Don't like the weight of your speed-sensitive power steering? Like to make it a bit firmer at all speeds? Have less than $20 to spend? Well, this modification might be for you!
This article describes how I modified the electronically variable power steering system in my 1998 Lexus to make the steering weighting adjustable. And, while the detail of the mod is specific to the Lexus, the same procedure can probably be followed in other cars that use a similar approach to electronically varying the steering assistance.
To give the best overview, I'll describe the step-by-step process that I undertook to identify how the variable power steering works, how the weight change was made, the development of the prototype and its testing, and the final result.
The first thing that you need to do to undertake any modification of this sort is to have a thorough knowledge of how the electronic control of the power steering works. IMHO this can be achieved in only one way - by reading the factory workshop manual.
The LS400 has what Lexus term 'Progressive Power Steering' - PPS. This uses a solenoid valve to vary the flow of hydraulic fluid to a reaction chamber - a fluid force that actually resists the power assistance. If a lot of fluid is allowed to flow to the reaction chamber, the steering effort is higher. If little fluid flows to the reaction chamber, then the steering effort is lower.
Key point: more fluid flow into the reaction chamber equals a higher steering effort.
Controlling the amount of fluid flowing to the reaction chamber is a solenoid. The solenoid (I assume - I haven't cut one open) consists of a coil, a return spring and the valve. When no current is applied to the solenoid, it opens, allowing more fluid to flow to the reaction chamber and so the steering to become heavier.
The current to the solenoid is varied by means of pulse width modulation - the current is pulsed on and off quickly. If it is on for only half of the time (ie it has a duty cycle of 50 per cent) the coil will 'see' only half battery voltage, and so will not close fully. If the duty cycle is reduced to, say, 30 per cent, then the valve will open a little more.
Note that unlike an injector, the frequency of the pulsing is so quick that the valve doesn't open and shut to the individual pulses - instead the plunger hovers at mid-points.
The PPS is controlled by its own dedicated ECU - a little box. It's a relatively simple box, too, with just one input - road speed. As speed goes up, the duty cycle with which it feeds the solenoid goes down and so the valve opens further, increasing steering heaviness. Or, to put it the other way, as the road speed falls, the solenoid duty cycle increases, closing the valve and so lightening the steering.
Key point: a lower valve duty cycle equals a higher steering effort.
Well, with a bit of reading between the lines and some examination of the diagrams, that was what the workshop manual told me.
The first test that I undertook to prove that this information was correct was to pull the fuse that feeds the power steering ECU. This would both disable the variable speed sensitivity control (the power assist hydraulics would keep on working) and also cause the solenoid valve to fully open, feeding lots of fluid to the reaction chamber and so making the steering heavier.
I removed the fuse and went for a cautious drive.
Yes, the steering was much heavier at all speeds - especially at low speeds. In fact, while at first I had considered increasing the steering weight simply by disabling the PPS, the resulting steering was heavier than I wanted. Not nearly as heavy as a non-powered system, but a long way heavier than the overly light standard system.
So, having the equivalent of zero per cent duty cycle (ie no power at all) going to the valve gave too heavy a steering system, while the factory (varying) amount of duty cycle gave steering that I perceived as too light.
What was needed was a system that reduced the duty cycle coming out of the PPS computer, so that at all speeds the valve would allow more fluid to flow to the reaction chamber.
(Or, what about increasing the speed signal going to the ECU? That way the steering would also be heavier. However, I discarded this idea, partly because the transmission/engine ECU, the trip computer, the ABS and Stability Control Systems - and some other systems - also use the speed signal input. Even if it was modified after the signal had been taken for these other systems, best to leave it alone. And furthermore, a signal would also have to be generated when the car was stopped - telling the ECU that in fact the car was moving!)
So how to reduce the duty cycle output of the power steering ECU? And how to give it a variable adjustment, so that steering weight could be dialled in to exactly suit my preference?
Such a modification of the output signal duty cycle could be made using a micro-controller, power output transistor, other circuitry - and lots of work. For example, the first things an electronics engineer would ask if designing such a product are:
....and lots of other similar questions.
All for which I had no answer.
However, could the modification of the output signal be done a lot more simply, using just passive components? After all, the coil of the solenoid sees the average voltage being applied to it - it doesn't open and shut to the individual pulses, because they're being applied in too quick a succession. So, the coil doesn't know that it's getting sent a 50 or 60 or 70 per cent duty cycle - it just sees the average voltage resulting from that quick succession of on/offs.
So, what if the duty cycle remained unaltered - but the actual voltage of the 'on' pulses was reduced?
The suggestion from retired electronics engineer and scientist, Robert S Edgar (who happens to be my father) was that a resistor placed in series with the solenoid pulsetrain feed would possibly give the same effect on the solenoid plunger position as reducing the duty cycle of the pulses.
(I might add that he was very reluctant to embrace the idea without having far more information - stuff like, 'What is the inductance of the solenoid?' and questions like that, again all of which I didn't have answers for.)
The solenoid has a resistance spec of 6-11 ohms (why the huge variation in permitted specification, I don't know) but other data about it is unknown. The voltage outputs of the PPS controller shown in the workshop manual are measured with a digital multimeter - and on a variable duty cycle output, the sampling behaviour of the meter will affect the resultant reading. So while these measurements can be used to test the PPS controller, they weren't so good from a design point of view.
Therefore, it was decided to carry out some experiments.
The resistor that was going to be placed in series with the coil was always going to have to dissipate the power that the coil would no longer be seeing. How much power was the question - and that was easier to find out by trial and error than any other way. (Note that placing a resistor in series with the solenoid coil in this way means that less current will flow in the circuit, so a heavier load won't be being placed on the power steer ECU.)
The first experimental resistor - a standard quarter-watt 5.6-ohm resistor - got too hot to touch in about 1 second, literally. So obviously a lot more than ¼ watts was going to be generated as heat!
Next up some large old resistors were used, wired in parallel until 4.8 ohms and 3 watts power handling was obtained. These grew only warm to touch, however the result on the steering was less encouraging - there was no apparent change in weight. I then went up in resistance, still binding lots of large resistors together to give the power handling, but again this didn't seem to have an effect. Getting bored in going up by small increments, I then jumped to 100 ohms and immediately picked a major difference in the steering.
It was much heavier.
Not as heavy as with the fuse out, but still heavier than I'd prefer in normal use. So on the basis of that testing, a 0-100 ohm variable resistor of about 2.5 watts power dissipation would give me a dial-it-up steering weight variation from normal through to heavy.
The easiest way to gain a variable resistor is to use a potentiometer, wired as a variable resistor. Wire-wound potentiometers with 3 watts power handling are available from electronics stores - so I went and bought one (R6909 from Dick Smith Electronics at $6.95). With this wired into series with the solenoid I could, for the first time, change steering weight on the move by twiddling the knob.
It's an uncanny feeling putting in rapid small steering inputs and at the same time turning the steering weight adjustment knob.
However, I found that to get the steering near to my weight preference, I was right up at one end of the pot's range. I attached to the pot a piece of cardboard with a 0-10 scale on it, and added a knob. I then went out test driving, assessing the full range of adjustment that I'd need - from the heaviest I ever wanted the steering to be, right through to the lightest. When this range of knob adjustments was compared with the scale on the cardboard, I found I was using resistances only from 0 ohms (ie standard steering weight) to '2' on the scale (ie 20 ohms).
So yes, that earlier test with a 4.8-ohm bunch of resistors (where I'd felt no change in the steering) was wrong - when the alteration could be suddenly input with a knob, you could easily feel the difference.
So a pot that had about a 0 - 20 ohms range would be suitable for the final installation. However, it would also need adequate power handling - and here's where I had a problem. The 3-watt Dick Smith pot was getting very hot. During testing, I mounted it in front of one of the air con vents and even then, at slow speeds (where there was a greater voltage to pull down) the pot was still very warm.
So, perhaps 3 watts of power handling was insufficient.
Anyway, the Dick Smith wirewound pot was also an unimpressive bit of gear - flimsy terminals and a lightweight metal case. Perhaps a higher quality pot would handle the heat better as well.
I looked through the RS Components catalog until I found a quality pot with the right range - cat no 162-754 which is a $15, 2-watt pot with a 22 ohm resistance. Only 2 watts? Yes, but that's at 70 degrees ambient, when most pots are rated for power dissipation at only 40 degrees C.
I wired-in the new pot. This gave a very good adjustment range (now I could use the full sweep of the pot rather than being up at one end) but the pot was still hot to touch. In most electronics applications, resistors do get very hot (ie much too hot to keep your fingers on them) when they are working near their max power dissipation. However, in this application I wanted the temp of the pot at its maximum to be just warm.
I added a smallish heatsink (about 40 x 30mm) and that settled it down a bit - but I then swapped to a larger heat sink (Dick Smith H3460, $9.95, about 75mm square) and the temp of the assembly stayed well in control. With this heatsink, the temp rises only perhaps 5 degrees C above ambient.
The heatsink'd pot was mounted with the heatsink fins aligned vertically, and then stuck with quality double-sided tape to the side of the body control ECU that lives under the driver's side of the dash. While I wanted access for setting-up adjustment (and perhaps long-term tweaking as the tyres wear) I didn't want or need on-dash control. Instead, with the car stopped, the pot is easy to get to.
[Note that if I had not wanted an adjustable control, I could easily have selected two 5-watt wirewound resistors and wired them in parallel, giving 10 watts of power handling. For example, two 22-ohm resistors wired in parallel would give 11 ohms resistance and 10 watts power handling - at a total cost of 60 cents! Mounting these from underneath the car (ie accessing the wiring near the solenoid) would take literally 15 minutes, start to finish.]
So what difference does the adjustment make? To measure the changes in an indicative way, I attached a spring balance to the outer end of a steering wheel spoke and measured the force required to turn the wheel, with the car stationary and parked on smooth concrete. With the pot set to the 'standard' position (ie no change over the normal Lexus steering weight) I measured a required pull of 4.5kg to turn the wheel at a constant rate.
With the pot set to its 'heaviest' end (and remember, that's 22 ohms - I did have it higher), the required pulling effort doubled to 9kg. In normal road use, I am about halfway through the pot's range.
On the road, the heavier steering makes a quite amazing difference. The actual feedback of the road surface is unchanged, the steering precision is unchanged, but the extra 'meat' of each input alters the driver perceptions a lot. I think perhaps it's the feedback to the steering input which is the big difference.
No longer do you just whack on some lock when entering, say, an 80-km/h corner. Now you feed it in much more progressively. It's a little like if you have driven an arcade game that has zero steering resistance - you don't feel that you can steer very well at all. That's not to say that the Lexus previously had arcade game steering, but in the same way as resistance in the steering wheel of a game makes the steering better, so a lift in weight improves the steering of the Lexus.
In addition to mid-speed corners, the difference is also very noticeable in freeway sweepers, where the required amount of lock can be apportioned much more accurately. Also, the chances of making an inadvertent steering input are lessened. For example, if you shift in your seat, or sneeze, or glance down at the instruments or at a roadside sign, the steering wheel more strongly resists the hand movements that you didn't intend to make.
Note however that even with this modification, the steering still alters in weight as it did when standard - more assistance at parking speeds and less when moving quicker. It's just that it's heavier all the way through.
But perhaps the biggest eye-opener is to snake down an empty road at 50 km/h, swinging the car from side to side as another person turns the steering weight pot - you suddenly realise with startling clarity that the amount of steering weight makes a huge and instant difference to how the car feels on the road.