Stroke to Rod Length Ratios and Building a Long Rod Stroker EJ205

Like resoling your favorite pair of shoes, we are going to refresh the engine of our project WRX. If you have been following our series on the remake of Ian’s 2002 WRX, you have seen us rebuild the suspension and the brakes already. So now we are going to turn our attention to the search for more power and are going to do some innovative stuff to the Subaru’s EJ205 with some help from Eagle Specialty Products.

Since we have a GD STI with a 2457cc EJ257, we will have easy access to do A to B comparisons of Subaru’s big engine vs its smaller 1994cc EJ205. The EJ257 with its 99.5mm bore, 79mm stroke and larger displacement has more torque and spools its turbo fairly quickly. However, it seems a bit rough and not especially happy to rev.  Conversely, the smaller EJ205 with its 92mm bore and shorter 75 mm stroke seems smoother and freer revving. It’s a very pleasant feeling engine even if it lacks the torque and a bit of turbo response. Both engines share a 130.5mm rod length. The EJ257 sports a stroke to rod length ratio of 1.65:1 while the EJ205’s ratio is an impressive 1.74:1.

So in the pretext of doing something different and unique, we decided to see what exactly we could do with the EJ205. We studied the EJ22G, the engine in the JDM 22B GC8 STI. This engine had a bore of 97mm and a stroke of 75mm for a displacement of 2.212 liters.  This looks great on papers except the 97 mm bore which makes it not too practical to emulate as it would require the EJ205 block to be sleeved. We have reservations about sleeving engines due to the blocks structural integrity and issues that we have had in the past with sleeves staying put in race engines and even hot street engines.

To keep our block intact we would have to resort to stroking. Welcome Eagle Specialty Products to the project.  They happened to have a slick EJ20 crank available with a very reasonable 83mm stroke. Usually, long strokes mean more piston speed and more stress on the engine’s internals. Although this is a whopping 8mm more stroke than stock,  the stroke is still relatively short. Remember that an SR20 has an 86 mm stroke, so we will still have decently low piston speeds.

Eagle Specialty Products also just came out with a longer 132.9mm rod for the EJ20. We plan on having JE Pistons make us some 92.5mm oversize pistons with a 26.3mm pin height to accommodate the increase in stroke and the longer rod. This will give us a displacement of 2231cc, which is slightly more than the EJ22G. The stroke to rod length ratio falls to 1.60:1 which is still greater than an SR20’s 1.58:1. This will be a very reasonable engine, as far as internal stress goes, with decent displacement for the numbers. Our hope is to boost the EJ205’s torque and turbo spooling ability while not killing off its free-revving character. The end result being a very well rounded engine.

What is the significance of all of this concern about rod ratios?  Read on and find out!

In the illustration above we have plotted piston position in the bore vs crank degrees. With a very short imaginary rod vs an imaginary long rod.  This is to exaggerate things so you can see it on the graph easier. In the diagram, the piston with the longer connecting rod spends more time around top dead center (0-degree crank angle) due to a slower acceleration rate towards and away from top dead center (TDC).

A longer dwell time around TDC makes better use of the combustion pressure and turning the pressure into torque. a rule of thumb is that you want peak combustion pressure around 15-17 degrees after TDC. This helps mechanical as well as thermal efficiency. This is because the combustion pressures are higher while the crank angle is lower. The longer rod positions the piston to push down on the rod while it is straighter and the angle of the rod to the crank is more acute resulting in less side loading of the piston into the bore and offering more leverage of the rod to the crank for better energy transfer. Less side loading results in less friction between the piston and bore, freeing up power and reducing wear. The reduced piston acceleration also improves piston ring life, as the lower acceleration equals less force on the rings. Remember F=ma?

More critically, the slower acceleration around TDC on the exhaust and intake strokes improves volumetric efficiency at higher rpm. This is because you have more time to fill and empty the cylinders on the intake and exhaust strokes. Other things we have noticed with longer rod engines is that the torque and power peaks become closer in rpm. The engines are less sensitive to timing, so you can often get away with a few degrees less advance with no loss of power. The engines also rev more freely, seemingly with less effort and vibration. We have found in experimenting with Nissan, Honda and Chevy engines that even small changes in stroke to rod length ratios can make very feel-able differences in how the engines react and how long they last under racing conditions.

A rod to stroke ratio of 1.7:1 or higher is a good general starting point for a performance engine. Although many respectable production engines don’t meet this. For example, the Honda B18C is 1.58:1 and the K20A2 is 1.62:1. The Chevy LS7 is 1.53:1 and the BMW S54 is 1.53:1. High revving 1000cc sportbike engines tend to be around 1.9:1 to 2.0:1 and typically have a redline around 13k rpm. 600cc sport bikes use a ratio around 2.1:1 to 2.2:1 and redline around 15k-16k rpm. Bespoke racing engines can have even higher rod ratios. For instance, the 2009 Toyota F1 engine had a ratio of 2.72:1! The F1 engines of that period spun to 18,000 rpm. With those extremely high performance naturally aspirated engines the longer rod to stroke ratio was very important to making power, because the piston spends so little time around TDC due to the extremely high revs. Anything that can be done to improve dwell time near TDC, improve mechanical energy transfer, and reduce friction will greatly benefit performance.

In this graph, we show that a longer rod slows piston acceleration in the region of top dead center. Less acceleration means more time to fill cylinders and transfer combustion pressure to the crank.  Less acceleration also means a lot less stress on the engines reciprocating parts.

Not everything with long rod engines is all advantages. We have found a slight trend in a long rod engines having a slightly greater propensity to detonate. Especially when trying to get an engine to run well with California’s awful 91 octane fuel. This is largely mitigated by a longer rod engine’s less need of ignition advance to make the best power.  Another thing is that when the torque peak and power peak become closer together, peak torque is made at a higher rpm. This can make the engines bottom end response soggy. We think that because the piston moves away from TDC slower, the intake and exhaust port velocity is reduced at lower rpm actually hurting VE at lower engines speeds.  This is partly mitigated because the engine wants to rev freely. Lastly, because exhaust port velocity is reduced at high RPM there is a slight tendency to spool the turbo slower.


  1. A long stroke EJ20 is always something I wanted to try for a street car.

    So, controversial opinion ahead – I honestly think that rod ratio is lately really overstated as something to shoot for. If you look at that BMW S14 graph, going from the 144.25mm rod (1.66 rod ratio) to a theoretical 125mm rod (1.44 rod ratio) results in a change of peak acceleration at TDC of what, ballpark it at 5%? And as you point out a lot of very successful high winding engines have pretty “low” rod ratios – not mentioned but the K24 has a rod ratio of 1.54, and I know drag guys stroke them even more than stock.

    F-1 stuff has other constraints driving the rod ratios – the bore/stroke ratios were driven by the drive to higher RPM enabled by the valvetrain, but the rod ratios really were driven more by packaging than anything. If you look at cross-sections of F-1 stuff of that era (pictures of Ferrari’s 2000 vintage engine are out there), the engines are basically designed as short as they can without having the piston skirts hit the crankshaft or come out of the bores.

    I think the big thing driving the characteristics of the Subaru engine are not enough port and not enough cam. The EJ25 could easily have had the valve centerlines moved to take advantage of 7.5mm more bore if they were doing a clean sheet engine but instead they basically kept compatibility to everything else. I may end up building a flowbench mule head and see how much can be gained by moving the centerlines around, as I’m pretty sure there’s enough meat (albeit the buckets may require weird stuff)

    1. Try building a long rod engine and driving it, you can really tell the difference. The engine even sounds a lot different. I do agree with Subaru’s having poor and unequal port flow.

      1. As we often call them, “destroked” set ups are much smoother and rev happy. Have less peaky power delivery, but generally can give more revs than the turbo has powerband. Big bore, sleeved, has always been the ticket here.
        Once I’m done setting up my back to back 6758 and 7163 turbo testing on my 2.12 EJ20 stroker, my personal 4″ bore destroked (1.8:1 rod ratio) will be getting a GTX4294R and 9180 test.

        Dan and I’ve been talking, he’s been heavy on cam analysis, and I’d LOVE to see a clean sheet EJ cylinder head to take advantage of the larger bore.

        I will say on the big power end of things, the cylinder heads aren’t the limitation, the integrity of the mains ends up being the issue.

          1. Hey Khiem, I’m aware and have anxiously been waiting to see how the G-series would grow. The TR30’s we used in the DP and GT program were awesome in comparison to turbine temp limit of the EFR turbos. I also have seen the EFR line expanding now too.

            I’m still waiting to see what comes out in the 30/35 range for the G series.

  2. Why not build a block with the EJ257 as a basis, and go the other way around of doing it? Same bore (99mm) stroke of the Ej205 (75mm). Same rod stroke length as before (althou thats still can be altered by moving the wrist pin further up), but a displacement of 2308cc. Still as rev happy, but also more torque (albeit higher in the rev range). No resleaving needs to be done and a lot safer option from a rod/stroke length perspective.

    As for detonating: Did you look into the cam timing? As the main problem lies there: More overlap is less detonation, because of a lower dynamic compression. It all comes down to revs basicly. The more you run, the higher the chances detonation occurs. Calculating dynamic compression is everything to prevent that, which can be done fairly easily.

    BTW, always wondered, as I’m from Europe: Is 91 octane RON or MON rated in the States?

    To mitigate responsiveness: Just measure exhaust flow. What you mentioned above is true, but can be compensated for by high intake timing and less exhaust timing, with a further improvement in pickup speed by leaving the exhaust as small as possible while improving the intake side.

    The one thing that really becomes an issue though: Is looking for the right turbo: As the redline is increased this becomes especially critical for low end response, as you actually making the performance envelope bigger. A turbo that is happy to spool low down will be less efficient up top and vise versa.

    As for Dan Rosia: No they are designed als a long block. Thats the only way to get that type of RPM. 18.000 RPM is still a limited RPM. The Toyota engine was capable of 21.000 RPM. How I know? I Was there as one of the engineers. Of course that doesn’t mean there is actual wasted space or weight. So yes, they are built as short as possible, but are still a relatively long block. Its only because the stroke is so short that they appear to be short blocks, but in comparison they are not.

    1. @Kevski – Speedhunters deleted my rebuttal, but luckily we frequent the same websites. Do you remember? I recommended a 5-lug setup if you’re going to run a big brake kit, in order to minimize deflection.

      So what can you do about knockback?

      There are many schools of thought on addressing knockback, each with their own pros and cons. We’ll list them here not in order of preference or recommendation, but rather to assist you, the reader, in making your own best decision.

      1. Minimize wheel end deflections during cornering

      While it may sound obvious, making sure that your wheel bearings are fresh and tight is the first major step toward addressing knockback. Following the wheel bearing itself, upgrading hubs and other suspension components to achieve less deflection during cornering will also serve to minimize knockback. At times, heavy-duty or race-specification components may be available for bolt-on installation. A little bit of research here can go a long way.

  3. I want to build a destroked long rod EJ257 too but we wanted to try messing around with the EJ205 as it gets overlooked a lot. There is no such thing as dynamic compression ratio. Compression ratio is a mechanical relationship that never changes, even when the engine is running. What you are talking about is calculating MEP which isn’t so easy to model as it is multidimensional. The OEM’s spend a lot of time trying to model this with pretty sophisticated math with varying degrees of success. Reducing cylinder pressure my adjusting cam timing isn’t a sure cure for detonation because you are simply playing around with the VE curve and on a turbo engine, you can get into reversion which can be disastrous for detonation. The best way to eliminate detonation is to fix the root cause which is fuel octane, combustion chamber design and compression ratio, then taylor the ignition timing and AFR around that. Our octane is by the R+M/2 method. You can’t generalize what effects cam timing will have on all engines, not all engines will respond well to more intake and less exhaust timing. Older American domestic pushrod engines, for example, will often prefer the other way around.

    1. Reversion should be the first thing to tackle on any car: NA or FI. Making everything smooth as a baby’s butt is a general misconception. Flow alsways needs to be in one direction and restricted in the other. Thats what grooves are for. For the people who don’t know: never match you exhaust manifold to the exhaust port, but always leave a lip of at least 1mm. And do that as much as you can troughout the whole exhaust. Same for the intake, except the other way around. Yes, there is a lot of math involved in doing everything right, but it always helps even if not calculated.

      Currently running a stepped intake and exhaust on a otherwise fairly stock engine as my daily which was cobbled together in a hurry using parts I had lying around which works like a charm (which reminds me I still need to do the math some time). Of course doing it the right way is a lot more involved as where talking about freqencies that need to be attuned for, and fitting it all in the same car as compaired to on a dyno is the worst.

      What I use as a general rule of thumb if someone doesn’t want to calculate there stuff but asks for advice is if they know their overlap and know how much overlap there is beyond TDC. Lets say there is 12 degrees overlap past TDC. Thats in effect 168 degrees of intake stroke. And yes, I know that isn’t the case actually, but its only meant for the fuels we have here (98 Ron premium fuel, so that would equate to about 93-94 of your method), only on sea level (as most of this country is (about 95%) and only as a guideline. It does work though, so long as you can tune the timing and you understand VE’s.

      168 / 180 = 0,93 or 93%. When you have a compression of 10:1 you actually have a “dynamic compression” of 9,3:1 I can’t say the same apply’s in your region, as the climate, elevation and fuel differ a lot.

      And sure: Good fuel is always key, but when its not availible you need to tune for it. And when it’s not economically feasible its the same (as you can get race fuel on either side of the pond, but running it on a day to day basis is a costly matter).

      As for the generalisation: Your right. But then again: Ever wondered how many cams are actually tested and are the absolute best for the application? because if you measure them correctly most are the same from car to car other then the lift. Lazy company’s I guess? Happens as much with the OEM as with the aftermarket companies. But let me specify the generisation a bit: So long as volume stays the same It’s always a good idea to look at exhaust timing. And to the exhaust side of the turbo of course. And your right about the pushrod V8’s or any long stroke engine for that matter. The need more exhaust timing. The longer the stroke in general, the more exhaust timing your need.

      And we all know there is no such thing as the perfect engine, because its always a compromise. You just cant have it all.

      As for the longer rods in your other comment: Totally agree with you on that. Just less sideways load on the bearings, with effects you can actually feel when driving. As rule of thumb allways get the tallest block, and the tallest rod you can get away with.

      1. @kevski

        Naturally, none of the components on our cars are 100% rigid. Therefore, as these dynamic lateral loads are applied, all of the components reacting the cornering forces must deflect to some degree. In English, when we throw our cars into a corner, stuff bends. It’s not desirable, but that’s the reality of the situation. You can pay more money to make stuff bend less, but it will always bend to some degree. This is where knockback is born.

        As the wheel, hub, and wheel bearing deflect during cornering, the rotor hat sandwiched in-between is forced to go along for the ride. Because the caliper (red in the illustration) is attached to a more rigid suspension component – the upright – the parallelism between the rotor face (gray in the illustration) and the brake pads (yellow in the illustration) is altered. In so many words, the deflection of the rotor relative to the brake pads actually forces the brake pads away from one another. This spreading action pushes the caliper pistons (blue in the illustration) back into their bores a tiny amount (horizontal green arrows in the center illustration) so that when the deflection goes away (when the cornering event is over) there is not enough springback in the piston seal to push everything back together again (green arrows in the rightmost illustration). The pads are now pushed off the rotor and will stay put until the brakes are next applied.

    2. hey you’re the guy that doesn’t understand what pad knockback is, and the stupidity of bolting a huge caliper to a 4-bolt wheel hub that was NEVER intended to take those loads. That’s why I made the comment about upgrading to a 5-lug. it’s too bad Speedhunters deleted my comment. I had a nice rebuttal for you.

        1. @Mike Kojima: Probably someone who has a bone to pick with me? I don’t know? Allthough I do know the topic thats he’s talking about, which is on a different site.

          @Joe: But in short: I do know what knockback is. Install floating calipers and its almost gone. install a better master cylinder and your good to go. Most of the problems occur when there is an error/mismatch in the system. What I always say is: “Do it the right way, or don’t do it al al.”

          But: That still has nothing to do with the amount of studs used on a car. It applies to part past that: (the bearings). because if studs move or have play, you have way bigger problems at hand. Read the article carefully, and you can conclude that it doesn’t say anything about the studs.

          And I’ll leave it at that, as this is the wrong site and the wrong article for these comments.

        2. I think Joe is clearly in the wrong topic. Probably thought he was looking at the big brake kit or something. Was talking about pad knock back up above.

          BTW, you going to PRI again this year? I’d be more than happy to dump a wealth of Frankenstein EJ build stuff on you.

  4. I think reversion is particularly an issue with turbo cars due to the backpressure the turbine creates. Add to that, most moderately sized turbos do not run in crossover so it’s easy to get backflow into the cylinders under some conditions if overlap isn’t carefully considered. I like to advance the exhaust cam so some blowdown energy can drive the turbo better and kill some of the overlap. At least I like to experiment that way first to see what the results are with a given system.

    Someone in the office just pointed out to me that the current Infiniti QX50 actually has variable dynamic compression! What a mechanical mess!

    1. Its really equally an issue with any engine. VE actually does improve on NA engines as well. It not so much backpressure as it is in wave tuning. Every exhaust systems expiriences waves. And the same applies for the intake system. As long as you try to prevent flow the other way, you can increase port velocity without losing power and have a higher VE to boot. And it isn’t that much more then creating ridges/steps. Yes you could also try for a dimple/golfball effect, but getting that right is a pain on straight pipe, let alone on an intake or exhaust manifold with all the bends involved. Slipping one pipe over the next to make an exhaust manifold is relatively simple. Just step it up one step the closer you get to the turbo. That way you also expirience a lot less lag.

      Thats not the worst of it for the QX50: Flat torque curve is. OEM’s need to stop messing with the ignition timing, as a flat torque curve is only marketing gimmick and actually downtuned. But then again: Marketing is probably the thing thats wrong with the world today, so its hardly a surprise 😉

  5. I have found that stepped headers and anti reversionary tricks don’t seem to help much on racing engines that have to run stock camshafts due to rules. Presumably due to less overlap. What do you think about that?

    1. It’ll still work!

      BUT: Do you use a mandrel bend step or a different pipe diameter? Because it’s highly dependant on the stop: Under 15 degrees inclination a step won’t have any effect. I always used the mandrel bend, 45 degrees, until I tested steps of 90 degrees. With 90 degree bends it works like a charm. You do need to run a file or dremel tool on every end of the pipes though, as leaving just a hair upright will impede flow quite a bit, and cause of lot of drag.

      The one thing that goes wrong the most is exhaust pipe diameter though: Inlet piping to the turbo needs to be slightly smaller then the exhaust size inlet. From there on use at least one step, but preferably 2 or 3. That means high velocity exit which loses velocity on each step and gains pressure. So working your way backwards from the turbo in a sense. That would interfere with the ports on the exhaust side of the heads though, so I usually start from the head, and get the correct turbo size once I’m done. Or just mill out the inlet side some more to get it bigger then the header dimensions.

      If its NA you are talking about, it depends on what your after and whats allowed of course? I have to see the rules to know whats possible. (you can email me if you want, as you can probably see the emailadress before approving the reply? ) Overlap isn’t so much the issue per se, but the actual cam timing is. Overlap is just a side effect of cam timing. The less overlap the happier an engine is in general, but the less timing you can run because of cycles in a 4 stroke. So if running a DOHC head with a separate cam for intake in exhaust you could always run the intake slightly early and the exhaust slightly late. Less overlap is a higher efficiency after al.

      Stepping a header will give a couple of BHP up top if you time it right, but moest is found in the mid range on corner exits. Its a higher curve of torque and HP troughout the rev range where it makes the difference. I mean, lets face it: Torque is the only thing that made a winning engine in F1. Power output is fairly close with all the top teams in de 3.0 V8 and 2.4 V8 era of F1.

      The thing that I’m actually getting from what your asking, is not so much the reversion but the header/manifold itself. Most people think of headers in same length and diametre, but thats wrong. Its the same volume of air. Thats the same thing if pipes are perfectly straight, but thats never the case. And flow behaves differently in anything more then 15 degrees of bend. So are the pulses timed exactly right in the exhaust manifold. Because on a turbo engine thats what creates most of the lag that people expirience. Imbalance on the turbo is never good. And in NA engines it usually causes scavenging issues where one cylinder is lean and the other is rich. Both can lead to all kinds of issues further down the road. A happy engine is a balanced engine after all.

        1. Yep, but thats part of the equation: every bend interferes with flow. Inside curve is high pressure low speed, where as the outside curve is low pressure high speed. That makes a lot of difference in actuall flow, but also in timing. Let alone if there a different types of bends involved between each header pipe. How thick a wall do you use? 1mm is the minimum, but i currently use 1,5mm. haven’t tried 2mm yet, as it’s hard to bend and collapses easily.

          1. Then there lies your main problem: Get thicker walled steel. And just regular will do. Stainless tends to get stress fractures. Fine for the exhaust itself, not so much for the manifold. Thats nearly half as much as 1,5mm, being a little less then 0,9mm thick?

          2. I don’t have a problem with breakage, 321 is stress corrosion resistant and I use slip joints and springs. No breakage in years of competition with any of our headers. I actually don’t have problems with anything other than FA20 bearings living! You don’t happen to have any ideas for that?

          3. Using slip joints: Then you won’t have any problem whatsoever. But remember that the slip fit needs to be straight for it to work as a anti-reversion. It cant be stepped in the slightest.

            FA20 bearings living: The age old problem of any saburau engine. Best option: Get an inline engine/V or as a minimum a 6 cilinder boxer engine as a minimum. Still wanting the FA20 to work:

            We all know what the problem is: Oil starvation and rocking couple. 4 cilinder boxer engine problems. Rocking couple can be helped, but is way to expensive to do right, so I’m leaving that out.
            First of get the oiling right: I’ve seen cornering and accel/decel problems, so getting the absolute best baffle with a windage tray is must. Or a dry sump if you have the cash.

            The bearings themselves are not the main problem, although I must say you can design a bearing that will not spin in the case. Why you want to is a mistery to me though, as I would rather see some marks on the case and the crackshaft, as opposed to a lot of marks on the crankshaft alone. Just let a bearing company make a point protruding downward on the underside of the bearing and a hole in the case in which it fits perfectly. I usually specify my own bearings from a large supplier less then 2km or about a mile away from here. Isn’t that expensive really, and still way cheaper to get then importing stuff from abroad. But you generally don’t need to do the bearings, as the problem doesn’t lie there.

            Cheapest sollution (although I haven’t tried that myself) would be to put more oil in the engine. But that usually means more trouble elsewhere, so at your own risk.

          4. Just had an epiphany: I can’t say this for a fact since its all still a theory now, as I haven’t seen on felt the vehicle.

            But since its a FI engine and the exhaust is anti reversion. How about the intake? Everything might be pressurised, but that doesn’t mean there aren’t waves. So isn’t it possible that what you are experiencing is happening on the intake side? And if so: Same rules apply, only in a differenent direction.

  6. I have nothing super technical to add. I just want to express my joy about being able to join a discussion about Subaru engines, that doesn’t involve curse words, name calling, infantile texting abbreviations (U for you, etc) Or any sentience that involves words to the like of “brah, My cousine gots a mad sti. He busts fools in Mclearn’s everyday.

      1. ahh I do miss the way forums were decades ago. They were full of information (although some of it wrong), but it was a great place to learn.

  7. As annoyed as I am by smoking. I’m annoyed more so by vaping.
    But you can get the smell of stinky hippy out of a car interior. Not so with cigarettes.

  8. Was just pondering: A 2JZ and a 1JZ are in a way interchangeble. But the 2JZ block is taller? So in a way, you could build an absolutely bombproof 2JZ with the stroke of a 1JZ and longer rods. Has that actually ever been done before?

Leave a Reply

Your email address will not be published. Required fields are marked *