DANGEROUS Honda Electric Steering (Severe Pull)

[@jessicav2031]

The computer is measuring inductance! Inductance is resistance to change in current; when a current is applied to an inductor and then released, the inductor continues to apply current to the circuit as it depletes its stored energy. Another way to look at it is the amount of coupling between the magnetic field and the circuit (when a current is applied, the inductor builds up a magnetic field). Inductance depends on magnetic permeability, the amount of magnetization gained by the area around the inductor, which is altered by moving a ferrous material closer or further from the center of the coil. To measure inductance the computer must pulse the inductor with a current and then measure time, looking for how quickly current decays (or increases, or both - a start and end time difference between thresholds). It could either wait a fixed amount of time and measure what the voltage/current is then, or count how long it takes the circuit to return to a threshold voltage/current. Either way it is likely looking for a value in the middle of the curve to avoid the ringing at turn on/off. The 'help' description in the scan tool is a clue too, it said the value was calculated :) Yes, finally some EE stuff :)

[@PineHollowAutoDiagnostics]

Love the detailed EE comments :)

[@richardcranium5839]

is this an RC circuit? or an LC circuit? maybe even an RLC circuit.\, and is the detection a voltage detection?, current detection? , or a change in frequency or frequency shift? and isnt resistance to current change a product of inductance not a producer? another thing is the scope showing the voltage over time shows both signals synced in time but different voltages. good fodder for thoughtful discussion. would love to hear your thoughts

[@jessicav2031]

@@richardcranium5839 No capacitance is necessary (or desired, it would be likely to add ringing). Microcontrollers can only measure voltage, so current is always measured by the voltage drop across a resistor. Either way it is the same. To determine inductance you must know 1) the voltage applied (or some combination of two of voltage, current, and resistance...Ohm's law) and 2) the time it takes before a certain amount of current is flowing. It works the same way whether you are charging or discharging the inductor; when charging, less current flows (the inductor presents an effectively higher resistance, which in the case of AC is called reactance). When discharging, current continues to flow even though you have stopped applying energy. These are the two sides of the curve on the graph. With more inductance, it will take the curve more time to reach the same voltage (once again, Ohm's law). So on the graph, given a fixed on-time, the voltage will not dip as low (since this is pull-down). If you lower the inductance, it will reach a lower voltage in the same time. This is essentially what the computer is measuring. Possibly interesting side note: the fact that current continues to flow is why every coil controlled by a computer (such as an A/C compressor coil) requires a diode. The current that is continuing to flow needs to go somewhere (through the diode) or voltage will rise until it arcs over or blows up the driver.

[@richardcranium5839]

@@jessicav2031 interesting so with the input to both being pulsed it is a dual tuned? or time? circuit i assumed it would detect a frequency shift caused by the change of inductance but i see after looking at the video and the waveforms, exactly what you are saying about using voltage drop to measure the current flow and timing. with the input being pulsed dc it is more of a charge/discharge timing circuit than an oscilator tuned timing circuit if ac was used

[@707061756c69]

@@richardcranium5839 There is no need for tuned circuits. It only requires that one end of both coils are connected to a reference supply through a high side current measurement and the other ends of coils are connected together (like middle point in transformer) and this middle point is pulsed to ground. The difference of coil currents is sampled at certain time of pulse and voltage of that sample is used for torque measurement.

[@inothome]

You're seeing the raw values from the inductors. While you are used to seeing the filtered and amplified 5VDC based signals. Those mV signal changes may seem small to you, but in the world of microcontrollers and electronics that is huge. Also, you can't go by just voltage alone, there is a current associated with that signal return as well. That is much less effected by resistance in connection. Which is why a lot of industrial controls use a 4 - 20mA signal for control and feedback. Doesn't matter how long a run or if shitty connections, they current will be driven to the appropriate level between 4 and 20 mA by the transducer. Up to a point of course, but takes away the voltage drop of long runs of not so good connections.

[@PineHollowAutoDiagnostics]

Excellent explanation!

[@inothome]

@@PineHollowAutoDiagnostics Not 100% certain, but looks like the driver for that circuit sends a pulse, that you saw, and monitors the value as the inductors collapse in voltage. Just like how a coil fires a spark plug, but this reads the primary collapse and the position of the magnet or iron core effects the inductance, which in turn affects the level of collapse. You can see the ringinig of the coil when the pulse shuts off. Even though the meter read 4.5VDC as you found out that was an averaged value form the pulses. I didn't do the math, but if you did the math on the time it would tell you what frequency the sensor was being read at, 10kHz, 50kHz etc.... It should calculate to a whole number. If you wanted to go firther down the rabbit hole, sure you can find the info on teh actual driver IC and that will tell you all the cool dork shit you wanted to know on exactly how it works.

[@tharakanewan3544]

Are u a mechatronic engineer ? By the way nice explanation.

[@inothome]

@@tharakanewan3544 Thanks, 30 years of power generation, electrical substation and industrial automation experience. Ohhh and a few years long ago as an auto mechanic, ASE Master Auto and Master Heavy Truck Tech.

[@donkaiser6704]

The two inductor-sensors will be identical by design. There will be a chunk of ferrous metal attached to the mechanical part of the sensor. Depending on direction and magnitude of torque applied to the steering, the ferrous material will move closer to one or the other sensor, thus changing its inductance (and rate of discharge). The ferrous material should be equidistant from both inductors when no torque is applied. Something in the mechanical part of the sensor must have become broken or misaligned, causing the ferrous material to be closer to one sensor, resulting in the bias you are seeing in the output signal. It would be interesting to see a tear-down of the sensor to see how that happened.....

[@johnfry9010]

I don't believe there is a dealership on the planet that would have gone to all the trouble you did to diagnose this problem , Excellent Job Ivan !

[@JT-lq4yd]

I think the dealership would have said "the repair comes to $XXX for XX months".

[@baxrok2.]

Fascinating comments on this one Ivan. There are some knowledgeable viewers watching you!

[@PineHollowAutoDiagnostics]

Hey John! Yeah the discussion in the comments is great 👍😁

[@gonzgarr1592]

Ivan you did it again never seen anybody going into such details thank you

[@brakedd]

I've seen two Civics in the past couple of weeks doing the exact same thing. Thanks for this!

[@Anonymous-it5jw]

You, and Jennifer V, the Electrical Engineer, know more than I do about how the reluctors and inductors and electronic components work, but consider this - Insufficient voltage in, garbage out, i.e. the overlooked variable. To explain what I saw - When you were checking the Advt1 and Advt2 voltages after the reset procedure, battery voltage as shown on your ThinkTool was staying well below normal - 10.23 volts (at approx.11:00 thru 12:06). Was there a voltage drop between the battery and the wiring harness, or just a bad battery. The difference between this number and a barely acceptable battery voltage and the "missing voltage" you were looking for is 1.3 volts. Where could it be? Consider one more possibility - When you back-probed the sensor and were getting unexpectedly high readings, were you using a direct battery ground for the DVOM or were you using the same local ground that the sensor was actually using. A bad local or wiring harness ground serving the steering sensor circuit, with corrosion or a loose connection somewhere between the sensor circuit and the negative battery ground could explain a lot about the problem. Consider the timeless beauty of Occam's Razor - address the simplest possible solution, i.e inadequate battery and bad ground, before deciding it's the most expensive possible problem and firing the parts cannon at it. Thanks for the video.

[@TheOmegaman69]

As a few others have commented, the EPS Torque Sensor could possibly be a 4-20mA loop circuit. Cannot say for sure. Some of the advantages of a 4-20mA Loop is it has much better immunity to electrical noise, and can support longer wires as a result. It would make sense to use this for a steering assist given it's importance. 4-20mA Loop is the predominate method used in many industries. If one wanted to measure this it would be good to have a meter that will measure 4-20mA percent. 4mA would be 0%, and 20mA would be 100 %. And of course, everything in between. One Multimeter that can do this is the Brymen TBM869s. It will display both the current and the %4-20mA on the same display. It can also measure many other things, such as VFD (Variable Frequency Drive), Duty Cycle, Frequency, has a dual display, is very accurate, and for DC Volts has 500,000 count display. I am NOT recommending anything here, just pointing to one Multimeter that can measure the %4-20mA Loop circuit. Jessica V has explained this sensor very well in the latest post made. One small thing i would like to comment on is that Inductance is not "resistance to change". Inductance is the capacity of the inductor to store current, just as Capacitance is the capacity of the capacitor to store voltage. Inductive Reactance is the equivalent to DC resistance (almost) for an inductor, just as Capacitive Reactance is the equivalent of DC resistance (almost) of a capacitor. This EPS Torque Sensor is made up of two (2) inductors, very fine wire wound in somewhat of a circle. One end of each inductor is connected together and connected to the 3-Pin EPS Torque Sensor, pin 2 (PVF, Blue/Red wire). This signal is driven by the EPS Controller with a FET (Field Effect Transistor) to 0 Volts, or 5 Volts when the FET is off. One of the coils is connected to pin 1 of this connector (VS1, Pink wire), and the other end of the other coil is connected to pin 3 (VS2, White/Green wire). These pins have pull-ups (resistors) to 5V inside the EPS Controller. Each of these two (2) coils should read 12-15 Ohms resistance. On the scan tool VS1 (ADVT1) should read 0.90V - 3.55V, and VS2 (ADVT2) should read 1.02V - 3.73V. Both should read 2.5V when steering wheel is centered (12mA if the 4-20mA Loop is being used). I do not think this is a 4-20mA Loop circuit, but it certainly could be. If it were, the repair/service manual would describe it that way. I just have one question and one comment to make. 1) Question: In the description it says this is a "Honda Civic Hybrid". Doesn't a Hybrid have both an electronic power steering and a hydraulic power steering as well? 2) Comment: When you did the Torque Sensor Learn it was improved by 1 lb/ft, and it was much farther out then after doing it, I would think that if you did the TSL again, 2 or 3 times if necessary, maybe it would learn center correctly. This does not mean the Torque Sensor is not bad, but would be interesting to see if that made any further improvements. Great video. Good job filming as well. Thank you.

[@PineHollowAutoDiagnostics]

Great comment! Now we're getting into reverse engineering 😁 unfortunately the software calibration only allows for a +/- 1ft lb error in the mechancal part of the sensor... I tried it several times in a row but it said "failed" every time 👍

[@TheOmegaman69]

@@PineHollowAutoDiagnostics Thank you for your response. Multiple attempts has worked in some cases. I guess it was worth trying anyway. Thanks again. Very interesting video, and well done.

[@Eddavids2]

I went to a muffler shop and when I left I couldn't steer my electric steering only wanted to go one way. Come to find out when they were welding they never disconnected the battery and fried the steering. Dealer replaced whole steering column, 1000$ later the muffler shop reimbursed me. Was still a pain to deal with.

[@PineHollowAutoDiagnostics]

Wow that is nuts!

[@ghettsumm7949]

I attended two years of trade school for a degree in EET. But I'm learning way more about VFCs and VFMs and their applications on KZbin videos. Heck, even my transfer case motor has an encoder !! Thanks Ivan.

[@d1bergman]

how integrated is the sensor with the steering rack? Why can't you just replace the sensor?

[@roberthenuber5958]

LVDT as in Linear Variable Differential Transformer. Used for many years in process control actuator positioning.

[@ThatEEguy2818]

That was my thought as well. I wonder if the core shifted resting position somehow.

[@ThatEEguy2818]

I wonder if the coils themselves are air core or ferrite. If ferrite, I wonder if one of them cracked.

[@davidhollfelder9940]

The offset explains the current draw when centered .. There should be some kind of “rough zero adjustment” for this (other than the lame small “calibration”). Something got bent in the steering rack?

[@Graham_Wideman]

It would have been interesting to hook a (variable) resistor from the pulse signal over to the VS1 signal, to pull VS1 down slightly more during the down portion of the pulse, to see if the system could be fooled into behaving. Ie: supply some additional bias from the green trace in the desired direction to "correct" the blue trace.

[@PineHollowAutoDiagnostics]

Excellent suggestion, Graham! That would make for some amazing Bonus footage :)

[@DKr475]

Same thinks from me when i watch this:)

[@Graham_Wideman]

Obtuse wording due to filtering: There's an interesting related article to be found by usual means regarding Honda S2000 steering hack. "Electric Power Steering Hack"

[@troutjunkie7330]

I was just gonna write that in comments till I saw your comment. If it was my car, I would install resistor on opp side to balance the signal and drive another 100K

[@raymondreiff8170]

I ran two 2003 Civic hybrids in Central Wisconsin with Electric Steering over 300,000 miles each never ever had a problem with the steering system.

[@annaplojharova1400]

That is just plain inductive sensor, controller is pulsing ground on the common and sensing current via drop over equal sensing resistors (towards thepositive) in the module. When no torque, the steel (ferromagnetic, so affecting the inductances) plunger inside of the coils is in the center, so inductances of both coils are equal, so are the currents so the voltages. When torque applied, the plunger shifts, increases inductance of one and decreases the second, showing difference in the response signal (the response pulse height)... And regarding the last note about not having this issue with hydraulic: Well, met a clogged up torque valve and it does exactly the same style behavior, just more violent (it was a 15 ton bus after all). Better to say the valve was jammed by remains of the disintegrating pump. The pump ended up that way because of a neglected fluid leak so bad it run dry for a while (few hours). Then once pump replaced and fluid refilled, some fragments that remained in the lines ended up jamming that valve, so it fortunately went crazy on the shop floor and not on the road.

[@Graham_Wideman]

Still remarkable that the signal change per amount of torque is so tiny. And what would cause one of the inductors to lose some inductance?

[@ferrumignis]

@@Graham_Wideman More likely some kind of mechanical shift than an inductance change IMO i.e. the torsion spring has taken on a permanent set. The ECM will be take a torque reading using many cycles to get the final data which can increase measurement resolution and noise immunity.

[@Graham_Wideman]

@@ferrumignis Well, the mechanical change would explain the loss of inductance :-). Also, no doubt the ECM is carrying out a running average as you say -- I would not expect it to be updating the steering assist 14,000 times a second!

[@thromboid]

Ah, so maybe one of the sense resistors had drifted in value?

[@annaplojharova1400]

@@Graham_Wideman The defect may not necessarily be directly the inductance shifted, more likely some extra resistance. The inductance is measured via The current during the pulse, so whatever influences the current may be the culprit. And with the inductance alone: A lose speck of some steel debris in the magnetic circuit of the sensor may cause exactly such shift. Or shorted turns of one of the coils. And then of course a defect in the control module sensor front end.

[@graywaveautodiagnostics9389]

Blue and red trace look like a resisted capacitor charge and discharge where the green trace is turning on and off the circuit or at least, on and off a small load discharging the capacitor with a variable resistance? I wonder if there are inductors in there as well since I see some ringing. The more you turn the wheel, the quicker or slower the capacitor discharges within a specified time base set by the reference voltage time base. Very Cool!

[@lustfulvengance]

I believe it's purely inductive, a magnetic core moves up and down within two coils depending on which way you steer the wheel. The green trace (the ref voltage) is a pulse that induces current in the two sense coils.

[@graywaveautodiagnostics9389]

@T.J. Kong Apparently there was and I missed it.

[@kylerees6681]

Late in the game I know, just watched this. Note AllData or the scan tool said inductance. If you probed the two outputs with two amp probes you may have seen what the computer is using when the wheel is turned to measure torque. The voltage will not change much with inductance measurement especially since the magnet is not moving fast through the coil and only because an increase in current the voltage changes. Makes sense considering mechanically the sensor uses a torsion spring and thread, moving a permanent magnet through two coils. (On Right/Left turns receptively) Obviously the spring is damaged or bad spring steel (we've see this often with cheap metal) and the magnet is not centering between the two coils which is biasing the current on one or both coils.

[@Craig1967]

I agree that it is probably the sensor. One thing I would suggest is to swap the two sensor wires (remove pins and swap temporarily) to see if the "voltage" flips or not. If there is no change, then its not the sensor. A bad module or oxidized connection could cause the difference in current. As other people pointed out in the comments, the current, not the voltage is what should be measured. A home made CT (Current Transformer) can be used. I made one out of a small induction motor that came out of a fan. I just removed the rotor and phase shunts. In any event thank you for the video, and knowledge on how this works. Please do another video after the rack has been replaced. I am really curious.

[@morriewyatt6234]

That would be a bad idea as it is a negative feedback system. Swapping the wires would make it a positive feedback system and slam the steering hard to the right.

[@bmulligan9572]

What might happen to the steering if one connection failed....would the steering slam in one direction?

[@chrisosiecki8050]

Can the rack be forced to manual operation by just unplugging it?

[@aldrinalmario1513]

Another great diagnosis Ivan! This is currently way beyond my level of comprehension and is quite interesting to learn. Thank you for sharing and walking us through the process. Cheers!

[@elisabethellis8851]

Same here

[@kain0m]

I'd try measuring the current. same as thos wheel bearings with the magenetic tone ring - you're not looking at a voltage signal, you're looking at a current signal. The difference in voltage just comes from voltage drop due to the current flow.

[@billziegmond4943]

Easier solution. Make it a NASCAR it only needs to go left. LOL

[@PineHollowAutoDiagnostics]

LOL it's a Hybrid Hotrod!

[@ryans413]

Even better solution have no power steering make people work for it build up so arm muscle

[@thromboid]

When life gives you lemons... :D

[@jimwhite5964]

Another great video... really admire your diagnostic skills, Ivan! There seems to be no problem you cannot diagnose! Thanks for sharing your talents with us.

[@PineHollowAutoDiagnostics]

Thanks for the motivation, Jim!

[@watlinsd]

The description and operation at the beginning of the video did say the the signal was amplified, so I guess those small measured values with the scope are significant enough for the computer to handle. The parts manufacturers win again in this case!

[@707061756c69]

The torque measurement seems to be kind of a LVDT measurement (Linear Variable Differential Transformer), though a quite highly modified one. LVDT measurement is kind of a measurement bridge and are quite commonly used in linear postion sensing in industrial machines. I think that the sensing is based on spring coefficient and torque compresses or elongates the spring depending on steering wheel rotation direction. Sensor core is attached to spring and is in the middle of both sensing coils when no torque is applied. There should be an excitation current to both sensing coils and the position of core defines the inductance of each sensing coil thus changing the current through the coil. (One's inductance increases and others decreases). The difference of those currents define the relative postion of the core in respect to rest position. If the spring doesn't have same spring coefficient in compression and elongation the sensor gives faulty information. I think the change in coefficients can happen if the spring is compressed or elongated too much. (beyond the deformation point of springs metal) Ivan... be careful with hybrids while the key is on, they can (and will) start on their own if battery voltage goes too low. I don't know if there is a service mode of the car which could be activated to prevent this.

[@mattbickford2640]

i have had a beat up one of those hybrid civics come in and it took a computer for the assist to work again for it, crazy! curious to see a confirmed fix on it, thanks for the share

[@canadatransistor]

Ivan, regarding small voltage differentials.... and also pull up / pull down it could be an amperage circuit. Industrial robotics and instrumentation uses 4-20mA calibrated instruments. When you send current to a fixed "known resistor" such as inside the pcm it gets converted back to 1-5V then, an analog to digital converter can assess it back to digital measurements (8 bit 10bit or 12bit a/d depending on required accuracy needed). If the amperage is 3mA, then the pcm understands it's outside of calibration. Another benefit of using amperage is that "line loss" doesn't exist where voltage drop is a common problem for long cable runs etc inside industrial complexes. An amperage measurement is the same no matter how long your cable length is, and no matter where your measurement takes place along the wires. I wonder if Honda utilizes this in any way in their designs. Remember you indicated some of the readings were calculated values.... I'm thinking current to voltage conversion. Anyways - it was a super interesting topic. I love electric power assist in my chevy cruze. I swear I never wanna go pack to hydraulic for a small car like mine. Take care bud!

[@PineHollowAutoDiagnostics]

Yup I think that's what is going on in this system. Thank you for the detailed comment!

[@InsideOfMyOwnMind]

Nicely done explaining the 4-20ma. A classic setup that will never go away. Honda just need to do it twice so they can vote each other out.

[@Graham_Wideman]

You are right about current-based signals and 4-20mA etc. However, if this sensor was feeding a fixed-voltage current sensor (on wires VS1 and 2) then there would not be much if any main arc of those signals as the current in the coil built up. So it looks to me like VS1 and 2 are just resistors to some high voltage reference, like 5V.

[@dustcommander100]

This is late and probably has zero chance of helping, but some of the signal labels led me to believe there is a chance that the torque sensor uses "LVDT" or Linear Voltage Differential Transformer technology to sense the movement of the torque bar. In this case, the input coil(s) are fed an AC signal, with the output coils producing AC signals having a phase or amplitude change in response to a small movement. These changes would be evaluated by a special circuit giving a DC variable signal output. This was the basis of some industrial equipment I worked on - in the filthiest and most miserable part of the plant I worked in. The premise was that there was no physical contact with the moving part - only the stroke was measured. In your case, replacement of the rack is probably by far the safest call because of liability. Good work on your part, Sir!

[@ryvull9099]

I have this exact problem and I just put a new steering rack in the car and it’s still pulling the same do I have to calibrate the new rack to the ecu ?

[@consaka1]

Bent suspension or rack components? I bet it took a hit on a curb or something.

[@eldoradony]

You buy a car at auction and you find out it has a problem that will cost over $1000 to make it safe and driveable. What's the next step? Take it back to the auction and resell it? Maybe make or lose a few hundred? I have seen this with one of my dealer customers. He bought a car on a Friday at the dealer auction. The check engine light came on on the way home. The car had a stretched timing chain. When the papers showed up in the mail on Wednesday, the paper trail showed the car had been through 4 dealers in 6 weeks. If he opted to fix the car, he would have been $2000+ under water. No choice but to take it back and auction it again.

[@dosgos]

Maybe rare situation as the few wholesale auction buyers would change professions quickly. But with the current auto supply shortage, my sense is that a lot of low-quality cars are migrating back to the market. They are keeping our mechanics busy.

[@weshawkins7165]

Most cars go to an auction for a reason. I’d be very careful buying a car from an auction. I would expect to have problems with a car bought from an auction.

[@dosgos]

There are some legitimate cars that went to auction. For example, a Lexus dealership takes an old Kia in on trade probably just wants nothing to do with the Kia regardless of condition.

[@cannbudo]

Curious to know if there is a confirmed fix in this case study. Exceptional diagnostic capability you have in that mind of yours. Thanks for sharing.

[@louiscates6975]

This pull problem can happen on hydraulic pwer steering systems too

[@AVCONSULTTX]

I don't buy that they are depending on mV changes between the sensor and PCM, in a noisy car enviro. I'd easier believe these are periodic current loops being converted to voltages inside the PCM. If true, you'd have to insert a small known resistor value in the signal wire and measure voltage across the resistor to sample the signal.

[@weshawkins7165]

I’m loving the comments here. There’re as informative as the video. To damn technical to be putting in something exposed to the extremes of temperature changes, vibration and corrosion. Based on these comments by some very intelligent electrical engineers, is the sensor sending back induction readings and the computer is reporting that as calculated voltages to the scan tool? It would be nice if the wiring diagram showed the circuitry of the sensor.

[@throttlebottle5906]

I think the sensor itself is fine and there's a mechanical issue or damage, say a worn bushing or bearing, possibly they hit and bent something inside. then again, I don't know what the sensor looks like it may have bushings that wear and cause misalignment and offset. if they replace it dissect the old unit? 🤔

[@Vlda2393]

Controller flips a phase of one sensor in relation to other before processing the signal. that way it gets a better resolution...

[@mcgama88]

An interesting fault with curious sensor range and referent voltage. As you traced the problem, I kept suspecting the computer element. But the backprobe was a method I don't use much, (but likely should). And the car on the auction block might signal problems that an owner simply discarded, got rid of. As the part (was that 1,000) plus the labor...(unknown but perhaps 8 hours), fluids, alignment, managed to be significant for the miles and age of the car. * in the end, buyer beware....and a nasty driving hazard as any owner, any road test. I did wonder, has this fault caused fatality in the near past? M.

[@bigcheese781]

Should have tried to recal it again, (might be that each time it gives you 1lbft closer to 0). Might have been a no-parts req... Sensor circuitry is likely a feedback system, so the control circuitry continuously try to keep those channels 'alligned', fighting the control-deviation to 0 by controlling the current. The change in impedance is what matters, not the voltage.

[@RoboGeeeeeek]

What if you turned the wheel FIVE degrees left when running Torque Sensor Learn? And would "EPS Steering Angle Sensor Value Clear" have an impact? I am curious as to how the failure occurs. Is the rack or steering mechanically binding and throwing off EPS calibration? Or is the angle sensor just out of sync with the torque sensor? As always, a fantastic and informative video.

[@elisabethellis8851]

Wow,you are the smartest tech I've ever seen!!

[@patburnsent]

At 5:45 when you tried the relearn procedure the angle you turned the wheel may have been critical to getting the correct results. It appears you only estimated the angle you turned the wheel. By the way, very entertaining videos... keep up the good work.

[@sebczajna700]

So the 100 mV difference in the signal was confirmed by the damage to the power steering column???

[@gillgetter3004]

My buddy who is a Honda kook was here while watching this( borrowing a tool) he said rack and sensors!! Then he left

[@calvinguan3256]

This is known as PWM (Pulse Width Modulation). The computer reads the Root Mean Square of the pulse signal. The green waveform is the raw PWM, the red and blue are integrated PWM signals(PWM pulse train passthrough a low pass filter, aka, integrator). The ringings are typical 2nd and higher orders ringing in under-damped conditions which is normal.

[@ramdodgetruck]

couldn't sleep. time for Ivan.

[@PineHollowAutoDiagnostics]

Good morning!

[@yitznewton]

Too bad. Like you say, our 2006 Sienna's hydraulic rack has been slow-leaking for probably 10 years. I top it up every 6 months and it keeps marching on.

[@psdaengr911]

I thought I saw in the description of the circuit that it was using inductance . If that's the case those two sensor outputs could be a coupled to resonant circuits and the output voltage could be the magnitude of the resonance as the frequency varies off a center frequency

[@Giggiyygoo]

I have an S2000 that the eps jiggles the steering wheel and shuts off randomly. Never had the confidence to try to tackle it until now.

[@ecaparts]

You may be over analyzing the voltages on the scan tool because that is just an interpretation of the change in inductance. I assume this configuration consists of 2 coils with the steering shaft acting as an iron core. The computer sends a pulsed DC signal, which the inductor changes the shape of the waveform based on flexing of the shaft. I’m wondering if this calibration failure is due to dirt getting inside between the sensor and the shaft? Great video Ivan!!

[@roxanneabbott8424]

What a case study! I hope you get the repair!

[@PineHollowAutoDiagnostics]

Actually I'm more interested in the diagnosis 😁😁

[@bonedaddy4670]

I’ve been watching car repair videos from many different KZbinr for two years and never have I seen any theory and operation of electric steering. Now I see two in a row from Ivan…mind blown! P.S. where the heck is that dealer getting these auction cars? The steering graveyard?

[@joysimvictoria3647]

i like the analogy of steering graveyard..and your right 2 steering diagnosis in a row..

[@thromboid]

speedkar99 has some (and plenty of other explanatory vids).

[@blake8297]

Lots of steering issues on hondas across a wide range of years if you google it. Parents just took a 2020 back to the dealer yesterday for, yep, steering issues. The rack is bad and would partially lockup when driving straight for a period of time. When you applied enough force, it would pop loose causing excessive overcorrection. Its under warranty, but its $2100 just for the rack….

[@PineHollowAutoDiagnostics]

Whaaaat a 2020? What happened to Honda Quality?

[@cliffweinan3907]

Thanks, great video Ivan. Learned a lot. All these small voltages demand a good ground which you checked. No way to verify a junkyard electric rack has good sensors, so you gotta go with reman. Can you imagine these kinds of dangerous electric rack torque sensor problems happening on self driving cars? You could be dead pretty fast. Awesome techs like you are what it will take to fix the next generation of transportation.

[@kennethnichols7263]

Just did some research on strain gauges. One type simply uses a circuit board resistor that stretches in its sense direction with the material. This stretches the individual wires in the sensor which causes the cross section of the wires ( Poisson’s ratio).to shrink. Thus you get a change in resistance with applied force to the thing that the sensor is glued to. If two sensors were put on opposite sides of the shaft this could be used to measure torque. Could they be using this resistance in an RC circuit to measure how long it takes an applied pulse to this circuit to reach some fixed value? Those outputs sure look like a capacitor being charged by some resistor to me.

[@engeneer_ru5827]

Привет Иван! Я склонен думать, что проблема не электрическая. Сердечник в обмотках датчика, движется не достаточно. Я бы, подключился непосредственно к обмоткам датчика и сравнил их индуктивности, вращая руль в крайние положения. Что скажешь? А в целом, вообще не понятно, как возможно измерить МОМЕНТ, таким датчиком. Больше похоже на датчик угла поворота руля...

[@NoelTelosa]

I have this same problem when my steering rack assembly was repaired, no error codes. The mechanic resolved it by recalibrating the torque sensor

[@fredflintstone8048]

I have no idea what the signal should look like but I can say that current relationships as well as phase relationships are far more reliable than voltages in any instrumentation circuit. Voltages are very subject to the environment. Corrosion, resistance, many things can impact voltages, but we know that current is always the same everywhere in a circuit, and if the current signal transmitting circuit is monitoring current where it's located and most all current signal transmitters do, the transmitter adjusts the current at it's location to account for changes and therefore the current at it's destination MUST be the same. Phase shifts between signals are also reliable because nothing in the circuit will be able to change the timing between two signals or the timing in a single signal.

[@victorianorman1848]

Eric O did one of these and the torque sensor was cracked and required a new rack. Was the subject of an almost or secret recall. Many owners got this defect covered. There is a TSB apparently.

[@victorianorman1848]

Oops it was a Subaru with a similar problem.

[@The_Redkween]

I believe this was a recall for this vintage.

[@kdmq]

Could be that the torsion bar, that they talked about in service info, got permanently twisted somehow. Maybe someone got anxious and turned the wheel against the lock really hard, or hit a curb.

[@PineHollowAutoDiagnostics]

Yup possibly 👍

[@rp5029]

My first thought is it may be acting like an AFR sensor and the voltage reading on your scan tool are simulated based off of the current flow variations? Yup, I'll end my statement with a question mark lol

[@josephwash109]

I wonder if the digital meter would have agreed with the scan tool if it was set to measure AC volts, instead of DC volts. If I had to venture into suspecting the individual component at fault, I'd suspect either the torque sensor itself or the torsion bar in the rack. Maybe the fasteners that secure the torque sensor in position have loosened, allowing the sensor to shift off-center. If it's the torsion bar, I suspect it has fatigued over time and no longer returns to center, which skews the torque sensor reading.

[@billsimmons7754]

Interesting problem. My guess as to how this circuit works and maintains an insensitivity to environmental issues is as follows. The sensor is essentially a current transformer with movement of the core (because of torque) affecting inductance of the magnetic circuit. As the inductance changes the frequency response is altered. The pulse generator (in the module) would need to provide a current regulated pulse as opposed to a typical voltage regulated pulse generator. The current of the output of the transformer would be measured just before the end of the pulse. As others have stated the circuit is immune to series resistance variations (although shunt resistances from moisture are still an issue) .. Because this is rapidly repeated measurement, the output current measurements of the transformer could be averaged (in the module) over several pulses to further remove noise.

[@RobertHancock1]

Not sure about these rack mounted EPS systems. I think they have moved towards those because the steering feel is a bit better compared to column mount, but the components are exposed to a lot harsher of an environment and are harder to service. Also expensive obviously..

[@fredsalter1915]

Disappointing. Thought you were going to remove the whole rack-pinion along with the steering column itself and perform a bench top solder re-flow repair :-P Just kidding! Yet another fantastic vid, sir!

[@charlesronk2989]

They don't sell just the sensor. I had to replace just the sensor in a Malibu. It wasn't that bad.

[@joychung3639]

did you finally found the fix? I buy one recently and it auto-pull to the left. we change a used steering rack and it still do the same thing. send to dealer and they say the module refuse to reset or recalibration and need to change the whole new module. *slap*

[@bbogdanmircea]

That is one misterious sensor, I can't believe it uses voltage, must be current as the voltage different is very small. I am sure you will find out how it works in the end, you are much too good.

[@samsoft9390]

Hi, I know this is an old vid, but do you happen to have an update on this? How was it fixed? Thank you

[@CajunShrek]

Wandering if you did a good flush maybe some debris messing with sensors information?

[@kingofbrutaltheocracy9201]

Kinda want to put a joy stick where the torque sensor goes...also run the the accelerator pedal position sensor threw it. Be pretty sweet.

[@markkillian5154]

Where do you find wiring diagrams for this

[@ethanpotter1435]

Is it possible to only replace the sensor?

[@AP9311]

My brain hurts after watching this. Trying to comprehend how this works!! For life of me, i cannot! But otherwise cool diagnosis!! Excellent video!!! Hope there was solution in the end? A sensor or a rack? Let us know!!

[@VedafoneYT]

@PineHollowAutoDiagnostics Dear Ivan, could you confirm that there is no phase shift between the two signals it could be the computer senses the difference in angle between them then convert it to torque

[@rickjljr11]

Do you think this is some form of a Linear Variable Differential Transformer (LVDT)? an excitation winding and two coils of wire and that ring in the original Honda info is the core? Is there any continuity between the wires?

[@richardcranium5839]

ivan- you have some excellent discussion in the comments here on the engineering going into this electric power steering. you may want to do a follow up video with a better explanation on how it works. just a thought for you.

[@burtaverde8963]

I am sorry, but I think you got this wrong. You did notice that the steering angle was a few hundreds degrees when the steering was centered. That is an implausible angle. Out of the range. But it should be 0 when the steering wheel is centered. So the reset of the steering angle sensor was necessary, before the torque sensor calibration. You had that option in the diagnosis tool, just under the torque sensor calibration. I think that if you would do that, the torque sensor calibration would succeed!

[@PaulHigginbothamSr]

Hey Ivan: Were this mine, just looking at all the corrosion, it seems to me I could cheap out, pull the old one off and clean all the corrosion and lubricate and clean every connection saving me a thousand bucks. Would I trust it at 70 mph on a snowy road? Not one bit. I wouldn't even trust a new one.

[@justsumguy2u]

My old Yaris has electric power steering---hate it. There's no steering feel at all, and the steering wheel doesn't return on it's own like a traditional hydraulic rack would. I wonder if the customer will approve the repair on this? After all, it's an auction car that they no doubt picked up for cheap---will they want to put that kind of money into it?

[@neilmurphy845]

I heard the newer one's have more feel but I still don't really like them for feel

[@justsumguy2u]

@@neilmurphy845 It's like Ivan said, just more stuff to break

[@neilmurphy845]

@@justsumguy2u Yah I think they nailed it on the hydraulic power steering systems

[@thromboid]

Ah, so the "reference voltage" is actually a pulsed signal, and the control unit is measuring the varying inductance of the coils based on how they react to the pulses (as in *reactance*). A bit like the old PC joystick interface, only that used the charging time of a capacitor to determine the resistance of the pots. How frustrating that none of your documentation mentions that, and only talks about the signals as if they're just voltage levels. I wondered what the signals look like at the other end of the cable, at the EPS control unit. If it's measuring inductance, maybe pickup of EMI along the cable or rectification could cause issues. I think you're right that any unusual resistance in the wiring could be problematic, with such tiny variations being measured. It's reassuring that they have redundancy built into a critical sensor like this (similar to the accelerator pedal position sensors with dual pots and opposite voltage sense).

[@PineHollowAutoDiagnostics]

Actually the redundancy didn't help in this case 😬

[@thromboid]

@@PineHollowAutoDiagnostics That is sad but true! Makes me wonder if they never got around to implementing that, or if the discrepancy hasn't hit the fault threshold (though you'd hope it would have).

[@mykofreder1682]

The inputs make sense since a coil will do nothing when activated by a constant voltage it's just wire, you get a reaction from a coil with an AC or a square wave. Thery are using coils within the sensor, activated by the input square wave, to determine the coils relationship to each other (aka position). Probably not the coils since you get signal through it, probably other circuitry that has problems. If you had a wiring diagram for the sensor you might find it is a simple LRC circuitry with a faulty capacitor, the wiring could be part of the R value or it could be solid state.

[@mg8718]

How hard is it to swap out that steering unit?

[@cullenmiller8170]

Ivan you are killing me. I thought I was making some progress of understanding some of the new cars and you post that video. I had to bash my forehead on the table to get my eyes straight again since they had rolled in the back of my head. I miss my 1959 Mercedes Diesel more and more.🤣🤣

[@PineHollowAutoDiagnostics]

Haha yeah this one went a bit deep with the scope patterns 😅

[@laser-sj]

I would swap the signal wires to see if the fault moves. 😉

[@PineHollowAutoDiagnostics]

Excellent idea! I'm sure it would pull to the other side :)

[@morriewyatt6234]

@@PineHollowAutoDiagnostics Actually it's a bad idea, as you would cause positive feedback and it would slam the steering all the way to the right. As others have commented, the sensor is a variant of an LVDT or RVDT in this case. (Search for "LVDT principal of operation".) As you surmised, when the steering wheel is centred, the two signals should be identical. The sensor consists of three coils and a movable iron slug. There's one coil that is excited by the reference voltage pulse. The resulting magnetic field excites the other two coils. By moving the iron slug, the strength of the magnetic coupling will be stronger in whichever of the two output coils it is closest to. A differential amplifier compares the voltages of the two output coils. If there is any way to adjust mechanically the position of the iron slug in the sensor using a non-ferrous tool, you could null out the difference very easily. The sensor calibration function is there to electronically null out the differential if it is slightly off centre. A normal LVDT has the two output coils wired to oppose each other's polarity, making the output signal a tug of war, with the strongest determining the outcome. The normal LVDT is excited by a sine wave reference voltage rather than square wave pulses, but all said, it's still the same principal of operation.

[@pb3033]

Interesting !!!! Asking for part 2 if possible 👍

[@artcraftstudio560]

My hybrid just pop up power steering light and wheel can’t move and after 5 minutes it’s gone can know problem?

[@jamesbruno5896]

That was interesting thanks Ivan!

[@pl5624]

These modern cars won't live 10 years now without major expensive issues...

[@throttlebottle5906]

ten? most are chock full of issues before year three and full on abortion levels by six years.

[@blueghost3647]

Having to sit here and watch an add for 7 minutes not being able to get rid of it makes me want to go to other videos almost

[@paulk437]

Are there 2 sets of windings in the sensor coil and does one of them have some shorted turns?

[@vladchan]

I am surprised there is no fail safe tech engineered into this thing. This is such a glaring safety problem but the computer thinks everything is A OK. For Honda to put these cars on the road and hope for the best is criminal. Where is Nader when you need one?

[@PinBall3]

I Agree As a Corvair is turning over in the bone yard where is the safety (unsafe at any speed Nader) police when you need em` ?

[@vladchan]

@@PinBall3 I have a car with electric steering and am worried now it will steer me right into an oncoming garbage truck some day.

[@RobertHancock1]

Presumably if the values were too far off it would disable the system like on the last EPS video. But it seems like this system isn't necessarily designed the best.

[@ticesine6589]

Can anyone help me out ? I have the exact same problem , mechanic says it prob needs a new rack, I cant even find one online , my car is a 2008 civic dxg coupe manual , and everyone is saying the SI and the hybrids racks are different? Those are the only ones u can find ...

[@andycnm]

The voltage you are measuring are to small to work in the environment of a car. I believe this system must work on current loop. Probably 4-20 ma or 0-40ma this would be far more accurate a reliable in a car.

[@keithhibell7614]

I have a 2012 honda civic when it rains i have no power stering, its hard to steer in both directions, to drive it i have to unplug the moudle so it wouln't fight me.

[@samuelgranados776]

I just replaced my brothers 2019 Honda Civic steering rack and pinion. It’s Shatter inside and was making Grinding horrible noises lots of lights on dashboard, car will not drive at all.

[@nellyxavier8923]

Hi Ivan ..Good work as usual. I need to know what was the final call wen you realized that the signal was not getting pulled -down correctly. Is the sensor or the steering column?

[@RCTPatriot75]

I had a 94 Legend that would spin the steering all the way left than all the way right very rapidly.