Supercharged Honda K20 running issue
Discussion
Morning all
I have a TVS1320 supercharged Honda K20 in an Ariel Atom with other supporting mods, it has previously been custom mapped on a hub mounted dyno using Hondata Flashpro.
The car has recently been stripped down for chassis work, including engine removal/re-fitting. It's since getting the car back that a problem has manifested, whether that is a coincidence or not is yet to be determined…
The car has started losing power towards the top of the rev range, most noticeable in higher gears. The car also feels like it cuts out for a split second during these episodes. I have data logged the ECU and what is happening is that under high load/rpm (eg WOT in 5th gear), at around 6K RPM the ECT (engine coolant temperature) and IAT (inlet air temperature) signals to the ECU suddenly drop to -40 degrees C and rapidly fluctuate before eventually returning to normal once off the throttle. The Honda K20 engine uses VTEC. One of the parameters that must be met for VTEC to engage is that coolant temperature must be 60 degrees C or above. When the ECT signal goes faulty and falsely reports such low temperatures, VTEC drops out, throttle is restricted, and cold fuel and ignition trims are activated resulting in loss of power etc.
Suspecting a possible earth or power issue, I have checked/cleaned all grounds and power feeds. I have also run two new additional ground feeds from the cylinder head and block. This has made no difference.
My car then went on the Dyno at the garage that originally mapped it for me a few years ago. They "believed" that the issue was a poor/bad connection in a multi pin plug, and after attending to that the car apparently behaved itself on the Dyno. However, they also report that boost pressure is down resulting in a power loss of circa 25bhp across the rev range. Max boost pressure was originally around 14psi but is now around 9psi.
I took the car away and first time out it's doing exactly the same as before. New data logs show exact same issue ie ECT and IAT signals giving spurious readings under load at high rpm with the VTEC dropping out.
I've had another look myself. I have checked for boost/vacuum leaks and cannot find any (I sprayed brake cleaner everywhere relevant and could not get any hiccupping etc from the engine. I have adjusted the supercharger bypass valve which looked to be not closing fully under boost so hopefully the low boost issue is resolved, further data logs support this but I will wait until it’s back on the dyno for confirmation.
I disconnected the ECT sensor from the loom and checked continuity of the wiring with a multimeter. I'm getting a max resistance reading of 0.3 ohms for both wires from the sensor connector back to the ECU connector. These readings do NOT deviate when I twist/pull/wiggle the loom, so it appears that the integrity of the ECT sensor wires is "good", although I appreciate that I tested this without the engine running and therefore nothing was under load. Also these readings obviously exclude the actual ECU from the circuit.
I repeated the above with the IAT sensor wiring and had the same results, so again the wiring integrity appears sound. The three IAT sensor wires go through the same middle grey ECU connector as the ECT wires and share the same signal ground wire. The connector which the garage believed to have issues was NOT this connector.
The Honda ECU is stamped 37820-RRB-315 and I am using an ECU pin out diagram for the Honda Civic FN2 which so far is correct.
I've ordered a brand new genuine Honda ECT sensor to try when it arrives, but due to the nature of the fault (both ECT and IAT sensors giving bad readings under exactly the same very specific conditions) I very much doubt it's a sensor issue, but at least I will be able to rule it out.
The car is booked back in for further dyno investigation in a couple of weeks’ time.
In the meantime, what else can I do or check? I’m wondering if it’s worth severing the existing shared signal ground wire for the ECT and IAT sensors and running a new ground wire, just in case there is an issue (even though it checked out “OK” with my limited multimeter tests).
I’m really scratching my head. Dyno time is not cheap so any advice or help would be much appreciated, if there is anything else I can do myself to help narrow down the issue I’d be grateful for any suggestions.
Many thanks
I have a TVS1320 supercharged Honda K20 in an Ariel Atom with other supporting mods, it has previously been custom mapped on a hub mounted dyno using Hondata Flashpro.
The car has recently been stripped down for chassis work, including engine removal/re-fitting. It's since getting the car back that a problem has manifested, whether that is a coincidence or not is yet to be determined…
The car has started losing power towards the top of the rev range, most noticeable in higher gears. The car also feels like it cuts out for a split second during these episodes. I have data logged the ECU and what is happening is that under high load/rpm (eg WOT in 5th gear), at around 6K RPM the ECT (engine coolant temperature) and IAT (inlet air temperature) signals to the ECU suddenly drop to -40 degrees C and rapidly fluctuate before eventually returning to normal once off the throttle. The Honda K20 engine uses VTEC. One of the parameters that must be met for VTEC to engage is that coolant temperature must be 60 degrees C or above. When the ECT signal goes faulty and falsely reports such low temperatures, VTEC drops out, throttle is restricted, and cold fuel and ignition trims are activated resulting in loss of power etc.
Suspecting a possible earth or power issue, I have checked/cleaned all grounds and power feeds. I have also run two new additional ground feeds from the cylinder head and block. This has made no difference.
My car then went on the Dyno at the garage that originally mapped it for me a few years ago. They "believed" that the issue was a poor/bad connection in a multi pin plug, and after attending to that the car apparently behaved itself on the Dyno. However, they also report that boost pressure is down resulting in a power loss of circa 25bhp across the rev range. Max boost pressure was originally around 14psi but is now around 9psi.
I took the car away and first time out it's doing exactly the same as before. New data logs show exact same issue ie ECT and IAT signals giving spurious readings under load at high rpm with the VTEC dropping out.
I've had another look myself. I have checked for boost/vacuum leaks and cannot find any (I sprayed brake cleaner everywhere relevant and could not get any hiccupping etc from the engine. I have adjusted the supercharger bypass valve which looked to be not closing fully under boost so hopefully the low boost issue is resolved, further data logs support this but I will wait until it’s back on the dyno for confirmation.
I disconnected the ECT sensor from the loom and checked continuity of the wiring with a multimeter. I'm getting a max resistance reading of 0.3 ohms for both wires from the sensor connector back to the ECU connector. These readings do NOT deviate when I twist/pull/wiggle the loom, so it appears that the integrity of the ECT sensor wires is "good", although I appreciate that I tested this without the engine running and therefore nothing was under load. Also these readings obviously exclude the actual ECU from the circuit.
I repeated the above with the IAT sensor wiring and had the same results, so again the wiring integrity appears sound. The three IAT sensor wires go through the same middle grey ECU connector as the ECT wires and share the same signal ground wire. The connector which the garage believed to have issues was NOT this connector.
The Honda ECU is stamped 37820-RRB-315 and I am using an ECU pin out diagram for the Honda Civic FN2 which so far is correct.
I've ordered a brand new genuine Honda ECT sensor to try when it arrives, but due to the nature of the fault (both ECT and IAT sensors giving bad readings under exactly the same very specific conditions) I very much doubt it's a sensor issue, but at least I will be able to rule it out.
The car is booked back in for further dyno investigation in a couple of weeks’ time.
In the meantime, what else can I do or check? I’m wondering if it’s worth severing the existing shared signal ground wire for the ECT and IAT sensors and running a new ground wire, just in case there is an issue (even though it checked out “OK” with my limited multimeter tests).
I’m really scratching my head. Dyno time is not cheap so any advice or help would be much appreciated, if there is anything else I can do myself to help narrow down the issue I’d be grateful for any suggestions.
Many thanks
richhead said:
it might be worth doing continuity checks from sensers to ect, and while checking have a wiggle of wiring etc.
I have. Did you somehow miss this bit?Toilet Duck said:
I disconnected the ECT sensor from the loom and checked continuity of the wiring with a multimeter. I'm getting a max resistance reading of 0.3 ohms for both wires from the sensor connector back to the ECU connector. These readings do NOT deviate when I twist/pull/wiggle the loom, so it appears that the integrity of the ECT sensor wires is "good", although I appreciate that I tested this without the engine running and therefore nothing was under load. Also these readings obviously exclude the actual ECU from the circuit.
I repeated the above with the IAT sensor wiring and had the same results, so again the wiring integrity appears sound. The three IAT sensor wires go through the same middle grey ECU connector as the ECT wires and share the same signal ground wire. The connector which the garage believed to have issues was NOT this connector.
I repeated the above with the IAT sensor wiring and had the same results, so again the wiring integrity appears sound. The three IAT sensor wires go through the same middle grey ECU connector as the ECT wires and share the same signal ground wire. The connector which the garage believed to have issues was NOT this connector.
vw_99 said:
Could it be that out on the road and under load that something is moving and could be pulling the plug or a wire.
Yes, quite possible. The only thing I'd say is the fault doesn't happen randomly. You can drive around without issue, it's only high loads/rpm, and mainly high gears, that this fault manifests. If it was bumps in the road, I'd expect the ECT/IAT signals to drop in and out randomly at differing rpm/loads. The actual sensors and loom etc involved are not "tight" so I don't think they are being pulled.GreenV8S said:
I assume that these sensors are both thermisters. These typically have the characteristics that the resistance goes down as the termperature rises. You'd be looking for resistance in the circuit to cause the very low reading. It's quite likely the -40C reading is the minimum it can record and you're actually getting an open circuit.
If that was me I'd measure the resistance across each sensor under normal operating conditions and connect the equivalent fixed resistance directly to the corresponding pins on the ECU connector. If the problem still happens you know you can rule out all the engine loom etc and are looking for a problem either in the connector or within the ECU.
I'd also eyeball the insides of the ECU to see whether the connection from the socket to the board was physically intact and whether there were any internal daughter boards or connectors that had become dislodged. Sometimes these connectors are strain relieved properly and can be damaged / dislodged by excessive force on the wiring, such as by pulling on the connector or letting the ECU hang on the wire.
Excellent info and suggestions, thank you very much indeed!!!If that was me I'd measure the resistance across each sensor under normal operating conditions and connect the equivalent fixed resistance directly to the corresponding pins on the ECU connector. If the problem still happens you know you can rule out all the engine loom etc and are looking for a problem either in the connector or within the ECU.
I'd also eyeball the insides of the ECU to see whether the connection from the socket to the board was physically intact and whether there were any internal daughter boards or connectors that had become dislodged. Sometimes these connectors are strain relieved properly and can be damaged / dislodged by excessive force on the wiring, such as by pulling on the connector or letting the ECU hang on the wire.
The ECT sensor is indeed a two wire thermister.
The IAT sensor has three wires. A ground (shared with the ECT sensor), 5v reference from the ECU, and a signal wire back to the ECU. I think that works by altering the voltage of the signal wire back to the ECU relative to the measured temperature, so I don't know if that can be "tricked" by somehow substituting with a resistor??
I will open up the ECU to have a look for any visible damage. There are no piggy backs etc, it's just a standard OE Honda ECU. The only alterations are software based via the Hondata Flashpro. It plugs into the OBD port and allows changing of the ECU tables and parameters etc.
It will be a few days before I have enough free time to have a look at this, but I will update the thread.
Thanks again, that's a really helpful post with some great "outside the box" ideas
Another little update.
I fitted a new genuine Honda ECT sensor (was cheap enough to 100% prove one way or the other if the ECT sensor was iffy). Same fault symptoms.
I disconnected the loom from the IAT sensor and inserted a resistor of a specific value to make the ECU think that it was seeing constant 40 Deg C (thereby effectively removing the actual IAT sensor from the circuit). Again same fault symptoms.
I ran a new signal ground wire from ECU connector block B pin B23 up to the ground connection on both the ECT and IAT sensors. I ran this new ground wire in parallel with the existing loom. Fault symptoms remain.
I had the car booked back in with the garage for more dyno time but when I told them what I had done they said that they had planned to run new ground wire so I had beat them to it. They believe what's been done so far has eliminated a loom or sensor issue and advise that it looks to be the ECU that's breaking down.
The cheapest/simplest option is to just get another standard Honda ECU and upload/transfer the existing map. The alternative is to go balls out and upgrade to a Syvecs standalone ECU and map from scratch but that is a very considerable outlay, even though it's undoubtedly a better ECU for tuning etc
I fitted a new genuine Honda ECT sensor (was cheap enough to 100% prove one way or the other if the ECT sensor was iffy). Same fault symptoms.
I disconnected the loom from the IAT sensor and inserted a resistor of a specific value to make the ECU think that it was seeing constant 40 Deg C (thereby effectively removing the actual IAT sensor from the circuit). Again same fault symptoms.
I ran a new signal ground wire from ECU connector block B pin B23 up to the ground connection on both the ECT and IAT sensors. I ran this new ground wire in parallel with the existing loom. Fault symptoms remain.
I had the car booked back in with the garage for more dyno time but when I told them what I had done they said that they had planned to run new ground wire so I had beat them to it. They believe what's been done so far has eliminated a loom or sensor issue and advise that it looks to be the ECU that's breaking down.
The cheapest/simplest option is to just get another standard Honda ECU and upload/transfer the existing map. The alternative is to go balls out and upgrade to a Syvecs standalone ECU and map from scratch but that is a very considerable outlay, even though it's undoubtedly a better ECU for tuning etc
stevieturbo said:
Presumably the ecu offers more than one sensor ground pin ? Wire your sensors to the other, or one of the other sensor grounds.
Yes the ECU does have more than one sensor ground wire. It has three multi-pin plugs connected to it. On each of the three multi-pin connectors, there is a sensor/signal ground wire.The IAT, ECT and MAP sensors are wired onto the middle (B) multi-pin connector. The other two multi-pin connectors have there own array of sensor signals going to them (TPS, lambda, crank position etc etc).
If I were to wire the ECT and IAT ground connections to a sensor ground wire on a different ECU multi-pin connector, would that work? Are the different sensor ground wires on different ECU multi-pin connectors not independent of each other? I assumed the ECU is compartmentalized.
Hopefully this link shows the ECU pinout:
https://jumpshare.com/s/DXD9nqpfyjTvMnCFzdkJ
Edited by Toilet Duck on Friday 16th August 21:30
richhead said:
Toilet Duck said:
richhead said:
it might be worth doing continuity checks from sensers to ect, and while checking have a wiggle of wiring etc.
I have. Did you somehow miss this bit?Toilet Duck said:
I disconnected the ECT sensor from the loom and checked continuity of the wiring with a multimeter. I'm getting a max resistance reading of 0.3 ohms for both wires from the sensor connector back to the ECU connector. These readings do NOT deviate when I twist/pull/wiggle the loom, so it appears that the integrity of the ECT sensor wires is "good", although I appreciate that I tested this without the engine running and therefore nothing was under load. Also these readings obviously exclude the actual ECU from the circuit.
I repeated the above with the IAT sensor wiring and had the same results, so again the wiring integrity appears sound. The three IAT sensor wires go through the same middle grey ECU connector as the ECT wires and share the same signal ground wire. The connector which the garage believed to have issues was NOT this connector.
I repeated the above with the IAT sensor wiring and had the same results, so again the wiring integrity appears sound. The three IAT sensor wires go through the same middle grey ECU connector as the ECT wires and share the same signal ground wire. The connector which the garage believed to have issues was NOT this connector.
Edited by Toilet Duck on Saturday 17th August 09:01
stevieturbo said:
In general, internally usually they will all be linked. Either way, it should work fine
You could certainly unplug the ecu and bell the relevant pins for continuity. I would expect them to be all linked.
But it's certainly worth a try at this stage either way, because it does sound like one pin must be down. Very unusual though.
Ok, so I looked into this earlier today. I removed the ECU and buzzed out the pins, unfortunately the signal/sensor ground pin used by the ECT and IAT sensors have NO continuity to the other signal/sensor ground pins on the other ECU connections, they are electrically independent with the ECU in a de-energised state.You could certainly unplug the ecu and bell the relevant pins for continuity. I would expect them to be all linked.
But it's certainly worth a try at this stage either way, because it does sound like one pin must be down. Very unusual though.
It's looking like the end of the road for this one, I fear I am going to have to bite the bullet and replace the actual ECU.
Thanks for everyone's help and advice, it's certainly been a learning experience for me.
stevieturbo said:
I would still be adding a sensor ground from another pin. If you believe the ecu is damaged....what's the worst that can happen ? Not a lot really.
Very true, I can give it a go. In my head I can't see how it would work though if the cct for the IAT and ECT sensors is now being made using an "unrelated" signal ground? If there is no continuity between the different signal ground pins on the ECU, then using a different pin will surely break the circuit? Unless I'm being thick?GreenV8S said:
If it was me, at this point I'd be inside the ECU case and following where those ground connections lead. You're looking for an open circuit fault and you seem to have found it within the ECU.
I did undo the four screws and tried to seperate the two halves of the metal case but it didn't want to open. I'm not sure if there is some sort of sealant in there? I thought about using a hot air gun to add a bit of heat to melt any glue but thought better of it. I'll give it another go tomorrow. It must come apart, I was just wary of totally destroying itMorning all, I have an update!
Having exhausted pretty much every avenue, I managed to get hold of a used OE Honda ECU from an Atom owner who had upgraded to Syvecs. I uploaded my custom map and the car now drives without issue, data logs show clean ECT/IAT signals etc with no loss of VTEC. So it was an ECU fault.
I'm still mulling over switching to Syvecs myself. The extra fail safes and engine protection are a big appeal.
Thanks again for everyone's help
Having exhausted pretty much every avenue, I managed to get hold of a used OE Honda ECU from an Atom owner who had upgraded to Syvecs. I uploaded my custom map and the car now drives without issue, data logs show clean ECT/IAT signals etc with no loss of VTEC. So it was an ECU fault.
I'm still mulling over switching to Syvecs myself. The extra fail safes and engine protection are a big appeal.
Thanks again for everyone's help
Gassing Station | Engines & Drivetrain | Top of Page | What's New | My Stuff