A two valve versus four valve debate is certainly nothing new for me. Anyone who has spent time talking with me on this subject soon discovers I am not really a fan of four valve engines for the true street power plant. But they are out there in the millions and rather than fight the situation I thought it might just be more practical to look for ways to make a four valve engine that much better. So here is the patented PolyQuad head design that David Martin and I worked on for a number of years before taking out the patents in 1999 – 2003.
Granted there are plenty of variations on 4 valves per cylinder and this begs the question as to what a PolyQuad head does better than a current and typical four valve per cylinder head?
Question: What does a PolyQuad head do better than a current and typical four valve per cylinder head?
Answer – Everything ..
Compared to Honda’s VVT and other such variable valve timing system costs almost nothing and can be produced in a typical head shop! To best appreciate how the PolyQuad concept came into being making a start at the two versus four valve debate would be as good a place as any.
There have been some serious debates concerning combustion efficiency going on in serious hi-tech forums and from these it seems that it is well understood that there is more to producing torque and power than just filling cylinders with air and fuel. At low speed the engines ability to breath is hardly a problem as relatively speaking, there is plenty of time to fill the cylinder as full as it is ever going to get. The key to making low speed torque is two-fold a) how much charge is trapped at valve closure and b) how effectively the charge is burned.
Aspect a) is governed by the reverse flow properties of the intake valve seats and to a lesser extent the port area and the cam event timing. Better anti-reversion at low lift and higher port velocities are good here but at the end of the day it’s still largely down to the intake valves closing soon after BDC.
AS for point b) we find that how effectively a charge burns is down to mixture preparation and in-cylinder motion at the time of ignition. Assuming that mixture preparation is good then, at this point, best results would be achieved by the cylinder with the best combustion dynamics stemming from appropriate in-cylinder charge motion. The bottom line here is that a parallel valve two-valve bath-tub or wedge chamber design has inherent swirl properties. This means a head designer would have to work hard to eliminate the heads natural advantageous tendency to swirl the mixture (and believe me some have managed to do that). For a two valve head then the natural tendency is to swirl the mixture and this, in part, contributes to a two valve heads normally superior low speed torque.
On the other hand a conventional four valve head has no natural tendency to swirl – but it does have a tendency to ‘tumble’ the charge. The following illustration shows the difference between the two.
Not unreasonably a first inclination would be to assume that both types of mixture motion would work equally well. Up to a point they do but as the piston nears the top of it’s stroke these two forms of mixture motion diverge in characteristics. The tumbling charge motion breaks down to small tumbling eddies where as the swirl, as it gets compacted in the chamber smaller than the cylinder bore can speed up (conservation of angular momentum) while still retaining it’s swirling motion form. This swirling motion in conjunction with a well placed spark does a good job of effectively and rapidly burning the charge. On the other hand the tumbling action seen most often in four valve engines tends not to produce enough charge motion at low speed to produce as good results as the two valve engine swirl. Also tumble can be almost cancelled out when compression ratio’s much above 11/1 are used.
In addition to the lower combustion speed the inclined valve four valve heads tendency to allow the fresh incoming charge to cross flow out of the exhaust during overlap also cuts the low speed torque as well as increasing fuel consumption and emissions. With the appropriate proportioning of valves and bore/stroke, a two valve engine, with a well designed cylinder head, can out-perform a four valve engine to about 4000 rpm.
It is because four valve engines have less low speed torque that, in part, prompted the development of variable valve timing. That same challenge to improve low speed torque also prompted Ford’s port disablement. This technique used a butterfly to close off one runner of the pair leading into the cylinder such that the charge only exited from the operative port.
Although valve disabling can generate a great deal of swirl most of the motion is at the cylinder walls and the spark plug is still left in a relatively calm area of the chamber.
There are other means toward inducing swirl. A method I tried many years ago on a Cosworth BDA was to phase the intake cam profiles. By having one open earlier than the other the mixture was encouraged to swirl as well as tumble. There are many combinations that can be tried here. The first was to simply have two identical profiles phased about 5 degrees different. Although this induced swirl at the beginning of the induction cycle it induced reverse swirl at the end and cancelled out the initial swirl. This and the longer duration of the opening actually reduced low speed output for no real gain at the top end. Another example was to have one intake open before the other but close at the same time. This resulted in a slight increase in low speed and a slight reduction in high speed. The best results were seen when one intake lobe was about 4-7 degrees less duration then the other with each sharing the same intake centerline. This resulted in a moderate increase in low speed torque with only a minimal reduction in high speed output. The consensus at this point was that swirl might be a good idea for a 4 valve engine.
Added to the lack of swirl is the fact that, as mentioned earlier, a four valve head more easily ‘cross-flows’.
From this we can see that to get a given amount of low speed torque will take a cam of considerably less duration than it’s two valve counterpart. This negates much of the advantages of the four valve engine to produce better top end power.
Summing up the situation we find that a four valve head produces less low speed output up to about 4000 rpm than an equally well developed two valve head. As an aside this might make you wonder ‘why the predominance of four valve engines instead of two valvers’. This is a good question and one which warrants a story in it’s own right so this is about as far down this road as I intend to go here. The point is – we are stuck with 4 valve engines so let’s sum up the negatives and see what can be done to fix them.
The principle four valve drawback is the lack of swirl brought about by the typical port directed straight into the cylinder. Secondly we have the heads propensity to cross flow much easier and the negative impact that has on low speed torque, emissions and mileage. All of these reasons are primary candidates for the push to variable valve timing not to mention young engineers wishing to make a mark in industry.
The following points then highlight the factors we have so far discussed.
- Unless steps are taken to the contrary, a 2 valve motor will make more torque than a 4 valve unit in the 1000 – 4000 rpm band. Above this rpm the 4 valve motor will become progressively superior.
- With mid to large motors up to 95% of driving is done using 1500 to 3500 rpm so why is there still a predominance of 4 valve engines in this group! When was the last time you saw even a sprightly grandma driving her North Star Cadillac at 4000 or more rpm?
- A high torque output at low speed is the marker by which most drivers, especially those of luxury cars, assess the vehicles power output. On the street 20 lbs.-ft. at 1500 rpm is instantly discernible – 20 HP at 5000 rpm is not!
- Whether for a performance orientated driver or a ‘cruising’ driver a motor delivering high torque at low speed will be deemed a superior power unit.
- High specific torque outputs at low rpm are almost entirely the result of high compression pressure and effective combustion that is usually derived from good mixture motion. Compression pressure is generated by compression ratio, and short intake valve duration to produce an early closure. This second factor offsets some of the top end potential of the 4 valve design.
- Mixture motion – this, in a 4 valve motor, is usually generated by minimizing port cross sectional area’s, attention to squish and mixture tumble.
At present most mainstream manufacturers are pursuing low speed torque by optimization of more conventional parameters such as cam timing, compression, and port sizes that promote tumble. Chrysler USA is a leading proponent of this approach.
Some manufacturers have taken a more aggressive design approach and have introduced swirl levels consistent with those seen by two valve motors. Ford motor company has taken this approach by disabling one intake port of the pair as shown above. It’s success however is somewhat questionable.
Although VVT will always add to the power band width it is possible to emulate its use on an otherwise conventional head at near zero cost. This is what PolyQuad does and, shown below are it’s primary features. As you can see from the above drawing the PolyQuad layout for a 4 valve engine has four distinctly different valve sizes. This has many implications in it’s own right and together with other features that it may be used with. Let’s start with the most obvious.
From the above drawing we can see that the greatest amount of air will be coming from the larger valve. This causes the air flow to turn within the cylinder in the direction of the large green arrow. The natural tendency to swirl produced by the valve sizing difference is also supplemented by the port shaping on each of the two intake ports. The port for the large valve has a shallower bowl form on the long side turn so as to encourage flow out of the long side and discourage it on the short side.
The port for the small valve has a deeper bowl to encourage flow from the short side and discourage it from the long side. This has the effect of keeping the swirl active throughout the induction cycle.
Even though the focus here is on swirl the four valve heads natural tendency to induce tumble is still present and strong. What this means is the mixture has both a tumbling action and swirl combined (twirl).
The shaping of the chamber also influences the swirl generation. In this instance the effect is mostly at low lift and influences the mixture motion at the beginning and most importantly at the end of the induction event. Take a close look at the following drawing.
On the sides of the valve where flow is needed to augment swirl the chamber wall is well away from the edge of the intake valve (Wide Clearance A and B in above drawing) so as to eliminate shrouding and promote flow in that area. On the sides of the valves where flow would be countering the swirl direction the chamber wall is close to the edge of the valve (Close Clearance A and B in above drawing) so as to shroud it slightly to inhibit flow.
Cross Flow Suppression
During the overlap period there is an opportunity to utilize wave induced low pressure at the exhaust port to start the intake charge moving into the cylinder earlier than would be the case from piston motion alone. The problem here is that some of the intake charge can pass right through the combustion chamber and on out the exhaust. This is not good for mileage, power or emissions. Hemi and pent roof chamber designs are far more prone to cross flow than simple wedge or bath tube chambers. As such they cannot, for any given displacement of cylinder, successfully employ as much valve event overlap. To reduce this tendency the PolyQuad head uses a stepped chamber. This puts the intake nearer the piston and the exhaust further away. The result is a strong tendency for the incoming charge to pass over the front face of the exhaust valve rather than under it and on out the exhaust port. The following drawing gives a good idea of how this looks and works in practice.
In addition to the stepped chamber cross flow is also suppressed by the differential valve sizing. The big intake valve is opposite the small exhaust. This means the greatest amount of charge entering the cylinder is moving toward the smallest exhaust valve. On the other side of the head we have the biggest exhaust valve and consequently the highest exhaust activity aligned with the smaller and least active intake. This, in conjunction with the stepped chamber considerably cuts cross-flow to the extent that about 5-7 degrees more overlap can by used compared to a conventional four valve head before low speed output is compromised past acceptable limits.
PolyQuad in Practice
The first type of head that PolyQuad was applied to was the Arao 4 valve pushrod Chevy head conversion for small block Chevy’s. This was a good candidate as we could reference the results directly against a two valve head design. Essentially here is what was done. Russ Arao had already done a series of tests along these lines but with all the heads using the same 112 degree LCA camshaft. The heads used for the PolyQuad tests included a pair of GM Vortec heads. These heads are considered to be amongst the best two valve designs for producing strong low speed output. Anyone who has run back to back tests with a good aftermarket head will attest to the fact that at low speed they are almost unbeatable and are certainly far from shabby at high speed. The Vortec heads then seemed like the best to use for our low speed benchmark. To represent an aftermarket head situation I ported a set of a well known brand of heads that ended up at about 185 cc port volume and 280 cfm at 0.600 lift with the flow curve flattening out from there on up. The Vortec heads, the ported heads and the PolyQuaded Arao four valve heads were all run on the same short block. In each case the cams valve event timing was optimized for the heads concerned. The two valve heads both produced the best power curve on a 108 LCA cam. The PolyQuad heads were best at 112 to 114 LCA. The intent here is to see how much better the PolyQuad heads were at low speed compared to a two valve setup. With the wider LCA the intake would close later on the compression stroke. So there was a total match on the intake closure point the duration of the intake was shortened so that in all tests the dynamic compression was the same (as was the static compression). The exhaust was also shortened to match the new shorter intake. The tests then were done with each pair of heads having the same intake closure point and an appropriate LCA. The results are as per the following chart.
With 1.7 rockers on the intake and 1.6 on the exhaust the valve lift for the two valve heads was higher by about 0.05 inches than for the four valve heads. The difference in accessed flow in each case was about 40 cfm (nominally 235 for the Vortec heads, 274 for the ported heads and 315 for the PolyQuad Arao heads). The difference in power between the Vortec heads and the ported aluminium heads was 37 hp. The difference between the ported aluminium 2 valve heads and the PolyQuaded Arao four valve heads was 103 hp. This big discrepancy between the two types of heads in terms of power increase for flow increase can be accounted for by the far superior low lift flow of the 4 valve heads. At 0.250 intake valve lift the Vortec head and the ported head flowed 168 and 172 respectively. The PolyQuaded Arao head flowed over 230 cfm at this same lift!
The original tests with the Polyquaded Arao small block Chevy heads established, within reasonable parameters, that it was possible to achieve two valve head low speed torque without significantly sacrificing top end output. Along with this my aging re-furbished emission gear showed a reduction in all raw emissions. Most advantageous was
the reduction in NOX. This was ascribed to the reduced amount of timing the PolyQuad heads needed to get the job done. At the 60 hp @ 2000 rpm cruise test (all these tests were done with a 16/1 air/fuel ratio) the timing required dropped from 34 degrees to 24 although it could be reduced from here to as little as 20 before any real drop in output/fuel efficiency was seen. Mileage, using these same test parameters, also showed benefits to the tune of 3%. That may not sound like much but if this head design does make more low speed torque than current four valve heads that means it can pull a higher gear and that could easily account for another 3-5%.
At this point we have tested the validity of the PolyQuad concept in a two versus four valve shootout. It can make those low speed numbers. The next test was to see what it could do when it was a regular four valve versus PolyQuad test. The real question here was whether or not there was a down side. To do this test my partner in crime David Martin of Austec Racing in the UK ‘Polyquaded’ a 2000 model 2 litre Ford Focus head as far as could possibly be done without any welding.
Other than the prospect of a superior power curve the other aspect of PolyQuad is the fact that it emulates VVT for no ongoing additional cost if it was incorporated into a production line vehicle. This does not mean that Polyquad competes with VVT. The bottom line here is that a VVT engine would also benefit from the use of PolyQuad as it would just add to what is already there.
Using our moderately (it was really a half-way-house to being a full PQ deal) PolyQuaded Ford Focus head back to back tests were run at Roush’s UK establishment. Although the before tests were done very rigorously the after tests were cut just short of a proper full ECU calibration due to dyno problems. The whole deal was not a total loss though as some numbers were established. The test done showed as much a 7% increase in low speed output (1500 rpm) with no loss of top end output. That extra low speed torque would make this 2 liter engine drive as if it had 2.14 liters. Was Ford interested? Seemingly not – but there again I have been in this position many times. It seems an outsider coming up with a better idea than the guys on the inside threatens job security. That is the real reason the ‘not invented here’ syndrome comes about – but that’s life.
Real Race Stuff
While all the Ford Focus stuff was going on I was busy modifying a 2 liter Mitsubishi head for Trinidad drag racer Ryan Garcia. This head was not only an exercise on converting it to a PolyQuad design but one of pulling out the stops to produce the best results possible. To explain everything as fully as possible I’ll make a start at the intake manifold face and work across to the exhaust manifold face.
Firstly there seems to be two versions of the Mitsubishi head concerned and the point of difference we are concerned with is the size of the intake ports. For some reason the big port head is often favored although the port is simply too big for the job – any job that is. It’s too big for race or street, supercharged or otherwise. After doing some measurements I came to the conclusion that unless I could get the seats and bowl area to flow really efficiently the port of the small port head might actually still be a shade too big for best results. With this in mind I settled on the small port head as the basis for a PolyQuad conversion for Ryan’s 1986 turbo Mitsubishi.
The next problem to deal with was to get the differential flow difference between the primary and larger intake and the secondary smaller intake. The difficulty here was that making the small intake much less than 1 mm smaller than stock was not really on short of installing different valve seats. I did not have access to any and was in no mood to make some up and install them. As a result I settled for a secondary valve 1 mm smaller than stock. As for the primary valve this was made 1 mm larger than stock and to make sure there was the big flow differential at low lift (gives the charge a final swirl kick just before intake closure) I had the seat on this bigger valve made at 30 degrees. As for size the same deal was applied on the exhaust. One valve was bigger by 1 mm and one smaller by 1 mm. All the valves were made up by Ferrea in top grade material to withstand the rigors of 45 lbs of boost.
Cylinder heads for Ryan Garcia’s engine
Figure 9 illustrates two cylinder heads which were done for Ryan Garcia’s engine. The original one by me and the other by a well known performance engine shop. I was so busy doing the R&D on this project that it somehow slipped my mind to take photos until I was nearly finished. What you see here is the second head. At it’s current state it is rough cut and ready for detailing to finish the ports and chambers to their final form. At this point you might be thinking that a 1 mm up and down on the size of the stock valve may not develop much swirl. This might normally be the case but there is the port shaping to consider. By biasing the divided part of the intake port appropriately for the size of the valve concerned the swirl can be quite energetic. Add to this the 30 degree seat on the larger of the two intakes and we have swirl values about the same as a Vortec Chevy head.
Figure 10 – the left port in this shot flows about 15% more than the right but it will take a keen eye to spot the fact that the left hand port is bigger as the differences are subtle. This can be difficult to get right on the first time around and reference to the swirl meter is important. Once the form, subtle though it may be, is found, getting the other three the same is relatively straight forward. Note the port size. This is the smaller of the two Mitsubishi cylinder head variants and is still a shade on the large side. Next time around on this project the port will come in for about a 10% reduction in area.
Swirl Test Results
Figure 11 illustrates a view of the exhaust port. As might be expected, the throat (bowl) of the larger valve (right) uses a slightly different form to maximize flow.
Swirl Test Results
Figure 12 gives the results of a before and after flow/swirl tests. The stock port had almost no swirl. The small amount seen is inconsistent from port to port and is caused mostly by casting imperfections.
The big increase in low lift flow on the intake is due to the 30 degree seats inherent larger opening characteristics. From about 0.250 lift on up the flow increase is due mostly to a better port design. At low lift the exhaust flow was about as per stock. At about the 0.100 lift point the better port shape started to influence the situation and the flow increased significantly as the valve lift got higher. With almost 280 cfm from this head there is enough flow for a 2 litre engine, with a 13/1 CR to make about 300 hp normally aspirated.
The charge motion is a combination of the swirl and tumble seen here. This results in a twirl that spirals down the bore the centerline of which is displaced roughly a half an inch from the spark plug. This results in greater mixture motion at the plug than is generated by the deactivation of one valve.
One of the problems Ryan Garcia was experiencing was the fact the cylinder head, at high boost (50 plus PSI), would literally go into a meltdown phase. As a result heads were only lasting 2-3 events at a power level of some 700 RWHP. To combat this I had a thermal barrier applied on the intake right through the chamber and on out the exhaust.
DV Intake Ports:
Here is the finished DV intake ports (Figure 14 and Figure 15) with the thermal barrier coating applied. This was used from here right through the combustion chamber into the exhaust. The arrow here (Figure 15) indicates a ridge between the two exhaust valves. The top of this ridge was the level of the original chamber roof. This is about the amount the exhaust valves were lowered to reduce the cross flow during overlap.
Figure 16 illustrates a view down the finished exhaust port. Coatings here can cut the amount of heat dumped into the head casting by a very worthwhile amount.
Larger Intake Valve
Figure 17 – It is hard to see but the larger intake valve is on the right here and faces the smaller exhaust. The chamber coating cuts conduction into the head casting which is a distinct help toward head casting life in a highly boosted engine. The coating on the exhaust valve helps cut the valves bulk temperature. Keeping this to a minimum staves off detonation from a high temperature exhaust valve.
The second PolyQuad head used a later spec coating which is said to provide better thermal barrier properties yet over that used on the original DV cut head.
PolyQuad Mitsubishi Head
Seen here is the finished second PolyQuad Mitsubishi head. The red coating on the back of the intake valves has some insulating properties but it’s main asset is that carbon can not easily stick to it so the form of the back of the valve is not altered by carbon build up.
In practice the coatings appeared to work very well. The life of the cylinder heads was no longer dictated by thermal damage and to date, after the equivalent of half a dozen meetings, appears to be in perfect shape to the extent the back up head has yet to be needed.
When Ryan ran back to back tests on a chassis dyno in Florida on a very hot and humid day the PolyQuad head showed a gain of 98 hp and 105 lbs-ft on 5 lbs less boost. Ryan also later reported that the drivability (and this engine had some fairly big race cams) was greatly improved. The first dyno session netted a peak rear wheel output of 827 hp on wrinkle wall slicks. To put that into prospective it was a bad day to test if overall numbers where being sought. Also at 180 mph a pair of wrinkle wall tires absorb any where between fifty and a hundred hp depending on the air pressure and the diameter of the dyno rollers. Based on this it would be safe to say that had the car been equipped with a good high performance street tire it would have been shown at least 50 more rear wheel horses. Either way the 98 hp and 105 lbs-ft improvement stand. A later test with some more fine tuning brought the power up to 845 hp.
Allowing for the prodigious loss of power through wrinkle wall drag slicks plus some 50 horses of transmission loss it seems a fair bet that this 2 litre engine was definitely in the realms of 1000 flywheel hp. However not all this can be attributed to the PolyQuad aspect of the heads. Remember the PQ head being tested here has far more appropriate port sizing and the benefits of thermal barrier coatings. We are looking at overall gains of some 14% here. The Ford focus tests showed a gain of just on 7% at the low end with less PQ action than the Mitsubishi head so it is not unreasonable to conclude that about 50 hp of the gain was due to the PQ feature of the head. Time and the lack of a budget to research further means that this is as far as David Martin and I have gone. However the results are very encouraging.
Back in 1999 a patent application was started and these were, at great expense, granted about 2003. If you are a professional head porter and wish to use the PolyQuad concept a license can be issued. A test head is royalty free unless it is subsequently sold. A royalty of 7% will be charged on each PQ head sold. This will be based on the honor system but in case someone thinks it unlikely I will ever find out just remember I have not tens, but hundreds of thousands of readers out there that are, for all practical purposes, undercover agents!
Afterthoughts – Economy and at the Track
There is an irony here. The original purpose of the PQ concept was to increase the low speed output of a typical production line economy 4 valve engine and do so with no additional moving parts or, for that matter, cost if used as an OE design. Small, low cost 4 valve engines in the up to 1600 cc range such as used by KIA, Chevy and the like could benefit from additional low speed torque as delivered by a VVT engine but at no additional cost. Great for out of pocket dollar economy. Also since it is benign and no additional moving parts are involved nothing can go wrong from it’s use. If the 7% increase in torque is split to use a final drive ratio 3-1/2% higher (or there about) plus the other 3-1/2% to improve low speed drivability the PQ head could easily result in a 5% improvement in fuel economy. All achieved might I add at no extra cost on initial vehicle purchase.
If an engine is already equipped with VVT this does not negate any advantages of PolyQuad. What will happen is that the PQ’s ability to develop more torque and hp will simply be added to what ever advantage there was from the VVT.
Another thought to consider here is that a PQ 4 valve head looks like it is every bit a match for a typical (and much more costly) 5 valve head. Although we have looked at the PQ concept on a 4 valve head flow testing shows it to be equally effective on a three valve head such as seen of Fords modular motor. Fords 4.6 and 5.4 three valve heads are of a very sound design but for the most part very conventional. These engines are used in heavy vehicles where a gain in low speed torque would considerably improve the driving experience. The potential added fuel economy would not hurt also. In short it is early days for PQ and dyno results so far look good. This leads us to the question as to how well those dyno numbers transformed into race track success.
After dyno testing in the Orlando area (where he lived at the time) Ryan took the car to a conveniently close drag strip and did some shake down runs. All looked good so he trailer the car to Palmdale for a big final round import meet late in the year. All the top contenders for a championship were there including some really high dollar teams. Ryan’s presence there was as an unknown shade tree budget outsider. At the end of the day he proved to be both the upset and the star of the meet. He pretty much shut down the opposition to the point that there was no viable competition. In the predominantly pro classes of drag racing when you beat the next fastest car by 3-4 lengths in the lights that’s not a race, that’s annihilation.
If anyone wants a PQ head be it for a Mitsubishi like ours or any other multi-valve head then Ultra Pro Machining in the USA (704-392-9955 (who do so many cup car heads) or Austec Racing in the UK (01293 531080) are up to speed on this. The work is intricate to do on a one off basis so expect to pay about $600-700 a cylinder plus the cost of valves (which typically run about $200 a cylinder).
As of now my pick to do a PQ conversion on would be the 3 valve Ford heads. A PQ three valver would, in my estimation, outperform a ported 4 valve Cobra head at the bottom end and at least match it at the top. The increase in low speed output on a 4.6 liter engine would be about the same as increasing the cubes to 5.0 liters – but with no down side. A thought here, if you Ford guys bugged some of the big Ford tuners enough about doing a PQ three valver (you will have to refer them to this article otherwise they won’t have a clue what you are on about – maybe email them and put in a link?) and they got into a CNC porting deal on it the cost would drop a ton. It’s just a thought – depends on how fast you Ford power freaks want to go!
Share this Article with the world..
Click on the Stars (below) to leave your Star Rating and any comments.