Extractor Systems – How They Work, What to Look For
The workings of an extractor are frequently misunderstood and misrepresented, so herewith a bit of explanation.
For reading material, one of the best books I’ve found on the topic remains Phillip H Smiths “Scientific design of exhaust & intake systems” (even though it is over 40 yrs old) in which he presents a lot of detail about wave/pulse action in exhaust pipes.
In summary, the most important effect of a good extractor which is often omitted from the explanation is its role in maintaining separation between the various cylinder in order to prevent them discharging into each other (as happens in an all into one manifold). This has the effect of displacing fresh inlet charge with spent exhaust gas, thus effectively reducing the capacity of the engine….and we all know what that means!
There are three key characteristics of wave action in pipes that many people overlook/can’t get their head around:
- the wave/pulse is separate from and quite independent of the net movement of the exhaust gas moving along the pipe
- pulses turn corners even though the net flow doesn’t.
- on reaching the open end of a pipe the pulse is not only partly reflected back into the pipe (a startling concept to the uninitiated) but it does so as a negative (ie vacuum) pulse.
Looking at the first of these points, the only connection between pulse and gas is that the gas is the medium through which the pulse is travelling. Very much like the action of waves in the sea: The waves move in a direction which is independent of any current in the water.
The waves can move with the current, against the current or sideways across it for that matter as we are dealing here with two dimensions.
Another example: Sound travels through the air independently of wind direction. We can hear noises even though we are upwind of them. Of course, there is more attenuation in this instance because the sound wave has had to travel trough more air in order to reach you (and therefore dissipated more of its energy) than would be the case if the sound was travelling with the wind.
The second of these three points is highly significant: In a now outdated all into one manifold such as the MGA or Midget had originally, the high pressure pulse emitted when the exhaust valve opens travels not only down the exhaust pipe but also turns the corner and heads up the other branches of the manifold.
Now if you think about it a little, you’ll realise that the fact an exhaust valve has opened means that piston is nearing the bottom of its stroke … which means that another piston is nearly at the top and is about to close its exhaust valve. Our vigorous pulse gets there just in time to cram a whole lot of exhaust gas back into the cylinder just before the closing valve snaps shut – which of course means that there is that much less room for new inlet gas and its consequential power. As each valve in turn opens and emits a pulse, somewhere else a valve is just about to close cops a lungful of dead air!
This effect is far more deadly than “back pressure” will ever be, as it effectively reduces the capacity of the engine. It happens in not only the prehistoric Midget type of manifold, it can happen in all manifolds no matter how acute or streamlined looking the junction is that the pulse has to turn around.
The first and principal aim of a good extractor is to eliminate this effect and it’s relatively easy to do: make the distance from valve to adjacent valve long enough and the pulse arrives after the at-risk exhaust valve has closed – ie too late to do any damage. The effect is more marked in lumpier cams because the longer duration means that more of the pulse has time to get into the cylinder before the valve closes.
However, because the speed of sound (ie the pulse) is fixed for any given exhaust gas temperature the length needed to eliminate this effect will vary depending on engine speed. At high speeds you don’t need much pipe length because the crank’s turning so fast that the pulse doesn’t go very far in the shortened time between one valve opening and another closing. On the other hand, at lower or medium revs, the pulse has lots more time available and so will need a longer path to travel if it’s to find itself denied access to the cylinder by a valve which has just closed.
This is why all-into -one configurations (which generally have a limit on their primaries for practical reasons) usually aren’t so flash at mid-range but perform well at the top end, while a 4-2-1 (or 3-2-1 aka Long Centre Branch [“LCB”] for BMC Siamese port engines) performs well in the mid range because the extra length of the secondaries is enough to separate the offending groups of cylinders completely at the lower speeds. The MGB was one of the first volume production 4 cyl cars to make use of this with its split manifold and twin front pipe; now there are few that don’t take advantage of the extra effective capacity offered by it.
There is a further effect of the 4-2-1 setup (hereafter called LCB thanks to my MGB roots!): Each cylinder’s primary is paired with (what I call) its “mate” – ie the one whose piston moves up and down at the same time. The ONLY cylinder which will not have its exhaust valve open at some time in common is its mate, even with race cams (unless they exceed 360 deg duration!!). (Think about the cycle – when one is on exhaust, the other is on compression and nowhere near exhaust opening). This is why the siamesing of the exhaust in BMC engines made no difference to their performance at all (unlike, unhappily, the siamesed inlet ports!)
Thus in an LCB configuration, the branch to the mate cylinder is always (because the valve in it is closed) effectively just a blanked off piece of pipe. The pulse gets to the end and is reflected back down to the main pipe again. (There are also multiple subsequent reflections which can be significant but we’ll keep it simple at this stage.) The end effect of this is that instead of an intense pulse of short duration we get a flattend pulse – longer and less intense. This is of significance when we are considering the third of our three features shortly.
If you think about the above, you will see why the primary pairing is 1&4 to one secondary, 2&3 to the other. If it was 1&2, and 3&4 it would give poor results at low speeds because this pairing will ALWAYS result in the dreaded pulse interference between cylinders with opening & closing valves (unless the primaries are very long), with consequential reduced filling and reduced output at anything other than very high revs at best.
This inevitable interference is the same reason that BMC’s siamesed inlet ports with the same pairings produced such limitations on power output and why the scatter cams give more power by reducing the common time valves are open in the inlet port and allowing one to rob its neighbour.
Turning now to effect no 3 – the negative pulse reflecting from the tail pipe. Trick here is to get the negative pulse back to the head just as an exhaust valve is about to close …. yep, this time we DO want it there because it is a friendly vacuum pulse ready to suck out any remaining remnants from the cylinder, leaving even more room for new fuel mixture, and with any sort of respectable overlap to also give the said inlet charge a good kick in the pants to get it moving better.
The raw short peaky pulse initially produced by the opening valve would reflect to a very respectable vacuum pulse but of such short duration that in a given length of exhaust system its timing in arriving back at the valve would be only over quite a narrow rev band – good for tuning to peak power! On the other hand the long low pulse would result in longer periods of negative pressure in the manifold (though much less juicy in its strength of vacuum as the short pulse of course) so its effect be spread over a wider rev band – ie be good for lower and mid revs while still respectable at the top end.
All the above indicates why LCB’s are good mid range systems while still delivering respectable at peak power:
- They have secondaries as well as primaries so there is lengthened separation between “at-risk” cylinders compared to an all-into-one.
- The long low vacuum pulse helps over a wider rev band.
The lengths of the primaries and secondaries are the principal feature in determining how good an extractor will be, aided and abetted by the pipe diameter. Smith’s book specifies 18″ as the length of primaries in an LCB type system, with at least 2 feet in the secondaries. This is close to what the standard MGB system has.
Many so-called performance systems don’t produce much more, if any, power than the standard system because their pipe lengths and internal diameters are basically the same. Most commercial extractors have reasonable pipe diameters, but are a bit short in the secondaries.
If you want to significantly improve your exhaust on a B, then the best approach is to go up a size (to 1.625 ID) in the twin front pipe and also increase the pipe size in the rest of the system to either 1.75 ID or 1.875 ID – bearing in mind that the bigger the pipe the harder it is to silence so there will be practical considerations.
For a Midget MGA, or T-type, fit an LCB type system with the secondaries as long as the chassis will permit (up to that 24” mark).
Finally, a couple of points about silencers:
- Straight through silencers as commonly used on MG’s come in two general types. In the good type, the holes in the pipe trough the silencer are neat round holes that look as though they have been drilled. The bad type has the slots punched in from the outside of the pipe, leaving a series of vanes about 3mm deep protruding into the pipe. Why bad? Rolling road tests show they knock off 5-10% of power compared with the good type. And the sound doesn’t have that same nice MG tone.
- Silencer volume is important in achieving suitable attenuation of noise. If a silencer is too small, it won’t work effectively.
- For the Midget, MGA or T-type, the standard MGB rear silencer is just right for the job – suitable pipe size, nice tone, and enough volume for reducing the noise to unobtrusive levels and also reasonably priced. It is however a little small for an uprated B system, though fine for a standard one.