Increasing the power of your iron horse is always faced with a dilemma: the larger the turbine and power, the higher the output speeds for boost and longer spool.
The output was invented a long time ago - this is a twin-turbo.

The most effective, according to the scheme, implementation of the two-turbine system is Staged Compound Compressors.

The potential of this scheme is such that it is calculated that it is possible to obtain a pressurization of 1.5 bar already at 2000 RPM and 2.4 bar starting from 3000. and further to cut-off

The prepare for TT:


Trionic with integrated CombiAdapter

4BAR MAP sensor integrated in stock sensor.

In connection with the transition to MAP sensor, very quickly the question arose - and what to replace the stock MAP sensor, which is only 2.5 bars or 1.5 bars of supercharging.
Almost a full-time solution for 9k, namely 3bar MAP sensor did not suit because of ambition to blow more than 2 BAR of boost.
Having studied the market of existing automotive products, I found that there is no MAP sensor which would work from 0.2 BAR (abs) to 4 BAR. And if with 4 BARs of the upper limit everything is more or less clear, then with a lower limit of 0.2 bar, everything turned out to be not so simple. The fact is that the lower limit is the work on idle and in gas release modes, which directly affects the stable operation of the engine in these modes, as well as the fuel consumption.
Plus, very tricky features of filters of ADC trionics, which are tuned to the frequency response of the stock MAP sensor.
Yes, yes, having recorded a sound signal from the MAP's through the sound card, I heard such sounds there that in time to amplify and apply to the built-in acoustics

So, at the moment, already the 7th iteration of MAP sensors, thermostabilized, with correction of the frequency response, compatible with the standard of the breakdown and short-circuit diagnostics

And most importantly, that, past the upper limit of 4.1 bar, it was possible to make a lower limit of 0.1 bar, which would have a very positive effect on fuel economy and overall fuel accuracy.

Time ago I began work on the twin-turbo system on saab
Despite the seeming simplicity, the embodiment in a gland on a gasoline engine, without using any factory twin-turbo system for a diesel engine, involves the problems of finding and manufacturing TT elements.
In addition to welded products, the TT system consists of
A large turbine, a small turbine, a flap on the hot part, a bypass valve on the cold part and a highgate.
In the course of time, a lot of spare parts were bought, which could be suitable for their characteristics, their testing, testing and rejection.

With a large turbine, there was no choice - only its cost
Wastgate is a proven solution from Tial
A small turbine - came from younger brothers by brand
The bypass valve on the cold part brought a lot of trouble, but in the end there was a product of the concern VAG.
And the most difficult thing is the damper on the hot part. In order to find the optimal solution, it was necessary to go through all the circles of hell - buying valves from the Chinese, trying to draw and produce in Khrunichev space rocket plant (refused because of the difficulty, having already undertaken, feeding breakfast for six months), buying a product of the VAG concern, trying to manufacture a metal 3D printer ... ... And Finally - found a working solution from the Japanese.

And now, the whole kit is assembled, it is based on a fitting of geometric dimensions:

Preparing for TT: Carrying the battery into the trunk.

The release of space under the hood for the twin-turbo
More precisely not for the turbines themselves, but for a short, cold intake and heat exchanger for a water intercooler. (As well as if there is room for a pre-heater)

The wire is 50 squares or 0GA of copper, both + and - are transfered.
The estimated total voltage drop on both wires at the starting currents is less than 1 volt.

Everywhere outside the box in the cabin, on the plus wire is put on a protective poliamid stocking.
In particularly responsible places, in addition to both wires is a shrinkage with glue
Under the hood on both wires is clad corrugated and wrapped with electrical tape
Plus is connected up to the starter, minus to the bolt on the gearbox, where the mass wire comes in
Protective sleeves through the motor shield are made of a hydraulic hose for 2700 bar - will not be overdriven
Under the hood has not yet modified the standard wiring and the battery terminals are not removed
In the trunk is not yet made a fastener for the battery.
The cost of the wire was ~ $120 , it took with a huge margin.
Connectors, stocking, corrugation, heat shrinkage ~ $50.

9.5 hours without interruption.
1. The battery is connected in time, with disgusting terminals, is not fixed at all, not insulated, and so on. There will be another battery, LiFePO4 and it will connect via a white connector - visible on the photo next to the battery. Fuse and charge control - electronic.
2. The two wires are transfered to not violate the correct power distribution scheme with a single point + and a single point -
3. To filter out noise and jumps in the event of a sudden change in consumption, a supercapacitor for 20 farads is ordered and already shipped (for reference: in 84 farads you can start a car without a battery at all)

Shit what a build! Sub on this!

SAAB 9-5 Twin-Turbo: Layout Development

Usually projects of such complexity become long-term construction. A cheerfully initiated project is stuck in some technical problem and until it is solved - the car is in the garage, half-disassembled.
I do this project on the principle: All technical problems must be solved in advance, the system must be made in hard, and then installed in the machine for 1 time.

Because this project had a residual priority regarding the purchase of housing and basic work - it lasts long enough. But! Nothing in the garage is not worth it, all technical and engineering problems are solved without machine downtime

The house is already there, the renovation is done - you can activate on a pleasant
There was a final difficult stage - the layout and welding of the collector under the twin-turbo.
To do this, I made a 3D scanner, scanned the engine, turbines and other auxiliary things.
and now, in your favorite CAD select the best location of the units in the engine compartment
As soon as the collector converges - it will be immediately ordered to weld it according to the ready drawings.

SAAB MAFless: more than 1000 km on one gasoline refueling
In addition to the long road, there was traffic jams at outrance and traffic jams at the entrance to Moscow, two cold starts, four warm starts, while on business I drove around the city and   Slippage in the mud (got out from mud by the swing).
The result is in the photo. As you can see, it does not even have a gasoline arrow are not rest down, so the 57 km balance is quite realistic.
All the way the rest in the amount with the mileage showed a little more than 1000 km, which eventually turned out.
Of course - i waiting for the round figure and stopped specially for the photo

Finally something happened that I did not have time for about six months. 
I install the latest version of MAFless firmware (changed physical model for AIRmass calculation on MAP , noCEL, ) , moved to 4BAR MAP sensors newest version, and tried out a new version of my old develop - supersport mode: On the sportmode the RPM are kept, not lower than 1800
And when braking - automatically shift down gear
At any time, I get performance without delay   
This is most noticeable when cornering and passing speedbump 

continuation of work on SAAB TwinTurbo:
Cylinder head preparation. Test of dimension.

Why the MAF on the turbo is bad solution.

It would seem that the MAF considers the exact mass of air and there are no problems. Let's try to understand.
The turbo system looks like this:
filter --- MAF --- turbo intake volume --- turbine --- volume of pipes and intercooler --- throttle valve --- intake manifold --- cylinder.
it is immediately evident that the engine needs an airmass falling into it from the intake manifold, and the measurement takes place so far away (by the way, the measurement with the help of MAPs occurs just in the intake manifold ...)

consider the situation when the car engine on idle and then the driver sharply tap full gas pedal, the turbine spun and inflated, say, a BAR of boost.
what does the MAF consider? By its own consumption of air by the engine, the turbine increases the pressure in the system volume of the pipes and intercooler.
with the help of the magic formula P * V = m / nu * R * T we get that the MAF additionally calculated the air mass m = P1bar * C * T - Patm * C * T or in this case the mass of the additionally transmitted air is equal to the air mass in the pipes and intercooler at atmospheric pressure. The liter of air under normal conditions weighs about 1.25 grams, the volume of pipes and intercooler is about 8 liters, i.е. for a time of the order of a second we will receive an additional 10 grams or 10,000 milligrams of air. it's a lot.
In addition, the turbine makes a noticeable jitter in the readings due to various acoustic effects

How did SAAB engineers solve this problem?
to get rid of the jumping of the MAF readings for any turbine boost change, they imposed five different filters, which are engaged in that they smooth out the MAF readings separately for ignition, fuel calculation, limiters for Airmass, etc. now the MAF readings are relatively flat, but ... they began to arrive with a considerable delay.
so that the MAF readings were not much different from the actual ones, they limited the speed of opening the throttle to the delay value of the MAF. (tick the Fast throttle Responce in the suite)
But that is not all. there were modes of the moment of sharply tap full gas pedal and switching gears on the AT - at the time of switching we can not change the position of the throttle for a long time.
then the engineers of the Saab used the old, T5-tested air-flow calculation algorithm for MAP-as soon as the rate of change of pressure exceeds a certain value, the system of calculating air for a given time switches to the algorithm for calculating air in DBAs. this is indicated as Transient mode and is controlled by the MAFCal.t_PosTransFreezTab tables on the pressure set and MAFCal.t_NegTransFreezTab on its fall.

It would seem that everything is fine, but no, glitches start to creep: at the moment of switching between the two systems, the fuel mixture composition and ignition experience abrupt changes, the engine control algorithms begin to get confused in the readings and the car "does not go." this is well manifested in this effect: slowly push the pedal into the floor - drove perfectly, quickly down - was blown away and did not go.

but the biggest problems begin when tuning:
the switching time between the air calculation models is strictly normalized in the tables, install zero-filter is change this time!
The stock air pressure calculation table MAFCal.RedundantAirMap is marked up to 1100 airmass, which is prudent for Aero.
interpolation algorithms in trionics, when the requested value is outside the bounds of the table - they stupidly give out the last valid value.
those. The tuner sees the limitations of the airmass lying on the surface of the map and simply takes and increases them. now the machine can blow out, say 1200 ayrmass.
as long as the patient travels smoothly - the algorithm works for the MAF and everything is fine. but as soon as driver sharply tap gas, the regime on MAP switches on, and there is a maximum of 1100. The mixture turns out to be 10 percent lean, which leads to a short-term knock process or a "sneeze" of the engine-if you blow even more and the mixture is so poor that it does not burn. these are the most difficult to remove glitches, over which all the tuningists of the Saab fight, not understanding the reason: the tuninger looks - like and lifted up the fuel, but the knock on these condition and airmanses is still there, although he sees that it should not be. and so on.

and the engine, meanwhile, is blown

To solve these problems - you can go in two ways.
1. go to the MAP/MAFless. The path is good, but it takes an order of greater of knowledge for the tuner.
2. Move the MAF closer to the throttle. It partial solve the problem:
Initial system: filter --- MAF --- volume of turbo intake pipe --- turbine --- volume of pipes and intercooler --- throttle valve --- intake manifold --- cylinder.
we transfer the MAF and obtain:
filter --- volume of turbo intake pipe --- turbine --- volume of pipes and intercooler --- MAF --- throttle valve --- intake manifold --- cylinder.
It can be seen that in this way the influence of considerable volumes of the intake system is excluded and the machine begins to work much more stable, on heavily loaded machines.

However, the transfer of the MAF immediately gives a bunch of side effects: in hot air, the accuracy of the air measurement drops (although the increase in the measurement error is still ten times smaller than the crooked counted air, when the MAF stands in front of the turbine), the flying oil pollutes the MAF and it is required once every few months wash it.

Finally. MAF is needed for environmental reasons. Without him on a turbo car is much better

So. on the turbo car there is a problem of the receiver. At the time of opening the throttle with idle, it is filled with air through the long-term intake. it takes time and brings a significant contribution to the turbo lag at the start.
Also, to save the automatic gearbox on the shifts, it is necessary to low down the torque at the moment of gear changes. on large boost it's sooooo long: after the throttle is closed, the long time must pass before the engine picks up the contents from the receiver. as well as after switching, the time must pass before the receiver is refilled.

The solution to this problem is the individual throttle bodyes for each cylinder.

VGT actuator control test

SAAB 9-5 Twin Turbo: First fitting in hardware

As it was determined in the model - we even get into almost a regular place between the engine and radiators
Of course, all this iron weighs a lot ... 35kg: a collector, two turbines, a hot bypass and all the flanges.


For comparison - the 3D model is about the same view angle

Slowly go work on welding twin-turbo ....
And so far, the display has arrived for the LSD dashboard.


(a photo of dashboard displayed on LCD)


Design is planned in the style like Oblivion movie

SAAB Tech: B235R + AF40 (TF80SC)


With any tuning of the 9-5 for power, the question immediately arises - how can this power be transmitted to the road?
Our standard automatic transmission AW55-50 is quite reliable and can, with a good setting + cooling, workin very long time without damage risk about 450nm.
but, 1. this is not enough for me 2. need four-wheel drive

our automatic transmission cannot be reinforced. I suppose that this is still possible: there are several obvious things, such as an increase in the number of discs in clutch packs, but none of the specialists in the field of automatic transmission helped me in this matter.

In the process of searching, several options were considered, according to which I collected information for a long time:

Transverse ancient 4-speed automatic GM 4T65E. In fact, it is a longitudinal box located along the transversely located engine. It has tuning up to 1200 horsepower and 1000 Nm of torque.
and there are all-wheel drive versions - for example, from the Volvo XC90. in general, fits in and can even be placed at the right angle to the engine - but alas, the the front transfer case crawls out behind the front cover, at best the car will remain front-wheel drive. in addition, 4 stages are already small by modern standards.



The next candidate is the DSG-  Automatic Dual-Clutch Transmission. target gearbox DQ500, as a fit test we used DQ250, which has almost identical dimensions.
Alas, this option had to be discarded right away - the output to the driveshafts is on the same level with the crankshaft, which is generally high, although you can survive, but the the front transfer case is also at the level of the crankshaft, which automatically cuts the steer-rail and the cardan must go through the cabin. Also, the box is 5cm longer than our boxes, so it is cut with a spar and sticks out robustly into the left wheel arch.
The next minus is the gear ratios. Unfortunately, only one modification of the box has the necessary gear ratios, which was installed exclusively on the Audi RS3 and TT-RS. the cost of such gearboxes reaches $3000, despite the fact that you will not find them with fire during the day.



And finally, the final version - AF40 with Saab 9-3 2.8V6 aka TF80-SC
The box digests the torque up to 640nm - according to the available information, there is a Saab that regularly drives around the track at such a torque for several years already, the gearbox has an all-wheel drive modification (turbo-X), fits into the engine compartment, because has dimensions not exceeding AW55-50 and has an outlet to the drive with the desired offset downward relative to the crankshaft. After the final choice, work was carried out to physically dock this box with our engine.








The next step is introduction into the car, adaptation of the upper box support, rear support, front support - and you can try to drive out
Now about the management. All automatic transmission control goes through the can-bus. A study was conducted of the can-protocols on 9-5, 9-3 2.8V6 and 9-3 TurboX. As it turned out, the communication between the engine ECU and the box is very similar to each other, which allows you to make a can-translator between AF40 and Trionic7.