This article was first published in 2005.
Last week we got the turbocharged hybrid Prius on the road. The turbo was on,
exhaust finished and air intake system working. But the initial testing showed
that the air/fuel ratios were much leaner than desired – and much leaner than
expected. So the next job was to sort out the fuelling – something more involved than it first appeared: solving the problem the first time was
easy; solving the problem the second time was a nightmare.
I must admit that I was almost blasé about testing the air/fuel ratios: I was
running a turbo’d 6 psi and the previously fitted supercharger had been
producing much the same boost level, so I didn’t expect any change in air/fuel
ratios. But after I inserted the MoTeC air/fuel ratio meter’s probe, I was
shocked to find that at high loads, the mixtures were much leaner than expected.
(The car used the approach covered in
Altering Closed Loop Mixtures to switch it out of closed loop and then adjust the resulting mixtures.
Basically this involves using a relay to disconnect the two oxy sensors at a
certain engine load and then setting the resulting mixtures with the Digital
Fuel Adjuster kit working on the airflow meter output.)
Clearly, 6 psi of boost from the turbo was sufficiently different than 6 psi
boost from the supercharger to affect the air/fuel ratios - or perhaps the new
intake and exhaust were making a huge difference?
Unlike when the blower was fitted, this time the lean mixtures were unable to
be richened by adjustment of the Digital Fuel Adjuster (DFA) working on the
airflow meter output. There were three possible reasons for this:
airflow meter signal was as high as the ECU would accept
fuel pump couldn’t cope
injectors were running flat out (ie 100 per cent duty cycle)
At that stage I thought the latter was most likely, so I used a multimeter to
measure the full-load duty cycle.
In fact, the duty cycle didn’t much exceed 60 per cent – it appears that
in this car, that’s the ECU-fixed maximum injector flow.
But this wasn’t initially clear and much fiddling later – including fitting a
new fuel pump and modifying the in-tank regulator – I realised that I either
needed bigger injectors, or higher fuel pressure to push more fuel through the
factory injectors. I temporarily borrowed four injectors from a current Corolla
(thanks Mr Avis for the hire car!) but found that despite their greater flow,
the lower duty cycles of the Prius system resulted in mixtures that weren’t
richer – they were in fact leaner!
I then fitted a ‘black top’ Malpassi rising rate fuel pressure reg, with its
diaphragm plumbed to the intake system before the throttle. To fit the
regulator, a return line to the tank also needed to be installed and at the same
time an external fuel filter was fitted. This upped off-load fuel pressure from
the standard 44-50 psi to 60 psi and then caused an increase to 65 psi when on
boost. Together with the oxy sensor switching, this resulted in stoichiometric
mixtures (ie air/fuel of 14.7:1) being maintained until a full-throttle 2-3 psi
was reached, with the mixtures then moving to 12.5:1 at higher loads.
However, I then found a full-load misfire. I changed the plugs to one heat
range colder and at the same time went to iridium (from NGK BKR5EY to BKR6EIX)
but the miss remained. In the end I realised that the still connected Digital
Fuel Adjuster had somehow been set with a minus 127 correction at a high load
site – I must have bumped the unlocked hand controller without realising it.
With the fuel map returned to zero correction, the miss went away... in effect,
I’d been running a high load fuel cut! Ooops.
Intercooler and Intake
So far, all the test driving had been undertaken with intercooler (and inlet
air) plumbing formed from copper 2-inch tube and fittings, insulated with thin
foam rubber sheet and then covered in large diameter heatshrink. The use of
copper pipe and fittings gives an easy push-together system that allows the
pipework to be assembled and then taken elsewhere to be brazed, silver soldered
or soft soldered. (The approach is covered in
Copper Intercooler Plumbing)
In the supercharger installation, the total length of copper piping was only
about 50cm, but with the turbo positioned between the engine and the firewall,
the pipe runs went up a lot in length.
And while I was happy with the appearance and functionality of short length
of copper pipe and plumbing fittings, with several metres of plumbing now
running around the engine bay, I wanted both a better appearance and also more
gentle, flowing bends.
Making easier the decision to redo the plumbing was the fact that I’d bought
a new friction cut-off saw (these days, extraordinarily cheap) and a secondhand
arc welder. I didn’t intend doing all the welding with the arc machine (it’s
very hard to weld thin gauge tube with an arc welder) but I thought that it
would be straightforward to tack the tube together. Tube? Yes, I bought a heap
of mandrel exhaust bends in both 2 inch and 1¾ inch sizes.
By using the cut-off saw and the arc welder, it was quite easy to come up
with three completely new pipe runs – with the packaging flexibility of the
bends (especially in the smaller 1¾ inch size), I decided to route the pipe in
different directions to the previous approach. The tacked-together pipework was
then taken to the local muffler shop for MIG’ing, before I spent a few hours at
my belt sander grinding back the welds.
I then had a local welder silver-solder brass fittings on the plumbing for
two crankcase breather hoses, a boost feed to the rising rate fuel pressure reg,
a boost feed to the wastegate, a water injection nozzle fitting, and the two
connections for the recirculating blow-off valve. The next steps were
sandblasting and then powder coating the plumbing.
Incidentally, if considering a similar plumbing approach, don’t disregard the
significant costs - $120 for the mandrel bends, AUD$90 for the MIG welding,
AUD$100 for the silver soldering, AUD$27 for the sandblasting, and AUD$50 for
the powder coating. (So these bits of bent pipe ended up costing far more than
the secondhand turbo!)
New Airflow Meter
Up until now I’d been using the factory airflow meter, cut from the airbox
into which it is normally integrated and mounted in-line after the new Falcon V8
airbox. However, its small size and the slightly bodgie way in which the
chopped-up airflow meter fitted into the intake plumbing caused me to reconsider
the approach. I decided to kill two birds with one stone and build a new, larger
airflow meter that had proper in-line plumbing connections.
The new airflow meter housing was made from a variety of PVC plastic plumbing
adaptors, all available from the local hardware store. These were cut, filed,
sanded and heated into shape, with the resulting airflow meter having a minimum
internal cross-sectional area 31 per cent larger than standard. The airflow
meter was located behind the right-hand headlight. The Digital Fuel Adjuster was
then used to boost the airflow meter output signal so that the car ran as
More Testing... and More
With the new intake and intercooler plumbing installed and the larger airflow
meter in place, it was time to do some more testing. And to find still more
The first issue was that the air/fuel ratios at high loads were again back to
stoichiometric under full boost. That’s despite the increased fuel pressure, the
switching out of the oxy sensors and minor tweaking with the Digital Fuel
Adjuster. So why not do some major mixture tweaking with the DFA?
Well, even with the oxygen sensors disconnected, the Prius appears to have
the ability to revert over time to standard mixtures. I think perhaps it can
determine how much power the engine is developing via the hybrid control system
and has a look-up table that correlates expected injector pulse widths with
measured power. This results in gradual shift back to standard mixtures, even
with the DFA making major changes to the airflow meter output signal. This phenomenon was hard to isolate
because it could be seen occurring only after a whole bunch of full-throttle
events. Drive the car normally in closed loop and then nail it, and the mixtures
would go satisfactorily rich. But repeatedly hammer the car up and down an
(isolated!) test road and the full-throttle mixtures would gradually move from
mid-twelves back to mid-fourteens.
And there’s another complexity. The mixtures that occurred with the oxy
sensors switched out depended to some extent on the correction that they’d been
giving just prior to the switch-out. In other words, the long and short-term
fuel trim values would affect what air/fuel ratios resulted with the oxy sensors
switched out of the system. Perhaps it was this characteristic and the expected pulse widths for the
given power that would then determine what air/fuel ratios the car ended up
The result was that full throttle air/fuel values could vary anywhere between
11:1 and 14.5:1. Do two full-throttle runs and the mixtures would be perfect –
say 12.5:1. Then drive gently around the block and then hammer it... and the
mixtures at full throttle would have changed to 14:1....
But what was the problem - hadn’t the mixtures been sorted earlier? That’s
indeed the case – the problem reappeared only after the change in intake
plumbing and the new airflow meter was fitted. It appears that these changes had
further increased mass airflow – the engine was breathing better and better and
the full-load mixture problem was getting worse.
Then yet another problem appeared. On the occasions when the mixtures were
correct for full power, it was possible on the test road to hold full throttle
for many seconds. And in this situation, a bizarre event kept occurring. After
ten or so seconds, the electronic control systems of the car would momentarily
close the electronic throttle! This could be seen on the vacuum/boost gauge
which would dip from full boost to near full vacuum, before the throttle was
again automatically opened. Perhaps some part of the hybrid system was being
momentarily overloaded, which caused the system to reduce engine power?
Next week: solving the final