Honda / Acura K20a K24a Engine Forum banner

1 - 20 of 28 Posts

·
Registered
Joined
·
81 Posts
Discussion Starter #1 (Edited)
Disclaimer at the beginning, I don't have any of the parts for this analysis in hand, so all the weights come from findings on various forum posts and sales literature.

The Honda K-series uses a type II valve train design, this is an Overhead Cam (OHC) using an end pivot rocker arm:


The first thing to do is to get a handle on the various masses in the system to include the valve, keepers, retainer, spring, and rocker arm. These are all of the masses that the valve spring has to push against to maintain control of the valve position. Because F=M*A, lowering the masses means you require less force (spring pressure) to maintain control, which is always a good thing.

Stock intake valves are at 48g, exhaust at 44g.
Retainers for both are in the 14g range (I've seen 13-15g depending on design)
Keepers are 1.5g according to the Manley catalog
Stock springs (06 RSX-S, not a 200hp motor, if anyone knows the dual or single spring masses please let me know!) are 53.0g on the intake and 55.4g on the exhaust.
And lastly, the factory VTEC rocker comes in at a whopping 317 grams.

Now the valve, retainer and keepers all move together the entire distance from closed to max lift, however the springs and retainer both have portions that don't move at all, so we need to determine an 'effective mass' for these two.

For a standard coil spring, the top of the spring moves the entire lift distance, but the bottom of the spring doesn't move at all. These springs have a linear rate, which means they also have a linear mass distribution from top to bottom. This means the spring gets a factor of 0.5 so the effective mass of the stock springs is 26.5g and 27.7g respectively

The rocker arms are a bit more complex. Ferrea states the stock rocker arm has a total mass of 317g, and a "fulcrum mass" of 136g. To get the effective mass we need the rocker ratio (using 1.75 for the VTEC lobe) and the relative position of the center of gravity:

Mass Effective = Mass Total * ( Distance from pivot to CofG )^2 / ( Distance from pivot to valve tip)^2

I got a little creative using the information at hand and came up with an effective rocker mass of only 58.3 grams, quite a bit more manageable than 317 but still definitely a dominant mass in the system by far.

So the total effective mass for the stock valve train is 238.3g on the intake and 232.7g on the exhaust. These masses are able to be controlled effectively using the OE springs with a seat pressure of 55 lbs, and a 147 lbs over the nose at 0.483" (12.27mm) lift.

Aftermarket flat faced valves seem to come in a few grams heavier than stock. I also found a source quoting Supertech SPR-H1021D springs as 68.5g for the intake and exhaust, and their titanium retainers for the dual springs are commonly quoted at 7g. This means an aftermarket valve train has a mass of 150g intake and 146g exhaust. Slightly more mass, but the new springs have a seat pressure of 95 lbs, and over 250 lbs over the nose, depending on the lift of the aftermarket cam installed.

It looks like the hopped up cams all require springs with something around 100 lbs on the seat and a spring rate of 300-340 lbs/inch. This is because those cams will get more aggressive with the acceleration rates on the valves. If you don't have enough spring pressure, the cam could leave the nose of the cam at peak lift causing valves to float, or the cam could rotate faster than the ability of the spring to move the system. When this happens the rocker loses contact with the cam and becomes uncontrolled, then the valves bounce off the valve seat during closing.

I've been looking into ways to reduce these masses, and I'll probably end up going to a beehive designs for springs, as they have a few benefits. First, the mass of the spring does not have a linear distribution with length. Beehive springs are much smaller at the top, which makes the parts of the spring that move the most lighter, so we can use a ratio less than 0.5. Also, the springs are generally lighter overall for the same reason. Lastly, the size of the retainer is smaller as well. PAC beehive springs are resold under the Supertech name, but they come in at around 45 grams per the PAC catalog, and the steel beehive retainers weigh about the same as titanium retainers for dual springs but without the chance of galling damage.

Spit balling an effective spring mass ratio of 0.35 (pretty big assumption), the beehive spring setup would yield 201.8g on the intake and 201.1g on the exhaust, a 14-15% reduction in effective mass on the valve train, which means either the ability to run lower seat pressures to reduce friction in the valve train, or a higher engine speed before valve train damage if you miss a shift.
 

Attachments

·
Registered
Joined
·
1 Posts
Created an account just now to make this post.

Very nice analysis you've got here - it's definitely rare to see something this technical on public forums! Maybe the next thing to look into is some cam profiles, and calculate the force/acceleration of the system to see what RPM the stock springs would begin to float at? Versus the lighter springs?

I guess I'll wait until you get some of the parts first, lol. Definitely interested in seeing future posts and analyses though.
 

·
Registered
Joined
·
81 Posts
Discussion Starter #3
I started to do this already using some RSX and TSX cam profiles I found in an image. I'm crudely fitting polynomial curves in excel to get the formula for a curve that I can then differentiate to get velocity and acceleration values. It's a painstaking process, and I'll be the first to admit it's a ROUGH approximation, but it's interesting to see. At the very least it's an order of magnitude to be able to understand what kind of forces the valve train sees.

The biggest challenge is getting data for camshafts. Manufacturers specify a duration and lift. If we're lucky duration is defined at 0.050" (or 1mm, 0.039"). Really you need lift per degree data from a cam measurement system to really compare velocity and acceleration curves.

Truth be told, there's a lot of different methodologies for developing cam profiles using acceleration and/or jerk limits and curve shapes. Without a measurement device and a few cams to look at, I doubt I'll be able to get that deep into the topic, but stay tuned!
 

·
Registered
Joined
·
81 Posts
Discussion Starter #4 (Edited)
Ended up buying a rocker arm on eBay for $20 to examine. I can confirm that it does have a median rocker ratio of approximately 1.75 for both the primary/secondary and VTEC cams. I say median because as the rocker moves, it's contact point against the valve tip changes. The rocker ratio at the valve opening is 1.66:1, and at max lift (for what I'm assuming is a stock K20 cam) the ratio gets up to 1.8:1. I've seen reference that the primary/secondary have a ratio of 1.7:1, which is probably valid because they don't reach the higher lift and rocker ratio that the VTEC profile will.



I have also weighed the rocker arm and had to change some calculations. The "fulcrum" mass is actually the end weight of the side that is not in fact the fulcrum, so effective rocker arm mass is a lot less than I had figured. I just realized today that there's a helper spring acting on the bottom of the VTEC link that I need to get a spring rate for as well. This spring keeps the VTEC link in contact with the cam, but it will have the effect of countering the rocker mass, but not the valve/spring/retainer/keeper mass.
 

Attachments

·
Registered
Joined
·
81 Posts
Discussion Starter #5
It's been a bit. I thought I accidentally deleted the file I was working with on my home PC when I rebuilt it, turns out it was hidden among some files on my work computer.

So I wanted to know what sort of magnitude of acceleration the valvetrain experiences. Without actual measurements from a camshaft, I'm using what I have available:


It looks like this image got compressed when I uploaded it, but what I did is fit three curves to the TSX Highcam. One curve for the linear ramp that takes up the valve lash (-90 to -30 degrees), a 2nd order curve for the major acceleration event (-30 to about 7 degrees), and a 4th order curve to approximate the nose of the cam (7 degrees to 210 degrees).

The polynomial curves allow me to do simple derivatives to determine velocity and acceleration of the rocker. Because the base curves are plotted as mm per degree crank angle, I had to do some math and include an engine speed to get meaningful acceleration values:



The three plots on the left are lift with respect to crank angle, and the three on the right are the same data converted to lift with respect to time (I used 7000 rpm because that would be peak power for the TSX highcam).

It's pretty obvious looking at the velocity curve that there's discontinuities where the three curves meet, but the goal for me is the acceleration plots. The first line has constant velocity, so no acceleration, but then there's the opening ramp with a very pronounced acceleration. This value will be roughly similar on the closing side, so try to imagine it mirrored over, like this:



So roughly 1250 m/s^2 peak acceleration at 7000 rpm. This value goes up to 1375 m/s^2 at 7300 rpm, the engine's redline. This is the acceleration that's acting on the mass of the valvetrain that I've already looked at, so you can start to get an idea of the force required by the valve springs to counter this acceleration, and get an idea of the safety margin the stock springs have. That'll be my next step.
 

Attachments

·
Registered
Joined
·
81 Posts
Discussion Starter #6 (Edited)
So I got a K24A2 head, RBB-1 so an early TSX head with 35mm intake valves. I spent an afternoon tearing it apart, cleaning and measuring component weights and spring rates. In case anyone is curious:

Intake valves: 50g
Exhaust valves: 43.8g
Keepers: 1g
Springs: 41.3g
Retainers: 9.6g

The spring rates for the intake and exhaust are progressive, you'll have to forgive me for the mixing of metric and imperial units, but the intake and exhaust springs started at about 6.5 lbs/mm at their installed height (0mm lift, 40.4mm spring length), increasing to 7.1 lbs/mm at 6.4mm of lift, and ending at 8.2 lbs/mm at 13.3mm lift (about 0.520") which is just before coil bind.

The helper springs which act on the center of the three rockers is fairly linear, coming around 9 lbs/mm on average.

As seen in the previous post looking at acceleration at the valve, the two largest peaks are during the intake and closing for the stock TSX cam. The more critical of the two is the closing peak, because this is where the springs have to overcome the mass of the valve and rocker arm to keep everything in contact and prevent the valve from bouncing or the intake from contacting the exhaust during overlap.

This would coincide with the range from 1-3mm of lift. At 1mm of lift, the TSX valve springs provide about 500N of force, and the helper spring contributes about 58N. At 3mm of lift the valve springs are at 640N and the helper at 85N. Dividing these forces by the effective mass of the valve/rocker assembly yields around from 2522 m/s^2 to 3280 m/s^2 at 1 and 3 mm of lift respectively.

That's quite a safety factor over the 7300 rpm approximation of 1375 m/s required at valve closing, although that's an average value, and the reality is there could be much greater peak acceleration hiding in there somewhere. Also, these accelerations increase exponentially with engine speed, so that safety factor could be eaten up in only a few hundred rpm which would probably be the real design limit of the engine.

Another thing I looked at was replacing parts of the rocker to lighten it. The rocker itself is a sintered metal injection piece and more than likely fairly well optimized. I'd hesitate to replace the lash adjustment screw due to the forces it sees. Titanium was initially a thought, but it would have needed a steel ball attached to it, and Ti has a fatigue limit, pretty risky. The last option is the lash adjustment nut. At about 1.9g you MIGHT find a fraction of a gram in going to an M7x0.75 titanium nut, but it would only represent a fraction of a percent of the reciprocating mass in the valvetrain overall.
 

·
Arouse the DAMPFHAMMER!
Joined
·
6,069 Posts
I started to do this already using some RSX and TSX cam profiles I found in an image. I'm crudely fitting polynomial curves in excel to get the formula for a curve that I can then differentiate to get velocity and acceleration values. It's a painstaking process, and I'll be the first to admit it's a ROUGH approximation, but it's interesting to see. At the very least it's an order of magnitude to be able to understand what kind of forces the valve train sees.
Hi Scider, I love the detail passion you have for the engine stuff. For your approach of designing cams or at least understanding cams there is a lot of math theory and engineering receipts available, e.g. from Prof. G.B. Blair, which is my favourite author concerning engine simulation and designing. I used it to program a 1D engine simulation SW. A rough but good approach for cam or valve lift profile approximation is e.g. shown in his book about 4 stroke engine simulation "Design and Simulation of Four-Stroke Engines" ISBN-10: 0768004403.

The biggest challenge is getting data for camshafts. Manufacturers specify a duration and lift. If we're lucky duration is defined at 0.050" (or 1mm, 0.039"). Really you need lift per degree data from a cam measurement system to really compare velocity and acceleration curves...Truth be told, there's a lot of different methodologies for developing cam profiles using acceleration and/or jerk limits and curve shapes. Without a measurement device and a few cams to look at, I doubt I'll be able to get that deep into the topic, but stay tuned!
Yuup, completely agree with you. I went through the process of designing a valve lift profile from a thermo-flow mechanical perspective, the translation into a cam profile is most complex, latest at the bending forces at higher engine speeds you need massive massive experience or massive experience and good tools to calculate a profile.

For me two approaches are more applicable:
  1. measuring a basis cam lobe profile, take it for development of understanding and designing the improvements
  2. Work out a valve lift profile for the specific application and ask a cam lobe design at a cam manufacturer like Schrick, and so on...or if you have luck like me I found an almost similar profile in the aftermarket.
I spent 2 years of design work into my DAMPFHAMMER engine development and around 2000 h into programming tools to be able to do so. But I would spent another 2 or 3 years to understand and design a proper cam lobe profile for a K-series application without student support from University, who write thesis within that project. I decided against it, because of the duration and the risk. The member jaydee went through that process and designed his own SC'able camshafts based on an existing design. May you can talk with him about it.
 

·
Registered
Joined
·
81 Posts
Discussion Starter #8
A lot of what I'm doing is applying lessons I've completed from my Masters program specifically to the K-series engines. It helps the information sink in when it's applied to a subject that I have interest in.

While I'm in this program, I do have access to Ricardo Wave through the university. I've toyed around with making a model, but there would be a lot of assumptions and I don't have any real idea where to even start with it. I've also thought about buying a simplified program like Performance Trends Engine Analyzer, but haven't committed to that yet.

One difficulty has been not actually having an engine or parts to measure, but that's slowly changing. Right now I'm focused on getting this RBB head on the flow bench. I'll come back to cams when I have something to degree.
 

·
Arouse the DAMPFHAMMER!
Joined
·
6,069 Posts
A lot of what I'm doing is applying lessons I've completed from my Masters program specifically to the K-series engines. It helps the information sink in when it's applied to a subject that I have interest in.
That must be an awesome opportunity for you to dive into the engine stuff (really, K-Series?). I hadn't that chance neither for my Diploma- nor my Doctoral thesis. That time University of Munich, one of the top Universities of Germany, didn't have a Prof. for Internal Combustion Engines for about 4 years...just some miles from the BMW tower away (Munich), not far from AUDI and MAN truck engines (Nürnberg, Augsburg) or one of the oldest engine companies of the world, the MTU (Friedrichshafen, I worked there for 9 years later on) :facepalm:. You must be really lucky about diving in stuff you have passion for :). So you are still in that Master Program?

While I'm in this program, I do have access to Ricardo Wave through the university. I've toyed around with making a model, but there would be a lot of assumptions and I don't have any real idea where to even start with it. I've also thought about buying a simplified program like Performance Trends Engine Analyzer, but haven't committed to that yet.
This is indeed the most challenging part, getting qualitative high value measurements. It cost me some 1000 Euro and uncounted hours of collecting and measuring it. Finally the valve lift profile is the most expensive part. Port flow is more easy. A lot of effort has to be put in understanding and simulating the combustion process of the K-series to be able to translate it into a well shaped Vibe heat release curve. I experimented with Vibe parameter calculation approaches, there are a lot of papers to that available, but non of them showed a proper fit as final free parameter to the collected reference torque curves. It wasn't easy to find the right set of parameters for combustion delay, efficiency of it and Vibe parameters. To much parameters, so I took much of the experience I've done with lean burn engines at the MTU, where I've designed combustion processes (like todays F1 style) among other stuff. Lambda 1 engines are a big difference here, but finally my set did work in some +-3 Nm for my reference curve, I had fully measured.

Give some attention to the wave behavior calculation when using Ricardo Wave. They offer different models. A simple acoustic model doesn't work in high end NA engines, what you need is a model which solves temperature also to include the steepening of waves because of different sonic speeds areas. I hope you understood my underdeveloped English :D.

One difficulty has been not actually having an engine or parts to measure, but that's slowly changing. Right now I'm focused on getting this RBB head on the flow bench. I'll come back to cams when I have something to degree.
I am looking forward to see more of your project Scider!
 

·
Registered
Joined
·
81 Posts
Discussion Starter #10 (Edited)
I am still in the Master's program, class of 2020 at the University of Wisconsin. Graduation will be December 2020 and it'll be a degree in Engine Systems. It's focused on exposing students to all aspects of engine development. It's a distance learning program that caters to all the players in North America (Ford, FCA, GM, Cummins, Mercury Marine, PACCAR, etc. The list is quite long).

I was fortunate to land in Detroit where I spent four years working for AVL as a technical sales engineer for single cylinder research engines and optical combustion analysis. Now I do gasoline engine mapping for Ford, an obscure term for developing the air flow and torque models (among several other things).

When things get slow at work I've been spending my time reading white papers (SAE, JSAE, ASME, NACA, JSTOR...) understanding various engine topics. Most recently intake and exhaust tuning dynamics. The exhaust side is fairly straight forward, but as you reference, the intake is quite complex, far too complex for simple calculations.

I'm waiting for my Exocet chassis to arrive, it's a week or two away so they tell me. Next year I'd like to see about swapping a K24A2, and then building a K24 with the target of 300 whp. The K-series offers a lightweight package that will turn big numbers without forced induction. Keeping it simple and light is my goal.
 

·
Arouse the DAMPFHAMMER!
Joined
·
6,069 Posts
I am still in the Master's program, class of 2020 at the University of Wisconsin. Graduation will be December 2020 and it'll be a degree in Engine Systems. It's focused on exposing students to all aspects of engine development. It's a distance learning program that caters to all the players in North America (Ford, FCA, GM, Cummins, Mercury Marine, PACCAR, etc. The list is quite long).
Sounds very interesting as some of them are competitors of my former company.

I was fortunate to land in Detroit where I spent four years working for AVL as a technical sales engineer for single cylinder research engines and optical combustion analysis. Now I do gasoline engine mapping for Ford, an obscure term for developing the air flow and torque models (among several other things).
So you know Prof. Spicher from KIT and owner of the optical investigation company MOT, who is pretty active in the field of irregular combustion phenomena like pre-ignition and knock? I saw a lot of issues with pre-ignition in the 14:1 CR engine, which is run at 1.88 of lambda and a MAP 4.5 bar...the highest CR in the market. I learned a lot with that issue, especially concerning turbulent kinetic energy management and self ignition delay. I was involved in some projects with AVL, also in one with VisioTomo and VisioKnock...was a Diesel ignited Otto lean burn concept.

When things get slow at work I've been spending my time reading white papers (SAE, JSAE, ASME, NACA, JSTOR...) understanding various engine topics. Most recently intake and exhaust tuning dynamics. The exhaust side is fairly straight forward, but as you reference, the intake is quite complex, far too complex for simple calculations.
Yuup, it's more simple due to some effects like Mach no is pretty all the time of EVO duration at 1 anywhere in the port and limiting mass flow. But there are still grains to win if it is developed right. I developed a technology to widen the bandwidth of the high level torque, its part of the DAMPFHAMMER engine technology. If you like we can discuss some aspects of understanding contribution to VE of the intake...maybe we can find some things :). Do you have free access to SAE, JSAE, FSAE (most interesting!)?

I'm waiting for my Exocet chassis to arrive, it's a week or two away so they tell me. Next year I'd like to see about swapping a K24A2, and then building a K24 with the target of 300 whp. The K-series offers a lightweight package that will turn big numbers without forced induction. Keeping it simple and light is my goal.
Sounds like an very interesting project Scider! The Exocet comes around 740 kg (1620) lbs? The 300 whp target, assuming 93 octane pump fuel, is challenging, but possible to achieve with aftermarket parts. Balancing the cylinders is one of the key factors, therefore a lot of those did it with ITB setups. What is your route to go, single TB (STB) or individual TB (ITB)?
 

·
Registered
Joined
·
81 Posts
Discussion Starter #12
So you know Prof. Spicher from KIT and owner of the optical investigation company MOT, who is pretty active in the field of irregular combustion phenomena like pre-ignition and knock?
No unfortunately, optical analysis is not used much in North America. My only interactions were with Ernst Winklhofer and Alois Hirsch at AVL Graz, mostly with regards to variations of the VisioKnock system or optical access engines.

Do you have free access to SAE, JSAE, FSAE (most interesting!)?
As a student I have access through the University Library system. I also have access to several libraries and journals through Ford. I'd enjoy sharing what I have learned and decided upon so far with regards to intake and exhaust design. ;)

Sounds like an very interesting project Scider! The Exocet comes around 740 kg (1620) lbs? The 300 whp target, assuming 93 octane pump fuel, is challenging, but possible to achieve with aftermarket parts. Balancing the cylinders is one of the key factors, therefore a lot of those did it with ITB setups. What is your route to go, single TB (STB) or individual TB (ITB)?
I have an aggressive target of 681 kg (1500 lbs) with the K engine. The Exocet that weighed in at 1620 lbs has an iron block and turbo kit. I'm envisioning three stages of development:
1) Sort the chassis with the Miata engine
2) Stock K24A2 swap, custom 4-1 header and RSP manifold (hopefully)
3) Built bottom and top end w/ cams

It will take several years to get there though and will probably come in more stages than what I have presented ;)

There's a few good build threads on MotoIQ and I've gone over Mrsideways S2000 history with a 4Piston K360 and RR3 cams, although I feel his intake manifold has oversize runner diameters.
 

·
Arouse the DAMPFHAMMER!
Joined
·
6,069 Posts
No unfortunately, optical analysis is not used much in North America. My only interactions were with Ernst Winklhofer and Alois Hirsch at AVL Graz, mostly with regards to variations of the VisioKnock system or optical access engines.
I am a bit surprised, the demand to understand the combustion, especially to find out the best compromise of low load efficiency (= high CR) and knock and pre-ignition free full load operation is a huge compromise for downsized engines. Does e.g. Ford not develop the EcoBOOST series also in the US or only here in Germany?

As a student I have access through the University Library system. I also have access to several libraries and journals through Ford. I'd enjoy sharing what I have learned and decided upon so far with regards to intake and exhaust design. ;)
Oh, wow, this may an opportunity to extend knowhow based on furter literature for those discussions. I would appreciate it. I am not sure if anyone is interested here in that beside us, may we better switch to email for that? I will send you mine.

I have an aggressive target of 681 kg (1500 lbs) with the K engine. The Exocet that weighed in at 1620 lbs has an iron block and turbo kit. I'm envisioning three stages of development:
1) Sort the chassis with the Miata engine
2) Stock K24A2 swap, custom 4-1 header and RSP manifold (hopefully)
3) Built bottom and top end w/ cams
Very though weight target. If I remember right, the C20XE engine (280 Nm/315 hp are top values for the 86x86 engine) has an iron block and just weight 5 kg more than the K20A2, which has the iVTEC system in addition. It was 130 kg vs. 125 kg for this two engines.

It will take several years to get there though and will probably come in more stages than what I have presented ;)
Moving target project...seems a common K-series infection topic :D

There's a few good build threads on MotoIQ and I've gone over Mrsideways S2000 history with a 4Piston K360 and RR3 cams, although I feel his intake manifold has oversize runner diameters.
Isn't the one with the K360 engine and RR3 cams K20a.org member Mixwell? For your information: https://www.k20a.org/forum/showpost.php?p=3293257&postcount=350...my remember on engine setup is sharp like a Ninja sword :D. Beside that, 4Piston has a strong Drag race background, they tend to spread this in every application...seems all application has Drag requirement specifications :D. Yes, I would agree with you...but I don't know your application of the engine, yet.
 

·
Registered
Joined
·
81 Posts
Discussion Starter #14
I am a bit surprised, the demand to understand the combustion, especially to find out the best compromise of low load efficiency (= high CR) and knock and pre-ignition free full load operation is a huge compromise for downsized engines. Does e.g. Ford not develop the EcoBOOST series also in the US or only here in Germany?
Optical engines are used predominantly for cold start mixture formation on gasoline, to reduce particulates. There are brute force methods to calibrate this that don't require expensive optical engine contracts though. VisioKnock can do some of the same work as an optical engine as well as knock investigations, but unless you had one or two cylinders requiring drastically different spark values, I doubt there'd be much to warrant spending the money on testing. The 3-cylinder engines (1.0L Fox and 1.5L Dragon) are the ecoboost engines developed in Dunton UK. The rest of the ecoboost engines are developed in Detroit and are much lower stressed and tuned designs.

Oh, wow, this may an opportunity to extend knowhow based on furter literature for those discussions. I would appreciate it. I am not sure if anyone is interested here in that beside us, may we better switch to email for that? I will send you mine.
I just reorganized what I've downloaded and would be happy to share what I've learned ;)

Very though weight target. If I remember right, the C20XE engine (280 Nm/315 hp are top values for the 86x86 engine) has an iron block and just weight 5 kg more than the K20A2, which has the iVTEC system in addition. It was 130 kg vs. 125 kg for this two engines.
The KMiata website claims 15 kg (30 lbs) weight savings from swapping from the normally aspirated 1.8L to the K24. Maybe a third of that resides in the tubular subframe though.

Here's the info I was referencing from mrsideways. Turns out I found it on a different forum:
https://www.s2ki.com/forums/automotive-builds-284/k-swap-227whp-313whp-5-years-1175851/
 

·
Arouse the DAMPFHAMMER!
Joined
·
6,069 Posts
Optical engines are used predominantly for cold start mixture formation on gasoline, to reduce particulates. There are brute force methods to calibrate this that don't require expensive optical engine contracts though. VisioKnock can do some of the same work as an optical engine as well as knock investigations, but unless you had one or two cylinders requiring drastically different spark values, I doubt there'd be much to warrant spending the money on testing. The 3-cylinder engines (1.0L Fox and 1.5L Dragon) are the ecoboost engines developed in Dunton UK. The rest of the ecoboost engines are developed in Detroit and are much lower stressed and tuned designs.
Wow, thanks for the insight Scider! We used it to examine the WOT combustion to understand knock locations. A great recognition out of that was that the rumors knock happens on hot valves first now was replaced by the opposite, which is also the fact you learn at University. But if 100 k€ help to teach a Diesel loving company about Otto combustion, why not :D.

I just reorganized what I've downloaded and would be happy to share what I've learned ;)
I studied the mixture formation and its influence on VE and combustion duration. I think this is a point of potential. Therefore my DAMPFHAMMER engine has two planes where fuel is getting injected, a nearer and a farer position from valve. I will do a row of different spray patterns and injector sizes to see the influence of wall-to-time-mixture formation. This would be an exciting topic to discuss. Do you do any studies or readings about this?

Here's the info I was referencing from mrsideways. Turns out I found it on a different forum:
https://www.s2ki.com/forums/automotive-builds-284/k-swap-227whp-313whp-5-years-1175851/
Amazing work Mrsideways did. Especially the 300 [email protected] rpm goal. I really appreciate your link.
 

·
Registered
Joined
·
283 Posts
Interesting read.
I always had the understanding that knock in petrol engines first occurs near the edge of the piston crown. This is caused by positive interference of incoming and reflected pressure waves induced by the beginning spark induced combustion process. At these pressure peak zones near the edge of the crown temperatures raise and lead to ignition of the unburned fuel air mix. These new flame fronts progress inwards towards the initial flame front. Once these hit, even bigger pressure peaks occur leading to shock waves ringing around the chamber causing the characteristic knock sound.


I have based my positioning of the water/methanol injection jets on this hypothesis. I lead me to position them right ahead (in terms of air flow) of the fuel injectors on top of the intake manifold. This was with my former supercharged Rover K.
This position should allow most of the spray to pass over the top half of the intake valve, where the majority of the flow passes, and distribute itself uniformly in the cylinder. I have always assumed a simple swirl motion front-down, back and up again as a first order approximation.
I wanted my droplets uniformly near the edge of the entire cylinder’s circumference.

It seems to have worked well doing a fair bit of track work. Piston crowns were mostly uniformly brown and very litte knock was detected.
Initially I even had a 11.5:1 static compression ratio.
Good for part throttle fuel efficiency, but tricky to tune, especially the IAT ignition corrections. I was only able to tune the upper range of the tables on track.
Later on 9:1 things were way easier to map and I got more power, but economy on the road tock a heavy hit.

If you want to measure valve lift profiles, probably best to build a rig from a old head and fit a linear draw wire sensors to the valve head with a litte tab under the adjustment washer. Then just turn the cam bit by bit with a degree disk and read the valve lift number straight from the PC. The low cost options offer 0.1mm resolution and decent accuracies for all your purposes of evaluating profiles.

Of course, next could be a stepper motor and automate it.
A chap I worked with uses these to measure the position of various guides and sheets in adjustable packaging production machines.

They are made by Microepsilon in Germany.

https://www.micro-epsilon.com/displacement-position-sensors/draw-wire-sensor/?sLang=en

Or use a old school dial gauge. :)
 

·
Registered
Joined
·
81 Posts
Discussion Starter #17
You're right that knock originates at the periphery of the cylinder. It happens something like this:

After the ignition event, there is a flame front that travels outward in all directions consuming the hopefully homogeneous air-fuel mixture. As combustion progresses, it increases the pressure in the combusted gases and the end gas (unburned air-fuel mixture near the periphery) alike. Knock is when the end gas mixture spontaneously combusts, and when it does it's damn near instantaneous. Octane is a fuels ability to resist auto-ignition, so a higher octane means auto-ignition is more delayed at higher temperatures/pressures.

When knock does happen, it creates a pressure wave that rapidly expands across the cylinder, which is followed by an anti-pressure wave. These waves reflect in every direction because the bore is circular. The pressure waves break down the boundary layer of gasses around the piston/bore/cylinder head that insulate the metal from combustion temperatures, so you rapidly heat up metal near the origination of knock, which can lead to more knock and eventually pre-ignition which causes massive knock events which do things like blow pistons apart.

Where knock happens is impacted by a number of things like how evenly the combustion chamber is cooled, there's a prevalence for knock to happen towards the hot exhaust valves. Modern high-tumble ports seen in downsized engines also have a lot of charge motion as the piston approaches compression TDC. This can push the flame kernel in undesired directions. Ideally you want the flame to consume the end gas near the exhaust side before the intake, but this doesn't always happen based on port geometry and valve events.

Petrol 4-valve designs induce tumble motion (right). Diesel 4-valve and petrol 2-valve designs are more prone to swirl (left)


You're running a turbo correct? I don't envy you trying to tune 11.5:1 with port injection LOL. We push up to 10 or 10.5:1 with ecoboost engines but at WOT near peak power I have to put so much fuel into it to keep the exhaust temps down it's ridiculous. The margin between the borderline spark limit and exhaust temperatures is so narrow...

Where is your IAT sensor located? We use a TMAP sensor on turbocharged engines to measure manifold air temperature directly at the MAP sensor.
 

·
Registered
Joined
·
283 Posts
I am familiar with the tumbling patterns.

I have converted to a NA 93mmx87mm stroker K20 2016.
The engine I was speaking about was a build 1.8l Rover K fed by a Rotrex C30-94 supercharger.
The IAT Sensor was mounted in a pipe feeding the intake manifold.
I mounted the TB before the supercharger omitting the recirc valve.
I preferred the seemless throttle response and the much reduced noise levels.
Wit hthe recirc valve, my car sounded like Darth Vader breathing during cruise.
As the car, a Lotus Elise is rather light, it tends to react on small changes in resistance while driving, be it wind or slope. Essentially you constantly modulate the throttle (without noticing) which lads to this noise.
Knock was under tight control with the help of a J&S safeguard vampire knock control system.
For folks running vintage ECUs or even carbs, this unit is brilliant. Saved my butt a few times.
I also do most things myself as knowledge when it comes to exotic things like supercharged and charged cooled Rover Ks that don’t blow up is rare.

I later changed pistons when the opportunity came up to get a custom 9:1 set.
The Rover K is a long stroke engine with a low deck height to allow for a reasonable rod length. There is normally not enough meat in the piston crown for a dish. At some point someone came up with a design with a deep dish, but with a dome in the middle to maintain strength. They worked well.

There are WAY LESS options on the market for low compression pistons compared to the Honda K-series world.

So far I don’t regret moving from the Rover K to the Honda K :)

Little titbit info: Rover and Honda cooperated in the 80s and 90ties. The Rover PG1 gearbox is essentially a Honda gearbox. Bearings and synchros are straight from the Honda shelfs.
Having had both the Honda NPQ3 GB and the PG1 gearbox open for rebuilds, they look very similar internally, even down to the reverse mechanism. In the Honda GB, everything is just scaled up.

I am no engineer, but a chemist with a Ph.D. in experimental physics. This is why I am less good in modeling and simulating. Math was never a big interest of mine.
I am still pretty good in understanding most relevant effects from the qualitative standpoint.
I enjoyed a few reads of publications or threads of relevant phenomena such as the transition from a acoustically dominated to a kinetically dominated flow in intake runners at high rpm and high flow speeds. During my Rover time, I mapped my SC engine with two different intake manifolds of different length and cross section and was able to observe these effects do a degree.
Mapping the peaks of the IM resonances was one of my few modeling efforts in excel using empirical formulas and measurements I did on the manifolds and cylinder head. It worked well enough to explain what I was observing during my mapping sessions (on the road).
Markus took these things 3 levels higher :)

Good luck with your next degree.
I recognize you feature the main ingredient any engineer or scientist needs, curiosity on how and why things work. Best coupled with a “can do” and “just do it” attitude.
Unfortunately, this is a personality treat more and more of my students are missing these days.

I believe one is an engineer or a scientist before one studies. The university is just there to professionalize these skills. Same for music.
If you don’t have the passion and interest in it from a fairly early age, you will hardly ever be good and enjoy what you do. As always with humans, there are always exceptions.
 

·
Registered
Joined
·
1 Posts
...Knock was under tight control with the help of a J&S safeguard vampire knock control system.
For folks running vintage ECUs or even carbs, this unit is brilliant. Saved my butt a few times....
Thanks very much for your kind words. I started work on the knock controller in 1983, while working as a technician at Hughes Aircraft. It was for my home brew Fiat 131 turbo. It was blow through carb, no intercooler, so "it had issues".

The knock controller was analog at first and did not do individual cylinders. I left Hughes in 1989, then released the digital SafeGuard in 1990, with individual cylinder knock control. The first version was for distributor ignition.

Over the years I have added features and interfaces for coil on plug ignition, but the knock detection, thresholding, and control algorithms have remained unchanged.

Thanks very much for the interesting and very detailed discussion. Still just a tech, I didn't have what it takes to make it through the program. Much respect to you guys.
 

·
Registered
Joined
·
81 Posts
Discussion Starter #20
Got my K24A2 head fixtured to measure some cams. It's pretty rudimentary using a degree wheel and dial indicator, there's a fair amount of error in the measurements so they're still crap for velocity and acceleration. I've done a deep dive into high dollar cam measurement devices like a Cam Doctor, but it was more of an exercise in curiosity. They're a $2k investment and I now know most of that is wrapped up in a high precision rotary encoder and linear gauge. At some point I'll throw the data on my work computer and see if I can't fit a polynomial spline to it.

101525


The lobe centers are a wild ass guess based on what some other motors look to be doing and what some of the better cam cards available show. There are no valve event specifications in any service manuals these days and I have yet to purchase a long block which makes life difficult. I've got some K20Z3 cams coming so I'll have a look at those next.

It's somewhat interesting to see the offset between the lobe centers for high and low speed. The exhaust cam retards 7.2 crank degrees and the intake retards 11 crank degrees when VTEC engages. This is probably designed around the factory VTEC switching point so the cam phaser doesn't have to move. When people lower the switching point I imagine it causes all sorts of issues with cam phaser schedules which is one reason there's .

It makes me curious what aftermarket manufacturers are doing with their cam designs, and what design philosophy they use for the low speed cam profiles.
 
1 - 20 of 28 Posts
Top