Automobiles Forum banner
1 - 20 of 24 Posts

47 Posts
Discussion Starter · #1 · (Edited)
Prompted by some test results that didnt live up to expectations ive endeavoured to find out why, this article being the product of some considerable effort and time.

Please bear in mind that this is still a work in progress and is subject to alteration, in no way do i claim to have all the answers.
I still have a fair few tests to run, at least as many tests again (lots!) with the heads ports being used in their usual 2 runner configuration, this all takes time.

There is as far as i can see almost no data published anywhere regarding these heads, hopefully this article can help to fill the vacuum.
I have done my utmost to ensure the tests are run exactly in the same fashion every time and have attempted to eliminate as far as possible any errors that can creep in.
Please also bear in mind that flowtest figures vary on different flowbenches so i also dont claim ultimate accuracy or ultimate specification, just what ive tested and found on my own equipment in my own time.
Exhaust valves and ports arent considered in this article as ive found them very easy to improve upon in comparison to the intakes.

Basic specifications of the head as follows.

Intake valve diameter=32.00mm.
Top angle- not well formed looks like a left over from machining the 45 seat.
Seat angle =45 degree.
Lower angle =75 degree.
Throat diameter=26.90mm.
Port width of controlling section at the splitter=24.50mm.
Port height at splitter =31.24mm
Total internal width of port at splitter= 49.00mm.
Port entry width= 44.70mm.
Port entry height= approx 31.00mm.
Valve angle =24 degrees.
Coil bind at 12.00mm lift. (inner/outer springs).
Intake guide oem depth =10mm (from spring seat machined platform).
Inner spring free length=34.00mm
Outer spring free length=40.00mm
Inner spring compressed height=29.70mm (approx due to valve fitted height)
Outer spring compressed height=31.00mm (approx due to valve fitted height)
Shim under hardened spring seat =0.62mm
spring seat thickness= 0.93mm
1.55mm total under spring.
Seat pressure-Inner spring =20.0psi
Outer spring =45psi
Total 65psi seat pressure per valve.

Component weights.
Valve- 49 gr.
Cap- 19 gr.
Collets-2 gr.
Hydro tappet-70 gr.

Testing was carried out on one port runner only unless otherwise indicated, to eliminate as many influences as possible from the other open port.

SSR (short side radius).

Initial tests indicated that the short side radius was somewhat insensitive to modification when testing for bare port flows.
For example, smoothing and cutting back into it in order to improve the flow, a method which works on other heads showed no apparent changes recorded when testing in this mode.
With stock intake port pairs flowing around 194cfm @28", the usual modifications such as reducing guide boss height and surface smoothing can push that to 214cfm @28" depression, much of the restriction being imposed by the guide bosses.
The data i have gathered indicates that modifying the short side radius by cutting into or laying it back it will not work as it damages the valve-in response at many points in the lift curve, rendering improvements gained from other modifications less than optimal to say the least, which was the entire issue with some flowtests results.

Flow actually benefits by lengthening the ssr, adding material to the port and raising the port floor by around 3.5 to 4mm where you get a good gain at 7 to 11mm lift, the maximum i personally recorded using this method was 228.8cfm @28".
I also found that applying a couple of plasticine "humps" at a critical point in the port has a similar effect to raising the port floor but i need to do more work on that aspect to confirm if further infilling works better or makes the port perform worse, i think at this stage given what ive found already it will improve slightly.
Looking at the port from the viewpoint of air going into it, it appears to be a little counter intuitive to see less of the valve in the modified port than the stock item, however looking at a sectioned head shows a different story.

Wood Gas Auto part Font Symmetry

In short i believe the ssr shape is already wrong as cast and needs material adding not removing in order to make it flow better.
In that case youre really left with two choices- A) leave it well alone and live with it- acceptable considering its adverse effects on flow and the fact that dyno results with it in that condition show its possible to still make great power or B) infill the port and attempt to optimise the flow, which is pretty hard i can tell you but still do-able.

The photo shows the stock shape of the port floor vs what works better, the reason for the turbulence (and its tell tale noise) becoming apparent.
It appears to me that airflow is staying attached to the ssr for longer and also causing air to be pushed higher towards the longside so utilizing more of the available valve area.

Cylinder Gas Tin Gun accessory Nickel

This is the same port just painted to better show how the profile would look without the distraction of the parent material and some measurements of material thicknesses.
Motor vehicle Automotive tire Gas Automotive wheel system Automotive exterior

Wood Asphalt Gas Road surface Font

As you can see, i contend the airflow separates just prior to entering the valve throat (red arrow), effectively shooting across the back of the valve rather than staying attached to the ssr due to its shape not being conducive to keeping the flow attached.
That creates turbulence which in turn chokes off flow and makes the port noisy.
Lengthening the ssr by raising it and the port floor allows the airflow to stay attached to the curve and as a result flow goes up and turbulence goes down.

Heres a drawing that also shows the desired shape as opposed to the red outline which is what is cast as stock.
This all points to the inevitable conclusion that the port needs to be raised in the head to give a straighter shot down the port, unfortunately theres no real scope for doing that on these heads due to way theyre cast.

Port oscillation is a very noticeable phenomenon on the 10 valve heads, whereby using a 30 degree backcut valve in a stock port and seat combination (which works on the 20 valve heads) makes a hell of a racket in 10 valve ports as the port oscillates/whistles very loudly and flow drops off well below stock levels.
The noise you can hear is purely caused by a mismatch in the port and valve combination, that being a minor 30 degree backcut on the valve.
That noise is indicative of a flow killing combo!

10v head test -Port oscillation.

Valves and seats.

Valves respond very well to a simple 30 degree back cut on the reverse side of the valve head leaving a contact face of 1.5mm.(gaining approx 10.0 cfm) at the mid point of lift.
Probably best not to be tempted to make it less than this as seat life will suffer for likely little improvement, indeed, valves removed from another head showed a pronounced wear ridge hammered into the remains of the contact face as the seats had been cut minimally 1mm width.
Audi actually specifies an intake seat width of 1.5 to 1.8mm.
A good rule of thumb dictates that 4.5% of valve diameter will yield a seat of around the required width for robustness and give close to optimum flow on intakes.
So 4.5% of 32mm = 1.44mm, just 0.06mm different from Audis own minimum dimension.
Refer to the photo for a closer look at how this backcut modification looks.
Tool Hand tool Composite material Nickel Metal

The red arrow on the left shows a ring of material in brown just prior to the seat.
The red arrow on the right shows the outer position of the valve seat.
Zoom in to the blue square and look at the edge of the valve.
Its important to note that the outer periphery of the seat doesnt go all the way to the valve margin, in other words, the valve is 32mm diameter but its acting like a smaller one because the seat O/D is smaller.
Its not much but its wasted area that can be utilised for free due to the intake seat insert being fractionally larger than the valve,(32.50mm).
The inner and outer ring of the seat can effectively be moved further out, and trimming the seat to 1.5mm then allows us to put in a more substantial 30 degree backcut, its not very wide, about the same as the 1.5mm seat but its wide enough to make a not insignificant difference.
Running the numbers gives us 31.42 square mm extra curtain area per valve or 314 square mm extra over the entire 10 valves- not to be sniffed at- its free.
The 30 degree back cut angle on the valve also works when used in an unmodified port.
The graph shows the results of this minor modification on a totally unmodified port and head, the gains being well worth it for what little it takes to achieve it.

graph lost

3 angle seats.

The stock seat architecture is a none too distinct top cut as far as i can see,looks like a remnant of maching the seat angle, with a 45 seat and 75 lower into the parallel section.

A 30 degree top angle, 45 seat, 60 degree additional angle then 75 into the parallel section of the seat insert gave absolutely no measurable difference when using a stock valve profile.

This didnt work any better than stock.

Automotive tire Light Water Fluid Bird

This is the stock type seat arrangement recut to new condition.

Kitchen utensil Rim Automotive tire Cookware and bakeware Gas

Around 2cfm was gained at 7 to 11mm lift when used with the backcut valve, but a similar loss was also recorded in the range of lift 4 to 6mm, a mixed bag.

Green trace is the addition of a 30 degree backcut valve with the 3 angle seat combination.

Graph lost.

Valve guides and bosses.

Using a modified seat- 30-45-60-75-90, a 30 degree backcut valve and an infilled port floor with the valve guide at manufacturers installed height- of 10mm.

The guide boss on this port had already been reduced somewhat due to a 3rd party modification of it prior to my aquiring the head along with the ssr being laid back, ie, cut into, so this test served primarily as a baseline to look for a trend rather than a gauge for optimal flow results, although a port otherwise unobstructed by the usual architecture was also a consideration to judge the impact of the guides and bosses.

With the guide depth at 11.25mm (higher out of the top of the head to effectively shorten it in the port) no change in flow over the lift curve at all was noted.

Removing the remainder of the guide boss to the surrounding material levels with the guide at oem 10mm height gained some flow at 1-7mm lift but the guide looks very wrong in the port, ie too long.

I would not wish to modify ports to that degree due to the lessening amounts of material supporting the valve guide.
There appears to be around 10mm of supporting material left inside the valve guides bore in the head on the shortest side of them when the boss is entirely removed and a couple of mm material under the valve spring platform.
It may create misalignment issues on fitment of new guides and less material to conduct heat away up the stem and into the head.
There may also be a risk of cracking the head if the interference fit is too tight ( with aftermarket items sized at anywhere between 12.06 to 12.09mm- The usual Audi O/D of guides on these heads is 12.04mm) as theres less material in the vicinity when the boss is entirely removed.
Probably the best way to approach this is to remove the material at the sides and ramp at the front of the boss and streamline it rather than remove it wholesale, even though that would be quite the easiest method.
Shaving the guide on a lowered guide boss gave a gain towards the top of the lift range but would be too short to support the valve for road use i feel and for not a huge gain.
Infilling the port at the point the splitter merges on the longside turn with the guide boss had no impact on flows, in other words virtually no activity in that area.


At first sight there appears to be some chamber shrouding on the intake valves but due to the borelines close proximity to it of approximately 0.5mm there seems to be little scope for improvement unless the bore was expanded.
Removal of material at that close point shows a small drop in flow at the higher lifts so maybe best to not proceed to removal of material in that area, instead favouring simple surface smoothing in the vicinity.
Some very small improvements lower down in the lift range were noted but were offset by the losses elsewhere.

Small gains in flow were seen when the ridges around the intake valve seats were reduced and blended, in the range of 2 to 7mm lift.
Further tests of material removal on the squish pad on the raised area on the ssr, which was cut back in order to attempt to promote further flow at low lifts failed to do so, in fact it promoted a flow loss at higher lifts of a couple of cfm, so not a modification to undertake.

Red trace is the result of cutting back the chamber wall in the red areas on the photo.
Product Rectangle Slope Font Parallel

The following graph shows a fully modifed port and was tested as supplied, complete with shorter guides, lowered guide boss, castmarks cleared, 3 angle seats, backcut valve and ssr cut back.
It was this kind of result that prompted me to try and find the reason for such a disappointing result, given that the effort put in to making the port flow better than stock would have been a not inconsiderable one as i know only too well.

Blue trace is the modified port and seat etc.(one side only).

Rectangle Slope Font Parallel Pattern

I also ran a few tests on a head that was entirely stock in all respects and using both ports, the results of which appear below.

Rectangle Slope Font Parallel Screenshot

The first graph is a typical kind of curve from one of these heads although its interesting to note that on other ports tested large differences appear such as sharper changes of gradient and differences in flow at mainly the upper end of the lift range, from around 7 to 11mm which i believe to be the result of casting differences and surface shape at the ssr.

Graph lost.

The second test shows the result of a backcut on a stock pair of valves which were then run in the same port as previously.
As you can see, for very little effort we get a good flow improvement off the seat and most of the way up the lift range where itll count.
The seat inserts in the head were of stock profile.

Lastly, the ports have a couple of "humps" formed from plasticine which divert the air to a more adavantageous path inserted which gives the final result.
What i believe is happening is the humps are making the air perform as it would if the port floor was raised, indeed even with a raised port floor its still possible to get a little more flow with this which leads me to think the floor could do with being raised a little higher, so thats just another series of tests to run again later.

Photo is of a previously modified port just to illustrate the "humps" installed for testing.

Ill add in further info as i get more on these tests but theyll be limited mostly to assessing changes to surface finish and streamlining the guide boss, bulletnosing the guides etc.
Widening the port at the splitter has never registered any gains in flow when testing in bare port mode, however, its quite possible that some differences in figures may show up under testing with valves fitted and performing the usual lift tests, its also possible that it may have an effect on flows at higher gas speeds, but for now thats something i cant test for.
As you can imagine, all this is extremely time consuming, its something i will report back on later when ive had more time to evaluate.

Finally, because the opportunity arose i checked the airflow distributions on the following exhaust manifolds.
Stock AAN, Wagner S2 and Wagner sport versions.

I also tested an AAN Intake vs ABY.

Thanks go to Brian Tonks for the supply of a scrap head ( it was battered to bits in the one chamber due to a departed valve head -very messy event!) which has provided many of the results shown here.

I hope this is a useful article.

688 Posts
Awesome data and an awesome read! Thank you and well done!

1,421 Posts
So looks like as far as headwork smoothing and backcutting the valve offer the most 'bang for your buck'. I'd really love to see the cheapo ebay exhaust manifolds tested against the rs2 and wagner manifolds, as well as the stock manifold. Someone should ship him one!

4,680 Posts
thanks for the effort. going back to turbulence, do you think it is possible there is some benefit for combustion's sake? maybe a more complete burn, fuel economy, emissions or detonation resistance?

6,004 Posts
One of the best posts on this forum in yearsssss.


162 Posts

47 Posts
Discussion Starter · #9 ·
Glad its useful! :)

PKW- I couldnt say regarding the turbulence, but theres always got to be a better way to achieve that better burn.
Its a little like the conundrum of silencing.
You can get a silencer that works by restriction cutting down on flow to get the noise down or you can go for acoustic mirrors and reflection/absorbtion to do the same thing without the losses.

The one idea that i saw from David Vizard that imparts swirl (usually dont happen on a multivalve head- you tend to get tumble) is to use a larger intake valve and exhaust valve on opposite sides of the chamber so it biases flow to take on the swirl that youd never normally see, but keeping with the best flow down the ports also.

I did also modify a toyota turbo intake valve thats got a 6mm stem and 34 diameter head with a dummy guide to see what effect that valves in an audi head...whats the world coming to? :D ...tends to boost that low lift flow more heavily so maybe thats an idea we could eventually incorporate into our beloved 20 valvers.
Only thing we dont know is if itll make more power by making more swirl.... "makes mental note to get a swirl meter for the bench"....

42 Posts
I'm running a 3B engine in my coupe with my variation of Wizard's polyquad head. I only have the different size intake valves and ports shaped accordingly. It's a fully built engine, only the exhaust valves and both cams are stock items, about everything else has been changed or modified to some level. Lot's of changes that should move the powerband up the revrange, but it still pulls very strong off idle. I made too many changes at the same time to be able to tell if it was the polyquad thing or just the other porting work and mods with the aim of peak power at 7000 rpms. Now I have a cheaper test mule with a 2.0 16V ACE-engine to play with and I'm going to test the head work only polyquad vs regular porting tricks.

1,395 Posts
What a great read; I know I'll be coming back to read more in the future.

Thanks for doing this work, and sharing with the community.

242 Posts
Wow this is amazing. Mad props

47 Posts
Discussion Starter · #16 ·
Interesting post that, dosent quite appear to have the same intakes as the aan and variants heads tho, seems wider on the splitter and maybe steeper but could be the photos i suppose.

90 Posts
THANK YOU for this awesome Thread!

Seriously, I love data like this, takes all the bullshit out of everything.

Glad to be here, just picked up a stock 95 S6 with 209,000 miles.

Can't wait to start modding.

4,947 Posts
Very interesting that the 3b intake not only flows a bit better than the AAN but also more evenly.

5,940 Posts
Justin517 said:
Very interesting that the 3b intake not only flows a bit better than the AAN but also more evenly.
Yes, very interesting indeed, particularly in the light of the vastly different numbers "other" tests have shown:

Perhaps the former was more intended to generate sales and not purely scientific in its intent.

Really, a 7 CFM imbalance (3B IM) is hardly worth going after for the home porter, considering the ever-present potential to make things worse. Keith (RockinV8Q) swears up and down that he experienced a dramatic improvement on his 3B when he has the #1 end of the plenum raised, which, if true, suggests that there was more than 7CFM to gain at the small end of the plenum.

...Just my ideas, wadded up in a spitball and hurled at the wall.

1,421 Posts
AHHHH!!! Please fix the pictures in this thread. This is of vital importance...
1 - 20 of 24 Posts