Free-Flowing a Miata MX5

This article was first published in 2001.

Testing, designing and building a simple DIY induction system for the Mazda MX5 Miata. Better performance, lower intake temps and more horsepower can be yours for very little cost.

Despite having already clocked up over 210,000 km, the 1.6 litre B6 four of my car was still in great condition, clean running and happy to rev. The timing had been bumped slightly from the stock 10 degrees BTDC to the popular 14 degrees setting as tested and used by numerous MX5 and Miata drivers around the world, providing a slight but noticeable torque improvement (seat-of-the-pants testing only). A 2¼-inch cat-back system and a set of 4-2-1 extractors was helping dump the exhaust gases better than the stock system, opening up the top-end above 6000 rpm. Still, more power could hopefully be gained by allowing the engine to breathe in as well as out.

The Stock Intake

Starting at the left of the engine, a 400mm long, approximately 50 x 55mm oval snorkel picks up air from inside the engine bay close to the shock tower - the designers at least thoughtfully placed this as far as possible from the exhaust manifold. A resonator tube is moulded into the side of the snorkel, presumably to reduce intake noise. The snorkel leads into an airbox with a 130 x 230mm rectangular panel filter. The outlet of the airbox is a small, square hole in the lid, with absolutely no flaring or tapering used to smooth the airflow into it. This is immediately followed by a sharp 90 degree bend into the airflow meter, which is mounted on the top of the airbox.

The airflow meter is a vane-style unit, with a 50 x 50mm opening and a 70mm diameter outlet. A fairly narrow, ~700 mm long duct (made of plastic and corrugated rubber) runs across in front of the engine from the airflow meter to the intake manifold. Hung underneath this duct, close to the throttle body end, is a resonator box which others have tested and found to smooth out a drop in the torque that would otherwise occur at about 4000 rpm.

Testing the Stock System

First, it was useful to determine whether the stock system was at all restrictive. Using a water manometer (as described in AutoSpeed - do a search to find lots of articles on this technique!) the pressure drop was measured just before the throttle body at the PCV return-line connection. In a 2^nd gear WOT run to redline, a maximum pressure loss of 42cm H₂O was found, occurring at about 6000 rpm and not at peak revs (7200 rpm) as one might expect. This pressure drop was considered significant enough to make improvements worthwhile.

The question, though, was where exactly this pressure loss was occurring. To determine this, pressure readings were taken at several points along the intake system, as follows:

1. In the bottom of the airbox (after the intake snorkel, but before the filter)
2. In the top of the airbox (after the filter)
3. At the inlet to the airflow meter (after the outlet duct/bend of the airbox)
4. At the outlet of the airflow meter (the start of the long duct across the engine bay)

The pressure loss found at those points was as follows (cm H₂O):

1. 8cm
2. 11cm
3. 25cm
4. 37cm

From this we can see the pressure restrictions caused by each component (cm H₂O):

Pick-up Snorkel:	8cm	(19%)
Air filter element:	3cm	(7%)
Top of airbox:	14cm	(33%)
Airflow Meter:	12cm	(29%)
Duct to Throttle Body:	5cm	(12%)

This quickly points out some interesting facts: Firstly, the popular drop-in, reusable type of filter is clearly going to have almost no effect, gaining at best a maximum of 3cm pressure improvement. (Note that the filter tested was already quite old and dirty - and yet 97 per cent of the restriction was not from the filter!) Secondly, the duct across the engine bay, despite its length, flowed quite well. The tuned length of this part, and the existing resonator box, were regarded as desirable to keep.

Since 59% of the total restriction came before the airflow meter, putting the common cone-style filter onto the airflow meter should improve matters by getting rid of the stock snorkel and also that nasty, restrictive, right-angle bend in the top of the airbox. However, this still leaves a noticeable pressure drop caused by the airflow meter itself. It also means that the engine will be sucking hot, low-density air from right next to the exhaust manifold.

An apparently popular upgrade done on these cars is to replace the airflow meter with a larger one of the same physical design. Such a meter is used on the 86-88 Mazda RX7 - it has internal measurements of 50x63mm (a 26% increase), and it plugs straight in to the wiring loom. However, this bigger airflow meter does not fit to the standard airbox, which warrants a new filter system. This means that the effect of swapping the airflow meter is hard to verify alone, as it also removes the restriction caused by the top of the airbox.

A digital thermometer with a remote probe was used to collect intake temperature data, with the probe placed at the mouth of the snorkel. It was found that in stop-start traffic, intake temperatures 30-40 degrees C higher than ambient could easily be obtained. During some particularly hot summer days (~35 degrees C) the thermometer could be fairly easily clocked off the scale, which read up to 70 degrees C! Cruising at speed gave temperature readings typically 20 degrees C above ambient. One of the easiest ways to overcome this in an early-model MX-5 is to pop up the headlights, which act as massive cold air scoops, but to the detriment of aerodynamics when driving at speed.

The Plan of Action

One popular aftermarket intake available in Australia for MX5s is the Loch Stewart airbox. (Pics of this design can be found at http://www.miata.net.au/loch.htm). Replacing the entire airbox up to the airflow meter, it picks up cold air from the cowl below the windshield through a large 3-inch tube, and feeds it in a straight path through a (stock size) flat panel filter to the airflow meter. The cowl is a known high-pressure area - notice how air is pushed through your cabin vents when driving at speed, even when the HVAC fan is switched off.

The theory behind these particular intakes is sound, though in my opinion they suffer two problems. Firstly, by being constructed from fibreglass over a mandrel, the raw external appearance is well below what one would expect from a product costing hundreds of dollars (the inside is however beautifully smooth). Secondly is of course the price - if something could be built just as effectively for less, then it would be worth doing.

Other aftermarket cold-air intake systems available from the US (eg Jackson Racing, Racing Beat) draw cold air from over the top of the radiator, but given the current value of the Aussie dollar, these are prohibitively expensive. The LHD Miata, it should also be noted, does not have the ability to be fitted with as simple a cowl-induction intake as the Loch Stewart, as the brake master cylinder and booster are located where the intake would go.

I decided to follow the same principle as the Loch Stewart system, which of course involves the painful decision to cut a big hole in the sheetmetal of your car. This panel is however non-structural, so not likely to cause any weakening of the chassis, nor cause possible rust as might a hole through the wheel guard. By keeping the end of the intake above the bottom of the cowl channel, there should also be minimal chance of sucking water into the intake.

Gathering the Parts

I also wanted to incorporate a larger airflow meter. Wreckers were searched for a cheap RX7 airflow meter, but the cheapest source turned out to be a fellow AutoSpeed member who answered the call when a Wanted notice was posted on the For Sale forum.

Working out how to fully enclose the filter (so that cold air could be ducted to it) was a dilemma, and about the only easy choice I found was a Simota Super Power unit that incorporates a plastic heat shield around a cone filter. Analysing and measuring photos from the web showed that the intake hole to this filter was approximately 100mm - coincidentally a popular plumbing size. The filter (with a carbon fibre finish) was ordered from the Autospeed Shop for $81.

From a large hardware barn, a neat PVC 100-to-65mm adaptor was found at a mere $2.50. A (minimum order) one metre length of 65mm PVC plumbing tubing was bought for another $10, the idea being to use a short length of this to go back through into the cowl. 75mm tube would have been nice, but I couldn't find any 75-to-100 mm adaptors. Given that the length of tube required was about 100mm, I figured this was a minimal compromise. When the filter arrived, it was found that the plumbing adaptor slid straight on like it was custom-made to fit! I was expecting to have a slight tolerance error that could be corrected by softening the PVC, but this was extra-good fortune.

The Simota filter uses the ubiquitous 3-inch outlet duct size. While adaptors can be bought cheaply (for about $25) to fit such a filter, I had the materials available to build my own. By notching and pressing a short piece of 3-inch aluminium tube in a vice, the end could be squashed into a match for the RX7 airflow meter's rectangular intake (the inner circumference of both pieces being almost identical). This provided a very smooth transition with no square edges. This tube was bonded to a flat plate that had suitable mounting holes for the airflow meter, and the adaptor was complete.

Installation

At last, it was time for assembly. With the stock airbox removed, a 2½-inch holesaw was borrowed and used to cut a hole through the cowl wall. This hole required enlarging, using grinding tools on an electric drill, until the 65mm PVC tube could fit through. A 2¾-inch (70 mm) holesaw would have been perfect, but I wasn't going to spend $50+ buying one for a one-off job! The bare metal was finished with touch-up paint, and a length of plastic edging strip fitted inside the sharp metal edge to prevent damage to the PVC tube.

A short piece of the tube was cut, and the inside of one end chamfered slightly to hopefully smooth the flow of air into it. The trick of heating and flaring out the tube into a bellmouth could not be used since the tube could only be fitted through the hole by inserting it from the engine bay side. The tube was fitted to the PVC adaptor, and both were given a coat of Penetrol (penetrating oil - to help the paint stick without chipping), followed by a coat of silver spraypaint. I thought silver would look the best purely for aesthetics, though theoretically it should also help keep as much heat out as possible.

The whole intake contraption (snorkel tube, PVC adaptor, shrouded filter, airflow meter adaptor and the standard [see breakout box below] airflow meter) was bolted together and fitted to the car. Two small metal brackets were used to support the airflow meter to existing mounting holes, now vacated by the stock airbox. The only other matter was to relocate the windscreen washer bottle to the wheel guard, since the hole in the cowl goes right through where the bottle originally sits.

Testing

Back on the road, the most immediately noticeable change was the loud induction roar - definitely a more angry sound than stock, and quite pleasant in an aggressive way at WOT. Seat-of-the-pants testing confirmed more urge in the midrange, and also a willingness to run right out to the engine cut-out at 7200 rpm. Fitting the manometer showed that the total pressure loss had dropped from 42cm to 26cm H₂O (a 38% decrease). This meant that the restriction upstream of the airflow meter had improved from 25 to 9cm H₂O. Potentially more could be gained by fitting the RX7 airflow meter, but as explained in the breakout box, that setup was not tested.

Using a stopwatch, and hand-timing some acceleration runs was carried out to compare the new system with the stock one. To avoid the launch variations of a 0-100 km/h run, I chose to start at a cruise at 2000 rpm in 2^nd gear, punch the throttle wide open, and run out to 6000 rpm. Redline would have been nice, but I didn't want to break the speed limit by too much... With the standard intake, over 8 runs back and forth I had averaged 6.1 seconds. Fitting the new intake system dropped the average time to 5.8 seconds, a 5% improvement. While the 0.3 second improvement may not seem much, it was highly repeatable.

Temperature measurements were made with the probe fitted at the entrance to the snorkel. Under nearly all driving circumstances, the intake temperature was within 5 degrees C of the ambient temperatures - obviously a small amount of heat-soak from the engine bay was still occurring. However, the improvement was still typically about 30 degrees C cooler than before. Using the rule of thumb that a 4 degrees C temperature drop is worth an extra 1% power, this would imply that I could expect up to 7% extra power over the standard system. Parked at idle, the temperature got closer to 10 degrees above ambient (hard to judge since I didn't have a reference thermometer, but still well below the stock setup), and immediately dropped when moving again.

Later, the opportunity arose to dyno test the car at SelectMaz in Epping, Melbourne, with the MX5 Club of Victoria. The graphs show that there is a moderate torque improvement in the midrange at about 4000 rpm, and also a good top end improvement above 5500 rpm. Peak rwhp went from 75 to 79hp - a 5% increase. I can tell you that the graph also compares very well with the aforementioned Loch Stewart system which was fitted to other cars tested on the same day, giving almost identical power outputs (very similar exhaust systems were also used on those cars).

Note also that the bonnet was 'popped' to the safety latch for both runs, giving slightly cooler air to the engine bay - if the bonnet had been fully closed, I would expect the stock intake to have sucked hotter air and given worse results, and the corresponding improvement to look even better.

The Case of The Uncooperative Airflow Meter As stated earlier, the plan had been to use an RX7 airflow meter. Since they are designed for larger engines and therefore greater airflows, they have a stiffer spring on the flapper door. Without adjustment, this would make the smaller engine run lean, since the ECU would think there was less airflow. This can be compensated for by loosening off the pretension on the spring. I have a Jaycar Air/Fuel Mixture meter fitted to my car, and used this as the basic tuning tool for setting up the airflow meter. In order to prevent the engine from running lean at WOT (when the ECU goes into open-loop mode), the spring needed to be loosened by a total of 9 clicks of the tensioning wheel. This, however, had the unfortunate effect of leaving the flapper door partly open when the throttle was closed, so the ECU thought there was notable airflow and injected fuel accordingly. The closed-loop compensation was clearly not strong enough to adapt for this, and the idle quality and fumes were unacceptable for what is a daily driver. In a race car, the poor idle may be acceptable if one is looking for the absolute minimum of flow restriction at the top-end. As an alternative to loosening the flapper so much, I tried richening up the fuel mixture by altering the air temperature signal (the sensor being incorporated into the flow meter, and used to calculate air density). By wiring a potentiometer in series with the sensor, I could add up to 10 K-ohms of resistance to the signal, mimicking an air temperature of down to about -20&�°;C. By doing this, I could adjust the spring tension back slightly to 7 clicks softer than standard, but this still gave a poor idle. Setting the spring 6 clicks softer than stock just gave an OK idle (with a dithering O2 signal), but still ran lean at WOT. Altering the coolant temperature signal in a similar fashion can also be used, but I was afraid that the same rich idle problems would occur, and also didn't feel like hacking into the wiring loom. Clearly, to implement the bigger airflow meter properly, it would be desirable to electronically intercept the flapper voltage signal and adjust it to suit, but this requires some detailed electronics work, or an expensive interceptor module. For now, the stock airflow meter is satisfactory.