Hasn't been any real updates on the vems firmware page since 1.1.96 being the last official release. Whats planned for the next few release? When are the knock tables being released for vmes
Yes, there will be knock tables and lot of new functions.
When we could expect that?
Bump to that
It's all very quiet, must be big changes ahead, not even a new vemstune. Hopefully be bery interesting
do we only know whats coming from the programmers by the actual releases? Or is there a page that says whats being worked on somewhere?
1.2.0 and 1.1.99 are out (not release) but I don't see anything about Knock
On the firmware page it says there has been a functionality freeze whilst issues have been addressed which is far enough
1.2.0 or 1.2.1 will be the next release hopefully we will be targetting to install!
Quote from: MWfire on January 31, 2012, 09:33:47 PM
Yes, there will be knock tables and lot of new functions.
so when we are going to get it? :)
Exactly, when we can expect new functions with knock control ?
I say dont expect anything regarding knock control , then you Will
Not be disapointed ,
I would just love the VEMS to be able to run my Subaru engine , but the ecu lacks input to run the engine right ,
So for me it a change for another ecu ,
Rob did you ever get that Subaru running the syvecs ecu you purchased ?
Thanks
Skassa
Quoteexpect anything regarding knock control
It is hard to compete with the knock detection function from the manufacturer of the car ecu. That manufacturer has a work bench, huge clean data to analyse, experience, engineers, everything to make a sensible recognition algorithm adapted to the engine.
I think that a universal solution is unreachable by Vems or any other. I simply expect some tools which helps to define and set some pattern (frequency, level etc) common to those series where a knock as been detected by the driver, listening from headphones, and marked as so onto the log data.
Again I should remind: in Russia guys already use knock code for four years. Knock code was prepared by Andrey from Ekaterinburg. And I do not read any complaints about it in their forums. They just use it with their not so good fuel quality....
In official wiki some time ago was 1.1.89 version. I knew who use this code with success. In addition those configs what I see with complaints all was with wrong settings under knock section.
Andrey has also much more stable engine phase angle reading algorithm what VEMS use for injector angle and cam control. Official firmware has some delay for one cycle or one revolution (!) what makes both functions bad at low rpms or high accelerations. (look at VT raeding for sec trigger angle under free rev - it floats a lot!) His reading stays constant like glued (I mean engine with fixed cam, not VVT)
In addition his firmware mod controls automatic transmission on their RHD Toyotas http://www.tourerv.ru/ widely used in those area.
As I knew he is welcome to share his code with developers. But as we see something did not happening there :(
Gints
The official line was the knock code caused the ignition timing to be incorrect? When can we expect to get all this back into vems? it seems weird the code is ready but not in a firmware for use
Quote from: Jamo on June 14, 2012, 01:55:21 PM
The official line was the knock code caused the ignition timing to be incorrect? When can we expect to get all this back into vems? it seems weird the code is ready but not in a firmware for use
That's right!
Does Andrey want to share his FW/settings. has he implemented it in newer FW Versions? We have also crap fuel here :)
Quote from: GintsK on June 14, 2012, 11:53:32 AM
Andrey has also much more stable engine phase angle reading algorithm
Indeed I noticed that the phase angle speed or rpm is too much noisy for a physical value. I guess (!!) that value is given through an instantaneous and simple computation of the time derivative of the phase angle.
Next, possibly, these values are averaged to make a reference fo the next revolution.
Equally we havegot some delay for the engine to start because during this starting mode the computed speed is much changing especially when we have few triggers.
If this is the case, this is an old story from the sixties ( I was dealing with dynamic ship positionning systems some years ago) People use Kalman filters (or so) to mild the data with a mixed of prediction based on a simple physical model (one mode) and of measured data, or even a serie of those filters with the combination parameters being scaled according to the speed.
Kalman was American, but Russian know the same technic very well.
Philippe
Still no news on knock alternative? I've begun looking around at what other ecu's offer in terms of knock detection and vems is lacking at the moment. Have to say i'm tempted to switch ecu
Quote from: Jamo on July 19, 2012, 10:45:55 PM
Still no news on knock alternative? I've begun looking around at what other ecu's offer in terms of knock detection and vems is lacking at the moment. Have to say i'm tempted to switch ecu
Quite usefull for all cars (motorsport and daily driven. Instead of iButton and stuff that almost nobody uses why don't they concentrate on this?
Quote from: billman on July 20, 2012, 04:31:47 PM
Quote from: Jamo on July 19, 2012, 10:45:55 PM
Still no news on knock alternative? I've begun looking around at what other ecu's offer in terms of knock detection and vems is lacking at the moment. Have to say i'm tempted to switch ecu
Quite usefull for all cars (motorsport and daily driven. Instead of iButton and stuff that almost nobody uses why don't they concentrate on this?
Really hope that developers prioritize Knock sensing. Here I live we have to drive 70 km to buy 98 octane gasoline. Is too little sales on 98 octane they say. :-(
Still no knock alternative code? Some news?
Do really want to try knock alternative :)
Emil asked me a few weeks ago if i had time to put a car on the dyno and torture into knocking, so i guess there´s something going on.
fw version 1.2.9 offered a new boostcontrol algorithm. Does someone know what was wrong with the previous one and the new performances expected from this one.
Does it only deals with possible overshot issue? How does it solve that case?
Of course, nice to know the test cases !!
The new boost is really cool.
The reference table is Boosttarget based, so when its tuned you´ll only need to change boost target and the ref-dc follows (good for anytrim etc..)
Also a new spoolup function, you can choose to max the boost-dc until it closes up to target.
And also a few more stages to give excellent boostcontrol. (PD only and then DI only with some overlapping).
The old strategy is still avaliable if someone doesnt like the new one
From description it seems now we have have possibilities to chose when which PID component comes in action depending from actual/target condition.
IMHO previous main problem was I component: it should be small or strictly limited. Because of that bias DC table was always below actual DC at target. I came in action early and made actual DC already disturbed during boost build. But we want I acting during close to target situations. Anyway boost was tunable very nicely and fast (with good hardware).
I think now we have very useful improvement.
Gints
Does it have the 3d P term I was wanting?
I´m happy it has the BoostTarget vs rpm DCref table I wanted, and which I believe will vastly improve boost targetting for almost all scenarios.
I haven´t had time to hook up a ecu and check it out yet.
QuoteI haven´t had time to hook up a ecu and check it out yet.
Neither did I. I spent some time on the bench with the previous algorithm and I have not yet fully understood its logic (2 sides refDC, minboost_minpressure,.) or vocabulary.
Indeed what we need is to know more about the validation tests used, or at least, config+ log data for 2 or 3 reference cases running standard boost control.
ID control is a new way which as Erikk said can be used close to the boost target to counter-react the fast increase of the boost, although the P action is usually much smaller that D action when entering that zone and that without P action to damp the motions, you get overlaps, or if D can be set exactly for a nice landing in some cases, a non robust regulator. When rallying, outside this regulation zone, of course I is shut and D action is useless
so how knock testing are going?
Quote from: Kamuto on February 23, 2013, 09:29:48 PM
so how knock testing are going?
I am also excited about how the knock testing are going?
Have an old Volvo B230k motor, which I can torture with Knock firmware :)