Sweet, yes steal it!
bogle's 1991 mild build thread
- Thread starter bogle
- Start date
Sweet, yes steal it!
The AC project is in the bag. I finally got the right part, and got it all back together and charged up this weekend. [emoji3587][emoji3587]
For context, what was my AC problem? After some thrashing initially to get it converted to R134a and working by a shop, it appeared that the compressor was bad. All goo all over
My assumption was that the compressor was not able to hold a seal somewhere, and it had leaked all the refrigerant out. The solution would be a new compressor and some new o-rings at the compressor. Seemed reasonable enough.
But when I went to remove the compressor, there wasn't, uh, an egregious leak. I suppose there was just not enough pressure to kick the compressor on.
With the compressor out, it was pretty clear that the source of the leakage was the pressure relief valve. See the green here:
Prior to seeing it in the flesh, I didn't even know this part existed. Basically it's an over-pressure failsafe. It's hooked up to the high side (discharge side) of the manifold block. Pressure over like 350psi? Open! Newer cars don't even have these things as it's not ideal to vent to the atmosphere, they shut the compressor off in an over-pressure situation instead.
The question is why. Why did this valve open in the first place? After some reading, here are a couple potential reasons:
* The valve it self was bad. It's possible there was an over pressure situation in the past and made it open even at lower pressure. Maybe it just went bad? Seems unlikely though.
* There could be a blockage somewhere in the system.
* It was overcharged?
* Some issue with the compressor?
I could sorta address all these things by replacing the compressor, the valve, flushing the system to make sure there is no blockage, then properly charging it.
The valve
Ok so I needed to replace the valve. This section is kinda long because this was the most painful part of the whole job. This dumb little $10 part is the reason the car was in the air for 3 weeks.
The stock 91 relief valve (38801-PT3-A01) is, like all the other R12 specific parts, discontinued. Well, what about the later-model one? I was trying to finish the job the day I discovered that I needed to replace the valve.
I called around and the local Acura dealership had the late-model part (38801-P9K-E01). It was $78 and like 1/3 the price on Amayama, but I needed it so I went and happily overpaid so I could finish the job.
Does it fit? Nope! The threads in the manifold block are totally different:
* 91-92 R12 - 3/8"-24
* 93+ R134a - M10x1.5
So I needed either a manifold block from 93+ car with the M10 threads or a valve with 3/8"-24 threads.
I tried to find a manifold locally, but nope. The best option was Amayama, but their lead time is often several weeks to a month.
Ok, maybe an aftermarket valve? With an aftermarket valve, there were other considerations beyond thread size. First of all, I needed valve with a "non-captive" o-ring. There is a recess in the manifold housing for the o-ring, so the valve can't have one
Aaand the last consideration is that some valves with the correct threads and o-ring configuration explicitly said "only for R12". Ugh, why? I read it may be that the max pressure is higher for R134a. It is possible that it was only because of the o-ring used. But I didn't want to risk it, only to have to tear everything apart later because of this dumb thing. I wanted to find a valve that indicated it'd work with R134a.
This was also a process, many websites that list these have pretty minimal info. What are the threads? Is it a non-captive o-ring? Is it for R134a? Napa had one, I ordered it and picked it up: captive o-ring, not it.
After tons of searching, I landed on Sanden T6951 / Airsource 5686A. It allegedly had the right attributes, and looked similar to the later OEM one. I bought two cause shipping was $18 for a $9 part. It finally came 3 weeks later:
Does it fit??? OMG Yes! Installed in the manifold block:
Everything else can go back together now!
Flushing
Moving on! Alright. I referred a bunch to @Big McLargeHuge's AC post for this. I focused on flushing things on the high side: compressor to condenser line, both condensers, the line between the condensers, and the line from the condenser to the drier. If the lines were super dirty, then I'd do the other 2 lines, which are hard to flush because of the way they mount to the firewall.
I bought this four uncles flush kit:
I wanted to be super clean so I bought some plastic tubing, an oil mat, and had some leftover water jugs to collect the solvent.
3/4" ID tubing and a short piece of 1" ID tubing covered all the fittings just right:
I flushed the lines first cause they seemed easier. They were all very clean. Cool, onto the condensers.
The drivers side condenser was easier to deal with because of the angles IIRC, so I did it first. It was dirrrrttttyyy, wow. But I flushed it until clean solvent came out. There was no blockage here, just dirty.
Then I went to the passenger side. It seemed like there was indeed a blockage. At first virtually no air or solvent came out the other side, then all of a sudden it did, all at once. It happened very fast, it's possible I had the spray gun at a weird angle or something at first. Or you know, there was an actual blockage. A plugged condenser, especially the first one from the compressor could explain an over pressure situation.
Here's the jug after flushing the drivers compressor, it's v dirty. You can also see the short 1" tubing. It slides over the 3/4 tubing for the big fittings:
A couple things I learned while flushing the system:
* The solvent flush gun will use all the solvent you put in the canister within 2 seconds of pulling the trigger. The canister holds a lot so you can go through a ton of solvent very quickly. I only bought 2 cans of solvent. I spent a lot of time filling the canister 1/4 of the way to conserve. Pull the trigger, is it clean? If not, run another 1/4 can through the part. I used about a half a can for each condenser. I ended up a little of the 2nd can of solvent left.
* The fittings on the flush gun would bust loose on occasion while under pressure and spray solvent everywhere. It's tough to get the canister (upright) and spray gun (contorted) oriented correctly, so there was a lot of moving and twisting. Sometimes the twisting would break the fittings loose. I learned to hold onto the fittings when moving the flush kit around. Also wear goggles omfg
* After flushing, especially the condensers, there was a bunch of solvent still in the part. I just blew air through them until they were dry.
The compressor
There are several options for the compressor. Which compressor? Which oil?
I made the decision a while back to use the 91 replacement compressor (Denso 471-1424) and ester oil for ease and compatibility all around. I get a new pulley and clutch, don't need to replace the bracket or connector, and I'll load it with oil that is compatible with the small amount of residual mineral oil in the system. If I were starting with all new parts, I'd go with the 97+ compressor, PAG oil and related parts. But I'm not, so 91 replacement + ester it is.
First up was draining all the oil out of the new compressor. It's mineral oil, and I wanted to get as much out as possible. I had a bit of a hard time getting the oil out and measured. It was pretty cold when I was doing this (in the high 40s), so it's possible it was higher viscosity than in a normal situation. I got about 40ml out and measured:
I was having trouble getting more out quickly, so I just left it upside down for a while, rotating the pump every now and then. I probably got another 30 or 40ml out of it that way.
I would be flushing a few lines and the condensers. I loaded it with 100ml of ester.
Cool. I got new o-rings and a new idler
Then everything installed and the belt back on. Yay!
Reinstall
Overall, reinstalling everything went smoothly. The epoxy on the motor mount nut broke loose, but I was able to get both hands in there and get it connected. I was even able to get the torque wrench on it.
Here's the torque spec page for the crossmember and friends cause I didn't take any pics. I like torquing suspension parts.
Car on the ground
Finally I could put the car on the ground. I started it up to make sure there are no oil leaks from all the oil work I did. No leaks! So far. We'll see after some driving and trips up to redline.
Drier
I bought a four seasons 33412 drier. I also got a new bracket cause mine was rusty and I read that the later drier was smaller. We'll this drier is a pretty fat boi, so I had to get a new like 50mm bolt. Titanium cause why not:
Then right before pulling vacuum, I installed the drier and connected up the lines.
Pulling vacuum
Next up was pulling vacuum. This will remove moisture in the system and tell you if there are any leaks. Like can it hold vacuum for 30 minutes? If so, no leaks.
I got this little thing
Then I hooked up my AC gauges.
Connected the air compressor with the engine off, opened all the valves, and we had vacuum.
My air compressor is a pretty little one ("the tank"). It would allow the pump to pull 25" at first, then when the tank was depleted, it'd only pull 23". Seems fine as long as it's under 20"?
I pulled vacuum for 15 min, closed the valves, then disconnected the compressor to make sure it held vacuum. I waited for about 30 min, and the gauges didn't move, so I figured it was fine to go ahead and start the charging process.
Before charging, you pull vacuum _again_, but for longer. Hook up the vacuum pump again without the car running, and plug the compressor in. I think I pulled it for about 45 min. The annoying part is the air compressor is running the entire time. "Why is it so loud out there?" my girlfriend asked several times. Progress, you know?
It seems like it removed a bunch of moisture. I put a rag under the pump and it was pretty soaked at the end.
Charging
I watched this video below and it jived with everything else I had read:
https://youtu.be/Pdq8JAlct6s
I measured all the cans. I was shooting for a tiny bit over 800g. My cans had about 330g of refrigerant in each.
I didn't get any pics while filling. And the video does a great job on the steps, so I won't rehash here. But there were some things I learned:
When you add the first can, there will be air in the yellow fill hose. You need to vacate air from the yellow fill hose so it's only refrigerant. I had no Schrader valve on the fill hose or my gauges. Before I opened the low side valve to start the charging, I burped the yellow line fitting at the gauges. In this pic, I would loosen the yellow line fitting until some refrigerant came out, then tightened it back up:
* I had trouble swapping in a new can. In the video, he swaps it out with the valve fully closed to hold Freon in the fill line. When I tried that, refrigerant went everywhere. I ended up just burping the fill line each time.
* After the first can, AC Compressor didn't consistently stay engaged. This is probably normal, but I dunno. It'd stay on for 10-20 seconds, then kick off for 20 seconds, kick back on, etc. Maybe cause ambient was cold? It was high 50s F in the garage. It'd only pick up refrigerant when the compressor was engaged, so it took a while.
* When I was finished and with the compressor engaged, the low side ended up being 25psi, high side was 148psi which is in line with all the charts for my coldass ambient temp. I ran it for a while after charging just to make sure those pressures were consistent over time.
* I ended up using 810g of refrigerant. Though, there may be less in the system as a bit came out while burping.
Done!
We'll that's it! I have AC finally! It's cold! Everything seems to work (condenser fans etc)! I’ve had the car up in the air again a couple times, and it seems like no leaks. Hopefully it lasts this time. I'm really glad this project is over.
I also made progress on the oil temp sensor. I’ll post about it soon
For context, what was my AC problem? After some thrashing initially to get it converted to R134a and working by a shop, it appeared that the compressor was bad. All goo all over
My assumption was that the compressor was not able to hold a seal somewhere, and it had leaked all the refrigerant out. The solution would be a new compressor and some new o-rings at the compressor. Seemed reasonable enough.
But when I went to remove the compressor, there wasn't, uh, an egregious leak. I suppose there was just not enough pressure to kick the compressor on.
With the compressor out, it was pretty clear that the source of the leakage was the pressure relief valve. See the green here:
Prior to seeing it in the flesh, I didn't even know this part existed. Basically it's an over-pressure failsafe. It's hooked up to the high side (discharge side) of the manifold block. Pressure over like 350psi? Open! Newer cars don't even have these things as it's not ideal to vent to the atmosphere, they shut the compressor off in an over-pressure situation instead.
The question is why. Why did this valve open in the first place? After some reading, here are a couple potential reasons:
* The valve it self was bad. It's possible there was an over pressure situation in the past and made it open even at lower pressure. Maybe it just went bad? Seems unlikely though.
* There could be a blockage somewhere in the system.
* It was overcharged?
* Some issue with the compressor?
I could sorta address all these things by replacing the compressor, the valve, flushing the system to make sure there is no blockage, then properly charging it.
The valve
Ok so I needed to replace the valve. This section is kinda long because this was the most painful part of the whole job. This dumb little $10 part is the reason the car was in the air for 3 weeks.
The stock 91 relief valve (38801-PT3-A01) is, like all the other R12 specific parts, discontinued. Well, what about the later-model one? I was trying to finish the job the day I discovered that I needed to replace the valve.
I called around and the local Acura dealership had the late-model part (38801-P9K-E01). It was $78 and like 1/3 the price on Amayama, but I needed it so I went and happily overpaid so I could finish the job.
Does it fit? Nope! The threads in the manifold block are totally different:
* 91-92 R12 - 3/8"-24
* 93+ R134a - M10x1.5
So I needed either a manifold block from 93+ car with the M10 threads or a valve with 3/8"-24 threads.
I tried to find a manifold locally, but nope. The best option was Amayama, but their lead time is often several weeks to a month.
Ok, maybe an aftermarket valve? With an aftermarket valve, there were other considerations beyond thread size. First of all, I needed valve with a "non-captive" o-ring. There is a recess in the manifold housing for the o-ring, so the valve can't have one
Aaand the last consideration is that some valves with the correct threads and o-ring configuration explicitly said "only for R12". Ugh, why? I read it may be that the max pressure is higher for R134a. It is possible that it was only because of the o-ring used. But I didn't want to risk it, only to have to tear everything apart later because of this dumb thing. I wanted to find a valve that indicated it'd work with R134a.
This was also a process, many websites that list these have pretty minimal info. What are the threads? Is it a non-captive o-ring? Is it for R134a? Napa had one, I ordered it and picked it up: captive o-ring, not it.
After tons of searching, I landed on Sanden T6951 / Airsource 5686A. It allegedly had the right attributes, and looked similar to the later OEM one. I bought two cause shipping was $18 for a $9 part. It finally came 3 weeks later:
Does it fit??? OMG Yes! Installed in the manifold block:
Everything else can go back together now!
Flushing
Moving on! Alright. I referred a bunch to @Big McLargeHuge's AC post for this. I focused on flushing things on the high side: compressor to condenser line, both condensers, the line between the condensers, and the line from the condenser to the drier. If the lines were super dirty, then I'd do the other 2 lines, which are hard to flush because of the way they mount to the firewall.
I bought this four uncles flush kit:
I wanted to be super clean so I bought some plastic tubing, an oil mat, and had some leftover water jugs to collect the solvent.
3/4" ID tubing and a short piece of 1" ID tubing covered all the fittings just right:
I flushed the lines first cause they seemed easier. They were all very clean. Cool, onto the condensers.
The drivers side condenser was easier to deal with because of the angles IIRC, so I did it first. It was dirrrrttttyyy, wow. But I flushed it until clean solvent came out. There was no blockage here, just dirty.
Then I went to the passenger side. It seemed like there was indeed a blockage. At first virtually no air or solvent came out the other side, then all of a sudden it did, all at once. It happened very fast, it's possible I had the spray gun at a weird angle or something at first. Or you know, there was an actual blockage. A plugged condenser, especially the first one from the compressor could explain an over pressure situation.
Here's the jug after flushing the drivers compressor, it's v dirty. You can also see the short 1" tubing. It slides over the 3/4 tubing for the big fittings:
A couple things I learned while flushing the system:
* The solvent flush gun will use all the solvent you put in the canister within 2 seconds of pulling the trigger. The canister holds a lot so you can go through a ton of solvent very quickly. I only bought 2 cans of solvent. I spent a lot of time filling the canister 1/4 of the way to conserve. Pull the trigger, is it clean? If not, run another 1/4 can through the part. I used about a half a can for each condenser. I ended up a little of the 2nd can of solvent left.
* The fittings on the flush gun would bust loose on occasion while under pressure and spray solvent everywhere. It's tough to get the canister (upright) and spray gun (contorted) oriented correctly, so there was a lot of moving and twisting. Sometimes the twisting would break the fittings loose. I learned to hold onto the fittings when moving the flush kit around. Also wear goggles omfg
* After flushing, especially the condensers, there was a bunch of solvent still in the part. I just blew air through them until they were dry.
The compressor
There are several options for the compressor. Which compressor? Which oil?
I made the decision a while back to use the 91 replacement compressor (Denso 471-1424) and ester oil for ease and compatibility all around. I get a new pulley and clutch, don't need to replace the bracket or connector, and I'll load it with oil that is compatible with the small amount of residual mineral oil in the system. If I were starting with all new parts, I'd go with the 97+ compressor, PAG oil and related parts. But I'm not, so 91 replacement + ester it is.
First up was draining all the oil out of the new compressor. It's mineral oil, and I wanted to get as much out as possible. I had a bit of a hard time getting the oil out and measured. It was pretty cold when I was doing this (in the high 40s), so it's possible it was higher viscosity than in a normal situation. I got about 40ml out and measured:
I was having trouble getting more out quickly, so I just left it upside down for a while, rotating the pump every now and then. I probably got another 30 or 40ml out of it that way.
I would be flushing a few lines and the condensers. I loaded it with 100ml of ester.
Cool. I got new o-rings and a new idler
Then everything installed and the belt back on. Yay!
Reinstall
Overall, reinstalling everything went smoothly. The epoxy on the motor mount nut broke loose, but I was able to get both hands in there and get it connected. I was even able to get the torque wrench on it.
Here's the torque spec page for the crossmember and friends cause I didn't take any pics. I like torquing suspension parts.
Car on the ground
Finally I could put the car on the ground. I started it up to make sure there are no oil leaks from all the oil work I did. No leaks! So far. We'll see after some driving and trips up to redline.
Drier
I bought a four seasons 33412 drier. I also got a new bracket cause mine was rusty and I read that the later drier was smaller. We'll this drier is a pretty fat boi, so I had to get a new like 50mm bolt. Titanium cause why not:
Then right before pulling vacuum, I installed the drier and connected up the lines.
Pulling vacuum
Next up was pulling vacuum. This will remove moisture in the system and tell you if there are any leaks. Like can it hold vacuum for 30 minutes? If so, no leaks.
I got this little thing
Then I hooked up my AC gauges.
Connected the air compressor with the engine off, opened all the valves, and we had vacuum.
My air compressor is a pretty little one ("the tank"). It would allow the pump to pull 25" at first, then when the tank was depleted, it'd only pull 23". Seems fine as long as it's under 20"?
I pulled vacuum for 15 min, closed the valves, then disconnected the compressor to make sure it held vacuum. I waited for about 30 min, and the gauges didn't move, so I figured it was fine to go ahead and start the charging process.
Before charging, you pull vacuum _again_, but for longer. Hook up the vacuum pump again without the car running, and plug the compressor in. I think I pulled it for about 45 min. The annoying part is the air compressor is running the entire time. "Why is it so loud out there?" my girlfriend asked several times. Progress, you know?
It seems like it removed a bunch of moisture. I put a rag under the pump and it was pretty soaked at the end.
Charging
I watched this video below and it jived with everything else I had read:
https://youtu.be/Pdq8JAlct6s
I measured all the cans. I was shooting for a tiny bit over 800g. My cans had about 330g of refrigerant in each.
I didn't get any pics while filling. And the video does a great job on the steps, so I won't rehash here. But there were some things I learned:
When you add the first can, there will be air in the yellow fill hose. You need to vacate air from the yellow fill hose so it's only refrigerant. I had no Schrader valve on the fill hose or my gauges. Before I opened the low side valve to start the charging, I burped the yellow line fitting at the gauges. In this pic, I would loosen the yellow line fitting until some refrigerant came out, then tightened it back up:
* I had trouble swapping in a new can. In the video, he swaps it out with the valve fully closed to hold Freon in the fill line. When I tried that, refrigerant went everywhere. I ended up just burping the fill line each time.
* After the first can, AC Compressor didn't consistently stay engaged. This is probably normal, but I dunno. It'd stay on for 10-20 seconds, then kick off for 20 seconds, kick back on, etc. Maybe cause ambient was cold? It was high 50s F in the garage. It'd only pick up refrigerant when the compressor was engaged, so it took a while.
* When I was finished and with the compressor engaged, the low side ended up being 25psi, high side was 148psi which is in line with all the charts for my coldass ambient temp. I ran it for a while after charging just to make sure those pressures were consistent over time.
* I ended up using 810g of refrigerant. Though, there may be less in the system as a bit came out while burping.
Done!
We'll that's it! I have AC finally! It's cold! Everything seems to work (condenser fans etc)! I’ve had the car up in the air again a couple times, and it seems like no leaks. Hopefully it lasts this time. I'm really glad this project is over.
I also made progress on the oil temp sensor. I’ll post about it soon
Last edited:
On these cars its normal for the AC compressor to cycle on and off like that so no worries there. That said, I have no idea how AC is configured with an AEM unit.
Great to know! Cool, I think the ecu is pretty dumb on the AC output. I think it just takes the request input and outputs as long as it’s not out of range on some param (e.g. TPS). So it’s probably doing whatever the CCU tells it
I'm curious on the AEM instructions for A/C as well. On my HKS, it was highly recommended to me early on to keep the factory ECU in place for certain key base level functions. One of them was AC. For example, the AC still shuts off during aggressive driving. i'm not exactly sure on the parameters in place but the compressor clearly shuts off after a certain RPM or perhaps load as well. Anyone have any info on this? I should do a search here on prime.
As far as the AEM unit, the AEMPro user manual doesn't even describe the AC options at all. They probably do in the later model software, but not this one. I looked into the whole AC pathway a while back when I was originally attempting to diagnose it. The brains are, as far as I can tell, really the CCU and fan control unit:
* Human turns the A/C on via the A/C button on the CCU
* CCU sends signal to the Climate Fan Control Unit (FCU; next to the main relay behind the center console)
* The FCU checks some sensors (pressure switch, coolant temp), then ultimately grounds the A/C request pin on the ECU (pin C3)
* The ECU makes sure the engine is running and not WOT, then grounds the A/C compressor pin (A15)
* ECU pin A15 makes its way to the A/C compressor relay (back, 2nd position from left)
* Relay clicks on, sends power to compressor clutch
* Clutch turns on
*
So I bet the AC request pin is being turned on / off during this compressor on / off cycling. The ECU doesn't even get some sensors, like the pressure switch, or cabin air temp sensor to make legit decisions.
The AEM software has basic options: min RPM, max TPS, then a couple options to tell it which input is request, and which output is the clutch. I will lower the TPS limit, but this is it as far as AC setup:
@MotorMouth93 probably has a better idea what the stock ECU does, but I bet it's not a whole lot smarter than checking a couple params, then relaying the request signal to the clutch output.
I think the OEM ECU A/C integration is limited to adjusting the idle when the compressor clutch engages to maintain a smooth idle.
Last weekend I got the navpod displaying the oil temp from the brand new sensor. Pretty cool to have all this in the navpod now. Oil temp is top right.
Results
I drove the car today really curious how it would behave. I'm not really sure what "up to temp" oil is.
I let the coolant warm up in the driveway, then took it out. After some chill driving for 10 minutes, it was hovering around 185F. I did some "spirited" driving getting into boost several times, spinning it close to redline in 2nd, and it plateaued at about 205F. Then some more chill driving, and it dropped down to the low 190s and hovered there. This makes sense to me: the water-cooled oil cooler would make it pretty closely mirror the coolant temp.
Seems legit? Can anyone share their oil temp experience? What should I be expecting?
Setup details
I hadn't tested the sensor voltage after building out the resistor board for the temp sensor. This felt a pretty dirty. A career of pushing code to production environments for critical services has beat into me that I should always triple check my work; it's easy to miss something small, and I've been bitten by plenty of dumb mistakes. Plugging my little board into the ECU and starting the car was "production" here. The risk was somehow burning out the input or otherwise harming the common sensor 5v or ground. The board for ref:
I _had_ done some resistance testing before installing the pan. Basically I plugged the sensor into my board and tested resistance at each of my 3 pin wires. Grab onto the sensor to heat it up, and test resistance again. It checked out: no shorts, resistance in line with calcs. But I was still paranoid, so I went through a bit of a process to make sure it was ok:
* I noted what the voltage should be at a couple different temps so I'd know what to look for, e.g. 60deg: 4.39v, 80deg: 4.0v, 100deg: 3.6v, 130deg: 2.9v
* I unplugged the temp sensor!
* With the engine off, I hooked up the laptop and turned the key to IGN to get ADC sensor readings
* Then I noted the other ADC sensor voltages like oil pressure, TPS, etc. I'd check them later
* With the key turned to IGN, I measured the voltage at the plug, just to make sure the wire colors were as I expected, and for the 5th time, made sure the colors matched up with those going into my board
* I've learned with the AEM series 1, once the software is open and connected to the ECU, I can turn the ignition off and back on and it will quickly reconnect to the ECU and show values. This was handy, as I could just turn it back to IGN, check the values, if anything was weird, immediately turn the car off.
* Turn the car off
* Plug in the temp sensor
* Turn the key to IGN while staring at the laptop
* When it connected, scan through all the ADC channels: do they look ok? (yes) Is my temp voltage showing about 4.4v? (yes)
* Relief, all good!
* Start the engine
* Then I watched the voltage fall as it heat up and passed the numbers I noted in the voltage -> temp scale above
Everything was ok! I was over-paranoid!
Here are some logs of it warming up. I'm running oil temp into the OG TPS input (pin D11), which, you'll know if you've been following for a while, is noisy AF. Well, it continues to be noisy, but it actually ends up not so bad in this case, fluctuating by only a couple degrees. With a little smoothing in RealDash it'll be fine.
Temp voltage is the green line, pink is oil pressure voltage
Sweet, next step was to get this data into the navpod as real temp in F. It took a bit to get it setup correctly in RealDash. This sort of temp sensor (thermistor) has a non-linear resistance scale. Here is the Rife sensor's scale. Most thermistors that end up in car temp sensors have huge resistance swings at low temps--literally thousands of ohms--then vary only by a few ohms at high temps.
This means the voltage scale is also not linear. These sorts of scales are well suited to tables with interpolation.
To build out a voltage -> temp table from the ohms -> temp table, I needed to know the total pullup resistance. I used a (measured) 2200ohm resistor in my little board, then the ECU has a 100k pullup resistor internally. These will be wired in parallel, and will result in ~2150 ohms total. I used a script I wrote a while ago to build out the table.
This sensor is on a generic ADC channel, so the ECU only sends voltage to RealDash, and RealDash doesn't support tables, only equations. That means I needed to figure out an equation that mimicked the table. High-school math has entered the chat. What I needed was polynomial regression.
I wrote a a quick script that takes the voltage -> temp table, uses a lib to do the regression, then calculates the error between the generated equation and the table so I can see how close it is. I did some experimentation with different algorithms and landed on this:
y = -1.1718x^5 + 15.4508x^4 + -80.2679x^3 + 207.863x^2 + -314.9237x + 400.148
It's most accurate from 100deg F -> 300deg F (< 0.5% error), which is where it will spend most of the time. Then as it gets colder, it's less accurate, but still close enough. Oil temp should never go below 40deg, and only on startup, so it's fine.
It took some fiddling to get this equation properly functioning in RealDash, but after a couple tries, I got it working.
Results
I drove the car today really curious how it would behave. I'm not really sure what "up to temp" oil is.
I let the coolant warm up in the driveway, then took it out. After some chill driving for 10 minutes, it was hovering around 185F. I did some "spirited" driving getting into boost several times, spinning it close to redline in 2nd, and it plateaued at about 205F. Then some more chill driving, and it dropped down to the low 190s and hovered there. This makes sense to me: the water-cooled oil cooler would make it pretty closely mirror the coolant temp.
Seems legit? Can anyone share their oil temp experience? What should I be expecting?
Setup details
I hadn't tested the sensor voltage after building out the resistor board for the temp sensor. This felt a pretty dirty. A career of pushing code to production environments for critical services has beat into me that I should always triple check my work; it's easy to miss something small, and I've been bitten by plenty of dumb mistakes. Plugging my little board into the ECU and starting the car was "production" here. The risk was somehow burning out the input or otherwise harming the common sensor 5v or ground. The board for ref:
I _had_ done some resistance testing before installing the pan. Basically I plugged the sensor into my board and tested resistance at each of my 3 pin wires. Grab onto the sensor to heat it up, and test resistance again. It checked out: no shorts, resistance in line with calcs. But I was still paranoid, so I went through a bit of a process to make sure it was ok:
* I noted what the voltage should be at a couple different temps so I'd know what to look for, e.g. 60deg: 4.39v, 80deg: 4.0v, 100deg: 3.6v, 130deg: 2.9v
* I unplugged the temp sensor!
* With the engine off, I hooked up the laptop and turned the key to IGN to get ADC sensor readings
* Then I noted the other ADC sensor voltages like oil pressure, TPS, etc. I'd check them later
* With the key turned to IGN, I measured the voltage at the plug, just to make sure the wire colors were as I expected, and for the 5th time, made sure the colors matched up with those going into my board
* I've learned with the AEM series 1, once the software is open and connected to the ECU, I can turn the ignition off and back on and it will quickly reconnect to the ECU and show values. This was handy, as I could just turn it back to IGN, check the values, if anything was weird, immediately turn the car off.
* Turn the car off
* Plug in the temp sensor
* Turn the key to IGN while staring at the laptop
* When it connected, scan through all the ADC channels: do they look ok? (yes) Is my temp voltage showing about 4.4v? (yes)
* Relief, all good!
* Start the engine
* Then I watched the voltage fall as it heat up and passed the numbers I noted in the voltage -> temp scale above
Everything was ok! I was over-paranoid!
Here are some logs of it warming up. I'm running oil temp into the OG TPS input (pin D11), which, you'll know if you've been following for a while, is noisy AF. Well, it continues to be noisy, but it actually ends up not so bad in this case, fluctuating by only a couple degrees. With a little smoothing in RealDash it'll be fine.
Temp voltage is the green line, pink is oil pressure voltage
Sweet, next step was to get this data into the navpod as real temp in F. It took a bit to get it setup correctly in RealDash. This sort of temp sensor (thermistor) has a non-linear resistance scale. Here is the Rife sensor's scale. Most thermistors that end up in car temp sensors have huge resistance swings at low temps--literally thousands of ohms--then vary only by a few ohms at high temps.
This means the voltage scale is also not linear. These sorts of scales are well suited to tables with interpolation.
To build out a voltage -> temp table from the ohms -> temp table, I needed to know the total pullup resistance. I used a (measured) 2200ohm resistor in my little board, then the ECU has a 100k pullup resistor internally. These will be wired in parallel, and will result in ~2150 ohms total. I used a script I wrote a while ago to build out the table.
This sensor is on a generic ADC channel, so the ECU only sends voltage to RealDash, and RealDash doesn't support tables, only equations. That means I needed to figure out an equation that mimicked the table. High-school math has entered the chat. What I needed was polynomial regression.
I wrote a a quick script that takes the voltage -> temp table, uses a lib to do the regression, then calculates the error between the generated equation and the table so I can see how close it is. I did some experimentation with different algorithms and landed on this:
y = -1.1718x^5 + 15.4508x^4 + -80.2679x^3 + 207.863x^2 + -314.9237x + 400.148
It's most accurate from 100deg F -> 300deg F (< 0.5% error), which is where it will spend most of the time. Then as it gets colder, it's less accurate, but still close enough. Oil temp should never go below 40deg, and only on startup, so it's fine.
It took some fiddling to get this equation properly functioning in RealDash, but after a couple tries, I got it working.
The weather was super nice this weekend and I spent most of Sunday with the car. I’ve been really happy with it lately. I’ve managed to fix everything annoying on the car. Still a lot to do, but no major frustrations.
I got a new camera (A7Riii) back in December but I hadn’t used it to shoot anything on the car yet—maybe you’ve noticed the steady stream of crappy pics on this thread. I brought the camera with me and got a few good shots.
This thing cleans up pretty good
I got a new camera (A7Riii) back in December but I hadn’t used it to shoot anything on the car yet—maybe you’ve noticed the steady stream of crappy pics on this thread. I brought the camera with me and got a few good shots.
This thing cleans up pretty good
Such a nice NSX. Cleans up really well! I'm a big fan of the BBSC (in terms of design- not engine management LOL) and it's so nice to see one being properly cared for and preserved.
Thanks! Its tough to really capture this car (any nsx) in pics for some reason. It looks so much more wild to me in person. Like the extremely low roof height doesn’t come across or something. Or the way the tires poke out around the bumpers just doesn’t make it through. I dunno what it is
I really need to remove the sticker on the taillight. I totally forget about it until I take pics of the car. It’s actually an old whipple CTSC, but from what I understand the basch dude installed it in the early 2000s, thus the sticker. Technically it is “supercharged by basch” I guess, but not with his thing
I really need to remove the sticker on the taillight. I totally forget about it until I take pics of the car. It’s actually an old whipple CTSC, but from what I understand the basch dude installed it in the early 2000s, thus the sticker. Technically it is “supercharged by basch” I guess, but not with his thing
Ah, I see. Most likely someone swapped the BBSC for the CTSC, which was a much more stable system back in the day (the BBSC was grenading engines left and right back then and folks were MAD). I'm fairly sure Basch only slapped that sticker on there if you bought his supercharger.
I agree the NSX is tough to photograph properly. I agree that it's the low height- I feel like most people have too high of an angle, which makes the car look more like a bubble. If you shoot it low, it really highlights the low stance, wedge shape and lines, which are the car's hallmark.
Alternatively the NSX sometimes looks funny next to modern cars for that same reason
A lot of things I decide to do to this car are pretty unnecessary. Does it need a navpod screen? No. A blingy rear strut bar? Definitely not. An oil temp sensor? I mean, not really. If I’m being honest, the whole thing kind of embodies unnecessary; it’s an art project I guess, a guilty pleasure all over.
Within the car sphere of impractical things I don’t really need, there exists an even more unnecessary, guiltier pleasure: wheel shopping. If I had the space and was willing to spend the relationship capital, I’d probably have 20 sets of wheels for this thing.
Well I bought another set of LMGT4s:
When I pieced together the first set, I had to buy a whole set of 4 fronts. I tried to sell the other pair of fronts, but fielded mostly lowball offers. For the $600 or $700 I was consistently offered, I might as well keep them, right? The 2nd pair were actually the nicer of the set.
They are still OG bronze and have minimal rash, so I feel a little bad about powdercoating them. I guess I could look for a pair of OG bronze rears, but when they come up they are easily double the price of the more common silver Z car set I ended up with, and are usually only for sale in Japan. (I mean if you have a lead on an OG bronze pair, I’m all ears)
I’ll probably end up powder coating them. I’m still deciding on color, but I’m heavily leaning toward champ white. Number 2 option rn is mag blue.
If you have color opinions, let me know! Or photoshop skills so I can visualize, that would be amazing
Within the car sphere of impractical things I don’t really need, there exists an even more unnecessary, guiltier pleasure: wheel shopping. If I had the space and was willing to spend the relationship capital, I’d probably have 20 sets of wheels for this thing.
Well I bought another set of LMGT4s:
When I pieced together the first set, I had to buy a whole set of 4 fronts. I tried to sell the other pair of fronts, but fielded mostly lowball offers. For the $600 or $700 I was consistently offered, I might as well keep them, right? The 2nd pair were actually the nicer of the set.
They are still OG bronze and have minimal rash, so I feel a little bad about powdercoating them. I guess I could look for a pair of OG bronze rears, but when they come up they are easily double the price of the more common silver Z car set I ended up with, and are usually only for sale in Japan. (I mean if you have a lead on an OG bronze pair, I’m all ears)
I’ll probably end up powder coating them. I’m still deciding on color, but I’m heavily leaning toward champ white. Number 2 option rn is mag blue.
If you have color opinions, let me know! Or photoshop skills so I can visualize, that would be amazing
- Joined
- 22 November 2014
- Messages
- 78
I really like the way these wheels look. Where are you finding most of them? And are you on stock fenders all the way around?
- Joined
- 10 September 2021
- Messages
- 156
Nice car and those wheels look great! I have major wheel whoring issues with any car I own. Currently sitting on two sets of regamasters, TE37s and advan 5s. If you’re on the hunt, yahoo auction Japan is a great site for used wheels with multiple places who will proxy bid and deliver to US. I don’t recommend it if you’re like me and love wheels. Lol. It gets me every time. I even saw a set of comp tech headers for sale the other day.
Thanks! The front set I bought from a post on Instagram, the 1st rear set I got from eBay, this latest rear set was from Facebook marketplace. Yep, stock fenders. These specs have pretty standard front spacing for there nsx with no spacers. They feel like they were made for this car. I wish I could run the LMGT4 concave fronts, but I’d need 20mm wider front fenders.
Thanks! That’s a great collection! I feel you, I have a few search tabs open to yahoo auctions I refresh every day lol. I went through the bidding process on one set of OG bronze CEs a little while ago with Jessie Streeter, who has great reviews as a wheel proxy. But they went for waaaay more than I was willing to spend
- Joined
- 22 November 2014
- Messages
- 78
What are the specs on the current set? And who do yall recommend for the best place to proxy bid and ship from those auctions?
The current set is 17x7.5+30 and 18x9.5+30. While I didn’t win the auction, I did pay and was refunded by Jesse streeter (https://jessestreeter.com/contact-jesse/), he was super communicative and friendly. He’s not a “sourcer” tho. You send him the listing to the thing you want to buy, pay him, then he buys it and ships it to you. There are a number of people who will source wheels for you. Give them a deposit and they look out for the spec you want. I don’t have any experience with those folks though.
Real life has been super busy lately, so I haven’t had any big blocks of time to work on the car. But I have had an hour here and there to knock out tiny projects. Some of it is without pics. For example a bunch of alarm related things: new fobs, turning down the sensitivity of the warning beep so it doesn’t randomly go off like it has separation anxiety, etc...
I did manage to take a few pics from a couple projects
Hatch trim
The hatch trim/seal was pretty faded, surprisingly really the only faded trim on the car. I did buy brand new hatch seals in a recent amayama order, but those probably take time to install. I figured I’d try to revive the OG seals with some solution finish
Before, v ashy
After
It looks a lot better as a whole
Harness bar bolt
There’s been an annoying metallic rattle on the passenger side since I’ve had the car. I traced it down to the washer on the harness bar mounting bolt: basically the bolt was not snug against the harness bar. This is an after pic, but you get the idea: washer was loose, sketchy. Seemed like a pretty simple project, but turned into an ordeal.
So like, just tighten it up, right?!?!? Lol, that would be too easy. The bolt was totally frozen in the bracket. It was a huge struggle to get it out. Wd40, the impact gun + a crescent wrench on the bracket, a 1/2 drive ratchet with a cheater, impact gun again and at least a whole ass hour. You can see I marred up the bracket with the crescent wrench. I was contemplating using a cutoff wheel and thinking maybe I wouldn’t reuse the bracket at all
When it came out, some of the threads were all boogered up near the end, but not all of them
There were still a lot of good threads left in the bracket, so I decided to reuse it. I just needed a tap to clean it up and a new bolt. Cool, what is the thread size? It measured out to be either M12x1.25 or 1/2”-20. Turns out they have the exact same thread pitch (25.4/20 = 1.27mm). Diameter wise, they are very close as well, and I was measuring from a jacked up bolt, so I wasn’t confident I either size. Also the M10 bolt that went into the pillar on the other side of the bracket used the same size hex driver: 3/8”, which is exactly 9.5mm…. Extra confusing, right? On top of that, no one local had anything in these sizes so I couldn’t test. Ugh.
I bought the M12 tap thinking it had to be metric. Everything on the car is metric. The tap came, but it felt a little loose in the bracket. Not it. So I got the 1/2-20 tap and a pack of bolts. 1/2-20 turned out to be correct
All done! Now with anti seize! This dumb little thing had the car down for 2 weeks.
A little bonus, the trim here has been loose forever.
Turns out it was too deep in the floor garnish, yay for my OCD
JP license plate tho
I know it’s not necessary to hide your license plate on the internet, but it’s ugly. When I’m hard parked or want to take booty photos, I feel like it takes away from the car. So I had a Japanese plate made for the rear.
They didn’t have tochigi prefecture as an option, so this one is Suzuka because race car. The vehicle class number (33) is correct for the period and engine size, but also like verstappen, you know? Then 91 for vehicular birth year. I was inspired by nsx_44 on ig, he has the double digit number only (44!). His is real tho.
I made some hasty brackets so I could just lay it over the Murican plate. Easily removable for a quick getaway
Here it is in its habitat. This JP plate is like 1/2” larger than the USA plates in all dimensions, so it nicely covers. You may also notice the lack of supercharged by basch sticker. It needed to go…
Cars n coffee
I took the car to its maiden cars and coffee in Santa Rosa a few weeks ago. I ended up among a bunch of muscle cars. Was a good variety, and the muscle car dudes were all very into the car. Several different people came up to me and assumed I/the car was from Marin. If you’re from the Bay Area, that carries some specific connotations. Lol, I’ll take it to mean the car was pretty clean.
I did manage to take a few pics from a couple projects
Hatch trim
The hatch trim/seal was pretty faded, surprisingly really the only faded trim on the car. I did buy brand new hatch seals in a recent amayama order, but those probably take time to install. I figured I’d try to revive the OG seals with some solution finish
Before, v ashy
After
It looks a lot better as a whole
Harness bar bolt
There’s been an annoying metallic rattle on the passenger side since I’ve had the car. I traced it down to the washer on the harness bar mounting bolt: basically the bolt was not snug against the harness bar. This is an after pic, but you get the idea: washer was loose, sketchy. Seemed like a pretty simple project, but turned into an ordeal.
So like, just tighten it up, right?!?!? Lol, that would be too easy. The bolt was totally frozen in the bracket. It was a huge struggle to get it out. Wd40, the impact gun + a crescent wrench on the bracket, a 1/2 drive ratchet with a cheater, impact gun again and at least a whole ass hour. You can see I marred up the bracket with the crescent wrench. I was contemplating using a cutoff wheel and thinking maybe I wouldn’t reuse the bracket at all
When it came out, some of the threads were all boogered up near the end, but not all of them
There were still a lot of good threads left in the bracket, so I decided to reuse it. I just needed a tap to clean it up and a new bolt. Cool, what is the thread size? It measured out to be either M12x1.25 or 1/2”-20. Turns out they have the exact same thread pitch (25.4/20 = 1.27mm). Diameter wise, they are very close as well, and I was measuring from a jacked up bolt, so I wasn’t confident I either size. Also the M10 bolt that went into the pillar on the other side of the bracket used the same size hex driver: 3/8”, which is exactly 9.5mm…. Extra confusing, right? On top of that, no one local had anything in these sizes so I couldn’t test. Ugh.
I bought the M12 tap thinking it had to be metric. Everything on the car is metric. The tap came, but it felt a little loose in the bracket. Not it. So I got the 1/2-20 tap and a pack of bolts. 1/2-20 turned out to be correct
All done! Now with anti seize! This dumb little thing had the car down for 2 weeks.
A little bonus, the trim here has been loose forever.
Turns out it was too deep in the floor garnish, yay for my OCD
JP license plate tho
I know it’s not necessary to hide your license plate on the internet, but it’s ugly. When I’m hard parked or want to take booty photos, I feel like it takes away from the car. So I had a Japanese plate made for the rear.
They didn’t have tochigi prefecture as an option, so this one is Suzuka because race car. The vehicle class number (33) is correct for the period and engine size, but also like verstappen, you know? Then 91 for vehicular birth year. I was inspired by nsx_44 on ig, he has the double digit number only (44!). His is real tho.
I made some hasty brackets so I could just lay it over the Murican plate. Easily removable for a quick getaway
Here it is in its habitat. This JP plate is like 1/2” larger than the USA plates in all dimensions, so it nicely covers. You may also notice the lack of supercharged by basch sticker. It needed to go…
Cars n coffee
I took the car to its maiden cars and coffee in Santa Rosa a few weeks ago. I ended up among a bunch of muscle cars. Was a good variety, and the muscle car dudes were all very into the car. Several different people came up to me and assumed I/the car was from Marin. If you’re from the Bay Area, that carries some specific connotations. Lol, I’ll take it to mean the car was pretty clean.
Last edited:
I have another set of LMGT4s that need to be powdercoated. I've been tinkering with photoshop to get an idea of what color I might want. I have a couple favorites, curious what y'all think.
The current set is bronze, which I do love. But I want something different to go along with the bronze set for fun. The original pic:
Theee options. The bronze set is matte, so pretty much all these options also look matte. Colors with way different brightness (white, black) were pretty hard.
Formula silver
Titanium gunmetal
Matte blue gunmetal
Mag blue
(mag blue in real life is super shiny)
Gold
Hyper red
White
These ones are at the end cause I really don't think they do the colors justice. After a lot of tweaking, this was the best I could do. Have photoshop skillz and want to help? I'd love to send you the PSD.
Flat black
The current set is bronze, which I do love. But I want something different to go along with the bronze set for fun. The original pic:
Theee options. The bronze set is matte, so pretty much all these options also look matte. Colors with way different brightness (white, black) were pretty hard.
Formula silver
Titanium gunmetal
Matte blue gunmetal
Mag blue
(mag blue in real life is super shiny)
Gold
Hyper red
White
These ones are at the end cause I really don't think they do the colors justice. After a lot of tweaking, this was the best I could do. Have photoshop skillz and want to help? I'd love to send you the PSD.
Flat black
Last edited:
I vote Formula Silver. Looks badass.
Yeah I was surprised by how much I liked it. I usually feel like 350zs with silver LMGT4s look pretty meh. But there’s something about the depth of formula silver that is appealing. I tried to replicate it with extra contrast and a tiny blue hue
My vote is Mag blue
