mark laser zapper was any good for real world applications... do not forget we are talking 3.5 micrometers back then, not nanometers.

However I guess that they got just the right temptation to some nerd R&D engineer at the right company post to forget a demo chip unlocked or so....

By the way, Sir @dannoxyz , I am impressed
 
The following users liked this post:
dannoxyz (09-05-2022)
  #26  
Old 09-05-2022, 07:09 PM
dannoxyz's Avatar
Senior Member
Join Date: May 2011
Location: Mesa, AZ
Posts: 1,987
Received 432 Likes on 391 Posts
Default

Why thank you!
 
The following users liked this post:
Mi_ka (09-06-2022)
Related Topics
Thread
Thread Starter
Forum
Replies
Last Post
letsride
CBR 954RR
9
06-09-2023 02:39 PM
BoRkA
F4i - Main Forum
9
08-22-2007 08:42 AM
OldFatGuy
Track Days & Riding Schools
24
06-20-2007 04:16 PM
Nick I.
CBR 600F2
1
09-19-2006 01:50 AM
OldFatGuy
Track Days & Riding Schools
12
09-13-2006 08:23 PM



Quick Reply: 2005 CBR 600F Track version



All times are GMT -5. The time now is 07:20 PM.