"Why is this engine so damn complicated??" Part 1: Sequential turbos demystified
 

Key words: vacuum, routing, simplified, sequential, boost, control, turbo

Introduction

We all know the flaws in the sequential turbo system. The biggest problems are that the components cannot stand up to the heat, and that the factory boost control system was designed for a completely stock car and needs a lot of tweaking to work right with mods.

But did Mazda sit down and say "ok, I want a sequential turbo system [that's a can of worms in itself]. How can I make it as unreliable of a clusterfuck as possible?" I doubt it. Every engineer must build a system to operate within certain tolerances (range of acceptable values) given their budget and level of technology. Mazda modified older designs from the second generation turbo cars and then put it all together. I can't defend the physical quality of the components. But I would like to present here Mazda's overall system design, first with a focus on the function of each actuator/solenoid and then a discussion of how the air moves in the vacuum routing system. I would appreciate any constructive criticism or corrections.

Mazda left us a host of clues as to why they designed particular aspects of the FD. We use as our sources the vacuum routing diagrams, the workshop manual, and the service highlights document.

Solenoids and Actuators

The sequential system consists of the following solenoids and actuators:

Precontrol
Turbo Control (2 solenoids, vacuum and boost)
Wastegate
Charge Control
Charge Relief


But why did Mazda feel they needed all that stuff for a sequential turbo system?

Design hurdles

Mazda had to overcome a series of problems/hurdles when it decide to design a sequential turbo system. Here are the two big overall issues:

1. How do we control the exhaust gases to the secondary turbo before it comes online?

For this, Mazda needed to bleed exhaust away from the primary turbo in a controlled manner in order to maintain a set boost pressure and spool up the secondary turbo--a function equivalent to a wastegate. That's what the "precontrol" system is for. They also need to make sure that both turbos get as much flow as possible as quickly as possible once it is time for both turbos to be online (after transition). That's what the "turbo control" system is for. The 87-88 turbo Rx-7's also had a type of turbo control valve, to open two different passageways in the turbine housing based on rpm. You can see from this old marketing literature for the 87 turbo models that there is nothing new about this type of design at all:

https://www.rx7club.com/attachment.p...1&d=1243179870

Mazda felt that the turbo control valve needed to open as quickly as possible. When you are moving a couch out of your hose, you usually have one guy pick up one end and the other guy pick up the other. One guy is applying a pulling force and one guy is applying a pushing force. That is exactly how the turbo control valve works. The guy pulling the couch is a vacuum force, and the guy pushing the couch is the pressure force. An engine does not produce vacuum when it is producing boost, so now we need a vacuum chamber to store that vacuum for use later. And if we want the level of pressure to be relatively predictable every time, we store that force in a carefully designed pressure chamber.

https://www.rx7club.com/attachment.p...1&d=1243180104

2. What do we do with the compressed air from the secondary turbo before the turbo comes fully online?

As the secondary turbo starts spinning up (but before it is ready to come online), the air it compresses can either go into the engine or not go into the engine (duh). Mazda chose to keep it out of the engine, perhaps because it minimized the amount of calculations that their 8-bit (8-bit like an NES with Duck Hunt and everything) ECU would have to perform. To keep secondary turbo's output from the engine, they need to:

1) make sure it does not combine with the air from the primary turbo. That's what the "charge control" valve is for. The charge control valve is a butterfly valve just like a throttle plate. They took the same type of throttle plate you have in your TB, modified it a bit, and put it in the Y pipe. That valve must not leak or crack open in any way, so Mazda needed to apply vacuum to it to keep it shut when the secondary turbo is not online. An engine does not produce vacuum when it is producing boost, so now we need a vacuum chamber to store that vacuum for use later. When it is time for the secondary turbo to come online, pressure is applied to the butterfly valve and it opens.

2) Since the air isn't going into the engine, it has to be vented back to the air cleaner. That's what the "charge relief" valve is for. The charge relief valve is basically just a blowoff valve. Blowoff valves open when vacuum is applied. An engine does not produce vacuum when it is producing boost, so now we need a vacuum chamber to store that vacuum for use later.

While the charge relief valve vents during the whole prespool process, the turbo control and charge control open at about the same time.

https://www.rx7club.com/attachment.p...1&d=1243182460

So at that point the engine sends more exhaust from the engine to the secondary turbo and the secondary turbo now combines its air with the primary turbo. Perhaps this sudden snapping-open effect of the two valves causes such a pressure drop during transition. Why don't they open gradually and progessively, which I believe is what the Mark IV Supra does with these functions? Well, if your computer was as powerful as the one that gave us such classics as Burger Time and Contra, that pressure drop would be... good enough.

Note that the RPM of transition from just the primary spinning to both turbos spinning varies with throttle position. This rpm value is actually adjustable in the PFC with a Datalogit. On deceleration, the rpm at which it returns to sequential operation (just the primary turbo active) is also adjustable with a datalogit. The PFC indicates that that RPM does not vary with throttle position, unless you program it to do so.

https://www.rx7club.com/attachment.p...1&d=1243187864

See where this is going? It's all very logical, albeit unreliable due to the quality of the components that were used.


Boost Control system

What is the point of these damn restricter pills?

The amount of air entering the wastegate actuator can vary based on a number of conditions--just like the amount of water flowing through a river can vary with the seasons. So what's one way we can predict how much water is going to flow through a given point on a river at any given time? We dam it up! Then we can build an entire hydraulic system around a predictable flow of water. The same principle applies to boost control systems. Restricter pills are used in lots of factory boost control systems, for example the WRX. Why spend money and time making an insanely complicated model of airflow (calculated on your Super Mario Bros. 8-bit ECU) through the actuator under every condition, when you can just put a half-cent little piece of plastic in there?

Now consider a tea pot. For purposes of our discussion, when pressure builds up, it whistles. Think of the wastegate and precontrol actuators as tea pots then. When the teapot whistles, the wastegate cracks open. So what if I didn't want it to whistle, or I wanted it to whistle at a time of my choosing? I'd open the lid and vent the pressure out in a controlled manner. That's exactly what the factory wastegate and precontrol solenoids do. With the restrictor pills in there it's a very simple and predictable design.

The precontrol and wastegate solenoids are simply passageways that open and close on command. All the ECU can do is tell it to turn on or off (by switching ground), usually in a rapid manner like flicking a light switch repeatedly. If we can predict how much air is coming in (restricter pills), then we will always know how often to open and shut the passage to get the amount of air we want. Think about flicking a light switch very rapidly but in a controlled manner so that it appears that the light is on almost continuously.

There's nothing new about this design. Mazda modified the boost control system of the series 5 turbo Rx-7. In that case, air was vented before it reached the actuator (which had one nipple on it instead of two) instead of the air being vented after it entered a two-port actuator. No restricter pills were used in the series 5 boost control system to my knowledge, but boost was only 8ish psi on that car.

Vacuum Routing

I have modified the simplified sequential diagram to illustrate how all the air flows. The charge control and charge relief solenoids switch back and forth from a barometric pressure (or boost pressure depending on engine condition) source to a vacuum source. By default, the charge control solenoid is applying vacuum (to keep the butterfly closed), and then switches to pressure when the ECU engages the solenoid (to open it up). By default, the charge relief solenoid is applying baro/boost and then applies vacuum when it is time to vent out air. The turbo control solenoids presumably switch their respective vacuum or pressure sources at about the same time based on the FSM pages I have posted here.

The wastegate and precontrol solenoids open and close rapidly (duty control) to vent air out of the wastegate and precontrol actuators. The less air is vented, the lower the boost will be (but no lower than the spring pressure allows):

https://www.rx7club.com/attachment.p...1&d=1243186262

All these solenoids are really the vacuum-routing equivalent of relays. Certain "passageways" in a relay open or close based on whether power and ground are applied to the switch (aka the coil) inside. The same principle applies to an air solenoid like those used in the FD sequential turbo system.

https://www.rx7club.com/attachment.p...1&d=1243185797


Conclusion

I'm not saying the sequential turbo system is reliable. But if I had to design one given their development budget and level of affordable computing power, I'm not sure I would have done it any differently, besides using components that could stand up to heat better over time.

Part two of this article will be a similar discussion of all the emissions control systems on the vehicle.

Appendix:

https://www.rx7club.com/attachment.p...1&d=1243187323
Another diagram of all the sequential actuators

https://www.rx7club.com/attachment.p...1&d=1243186414
Courtesy AVC-R manual

https://www.rx7club.com/attachment.p...1&d=1243186655
Series 5 Rx-7 turbo boost control system

PFC boost control behavior:

Boost settings:

Pr 1.00 boost 70% duty
Sc .90 boost 86% duty

https://www.rx7club.com/attachment.p...1&d=1242624130

Nice info, I especially like the analogies of the teapot whistling and two people pushing&pulling a couch to move it.

In the section on boost control:

There's a typo here. "more" should read "less" . higher wastegate/precontrol duty = more air vented = higher boost . lower duty = less air vented = lower boost

nice write up

Fixed.

I agree. While I believe the underhood temperatures and/or embrittling underhood components (same problem, just how you look at it) is a major mess-up by Mazda, a lot of people get upset that the stock boost control system is not adjustable. For a fixed-level system this is a very basic and sensible approach.

So, it seems like the primary and secondary PFC boost numbers do not actually control boost - the duty numbers do. (The secondary boost number does control fuel cut; the primary is ignored.)

Are there any rules of thumb when setting these duties?

For example, on a stock FD what PC and WG duty numbers are necessary to produce a 10-8-10 psi pattern?

On an FD with the standard bolt-ons (intake, ic, downpipe, exhaust etc.), how do these numbers change?

If I start with a pattern of 10-8-10 psi how do I get to, say, 12-10-12 psi?

Is a lot of experimentation required or is there a fairly direct relationship between duty and boost for a particular configuration?

Informative write up, thanks.

I will begin by saying that my analysis of the boost control system is based on 3 things:

1) a limited amount of empirical testing on an FD that I don't even own. All my logs are based on tuning a friend's car. I own a single turbo series 4 FC with the Banzai Racing PFC adapter kit. After the MAP sensor line blew off on my T2 I rebuilt the engine on it and I've been working steadily towards getting it on the road, a point which I will hopefully be reaching soon. So I can't do tests on a stockish FD and I can't do test. I can only occasionally do tests on this modded street ported FD with FMIC, 3" exhaust with no cat, a ported wastegate, and stock twins at 14psi (which will not hold to redline even with boost control duty maxed, probably due to mechanical reasons).

2) My working knowledge of control theory (don't ask me to do any calculus!). In any control system there must be a target or set point. The set point in the factory fuel feedback system is 14.7:1 (Lambda = 1), or .5v on a narrowband sensor. To say that boost is controlled by the set duty cycle of the solenoid is like saying that air/fuel ratio is controlled by the injector pulsewidth in a fuel map: it is based on a limited understanding of how feedback systems work. How would the ECU know whether the mixture is lean or rich if it didn't have .5V as the target value?

Continuing with the analogy, the set point in the fuel feedback system is .5V on the O2 sensor. The injector firing time/on-time/duty/pulsewidth, however you want to consider it, is first calculated by taking a base value (after temperature correction etc). Then the injector is fired. Then the mixture is measured as rich (greater than .5V) or lean (less than .5v. or did I get that reversed?). The amount the O2 sensor reading deviates from the target is then calculated (called an "error" calculation. Then an adjusted pulsewidth is calculated in an effort to reach the target.

Basic Injector pulsewidth --> resulting O2 sensor voltage compared to target voltage (error calculation)--> base pulsewidth is recalculated and then further corrected to get closer to target voltage --> o2 sensor voltage is compared to target again

This is a feedback system. For purposes of this discussion, the amount of correction could be calculated in two basic ways. There are basic logic statements: "IF mixture is too rich, THEN reduce fuel." This is called "Fuzzy logic" and is the control method employed by most aftermarket boost controllers, including the AVC-R and presumably the Power FC.

The second basic method is called PID (Proportional, Integral, Derivative) control. In the PID system of control the "Proportional" factor is akin to the "gain" setting on a lot of aftermarket boost controllers, which adjusts how responsive the self-correcton ability is. Too much "gain" and whatever is being controlled will overshoot the target, and too little gain will result in undershooting the target. I'd go into the mechanics of it further but it's not too relevant here, and it relies on calculus that's beyond me. Some variation of PID control is probably what's used in the factory fuel feedback system.

https://www.rx7club.com/attachment.p...1&d=1243321494

We know that Mazda employed a feedback system in the factory boost control logic:

https://www.rx7club.com/attachment.p...1&d=1243321494

It would be reasonable to assume that such a feedback system is employed by the Power FC. It would go something like this:

Basic boost control duty --> resulting manifold pressure compared to target pressure (error calculation)--> base duty recalculated and then further corrected to get closer to target pressure --> pressure is compared to target again


Look back at that boost control graph at the end of post #2. Why else would the duty cycle be varying? How would the Power FC know how to change the wastegate/precontrol duty cycle if it didn't know what to shoot for? From my testing, that target value appears to be the Secondary boost control value. Look at the blue line on the bottom of the graph (with yellow text I put in there). The measured peak boost before and after transition was about .90 kg/cm^2, 12.8psi according to the inaccurate PFC MAP sensor calibration and 14psi on an autometer boost gauge. Is it just coincidence that the boost value actually reached corresponds with the set secondary boost control value?

3) We can get some clues as to how the PFC boost control works based on reading through the documentation for the AVC-R, which also has a set boost value, a set duty value, a self-learning mode to correct the duty, and a gain setting to adjust the response of the self-correction logic (the PFC has no adjustable gain setting from what I can see).

https://www.rx7club.com/attachment.p...1&d=1243323190
https://www.rx7club.com/attachment.p...1&d=1243323190
https://www.rx7club.com/attachment.p...1&d=1243323190

I can only speculate here. All this academic discussion doesn't always translate into real-life results, especially as everyone's setup is so different and the MAP sensor calibration issue coupled with the somewhat undersized factory wastegate makes things more complicated.

A possible method for setting PFC boost control duty

1. Set both boost and duty cycle values the same. If we wanted 10psi, we would set it to say .65 kg/cm^2 or 9.25psi. By default, the setting is .70 (10psi), but the factory MAP sensor calibration reads lower than a mechanical gauge.

2. Adjust both duty values until the boost after transition reaches the target (or maybe it would hit the target value before and after transition if you're lucky).

3. If boost after transition is where you want it to be but boost before transition is not, adjust the "primary" duty value if boost is too high (subtract duty) or low (add duty) before transition. Keep the secondary boost the same. But that boost level after transition may change some anyway, in which case you would have to keep fiddling with it until it reaches a level that you can live with. There are good reasons why people throw in the towel and ditch the PFC for boost control and I'm not knocking that.

Example of tuning the PFC boost control:

Default settings:

Pr .80 56%
Sc .70 64%

Let's say we are overshooting some. So we'd try:

Pr .65 56%
Sc .65 56%

This might work (maybe it would take a couple pulls for the PFC to learn?), because the PFC can make adjustments as I've said. I've seen the PFC vary primary duty by 3-4% and secondary by 2-3% maybe.

But for the sake of example let's say the pattern is now 9-8-10 or something for whatever reason. So we need to increase the primary duty value some (Datalogit logs would help a lot here so you can see the whole duty curve). This is just an example I am pulling out of thin air, don't go putting these numbers in and expect a particular result! We could try:

Pr .65 64%
Sc .65 56%

Max acceptable duty value appears to be capped at 89%, at which point it will hit a ceiling and the PFC won't allow it go any higher. Maybe that would help. Maybe not. I don't even own an FD man, give me a break :dunno:

It would be the same basic tuning process. If you have the stock inaccurate MAP sensor calibration, you could try a target of say .75 kg/cm^2 (10.665 psi) or .80 g/cm^2 (11.376 psi). The factory ECU corrects the MAP sensor reading based on barometric pressure (altitude basically), the PFC does not. That's why I can't give people hard and fast values for anything, nor guarantee any results. Engineering is all about tolerances (acceptable range of values), and aftermarket ECU's (like most aftermarket components) do not have the tight tolerances of a factory component that was engineered as part of a complete system.

As far as how much the boost drops during transition, I'm not sure what to tell you about that. I'm not sure how you would adjust it, or if it is even possible due to the fact that the turbo control valve (which feeds exhaust to the secondary turbo) and the charge control valve (which feeds secondary turbo boost to the engine) snap open. You might able to adjust the "high" value under the turbo transition settings, but that may not really do anything. On my friend's car, it did 14-10-14, and for the life of me I couldn't make it 14-12-14, nor could I tell you exactly what determined how much the dropoff would be.

I hope I didn't lose too many people on that exhaustive discussion.

My understanding is that in the PFC, the duty % you set are the initial duty values only; the closed control system then adjusts the duty of the bleeding solenoid valves to meet the target. With higher initial duty you bleed more air at the beginning, i.e. higher/quicker boost.

- Sandro

Correct. From the logs I do have, once the actuators begin to open the boost control system will vary pre control duty ("primary" duty) by up to maybe +/- 4% and wastegate duty ("secondary" duty) by maybe 3%. Once the actuator opens, the PFC isn't going to change duty by 10 or 20% in either direction from what I can see, although I welcome any logs someone may have that indicates otherwise.

This is partially true. During the very beginning of spoolup, the wastegate and precontrol solenoids run at a high duty value that does not vary and cannot be adjusted. This bleeds off most/all pressure that would normally crack the actuators open, and effectively closes the respective flappers completely. The high fixed duty value continues until the PFC feels it is time to bleed off less air (drop to lower duty) first so that the precontrol may open. When the precontrol opens (duty drops) is probably determined in part by the set secondary boost level and a calculation of how fast boost has already been building so far.

The wastegate is held shut until right around the point of transition. This behavior is consistent with the factory boost control system:

https://www.rx7club.com/attachment.p...1&d=1243381431

Look at the flat PC and WG duty lines in the graph I posted at the end of the 2nd post, the part of the graph marked "WG shut." You will see that overall, the PFC boost control logic is mostly consistent with the way Mazda did it from the factory.

First, thank you very much for compiling your notes. I finally found the time to read them diligently and really appreciate your effort.

I went back to my logs and would like to ask your advice.

The attached log shows a typical start of an autox run. 1st-2nd gear shift, rpm ramp up, primary-secondary transition. As you can see, I have a very large drop in boost during transition (all parameters shown are scaled to multiples of 10 for easy reading).

Map monitor shows a drop from 0.74 Kg/cm2 to 0.42Kg/cm2. Questions:

1. Do you think such a large drop may be indicative of some problem in the control system?
2. What would you suggest for improving the transition? Note, in my application it only affects the start of the run, after that rpm usually stays above 3,000 and control system does not switch back to sequential.

Stock ports, stock turbos and (unported) wastegate, pills, DP, muffled MP, high flowing catback, Adam's intake, stock IC.

PFC boost settings are attached.

Note, my older logs show a relatively similar boost drop during transition even when the cat was still on.

Thanks,

Sandro

Great thread! Discussions like this is what this forum is all about.

Very nice. Now I want to chart mine . . .

Thanks,
:-) neil

i smell sticky!

Somehow related...As I was searching for info on how to improve transition I found this amazing old thread

https://www.rx7club.com/3rd-generation-specific-1993-2002-16/weird-boost-issue-341534/page3/

Note solenoids may get stuck open if pressure above 10 psi

- Sandro

You need to post up a log, I'm hesitant to make any comment based on the graph you posted. I need to see the precontrol and wastegate duty cycles and I need to see them in their own separately scaled graph. Those are "advance" parameters but you have posted up a log of "basic" parameters, the ones that are viewable on the commander. Make sure that you are logging "advance" under the monitor window, but also check "sensors" but only "CCN" (charge control solenoid) and "TCN" (turbo control solenoid). Then we can see exactly when those solenoids are kicking in.

So under your monitor window, the following should be checked:

in the bottom right corner,

Advanced
Sensors

Under the sensors section,

Sensors --> TCN
Sensors --> CCN

and nothing else

duck hunt
 

Why you knockin' on 8-bit duck hunt? I loved that game :) j/k, thanks again for compiling this kind of info for us, again :D

I don't have historic advanced logs to post at this time. I will come back after logging the parameters you indicated.

Thanks,

Sandro

So how does Mazda's design compare to say, Toyota's twin turbo set up? The 1j/2j system seems to be reliable. Anyone know how they work, compared to ours?

^ there has been a thread or two comparing them...

They are very similar. I believe the difference is in the transition - the Supra has one less valve in the prespool.

I think it boils down to the heat in the RX-7 engine bay is much higher. It causes the hoses to harden and loosen, solenoids to fail, wiring to get brittle, and actuators to leak.

Dave

First off, EXCELLENT post. This is a grey area that a lot of people have. If you truly understand something, then you can really work with it.

I have a few points to make from my own experience.

The stock boost control. You have to remember, like Arghx said, the parameters Mazda engineers were working with. Back in the late '80s/early '90s, they hammered out this system based on a STOCK car. Precat, big muffler, restrictive airbox, all the good stuff. With a stock car, the boost control solenoids and pills work quite well. But, the factory ECU has a VERY limited set of parameters to work with and, as many blown motors can attest, can't deal well with mods.

When you change the flow characteristics of the motor by opening the intake and exhaust up, the factory boost control CANNOT compensate. Mind you, the sequential control system is doing just fine with the turbo control door and what have you, it's the wastegate control and turbo precontrol solenoids and pills we're talking about. When you open things up, you start getting creeps and spikes. The engineers didn't account for dramatic changes in flow.

Anyhow, this is a long roundabout way in saying the stock boost control is good for a STOCK or NEAR STOCK car. Once you modify the car, the system has to go. Would you put a single turbo and big fuel system on a car then pump that all through the stock intercooler? No. You have to realize the limitations of stock parts and know when they need to move on.

Even with the PowerFC being able to actually control the duty cycle of the stock solenoids, you are still limited by the pill design. You will soon have to drill out pills/swap pills out, and that ain't boost control, that's voodoo. You can pick up a used electronic boost controller very reasonably that will do the job better than the factory setup and keep your boost where it should be.

The object of boost control is in the name - to CONTROL BOOST. If you can't, time to do something about it.

Dale

^ I have used the PFC boost control on a friend's significantly modded sequential car with some success (14psi, street port, FMIC, ported wastegate). I presume he has no restrictor pills but I'm not sure as it's not my car and I didn't do the vac job. I think the jury is still out on this matter, although most people have understandably ditched factory boost control solenoids and not looked back. That's why I say that if you want to play around with factory/PFC boost control, do it, but if you don't have the nerves for it just go aftermarket.

One of my latest projects has been trying to figure out if it's possible to control boost on a single turbo car using a factory FD wastegate solenoid and sequential turbo control enabled. This is utter madness in most people's minds I'm sure. I did a bunch of testing tonight, and if I actually get anywhere I will let everybody know.

I am attaching the log of a recent run (National Tour in Seneca, NY last Sunday).
As suggested, I logged advanced and sensors.

The course had several turns at the start. Could not stretch the rpm at the start, so this run is not that indicative of the turbos transition pattern.

However, I have another question as I do not understand the following.

You have know that my last PIM row (19600) is set to stop boost creep by pulling the timing to zero deg. As you can see this works well for my needs - although for some reason the IGL does not drop to zero but sets at 4 deg - any clue on why?

My main question is the following: if you look at the overboost conditions, e.g. at around 350 sec, the WG duty is only around 65% (@160-170 range divided by 255, by the way, why 255?). Note that during the run, it can be as high as 88% (@225). Why? I would expect the PFC to bleed the maz amount of air when the pressure error is the highest. i.e. in those overboost conditions. I had set Pr and Sec boost targets both at 0.90 Kg/cm2 (approx 18900 PIM I believe), with initial duty of 62% and 76%, for Pr. and Sec., respectively.

I am attaching an excel spreadsheet were I have hidden non-relevant rows (a lot of time spent on idle) and columns, as well as the raw log (with non relevant information deleted).

Thanks for looking,

Sandro