VGT Turbo M20 Sleeper ('87 325 Sedan)

[Panici]

During testing I definitely found the surge limit of the turbo a few times, as seen by the jagged MAP signal. This was a pretty aggressive vane position, a little over 90% closed:

Pretty neat to see the surge limit.
Do you have a compressor map for your setup to reference?

[mikey.antonakakis]

Did a lot more VGT characterization testing last night. Basically holding a steady RPM and setting a fixed vane position during spoolup, and letting it spool to its limit, and repeating, varying RPM and vane position. Logged 42 distinct combos to work on characterizing the relationship between RPM, vane position, MAP, drive pressure ratio, and rate of building boost.

During testing I definitely found the surge limit of the turbo a few times, as seen by the jagged MAP signal. This was a pretty aggressive vane position, a little over 90% closed at 3000rpm. In this sample, the surge is mostly caused by the small VGT opening restricting flow through the engine - which is cool, because that means I'm not actually at my boost limit at this engine speed of 3000rpm - I can open the vanes a little more and get the surge to go away.

You might also notice the wastegate dome pressure signal going a little wonky during the surge event, but it's unrelated. It's doing this because I'm getting close to the boost target so it starts to modulate its pressure to avoid overshoot (not really important at 3000rpm, but makes a big difference at higher RPMs). During spool-up, I keep dome pressure quite high due to the high exhaust manifold pressure resulting from the VGT clamping down. For instance, in the data below, the steady state condition during the surge has EMAP of 278kPa. MAP 158kPa, and dome pressure 250kPa (all in absolute pressure). So there's a 92kPa net pressure on the wastegate piston, in addition to the wastegate spring which by itself gives about 51kPa boost over barometric pressure; opposing that net force is the nearly 200kPa pressure differential across the wastegate valve in the manifold (EMAP minus downpipe pressure, which should be equal to baro), which is trying to push the wastegate open.

In fact, you can see the wastegate start to open at the cursor (vertical blue line) when the dome pressure drops down to about 200kPa - notice how EMAP (salmon colored line) drops at that point, while the vanes remain constant. This drop in EMAP (and drive pressure ratio) is due the wastegate valve unseating itself and relieving some exhaust manifold pressure. Looks like it actually starts to happen at about 230kPa dome pressure.

Time to process a pretty big set of data (multiple samples of each of the 42 test points).

[mikey.antonakakis]

Quick update for now. I've done a lot of datalogging and data analysis to get the VGT working more optimally during spool-up. I've made a huge improvement, as shown below in datalogs starting around 4000rpm in 2nd gear.

The lighter colored lines are the previous calibration/algorithm, and the bolder lines are today's setup. MUCH faster spool-up, even with the starting conditions being very different. Today I just stomped the throttle after coasting in 2nd gear, vs. the older data being a 1-2 WOT upshift. In other words, turbine shaft speed with the new data started at 13k, vs. 45k with the old data.

Despite that, with the new setup it spooled much more aggressively and hit target boost sooner, and the difference was much more pronounced at the 2-3 shift with a more even playing field (although the new tuning still had a little disadvantage due to a bit of a short-shift). For the 2-3 shift, 0.7s vs. 1.7s to hit target boost. I didn't flatshift, but it would have spooled faster if I did.

[mikey.antonakakis]

Okay, time for some fun stuff, more data! The big updates are that the car now has EGT and exhaust manifold pressure measured at the turbine inlet. It's also running the turbo control system (wastegate dome pressure control and VGT control) on the new PCBs I designed.
So now with EMAP I can calculate drive pressure ratio and see the effect of the VGT. I can even do something like alter my VGT algorithm to seek a target drive pressure ratio (say, 2:1) during spool-up, rather than having an open-loop setup.

Quick sample datalog below, 1st through 3rd gear up to ~30-105mph in about 10s, not flatshifting. Note that the bottom plot is recording wastegate dome pressure, turbo speed, VGT vane position (calculated as nozzle size), EMAP (exhaust manifold pressure), EGT, and drive pressure ratio.

I think I can get more aggressive with the VGT actuation, drive pressure ratio starts to drop immediately as boost builds, peaks around 1.4:1 near redline with vanes (almost) fully open. Dome pressure control is working great, within 0-5kPa of boost target, and it's in a sort-of open-loop mode (not trying to directly control MAP). Note also that wastegate dome pressure is kept high during spool-up to ensure the wastegate stays closed.

[Panici]

Z3 rack (“031”) install finished, also swapped the solid aluminum transmission mounts for 95a Garagistic poly and replaced the steering flex disc with aluminum. Quick and rough home garage alignment and a 5min drive, whoa, what a difference! Partially from the quicker rate, yes, but also the lack of slop now. Huge improvement!

I have the Z3 rack on my E30 as well. Makes a huge difference coupled with a slightly smaller steering wheel. (370mm vs 385mm)
Difference between feeling like driving a bus, and a sporty car.

Definitely go for an alignment. I basically toasted a set of front tires driving around with my garage alignment for a few months.

[mikey.antonakakis]

Z3 rack (“031”) install finished, also swapped the solid aluminum transmission mounts for 95a Garagistic poly and replaced the steering flex disc with aluminum. Quick and rough home garage alignment and a 5min drive, whoa, what a difference! Partially from the quicker rate, yes, but also the lack of slop now. Huge improvement!

[mikey.antonakakis]

Rack came out without a fight this morning, lifted the engine about 1-1.5" and there was plenty of room (not too hard, undo downpipe-to-midpipe v-band, pop air filter off turbo, loosen intercooler supports, and lift). Had to do a little persuading to get the knuckle assembly off the steering column. Ready for modification/rebuild, Garagistic kit out for delivery. We'll see how well the splines go back on...

[mikey.antonakakis]

Biggest pain on removal: splines didn’t want to let go on the rack end of things, but not too terrible. The rack is totally free now, and almost has enough room to escape its confines behind the oil pan, but not quite. Don’t want to bend the tabs, so next I just need to jack the engine an inch or so, but wife and baby are asleep above the garage some I’m not going to make a huge racket rolling the floor jack around tonight.

[mikey.antonakakis]

Z3 rack swap started, about 30min and close to pulling the rack out. Just need to loosen motor/trans mounts, lift the engine an inch or two, and then should be able to pull it. Garagistic swap kit should finally arrive tomorrow. They were backlogged on delrin steering couplers, with they delay they offered to swap me for aluminum and they discounted me without me asking or without them even mentioning it. Such a great supplier!!

Hey Future Self: this last part of pulling the old rack probably didn’t end up being as easy as you thought!

[mikey.antonakakis]

I'm in the same boat, though - I think my timing map is pretty conservative at the moment, except for maybe at peak torque at a little over 5000rpm where I can hear the slightest knock if I add any more timing. I'm looking at getting some dyno time in the near future, turns out these guys are not too far from me and have more old BMW experience than any other shops around here, and seem laid back about letting me do the tuning myself. Also seem quite confident that they've developed good timing maps for turbo M20s specifically. Mine might be a little more limited than most due to eta pistons with 885 head.

[Panici]

Don't know if I mentioned this, but I did build some DIY "knock ears" - since most knock sensors are piezoelectric microphones essentially, you can just amplify the signal and listen it. Pretty cool to listen to in real time, especially on an engine as noisy as an M20 lol.

I did try to build those, but my hearing isn't that great with tinnitus. The Knock Monitor Pro I bought has a software interface to detect knock and a set of headphones if I want to listen.

[mikey.antonakakis]

Wow, that's what I call a comprehensive answer. Thank you!


Very interesting, and it does make a lot of sense.


Seems like an interesting application for an electronic wastegate to facilitate an Economy mode.
Keep the gate open all the time in Eco mode to reduce pumping losses at the expense of power.


Think I will try to explore some tuning for fuel economy in my Miata. From what you've said I should be able to make some gains, as it has a LARGE eBay special turbo and a tiny 1.6L engine. The large turbo shouldn't be a restriction under normal driving.

I did finally get a Tuner Nerd Knock Monitor Pro in the mail this week, so time to get my very conservative timing map up to snuff! I've read that more aggressive timing coupled with a leaner then stoich burn in the cruise area can make some significant gains.

I should point out that I did not directly calibrate turbo powertrains while at Toyota (did do some evaluation, though - BMW's modern turbo engines are pretty darn awesome, they do some fun stuff to maximize drivability). I also did this circa 2014-2015 and there may be advancements in technology I'm ignorant of. But I think my answer should hold water for older port-injected engines at the very least, and those without variable valve timing/lift or only simple VVT systems.

Yeah, I think a big turbo on a 1.6 should not be any significant detriment to FE for normal driving. And a little knock at light loads might be no big deal, I think plenty of OEM stuff allows for some knock at light load, although I was a transmission engineer, not an engine calibrator, so don't rely on that as sound advice!

Don't know if I mentioned this, but I did build some DIY "knock ears" - since most knock sensors are piezoelectric microphones essentially, you can just amplify the signal and listen it. Pretty cool to listen to in real time, especially on an engine as noisy as an M20 lol.

[Panici]

Wow, that's what I call a comprehensive answer. Thank you!

So, downsized turbo engines DO get better FE (thanks only to the downsizing). Maximum FE benefit is when they are putting the least possible power into the turbine since this reduces backpressure/pumping losses

Very interesting, and it does make a lot of sense.


Seems like an interesting application for an electronic wastegate to facilitate an Economy mode.
Keep the gate open all the time in Eco mode to reduce pumping losses at the expense of power.


Think I will try to explore some tuning for fuel economy in my Miata. From what you've said I should be able to make some gains, as it has a LARGE eBay special turbo and a tiny 1.6L engine. The large turbo shouldn't be a restriction under normal driving.

I did finally get a Tuner Nerd Knock Monitor Pro in the mail this week, so time to get my very conservative timing map up to snuff! I've read that more aggressive timing coupled with a leaner then stoich burn in the cruise area can make some significant gains.

[mikey.antonakakis]

I have little knowledge of VGT turbos; keeping the vanes closed mimics a smaller exhaust housing?

I always thought that (all things equal) having the turbo sized to spool during cruise would increase engine efficiency and therefore increase fuel economy. Is this incorrect?

I feel like an imposter a lot of the time, but this is a topic I'm actually qualified to opine on (former Toyota calibration engineer)! Warning: strong opinions and technical stuff to follow. But first, your first question: basically changing vane position changes the AR of the exhaust housing, yes. There are different VGT structures out there, mainly rotating vane and sliding wall. Holset VGT uses sliding wall. Here's my turbo when I was first figuring out how to control the VGT position via CAN bus. Sorry for the poor lighting, but you can see the vanes/nozzle wall moving:

Second and longer answer:
Best thing for fuel economy in terms of exhaust side, all other variables being equal for a given engine, is getting the least back pressure possible, minimizing pumping losses.

Downsized engines with turbos being better for FE is at best intentionally misleading and at worst a bald-faced lie by auto manufacturers, just like "AWD is so much safer!". It's only the down-sizing part that's making an improvement in FE. It can be a really complicated topic, because engine usage varies so much with driving style or fuel economy test cycle, and modern engines have so many tunable systems. But there's a really easy way to "normalize" things: power.

At a given point in time when you're driving a car, your right foot is demanding a certain amount of acceleration from the car, or in other words, longitudinal force at the drive tires' contact patches - i.e. you're asking for a certain amount of power from an engine. You make think "no, you're demanding torque," and while it's a little pedantic I'd insist that power is the key real-time parameter:

• You, as a human in a seat in a car, feel acceleration, and at a given time you are going a given speed.
• The car has mass and drag and rolling resistance, etc., all of which are either speed-dependent or constant.
• F = m*a, and the "a" is what you're actually feeling (although our perception of acceleration is speed-dependent too! But that's almost getting into physiology/psychology or something so we'll ignore that one for now).
• P = F*v, or P = m * a * v. All the "losses" can be calculated for a given speed in terms of force, like aero drag and rolling resistance, which means they can be subtracted out from that equation, which after that's done I'll refer to as "Pnet".
• Solving: a = Pnet / (m * v)

This is not only the "right" way to think about things IMO, it's also super handy for a couple of reasons: one, because almost all cars have some means of changing gear ratios, thereby making a torque-based approach useless. Two, brake-specific fuel consumption, which is a parameter that gives the fuel consumption in terms of power, i.e. power-based fuel efficiency. Fuel flow rate is a useless parameter without additional information (just like torque - an engine with 1000ft-lb that only spins to 1000rpm won't make for a quick car). Since we've established that the best real-time parameter to base things on is power, at a given instant when you're demanding a given power from your engine, a lower BSFC will give you better fuel economy. BSFC maps are plotted against torque and engine speed, and for cars with multi-gear or CVT transmissions, this is great, because you can use that map to decide which gear is best for fuel economy at a given instant. Note the "constant power" line on this plot and the discrete gear ratios. 4th gear is best for FE here:



As you might notice, for the engine in this plot, and most other naturally-aspirated gasoline engines, best BSFC is at relatively low engine speed and relatively high torque. With direct injection and variable valve timing, best BSFC is closer to/at peak torque and low RPM. THIS is the reason for downsizing engines - most of the time when you're driving around, you only have to roll the tires and push air out of the way. You spend very little time increasing the momentum of the car, so your power demand is quite low the majority of the time. And you hardly ever ask for high or peak power. So to get good fuel economy, the car ends up using the tallest reasonable gear ratio to get to the low-speed/high-torque efficiency island. If your vehicle has a big engine to cope with intended usage (like a Corvette or a work truck), this is why cylinder deactivation is a thing. By itself, it loses efficiency since you're spinning a bunch of extra bearings and pumping air for no reason (just creating heat) but it makes a big enough change in BSFC usage that it is a net benefit to FE.

Now the obvious issue with just downsizing is that you lose peak power, making the car slower. This makes the customer unhappy, especially in the USA (not a political thing, it's the nature of our roads/highways/car-centric culture). US drivers legitimately use a lot more power and demand more acceleration in everyday driving than most of the rest of the world. To close this gap, add forced induction! Turbos don't sap mechanical power, great! That said, it does take a serious amount of power to drive a turbo at high rpm and full boost - no free lunch, that power has to come from somewhere, and it comes from exhaust energy, specifically by creating a pressure drop across the turbine wheel, i.e. increasing exhaust manifold pressure. In terms of FE, this increases pumping losses of the engine and therefore reduces BSFC. Add to that the fact that in boost AFR typically needs to go richer than stoichiometric and your FE absolutely plummets in boost. Direct injection helps run leaner by reducing/eliminating knock, but you still have a big pumping loss, since turbine drive power at full boost is somewhere in the range of 10-20% of your engine's power output!

So, downsized turbo engines DO get better FE (thanks only to the downsizing). Maximum FE benefit is when they are putting the least possible power into the turbine since this reduces backpressure/pumping losses - in other words, when it's acting the least like a turbo engine, when the turbo is doing nothing. Knock-on effect is now you're just carrying extra weight around for no reason. OEMs that use this strategy are very careful about calibrating around the EPA/CARB test cycles (which are remarkably low-demand - seriously, if you drove on most US roads like the test cycles, you'd get run off the road). They are able to keep the engine out of boost the majority/all the test cycle. This is why there's often a huge gap in window sticker vs. real-world FE for these engines - although it gets minimized if you drive gently on the highway.

So for my VGT setup, closing the vanes any amount from full open decreases fuel economy. Closing them too aggressively clearly shows up as richer AFR due to less mass flow of air due to increased exhaust manifold pressure (pumping loss). Even if I compensated my fueling to account for this and get AFR back to where it should be, I'm making less power (can feel in the seat of your pants, it's obvious). I'd guarantee my BSFC is lower with closed vanes - although there's probably a range from fully-open to "not very closed" at lower engine speeds that probably only has negligible impact. I'm going to try to document all of this on video/data logs as I go through the tuning process, if I can.

[Panici]

One mode has the vanes wide open while cruising (for best fuel economy) and the other closes the vanes a good amount to give the turbo a little bit of pre-spooling, so when you do get on the throttle and demand boost you have a little head start on spool-up. The party trick is that it works like an electric exhaust cutout, closing the vanes most of the way makes the exhaust noticeably more quiet.

I have little knowledge of VGT turbos; keeping the vanes closed mimics a smaller exhaust housing?

I always thought that (all things equal) having the turbo sized to spool during cruise would increase engine efficiency and therefore increase fuel economy. Is this incorrect?