I have always considered my water pump pulley (made from soft aluminum) to be an annual replacement item. The pulley is driven by a tiny contact patch on the back of the belt and the striations quickly wear out causing the pulley to slip at higher RPMs. It is a good bet that your pulley is worn smooth and your belt slips with higher RPM.
One of the "remedies" is to really crank on the belt adjustment. A really bad remedy. 75% of the motors coming in for rebuild show copper/wear on the front main bearing at 11 o'clock as a result of the front crank snout being levered upward by an over-tightened belt.
Here's the well engineered solution from Full Function Engineering...
As you know, the 1993 RX7 came with one knock sensor. When combined with an Apexi Power FC/Datalogit the knock system worked well. If your setup was properly fixtured and tuned the sensor issued readings around 40 while driving around town and under boost the readings normally drop to around 10 or less. Any departure from those numbers and you'd best be on high alert.
Knock sensors are microphones with a difference. They are designed to pick up only a narrow frequency of sound waves. Each engine block resonates at a particular frequency. Hit the rotary block with a hammer and if you have the right equipment you will register sound at approximately 3500 cycles per second/Hz.
Around the turn of the century Bosch came out with a new improved knock sensor. This sensor was adapted for use (tuned) for the RX8 and works well for the 13B-REW resonance.
Mazda part number: N3H118921. There are lots of knock offs and I recommend you buy the real thing from Mazda.
I adapted one per rotor in 2012 and find they work perfectly.
It is essential that you employ an ECU that has a good knock system as knock will help you towards a proper timing map.
If you don’t have excellent knock data you are tuning blind. While not having knock data might work on a low stressed LS motor, it is fatal when dealing with the enormous CCP and CCH of the rotary.
One of the important features of a proper knock system is “windowing.” An important objective when monitoring knock is the elimination of engine block noise NOT caused by detonation.
Detonation/knock only occurs in a narrow (approx 30-40 crankshaft degrees) window of crank rotation. A good system has the capability of not listening during the other 320/330 degrees.
Another important feature of a good system is allowing the system to not report at low throttle position. Every now and then when just cruising at light throttle I will see a knock recorded. Given the light load I doubt if it is detonation. By setting a throttle position qualifier my system will not react and cut timing.
A proper active knock system will deliver two essential items:
cut timing to zero on the next rotor face if encountering knock
provide the knock data necessary to construct a proper timing map
Timing reduction can be the only response to knock as the detonation needs to be eliminated on the next rotor face. Rotor faces could be spinning by the spark plug at 40 to 60 per second. Obviously cutting boost or fuel, both of which take seconds, won't get it done. Timing is electric which moves at the speed of light.
Simply put, if you don't have a 21st century ECU that includes an active windowed qualified knock system with a single turbo setup or if you do and don't have it set to take action you will be calling your engine builder.
Apex seals have improved over the years so generally you may not break an apex seal if you encounter knock. Newer design (more malleable) apex seals such as Goopy, E & J, Rotary Engine, I-Seals or RXParts generally won't break. O E apex seals, properly designed (of course) by Mazda for 215 rwhp or Atkins (too brittle) generally will break. Broken apex seals generally take out a rotor, rotor housing and the turbine wheel.
This leaves corner seal springs as the weakest link in the chain. Flattened springs won’t wreck your motor but you can expect to lose 30% of your compression. Your engine will have to come apart.
Check out these corner seal springs that came out of one of my customer motors running at about 400 RWHP. While my customer had installed an AI system he hadn’t quite gotten it up and running… oops. Corner seal springs should be .18 inches free height..
Here's the payoff... 585 hp and ECT at 67 C! (fourth from the bottom on the right column..
INSTRUMENTATION MAKES UP FOR GENIUS
When I first received my GT3 spec RX3SP from Roger Mandeville the tuning advice was… jet the Weber dual downdraft carb so you reach 1750 F at the end of the longest straight.
We have come a long way from that. Thank you Intel and the digital revolution. Nowadays there isn’t one single engine metric that we can’t log and modify.
Knowing what is going on is the first step.
The Ignition Map is the backbone of tuning. And it isn’t as simple or as constant as you might think.
The knock sensor/system is the key.
You can see the MAP sensor on my firewall above the throttle body as well as the fuel pressure sensor. I use the same absolute pressure sensor for my oil pressure and exhaust manifold backpressure. You
also probably have spotted the GM 3 BAR sensor which is part of my AI system.
Recently i made two additions to the coolant system, a Mishimoto MMTS-RX7-89 lower temp thermostat and a Full Function Engineering Idler Pulley. With no other mods my ECT is now 66-67 C/ 154 F. i didn't believe it. After a very careful look see i believe it. It is amazing to open my hood and not be met with a blast of hot air. I credit most of the reduction from the thermostat. Maybe i was the last person to find out about it but if you don't have one i suggest you remedy the situation.
Road tuning near year end 2017 produced a knock encounter that should be instructive.
i was evaluating my Borg Warner SXE 62 turbo around 18 psi on 93 octane and 1000 CC of methanol as Auxiliary Injection. i had completed a pull and was examining the log prior to doing another run and found knock. It was modest but significant and the point i need to make is i had zero knowledge that it had happened. Only the log. Here are two logs of the same run, one at 3910 RPM, 18.1 PSI, 12.0 AFR, 6.8 Lead timing (11 split) and the second at peak torque 6644 RPM, 10.8 AFR, 18.1 PSI and 10.9 timing 11 split. Peak torque is peak CCP and presents the most difficult knock challenge. My motors make peak torque around 6400-6600 and as such timing should be lowest in this area. Peak torque is where you are most likely to find knock. .
Here's a log showing the intake air out of my turbo and into my motor in real time.
I recently switched to an Innovate (TC-4 Plus) 4 channel thermocouple amp. (P/N 3915) The module converts the tiny voltage signal from the thermocouple to zero to 5 volts for logging and also linearizes it. .
I use the more traditional thermocouples for my two turbo manifold exhaust runners, also from the The Sensor Connection and is P/N EGT-IP-XXX-0001_SS-N.
Reaction time is an eighth of a second and accuracy is within 16 degrees F at a 1650F reading… (four hundredths of one percent).
After seeing the initial log I cut the timing on the next run (log 2) and found virtually no knock. Boost was exactly similar at 18.1, Air Fuel was virtually the same at the 10.9, and timing was lowered to 8.5 which virtually eliminated knock. If i had not been able to log and identify knock i would have eventually had a broken engine. (Major knock on my scale is around 350)
The log is stopped at the last 100% throttle plot. 7183 RPM, 14.5 PSI boost. The lower panel is the intake air temperature at that point just out of the turbo before the intercooler and under the UIM in the stock location. On this log it is C but converted to F the air out of the turbo is 316 F and into the motor is 108F.
Note the large drop and then a steady temperature into the motor generated at the OE location. This is the cooling magic of methanol.
A combination of the wonderful stock location intercooler (on the dyno no less so not much airflow) and the methanol.
Here’s a picture of the thermocouple between the turbo and intercooler:
This sequence clearly shows the value of a good knock system.
Absolutely no knock was observable from the driver’s seat nor on any other of the 50 items I log. Knock was only clear looking at…. the knock log.
Ignition was 10.8 at knock and was lowered to 8.6 on the second run. (11 split)
While ignition advance differed, power was quite close on both runs at a similar RPM. At 6530 on the first run with higher knock power was 446. Power was 451 at 6549 on the second run with no knock.
Both boost, 18.1, and AFRs 10.8 were exactly the same on both logs.
Neither of these are peak power at 18 psi as they are about 9% below peak which would put the Borg Warner SXE 62 at 491 at 18 psi. Max SXE 62 power should be 505 at 18 PSI referenced from the BW compressor map..
Unfortunately, winter arrived the next day so my car was parked. The motor was not injured so i had all winter to consider the knock event.
I called my go to tuning guy (Luke Stubbs, Beyond Redline, Green Bay, Wi) whom I have tuned with since 2003 and asked him why my old/trusted timing map was creating knock. His immediate response was
"winter" gas quality. He also said that generally the quality of gas has been reduced over the last 10 years. Luke uses only one gas station in Green Bay currently.
Another suggestion comes from an equally accredited source, Julio Don, Proprietor of AlkyControl. Currently Julio has an 8 second 1200 rwhp Fox body Mustang running on 93 pump and his Alkycontrol methanol AI system. Julio mentions that pump gas loses over one octane point for each month after refinement.
Given octane lossage with age, finding a high volume 93 octane gas station should be important. I had bought the Mobil 93 from a small station outside of a small town, in retrospect a no no.
Further, being December in Wisconsin, i had “winter” gas which is not as friendly as “summer” gas.
Revisiting the above in August 2018 after extensive additional road tuning, it appears i had bad gas as i have slowly moved my timing back to normal levels and see zero knock. Better gas, no knock at 585 hp on 93 octane and 1000 cc of meth AI. Timing was 11.6 around peak torque.
Just like knock data delivers “sight” to the tuning process a digital fuel pressure sensor alerts to any fuel delivery issue. Whenever I see a small analog fuel pressure gauge screwed into the adjustable fuel regulator I want to laugh. The gauge does allow you to set static fuel pressure (43.5 psi) if you believe the gauge is accurate… then you close the hood and go for a drive to 8500 RPM and… WHEN IT COUNTS… you have no idea as to fuel pressure.
Enter the 21st century and replace the stone age gauge with a Honeywell or AEM 30-2130-150 digital fuel pressure sensor and get logging. (disregard PN in picture please)
Here's how it looks on my engine. Goodby to a slipping water pump pulley and a coppered front main bearing. Hello Full Function Engineering
While I am on the subject of not being aware of something important if you don’t have the proper tools, in this case an excellent knock system, another related subject comes to mind…. Misfires.
Misfires are not knock, rather a lack of ignition on a rotor face. I don’t think they show up anywhere in my logged data. They do not show up on the dyno UNLESS you go to zero smoothing. Misfires generally are greatest at peak torque because that is where we find the highest fill.
Misfires/partial fires look like tiny wiggles:
If you set static at 43.5 and you are at 20 psi boost your fuel pressure should be 63.5. Check your log. If it is less perhaps your fuel filter needs replacement, or your pump needs more voltage (it must be hard wired for 13.5+ volts which is essential). Perhaps your fuel pressure is more than 63.5 which would indicate a fuel return constriction.
While you are at it I suggest you buy another and use it to log your oil pressure. I use a third pressure sensor to log exhaust back pressure…
Since these sensors are drop dead accurate, Honeywell doesn’t do sloppy, I suggest you also replace your MAP sensor with an AEM MAP pressure sensor. 30-2130-50.
Finished with pressures, let's discuss heat
Heat is of course a big deal. We have already mentioned CCH or Combustion Chamber Heat. Mostly a function of flow into the motor and generally very high due to a small displacement and a LOT OF FLOW.
Incoming air temperature is a major factor. Ideally we want air from outside the engine compartment. I am not quite there yet. What we absolutely do not want is to have the air filter anywhere downstream of the intercooler exhaust.
Most people do not know what their intake air temperature is as they do not have a proper sensor.
I remember looking at the air temps from one of my dyno logs and it showed 27 C from 2000 to 9000! (Power FC days).
I resolved to solve the lack of actual temperature data. The answer was right under my nose… my Type K thermocouples that I was using for EGTs.
A specialized Type K thermocouple is available to read intake air temps available at The Sensor Connection or EGT Technologies.
I log IAT just out of the turbo, before the intercooler and at the stock location under the upper intake manifold. Most would find just the UIM location helpful.
All of my current thermocouples are sourced from The Sensor Connection in Michigan. All are made in the USA and are both high quality and fairly priced.
The IAT thermocouple is P/N EGT-AP-072-0181-SS-N. currently $52.75. Reaction time is a quarter of a second and +/- .4 tenths of one percent error over the entire range.
Intake Air Thermocouples are certainly not necessary but they can come in handy especially when designing systems as well as fuel choice. My next visit to the Texas Mile will be powered by gasoline not E85..
VP suggests that I run Q16 if my intake air temps are below 140F and C16 if they are above. Thanks to my instrumentation I can answer that question.
I also find it interesting that a combination of my intercooler and methanol AI lower charge air temps 208 F!
I will use my accurate IATs to add more methanol until my IATs are below ambient.
A more common use of thermocouple sensors is for zeroing in on exhaust gas temperature. I run one on each rotor. If all is well they pretty much provide confirmation. If one is significantly different it would be time to find the answer.
Misfires can often be fixed by gapping plugs (10 heat range or colder) at .023.
If they still persist you need to look at your coils (IGN-1A, AEM Smartcoils), dwell setting (approx 4 ms), and plug wires (Magnecor recommended).
Keep in mind that the same very high CCP requires a bullet proof ignition.
I run a Kenne Bell Boost A Spark system set at 16 Volts to my coils (only) under boost.
You can also see my exhaust back pressure probe in this picture.
Since I have designed what I think is the highest flowing, least restrictive turbo manifold exhaust back pressure data is the acid test.
Backpressure is as evil to the motor as front weight is to the chassis.
Backpressure reduces turbo drive
Backpressure increases exhaust being carried into the next intake stroke!
Imagine: you are trying to get as much oxygen into the motor to burn. To the extent it is cold it is more dense and therefore more powerful. To the extent it is cold it is less likely to detonate.
To the extent you have backpressure a volume of hot, non oxygen bearing gas is mixed into your cool intake charge.
Backpressure stacks back into the exhaust phase raising the temperature and overheating your apex seals… warp.
If you review my manifold details (pages 85-) you will see short, large area runners and a perfect straight blend to the T4 flange.
I log oil temperature by gutting the conveniently located aluminum oil level sensor mount, adding a 1/8 NPT aluminum bung and an AEM 30-2012 coolant sensor. It was really nice of Mazda to provide such a helpful easy location in the oilpan.
For almost 20 years my ECT has been around 87C... I run a 2 inch aluminum radiator with the OE ducts in the stock location. I also have had, for the last 10 years, Rick Engeman's water pump featuring his trick impeller. Note the FD impeller looks less than trick.