Trying to Survive - '92 Suzuki Jimny Ownership Log

[Xulfiqar]

Was the system properly vacuumed before?
What is the charge state?

[pwuser162557941033]

Probably not, and no idea about the charge state. That's why I'm tempted to get couplers and a gauge set and investigate myself. Instead of paying Rs. 500 for a chuski and a low-side only gauge reading, especially from a kid that inadvertently vents a ton of refrigerant to atmosphere in the attachment process.

At 1000rpm idle, 22°C vent temp. It slowly rises towards 24°C. Can hear TXV spray.

At 3000rpm, 17°C vent temp. System is functioning more as an inefficient dehumidifier than an AC.

Driving brings it closer to 15°C or 14°C.

[Xulfiqar]

You are probably low on charge and also adulterated..

Fix it up yourself

[pwuser162557941033]

Yeah makes sense. I can't recover anything from the system, I don't think - otherwise I'd weigh it and replace the correct quantity. But I can check superheat with thermocouples on both sides. I don't particularly want to vent to atmosphere, but what else can I do? Buy the equivalent of an AC Pro can and put some more charge in?

i.e. R134a small can, R134a couplers, manifold gauge set with lines. And check pressures then see what to do?

Nissens' R134a HVAC PDF is a great resource.
Nissens_AC_System_Diagnostics_Poster_98x68 - Operating Pressures R134A - ENG.pdf (873.9 KB)

[pwuser162557941033]

Asked the boys at M/S Autos for a manifold gauge set. Readings at hot idle, 1000rpm, approx 35°C ambient.

Condition:
LP 65 psi (too high) , HP 230 psi (normal).

Cause:
Seems like scenario 1.

Correction:

• Check expansion valve ("it's spraying, so it can't be blocked" - shop owner)
• Make sure system is clean/properly flushed.
• Ensure correct refrigerant charge (too low possible)

---

The owner has a Robinair vacuum pump that he has never used, and was willing to try using it. We immediately pulled 10 inHg (25 microns) and after 15 minutes we have approx. 12.5 inhg (31 microns) of vacuum.

---

Pulled vacuum for 36 minutes, then closed the lines off and set a timer to see if the vacuum holds.

After 11 minutes, the needle has moved, indicating a leak somewhere.

17 minutes.

There's clearly a leak, not sure if it's in the gauge lines/fittings or the system. I've started to pull vacuum again. Will charge up immediately once the refrigerant arrives.

The service manual I have is for the R-12 version of this vehicle - they specify 600g of refrigerant.

---

After the first vacuum wouldn't hold, I decided to do another. Ended up taking 45 minutes until we got the refrigerant cans, 340g and 320g. Held the lowest possible vacuum, purged the line of air, and started charging the system.

I ended up charging until the static pressure reading on the low side showed 37°C, which was the ambient temperature. At that point, I'd used 550g of refrigerant.

The vent temperature is lower than before, and the air feels cooler, but not by much, and not at idle. I still see the low side too high and the high side normal.

The owner's proposed solution is:

• New drier
• New high side line to condenser
• New condenser
• Used Toyota Duet compressor, which is apparently a Denso scroll-type unit.
• Kharadia-made pulley, to fit the Jimny's 4PK belt.
• Welding-guy-made bracket, to mount and secure the compressor.
• ND-8 oil, don't know the exact amount.
• R134a refrigerant, don't know the exact amount.

His "flush" consists of filling the system with R134a and purging it once, then vacuuming. I'd insist on holding vacuum for 30 minutes to ensure there are no leaks. Then, charging the system. He expects a vent temperature of 9°C, up to 12°C in the day time.

---

Is it finally Duet compressor time?

[Xulfiqar]

Leak check is done under pressure, not vacuum.

[Xulfiqar]

If your r12 system called for 600g and you are charging r134a as retrofit you need to keep it at 80% of total charge otherwise you will liquid flood.

Is your expansion valve of an r12 system? Was it installed and insulated as it should?

Is the evap sweating?

[pwuser162557941033]

Yes that's exactly what I did. grams to pounds, times 0.9, subtract 0.25 lbs. However, it's an R134a compressor. It's just the service manual that I have that's for R12. My vehicle's system, I suspect, was always R134a.

Will look at the evaporator and TXV to confirm the position of the sensing bulb and the insulation you're talking about.

Yes, the evaporator is sweating. AC is nice and cool (not cold), even triggers the mechanical thermostat on the evap once or twice, but only when cruising. At 1000rpm it doesn't do much of anything. The compressor is very noisy, too.

Out of interest, why isn't a vacuum hold check good enough? That's also what the service manual specifies, i.e. see if it holds the vacuum for 15 minutes to check for leaks. I don't think I could find nitrogen to use to check the system under pressure.

[Xulfiqar]

The system runs under pressure, the liquid side discharge can reach upto 350 psi. The expanding stress on joints and pipes can expose microleaks which would be sealed shut under vacuum.

The repair manual is mute at best for ac repair on most japanese cars, the most comprehensive ones are found in american brands because instead of developing an ac system they would literally buy a refrigeration development firm and use that, like ford bought philco, GM did Harrison, chrysler developed their own with help of both.

Those old literature tell you in fine detail of how to test for leaks and also teach you best practices etc, while the famous 2 were fixed ccot systems, chrysler would literally have the txv sensing tail outside the evap box making sure that the bulb is sensing all boiled off refrigerant.

[pwuser162557941033]

Electronic Boost Control.

With a PWM driven solenoid attached normally open between the manifold boost source and the wastegate actuator diaphragm, the results are dramatically better than the counterfeit ball-and-spring cheapo type "boost controller" I have been using all this time. Have been fine tuning freq and duty cycle to dial in the right "scaling" of the stock wastegate spring curve.

Boost builds linearly, doesn't spike at all (a big problem with the old controller) and holds better than ever before. I can start spooling at 1600rpm and be at 50 kPa of boost well before I even get to 3000rpm. More responsive in traffic, harder pulls in the torque and power bands, much more fun to drive. Peak boost is held at 100 kPa at full throttle, rather than sagging like on the old controller.

Will work on a PID control loop to set boost more precisely based on manifold pressure. Maybe even gearing, lambda or RPM later on.

Monitoring temperatures and pressures throughout, and all seems normal. The oil sump temperature is heating up faster to my eye, but the gauge stays below 100°C even in the midday sun.

[pwuser162557941033]

Clutch

A few days ago, I heard a "plink" coming to a stop at a traffic light and lost the ability to change gear completely. The clutch was locked solid against the flywheel, it seemed, while the car was in neutral. I pushed the car to the side of the road, decided to mess with the linkage, and it started working. Drove it home immediately.

The next day, it was behaving perfectly normally and naturally, stumping all mechanics. I knew it was the clutch. Decided to wait until I saved some money, during which it was very hard to shift between 3rd and 4th, like the clutch wasn't disengaging fully. Mechanics thought it was the cable. New cable, etc.

Decided to head to a competent mechanic and convinced him to take it apart, and that it was my clutch. Here's what we found, courtesy Alpha Motor Tech, Korangi, Karachi - the last people to lay hands on the transmission.

Alright, great. What's on the inside?

The plink was the spring, the bolt was nowhere to be found. This clutch lasted almost 12,000km before being completely torn apart. It's also the wrong pressure plate and the wrong disc - the disc is for an Alto VXR (F10A) and the pressure plate for a Chamber (F8B). Both have been modified to fit my flywheel. Excellent. Great, reputable workshops violating your trust - what's new?

Go to a clutch dealer, he seems to have the part. Oh, nope he doesn't. Has to yell at a customer first. Yes, alright. Yeah we have it. No we don't need the part number, we've been doing it for 30 years. Here's the bill.

Take the part to a nearby machinist. Using the right language, seems educated. Says it's the wrong clutch. Indeed, it's the wrong clutch. They don't have the right model. Machinist thinks for a bit, suggests I get the Aisin F10A pressure plate and a Valeo F8B disc, stating that he's noted the Valeo to be marginally thicker than the Aisin counterpart and he wants the best possible fit.

Got the parts, along with new bearings. He drilled indents into the flywheel and balanced it. Seemed happy with the job.

Got the car put together and went for a drive. Clutch bites. Car is actually moving and the shifts feel smooth - although with the fresh Shell Spirax not as smooth as they did with the AMSOIL GL-4 I had in there (which, I'm claiming, is the reason.) Great feel, no noise with the clutch disengaged, much less noisy in general.

Coupled with the results from the electronic boost control, I've never noticed such a dramatic change in my time owning it.

[pwuser162557941033]

I don't know why or how - possibly due to the awkward mounting and the vibration - but the entire armature for the solenoid ejected itself from the housing. Pictured next to a similar valve here.

Fortunately, because it's NO, failure reverts to the worn factory spring that's 50-70 kilopascals. The car doesn't even get to 100km/h.

Ordered some new solenoids.

[pwuser162557941033]

Wrote a PID controller and tested it while waiting for the solenoids and MOSFETs to arrive. Reads the 5V MAP sensor and outputs a PWM signal to modulate the solenoid.

Very simply:

• If the MAP sensor reads vacuum, the solenoid is de-energized.
  Wastegate sees manifold vacuum. Prolongs life of solenoid. PID stopped to prevent integral windup.

• If 1 kPa of boost is read, the solenoid is energized at 100%.
  Wastegate is blind for fastest spoolup.

• If <= 20 kPa away from the target boost value, PID control starts to run and adjust duty cycle.
  Wastegate may begin to open partially to sustain set boost value.

• If > 5 kPa over the intended boost setting, the solenoid is de-energized.
  Wastegate sees manifold pressure.
  PID still running, will re-engage once pressure drops to or below setpoint.

So far, examining the graphs show a reasonably responsive controller.

[pwuser162557941033]

Update on the Electronic Boost Controller

No matter what implementation of PID calculations I tried, or what gain constants I set, the wastegate responded way too quickly to duty cycle manipulations and it was impossible to curb the oscillations with any values for constants.. Besides that, the MAP sensor voltage needed to be filtered because of the fast response times of the ADC on the micro-controller I used for this project.

Today, after 100 kilometers alternating between hard full throttle runs and idling in the midday sun, during which the car was very well-behaved with no overheating, no funny business - only an anemic, struggling air conditioning compressor at idle - I decided to implement a simple bleed valve configuration for more predictable results and to "scale" the stock wastegate response curve.

The end result?

No PID, just simple logic statements that de-energize the solenoid at vacuum, and energize it with a fixed duty cycle percentage when boost is detected. Dramatic difference, no more oscillations, engine behaving as it should, massive low-end (2000rpm lol) torque with just under 100 kPa / 14.5 PSI of boost.

I neither have the fueling nor the tooling to go beyond that number, and it's already only achievable by defeating the stock overboost fuel cut built into the ECU with a zener diode voltage clamp.

I think I'm satisfied. The car feels like a different animal between 2500-4500 rpm with essentially double the factory boost, and a much earlier spool-up time.

Still a small animal, but different nonetheless.

To help make the welcome a little better, a splash screen.

Another task coming up: read a thermocouple and don't raise boost unless the oil/coolant has fully warmed up. Stock threshold until warm, after which it gradually increases, only allowing full boost at or above 82°C operating temperature.

[Xulfiqar]

You have gone to bmw level boost management on your little suzuki

[pwuser162557941033]

I'm very fortunate that the associated components are cheap, plentiful, and easy to put together, so I thought I'd give this senior Suzuki a little help with modern prostheses, if you will.

The MCU I'm using is 8-bit, just like the Jimny's original ECU, but costs less than $3.

With the car pricing, service, and parts climate the way it is, I really want to be able to count on this thing to drive on a daily basis. The radiator fan coming on independently to prevent heat soak and expedite thermal siphoning in the turbine, the DIY boost controller and its solenoid being NO by default, the all-LED lighting to reduce alternator load and help charge the SLI battery faster, all of these are minor QoL improvements I'm hoping help the car last longer and be more reliable.

So far, all well. Fingers crossed it doesn't blow a head gasket again, lose another turbo, or crack a manifold.

[pwuser162557941033]

Cardboard box prototype. Trying to tune it for fastest response while also maintaining linearity with the throttle. The response was a little too twitchy for my tastes with any 100% duty cycle threshold. It's hard to get the wastegate to "snap" open at a peak value, because physics - the boost usually runs away beyond 100kPa on the Chinese gauge.

I'm peaking at 87kPa according to the 5V Mitsubishi MAP sensor I'm using to measure. It shows 2.63V at atmospheric pressure in Karachi, which is what I'm using to calibrate the scale.

However, my Chinese boost gauge is constantly sounding the alarm at 100kPa like I've asked it to, peaking at 105kPa. That's quite high, and the ECU definitely doesn't have fueling for it - I might be in danger of running lean and having Mr. Rod come to say hello.

Granted, the boost gauge's sender is the very end of a vacuum tee tree coming from the manifold post-throttle-body, the order being the car's own MAP, the boost controller MAP, and the boost gauge sender; this might be contributing to the discrepancy.

One way to find out is to remove the voltage clamp on the ECU MAP sensor and see whether it's cutting fueling at precisely 0.9 kg/m² like the service manual claims.

[pwuser162557941033]

Relationship Between Injected Fuel Amount and Driving Conditions

From the Suzuki Jimny 660 JA11 Service Manual

A great looking graph from the service manual that I translated. Highlights the importance of having a functioning coolant temperature sensor with regard to overall fuel economy, and also all the other sensors. The ECU isn't as primitive as I had initially thought.

Some behaviors the manual specifies are below.

---

Synchronous Ignition Characteristics

Cranking

• When cranking, the injectors are fired 3x every 2x crank revolutions.
• Normally, it is 1x every 2x crank revolutions.

Startup

• When the engine is started, it corrects the injector pulse width based primarily on the coolant temperature, regardless of the intake pressure. Intake air temperature and battery voltage is factored into the correction.
• There is a drastic increase in pulse width below 0 degrees C.

Time Elapsed Since Startup

• Immediately after startup, the injection amount returns to the base amount, but post-startup increase correction occurs for based on time elapsed since start startup.
• The time increases or decreases according to the coolant temperature.

Coolant Temperature

• When the engine is cold, the injected amount is increased based on the signal from the coolant temperature sensor.
• The aim is to improve driveability.
• After warmup, basic injection time resumes.

Intake Air Temperature

• As air density changes when intake air temperature changes, there will be a slight difference in the measured air intake amount calculated using the intake pressure. Therefore, a correction is made based on the temperature.

Acceleration

• When accelerating (throttle sensor IDLE contact changes from ON to OFF, or when the amount of change in the intake pressure is large) the injected amount is increased based on the coolant temperature.
• However, the amount increases based on the magnitude of the change in intake pressure.

Idle Stability

• The amount of fuel injected is increased or decreased according to the amount of change in engine speed to stabilize the idling speed.

Battery Voltage

• When the power supply voltage decreases, the mechanical operation of the injector is delayed, resulting in a decrease in the amount of injection.
• To compensate for this, the injection time is lengthened when the voltage is low.
• It is highest at <= 11 volts, and lowest at 14.7 volts based on what I can see from the graph.

CO Adjustment

• The CO adjustment is a fine adjustment made during idle conditions to the basic injection time.
• An adjustment resistor is attached to the CO adjustment coupler on the left side of the instrument panel.
• There are 6 types of adjustment registers, allowing 7 levels of adjustment. (Later JA11 ECUs had more levels available.)

Atmospheric Pressure Correction

• Because atmospheric pressures decreases at high altitudes, fuel injection methods that determine fuel injection time based on pressure values from pressure sensors cannot calculate the appropriate injection time.
• Therefore, changes in atmospheric pressure are detected, the fuel injection time is corrected, and the fuel injection amount is adjusted to an appropriate value that matches the driving conditions. (Does this mean the ECU has a barometer or barometric altimeter inside it that performs real-time compensation? Very interesting.)

Recovering from Fuel Cut

• When returning to normal fuel injection after a fuel cut, if the injection amount requested by the engine is injected as is, the air-fuel ratio will become excessively rich.
• Therefore, immediately after the return, the injection amount is reduced and a correction is made to gradually return to the normal injection amount.

High Load

• The amount injected is increased during sudden acceleration.
• High load at high altitudes increase the amount injected when the throttle is fully open in areas where no increase in amount is required.

EGR VSV

• Reduces fuel injection amount during EGR VSV control.

Asynchronous Injection Characteristics

Apart from synchronous injection (normal injection), injection occurs immediately when the signals from each sensor meet the following conditions.

Acceleration
Asynchronous injection is performed when the idle switch contact changes from ON to OFF, and at locations where there is a large change in intake pressure thereafter.

Deceleration
Injects if the engine rotation deceleration is large when restarting injection after fuel cut.
(I have personally experienced it do this.)

Other Controls

Fuel Cut During Deceleration

• When the engine speed is high and the throttle valve is fully closed (during deceleration), fuel injection is stopped to prevent unnecessary combustion.
• When the engine speed falls below the threshold speed, fuel injection resumes.
• Note that the specified rotation speed for fuel cut varies depending on the temperature of the cooling water.

Fuel Cut at High Engine Speed

• To prevent engine overrun, the fuel is cut when RPM reaches 7400. It is resumed at 7350 RPM.

Fuel Cut at High Boost Pressure

• When boost pressure reaches 0.9 kg/m2, fuel injection in all cylinders is stopped. It is resumed at 0.8 kg/m2.

---

I highly doubt, @Xulfiqar, that the OX terminal on the ECU is for an oxygen sensor. I think we can lay that whole thing to rest, and conclude that that's just a thermal fuse on the cat.

[Xulfiqar]

So it seems that suzuki is using the OX terminal as a “telltale” of extra hot exhaust and will force the ecu to trim back everything. The early CA18 nissans did that.

Ox terminal was “party is over” trigger.

And yes ECT sensor is critical in not just this case but literally every efi unit in any car, even back in the early 70s

[Xulfiqar]

Btw

Your username is one of my favorites of all time, Lancia Delta integrale, it was the OG rally champ of middle earth.

Absolute batshit stupid power (too much coffee and lots of italian cussing went into its design)