Category Archives: Cars

Car Performance & Handling: Swaybars!

It’s been several years since I autocrossed or owned a high performance car. I still like fun to drive cars, I just don’t have the time for it anymore. I finally got around to doing the first performance upgrades that I’ve done in years. Back in my SCCA days, the first 2 mods anybody did was (1) tires, and (2) swaybar.

Swaybar 101

When a car has a swaybar, in order for the body to roll L or R it must twist the swaybar. Stiffer swaybars reduce L-R body roll without affecting the spring or shock rates. If both wheels hit a bump and move together, the swaybar does nothing. It only kicks in when the L and R sides try to move differently. When one side (L or R) tries to move up or down, the swaybar forces the other side to also move up or down too. How much, depends on the swaybar’s stiffness.

A stiffer swaybar reduces body roll, which reduces weight transfer, which reduces overall traction. Yet at the same time, it improves response and agility. So it’s a trade-off. Put differently, you need to allow some body roll to get good traction, but too much of it reduces response.

Adjustable Swaybars

When the car’s body roll twists the swaybar, it does so through connecting arms. Longer arms give the body roll forces a longer moment arm, making it easier to twist the bar, making a softer bar (lower rate) with less resistance to body roll. Adjustable swaybars usually have several points along their arms where you can connect the end links.

Must both of the swaybar arms have the same length? Imagine if one arm is longer than the other; that is, connecting the end links to different mounting points on each arm. Newton’s 3rd law says the torques exerted are always equal and opposite, which seems to suggest that an asymmetric setup could give symmetric roll response. In this case, an adjustable bar with 2 holes actually has 3 different rates, or with 3 holes has 5 different rates.

However, while the torques are always equal and opposite, even when connected asymmetrically, the moment arms are not. And the torque on the bar exerts a force on the opposite side through its moment arm. So connecting the swaybar asymmetrically would create asymmetric roll rates: stiffer to the L than to the R, or vice versa.

The conclusion: always connect the end links to the swaybar using equal length arms.

Tuning the Response

Most cars are designed to understeer: that is, under most conditions the front slides before the rear does. This is easy to control, especially for unskilled drivers. But skilled drivers find excessive understeer to be less fun and even annoying. Excessive understeer makes a car less responsive. As a general rule:

  • A stiffer rear swaybar reduces understeer, increases oversteer.
  • A stiffer front swaybar reduces oversteer, increases understeer.

This is all relative. Most factory cars are too soft overall and also understeer, so a stiffer rear swaybar is ideal. But if the car is too stiff overall and also understeers, you might use a softer front swaybar.

Mazda 3

My 2014 Mazda 3 is actually pretty fun to drive, for a FWD economy car. At 36000 miles I finally had to replace the OEM tires. While I was doing this I figured I’d also install a stiffer rear swaybar.

This is such a common car there are many options. I used a 22mm Progress bar. It was relatively inexpensive and came with new bushings and brackets to handle the larger forces. The OEM rear swaybar is 18mm diameter with a rate of 334 in-lbs. The Progress is 22mm diameter with rates of 772 in-lbs (about twice as stiff), and 1,015 in-lbs. (about 3 times as stiff). This bar has excellent build quality and perfect fit with the end links pointing straight up and down just like they do with the OEM bar.

The soft setting made a noticeable yet not a huge reduction in body roll. I quickly shifted to the stiff setting which is fantastic. The net result was a BIG improvement in handling. Less body roll, quicker turn-in and more precise handling. There’s no difference when driving down the road. But as soon as you start cornering it feels like a whole different car. While it is much more responsive, it’s also a bit twitchier, so it’s not for inexperienced drivers.

Subaru Forester

My wife’s 2018 Forester is a lot softer than our 2004 Forester. It drives more like a bus than a car. Over the years, Subaru softened the suspensions. Subaru has OEM swaybar replacements in 2 sizes: 19mm or 20mm. Stock is 16mm, so the 19mm is about twice the rate. I got this bar from Subaru Online Parts, cost about $100 and included new brackets and bushings.

This makes the new Forester handle more like the old one. It’s less like a bus, more like a car. It feels tighter and more controlled. But not too tight. If you go too stiff with an AWD vehicle it can impair traction off road. It’s perfect for my wife, who wanted less body roll but is not a performance car enthusiast.

NOTE: the end links on this Subaru were quite nearly frozen. The end link attachment bolt was corroded to the nut. And the car is only 2 years old, 4700 miles, and has not been driven on salted roads. I removed the end links from the car, so I could remove the entire bar with end links attached to it. Soaked the end link bolts in liquid wrench and moved the nut back and forth chasing the threads until it finally came free.

More generally, a stiffer swaybar applies greater forces to the end links. So if you more than double the rates, don’t be surprised if the end links eventually break. Keep an eye on them and be ready to replace them with more robust aftermarket end links.

Fixing Intermittent Car Problems

A few months ago Michelle’s car (2004 Subaru Forester), which has been solid & reliable since we bought it new almost 13 years ago, acquired an intermittent problem: it would not start when warm. Cold starts were always good, but after you drive it 5-10 miles, just enough for the engine to warm up, then turn it off, then come back 15-30 mins later, it would not start. The problem was intermittent, happening only about 10-20% of the time. When it did fail to warm start, remove the key from the ignition and try again. It would almost always start the 2nd try. The start failure was: engine would crank like normal, but would not actually start. If you modulate the gas pedal it would start and run smoothly but it wouldn’t idle. No check engine light, and no OBD-II codes were ever thrown – not even when it was refusing to start. When the problem started, the car was about 12 years old with about 78,000 miles. It had always been well maintained – oil changes, air filter, clutch, tranny, brake & diff fluids, belt tension, etc. and was still getting about 20 mpg in around town driving, same as when it was new.

I do all our car maintenance because it’s fun problem solving, I trust myself to take the time and do the job right, and it saves a lot of money. Intermittent problems can be frustrating, but the challenge to fix them can be fun.

Since the problem only affected idle, and was electronic and intermittent, the obvious culprit was the Idle Air Control Valve (IACV). But this is a $350 part, and if it fails the engine is supposed to throw codes – but it wasn’t. There are several far less expensive parts that could be causing the problem, and I’d feel like an idiot replacing a $350 part only to find that the real problem was an $8 set of spark plugs or a $25 sensor.

Here’s what I did, in order… after each step I gave it a week or so to see if it had any effect.

  • Replace the front O2 sensor (the rear ones had been replaced a few years ago).
  • Replace the spark plugs (new ones gapped to spec). The old ones were clean but gap was about 4 times higher than spec. It ran smoother but didn’t fix the problem.
  • Re-teach the ECU idle (disconnect battery, ignition OFF then ON pattern, etc.). This improved the idle but didn’t fix the problem.
  • Clean the IACV – idle air control valve. It was pretty clean to start with, but cleaned it anyway. Also tested its function – OK.
  • Check & clean the crankshaft & camshaft position sensor. Upon removal they were surprisingly clean, but I measured the proper impedance, cleaned & re-installed them anyway.

That last item is what fixed it.

Correction: Dec 2016 – no it didn’t fix it – problem returned!

Since the sensors were operational, I can only surmise that the problem was an intermittent or poor electrical connection to the sensor, that got cleaned when I removed & reinstalled it.

Since the problem came back – next steps on my list below. Since the engine has never thrown a code or lit up check engine light, I wondered if the OBD-II system was even working. When testing the IACV I unplugged it while the engine was running. It immediately threw 4 codes, one for each wire pin. So the OBD-II system and my code reader are both working.

  • Replace the fuel pump relay: sometimes with age, the point contacts get corroded and don’t provide enough power to the fuel pump. When my 15-year old Honda Civic developed a similar problem, this was the root cause.
    • Replaced in Nov – did not fix the problem.
  • Main relay: probably not the problem; everything else on the car works fine – radio, headlights, etc.
  • Clean throttle body: no. A dirty throttle body would cause problems all the time.
  • Clean/replace the MAF: this engine – 2004 2.5 liter Subaru flat 4 – has no MAF.
    • It has a TPS – throttle position sensor
      • Inspected OK – operates smoothly and measures 190 Ohm – 5 kOhm
    • It has a MAP – manifold pressure/vacuum sensor

Update: Jan 2017

Finally, I decided to do what the original symptoms suggested: replace the IACV. By this time I had replaced every other cheaper part that could be causing the problem, to no avail. I found an IACV on Amazon for $250, which is still ridiculous but about $100 cheaper than the local parts place wanted, has a warranty, and is probably the exact same part from the same manufacturer. Took all of 10 minutes to install it, and the difference was instantaneous and obvious. First start-up, engine spun up to 2,700 RPM (which is unusual but this is a brand-new sensor the computer is learning how to control) then slowly ramped down to a normal idle speed. Next morning’s cold start (ambient temp 31* F) engine fired right up, spun initially to 1,700 RPM then slowly ramped down to 750 as it warmed up.

Ah, give me the good old days when an engine’s idle was adjusted by cracking the throttle open a smidge with a simple set screw. There’s a reason airplane engines don’t use all these electronic controls.