• Protip: Profile posts are public! Use Conversations to message other members privately. Everyone can see the content of a profile post.

S2k rack on NA1

A tad late to the party again. This is something I need. The NA1 rack to me always felt a bit jittery. Holding the S2000 steering always felt sturdy to me. And as you mentioned, feels better over uneven surfaces. Could be the weight difference but, can't hurt to have a lighter feel at low speeds. Lmk when a kit is ready for purchase.

PS, should of looked you up when I was in Philly with the X a few years back and did a couple drives to the Poconos just for kicks.
Very cool. Is the only input the VSS because fixed torque assist is only provided at a certain vehicle speed or less?

The system is very simple but looks like Honda did put a lot of thought in to it. Yes it's using it's calculations based on VSS, RPM, and TQ sensor that is already build in to the rack.

The EPS controller is monitoring and controlling the EPS motor's assistance force to match driving conditions:
*Low vehicle speed: High power assist (for easy handling)
*High speed driving: Low power assist (for stable driving)
*Low speed to high speed driving: Change smoothly from high assist to low assist

The steering force from the steering wheel is sent to the pinion shaft. The torque sensor measures the difference between the force applied to the pinion shaft and the resistance to turning the wheels due to road friction. The torque sensor converts steering torque input and it's direction into voltage signals that are sent to the EPS control unit. The EPS control unit uses these signals along with the engine and the vehicle speed signal from the ECU to calculate the direction and the amount of assist needed. The EPS control unit then sends out the appropriate command to the EPS motor. The EPS motor transmits it's power through a ball screw mechanism to push the steering rack right or left.
I youtube short video to go over the 3 racks and demonstrate the travel distance on each. I get tong tie on video but below are the findings.

S2000 traveled 2.34" per each full wheel turn 67% faster
Accord traveled 1.91" per each full wheel turn 36% faster
NSX traveled 1.4" per each full wheel turn

Last edited:
Funny. Jrz sees the dead / no feel of the s2000 eps rack as a feature compared to the perfect na1 non eps go cart feel.
Funny. Jrz sees the dead / no feel of the s2000 eps rack as a feature compared to the perfect na1 non eps go cart feel.

We definitely drive different types of go carts socalx. As per NSXMUGEN's video. When does 67% faster translate to dead / no feel :confused:. Good for you though. Many of us prefer the steering feel and speed of the S2000. That's why we're inquiring.
Jrz. Two different issues. While the S2k rack is quicker it is dead feeling eps. Or stated another way the na1 non ps rack is go cart like true road feel while slow
Jrz. Two different issues. While the S2k rack is quicker it is dead feeling eps. Or stated another way the na1 non ps rack is go cart like true road feel while slow

From what I understand the EPS aspect is non-functional over some low speed (like 20 MPH).

I guess everyone's different, but I don't see what's important about having a go-cart like feel below 20 MPH.
The s2000eps rack gives a dead feel to the wheel at all speeds. I want you to understand this is coming from someone that owns 3 s2000s right now (a 2000, 2006 and 2008). So my point is I love s2000s but would not want to ruin the feel of the wonderful na1 steering with an s2000 eps rack.
I have no problem ruining the na1 steering, I can't wait to have an s2000 rack in my nsx! :biggrin:

Great job OP.
Any idea if there is a GPS VSS that will work with the EPS from the later NSX's (97+)?

The NSX EPS units already get a speed signal. On MT cars the speed signal comes form something called the pulse unit (mounted on the back bulkhead) which picks up a signal from a differential speed sensor. On AT cars the speed signal comes out of the TCM using a countershaft speed sensor.

The pulse rate is probably set out in the service manual. I am not sure why you would want to do this; but, if you are perverse you can acquire GPS antenna / radio modules pretty cheap and you could interface it to something like an Arduino or Teensy. The GPS antenna / radio module would update position data to the Arduino which would then calculate an average velocity and convert it to a suitable pulse rate For the EPS controller. I am sure there are many 14 year olds around who can do the requisite coding of the Arduino.

On second thought it is probably a bad idea as the differential speed sensor does not suffer from all the potential problems that can affect a GPS (avoid driving through tunnels!).
I am not at all familiar with the details of the NSX R final drive. If your primary concern is the error in the speed signal for the EPS caused by the change in the final drive ratio I personally would not be overly concerned. The change in drive ratio will mean that the pulse / mph is higher than with the standard final drive so the EPS receives a speed signal that is higher than true speed (the % error being the difference between the drive ratios). This should cause the EPS assist to taper off faster than normal which some people might view as a desirable feature.

Just for good measure, the 97+ EPS unit has two speed signals. VSS1 comes off the same VSS sensor as is used for the speedometer and ECU. VSS2 comes from the pulse unit from the differential mounted sensor. For double good measure, the VSS pulse rates for the two signals as measured at the terminals to the EPS control unit are different. According to the service manual VSS2 is 50 pulses per revolution and VSS1 is 4 pulses per revolution (presumably revolution refers to a revolution of the rear wheels). The service manual is not clear on what the EPS does with the two signals other than generating an error code if there is a discrepancy in the derived speed signal between the two. Perhaps it defaults to a minimum assist level?

I know physically where the VSS1 and VSS2 sensors are located - pretty much side by side on top of the differential housing. VSS1 is a mechanically driven hall / opto sensor (I am guessing that the drive is right off the final drive gear) and VSS2 appears to be a 'tooth counting' variable reluctance sensor (it only has two wires) presumably sensing the teeth on the final drive gear (or is there a separate tone ring?). Presuming that both VSS devices get their signal off of the final drive gear the speed error caused by the change in final drive ratio will be consistent and should not cause any error codes for the EPS. In such case the only change is that the assist may taper off faster.

If the problem is that the VSS2 sensor no longer works with the NSX-R final drive because of an absence of a tone ring (or whatever reason), that is a bigger problem. In such case, if you really want an NSX R final drive I would be inclined to try to fake the VSS2 signal by picking up the VSS1 signal and writing some pulse multiplier code for an Arduino that could then generate the higher frequency VSS2 signal. This will work just fine for steady state speeds. While changing speeds there will be a short time lag between the calculated VSS2 signal and VSS1 which creates a potential speed error between the two which might trigger an error code depending on the EPC speed error threshold. A GPS receiver with separate controller could generate the VSS2 signal and presumably do it fast enough to avoid generating speed mismatch error codes in the EPS.

However, GPS units are not without issues. I have a GPS speedometer on another car and I know that there are certain times of the day that signal acquisition can take minutes, presumably because of the available constellation of satellites. Once acquired, I have never lost the signal as long as I stay out of well shielded structures and tunnels. One additional problem is that the GPS based system is not affected by tire size. If you change your rear tire size this cause an error in the VSS1 signal while the VSS2 signal remains true which then creates the potential for a EPS error code. The faked VSS2 signal from the VSS1 sensor would not suffer from this tire size change error.

Not being familiar with NSX R stuff, the higher ratio final drive creates a potential problems for the speedometer on the R. Does the NSX R use a different VSS1 sensor to give a correct pulse rate or does it use a speedometer that has a different calibration to work with the different pulse rate associated with the NSX R final drive?
Last edited:
Thanks "Old Guy" I always like your excellent explanation how tings work, I read it couple of times it's very interesting how the OEM EPS works. Very, very good info, thanks again! I did not even know this was an issue with the NSX R final drive. After googling looks like Dakota Digital LINK has some converters that might help with correcting the signal for NSX R final drive and also list GPS unit but I'm not sure.
Because of this canbus VSS on TSX rack, I'll put it on back burner and revisit some other time. But for now will only focus on S2000/Accord racks.

I took out my car for last drive with S2000 rack before moving on to Accord rack. I try to do some quick turns but the camera fall out of my mount or was shaking too much.

This is turning into a bit of an unintentional thread high-jack.

The problem with the NSX R final drive 'appears' to be that you lose the signal from the VR pick-up (VSS2) on the differential when you install the drive, not that there is a difference between the VSS1 and VSS2 signals (which there is because there is no VSS2 signal). In this thread below there was some discussion by SOS about a project to fix the problem.


The links are dead; but, you may be able to suss them out with the Way Back machine. Clearly the project did not go anywhere because on their website SOS says 4.23 and the 97+ EPS is a no go.

SOS in a later post in the thread alludes problems getting a clean signal from the sensor. Perhaps the 4.23 final drive changed the spacing between the ring gear teeth and the VR sensor and VR sensors are notorious for being sensitive to the spacing between the sensor and the 'tooth'. As an observation, gear teeth with a small pitch are not the best thing to try and get a nice signal from with a VR sensor or a Hall sensor at the best of times. If that is the problem, then creating a VSS2 signal from a GPS unit or running VSS1 through a pulse multiplier to fake VSS2 may work.

Notice the fine print on the yellow box device for the S2000. It drops the pulse rate to correct for higher final drive ratio. The VSS2 pulse rate compared to the VSS1 pulse rate is 50/4. You need to raise the pulse rate from VSS1 to create a VSS2 signal. If you want to poach the VSS1 pulse rate and modify it to give you a VSS2 signal you are going to have to do some software trickery or create a totally new VSS2 signal from a GPS unit.
Last edited:
Last weekend was very productive, after about 2months+ the accord rack is in. I had to revise the rack few times and between sourcing for parts and waiting on materials it just felt like it took forever to get the car on the road again. Ok, so the first reaction to the accord rack in nsx was giggling like a piglet as the steering wheel feel was so effortless. I push the car harder to some corners around me and could hold the steering wheel with one hand. It is a fun to drive. It's def overpowered but this is compering to S2000 and to manual. The low speed mode makes it easy to park and just turn with ease, when it engages the second mode the steering wheel is more stiff but still lighter then S2000. The steering ratio is also slower but still faster then nsx manual rack. I'm running on 17x7.5" wheels with 215 wide tires so if someone has wider front tires it would feel different. I felt that I can take any corner at any speed as if the car was as light as the steering effort but it did not provide as much road feedback. Those two racks have two different personalities. If I can put it in perspective, the S2000 rack is all business, and accord rack is a party on the weekend. If someone tracks their NSX, the S2000 is better option but if the car does not see track, I think the accord rack is plenty of luxury that was not there before. I'll take some new vids driving it and need to do more testing, but so far I'm liking it.

Here is quick vid after just getting it installed.
I'm going to be at nsx River Run link in NJ if anyone wants to take it for a test run, you are welcome. I'm going to build two more kits.