Tricking a 4age 20v silvertop ECU

[Leon]

At least it isn't as bad as the guy on NZ Honda's who was told that he could get 30kw (on a stock silvertop) by fitting a Honda dizzy and reflashed Honda ECU (because Toyota ECU's were so bad)

[Leon]

Oh, and he'd actually gone and started buying all the stuff to do it, got to the stage of asking somebody for a base tune ...

[Flannelman]

I have done some quick time trials with a friend in the car doing the times.

Car was rolling, holding 1500rpm in first. Given a countdown, then throttle was mashed to the floor. This eliminates any inconsistencies in launching a manual.

TEST ONE - STOCK................................................... ~5-20 1.48sec ~5-40 2.68sec ~5-60 3.93sec ~5-80 5.83sec ~5-100 7.77sec
TEST TWO - PLENUM TOP OPEN.................................... ~5-20 1.26sec ~5-40 2.36sec ~5-60 3.58sec ~5-80 5.46sec ~5-100 7.41sec
TEST THREE - PLENUM OPEN, TRUMPETS FLIPPED UP........... ~5-20 1.17sec ~5-40 2.35sec ~5-60 3.65sec ~5-80 5.51sec ~5-100 7.51sec.

During test 3, the car began to start to feel flat in the top end. Doing some calculations based on a timing run that resulted in a 7.14sec ~5-100 time (I didn't show this as the 20 and 40km times were inaccurate from the recorder) and vehicle weight (1020kg base weigh + 180kg people + 10kg extra fuel = 1210kg) shows a Hp reading of 186Hp. (internet source torque stats calculator) On further investigation reveals that the stock injectors only supply enough fuel for 180Hp. This is a problem and showed up as the vehicle going flat above 7krpm. Even when using the 7.41 time brings up a calculated 178hp which is borderline on fuel starved (in my opinion)

So, just to sum up

TEST ONE - STOCK..................................................... 7.77sec 0-60mph = 164hp
TEST TWO - PLENUM TOP OPEN...................................... 7.41sec 0-60mph = 178hp
TEST THREE - PLENUM OPEN, TRUMPETS FLIPPED UP............. 7.51sec 0-60mph = 174hp (7.14sec from quickest 0-60mph =186Hp)


From this, I conclude the the AFM is restrictive for power over a wide RPM range, AND the Silvertop Plenum is restrictive, being too small in volume limiting the amount of air that can get to the trumpets.

As that have been stated on previous posts, that's science. Disprove my evidence and conclusions with science - NOT OPINION.

[Bling]

What hardware was used to measure the times consistently?

Not someone with a stop watch I hope. If so though, 0.2 seconds would surely be written off as accurate as that's not much of a delay between a countdown being spoken, a button being pushed by one person and someone else pushing their foot down hard on a pedal. If you've done the tests with a better method than the above though, i'm interested to know what method was used.

If you want to take it as accurate though, then that's your choice. I'm not sure i'd call it science when there are so many factors at play that could account for such a small difference in time recorded. That is just my opinion though, so is worth nothing because science.

[Flannelman]

I do hold these numbers credible. There has been many a record set by a stopwatch.

I've given track times, now these times. It's up to the individual reading this to accept the or discard them.

Until I either dyno record or 1/4 mile test this vehicle, I suspect the majority will reject it.

Might as well post up that blacktop ITBs loose midrange power to silvertop ITBs. No one will believe that either.

[Dell'Orto]

In the interests of science, how many times were these tests repeated, and what variation between each run was there?
I don't think the stock airbox is that restrictive, I can only think of Tom's who made a replacement airbox for thrm

[Leon]

I can only think of Tom's who made a replacement airbox for thrm

/hugs TOMS airbox.

"My precious"

[Lith]

I would take it to a dyno operator who will just take your money and find out whether it's worth it, at this stage it's all fairly circumstantial evidence and if it turns out to be basically no gain then I'd call it not worth it - in my opinion. You could even do some kind of ultra rough testing, get a passenger who is reliable with a stop watch and do a 3rd gear pull from say 2500rpm - find out how long it takes to get to peak power, once with the "normal" setup and once with your new setup.

There is software out there (I've written some of my own) which uses datalogs using speed Δ versus time to calculate power throughout a pull and plot against rpm or speed - you can definitely get useful data using this kind of method, even if its not going to show where exactly in the rpm the gains are. Given how casually a 20v tends to accelerate in 3rd gear, the time scale of the entire pull (and as such, any changes) should be enough for a human to be able to measure if they are worthwhile - so if you can't get a measurable difference in a FULL 3rd gear pull, I would call it a waste of time.

TEST ONE - STOCK................................................... ~5-20 1.48sec ~5-40 2.68sec ~5-60 3.93sec ~5-80 5.83sec ~5-100 7.77sec
TEST TWO - PLENUM TOP OPEN.................................... ~5-20 1.26sec ~5-40 2.36sec ~5-60 3.58sec ~5-80 5.46sec ~5-100 7.41sec
TEST THREE - PLENUM OPEN, TRUMPETS FLIPPED UP........... ~5-20 1.17sec ~5-40 2.35sec ~5-60 3.65sec ~5-80 5.51sec ~5-100 7.51sec.

From this, I conclude the the AFM is restrictive for power over a wide RPM range, AND the Silvertop Plenum is restrictive, being too small in volume limiting the amount of air that can get to the trumpets.

Seems like a pretty plausible conclusion - the results definitely indicate a power increase across the board, good work on putting some scientific testing to it to quantify you experimentation, and thanks for sharing

As that have been stated on previous posts, that's science. Disprove my evidence and conclusions with science - NOT OPINION.

This is the most scientific post so far - imho, and you asked for opinions in the first place. I'm not taking back any of my concerns, I suggested adding the science to it (as above) to quantify the changes, and the 5-10% increase your testing indicates may have been seen at full throttle is solid but also very unlikely to knock seconds off lap times by itself. This gives a good idea of the restriction, if measured accurately and I don't find your findings hugely surprising

The thing I don't particularly like (as I feel I made clear is my only concern) is the part throttle implications in terms of tuning however I understand you are happy with the performance, tuning wise I am OCD and that's not going to change so probably a moot point. I've seen tunes with aftermarket ECUs which are probably no better, if not worse I guess.

[Mr. Mainstream]

Good on ya Flannelman,
At least your doing it

Dont let these keyboard warriors discourage you
Id love to see what other results you have discovered.

[Flannelman]

Possibly the biggest discovery is the CFM flow rates for a stock Silvertop throttle body. Its a phenomenal 270cfm. My flowbench is a floating pressure drop one, so what this means as compared to a steady state it reads LOWER. (270cfm for a 4 cyl is equivalent to 270Hp) The ST manifold is good for only 210cfm with throttle in place.

Compared to a Blacktop, it has 233cfm. The improvement in cfm V flow area are virtually the same.

BUT When the Silvertop manifold is ported to Blacktop specs, cfm goes up but area in throttle bore stays the same. The result is increased airspeed. The engine responds with more torque everywhere in the rev range.

So, don't fall for the "bigger is better" mentality. If so, your engine will feel very flat below 5K.

[RomanV]

Hey Flannelman.

Interesting thread, I've done a lot of investigating factory ECU behaviour via OBD2 monitoring (which doesnt work for silvertop unfortunately)

Honing out the intake runners though after the throttle plates makes for a lower airspeed towards the port, not higher.

Regarding the relationship between airflow meter and TPS for determining ignition/fuel.
Think of the airflow meter as a high accuracy reactive sensor.
Then the TPS as the low accuracy sensor that anticipates change.

Since the airflow meter can only RESPOND to changes after the engine has already started sucking in extra air, when you mash the throttle open there is a momentary lean spot.

So the TPS (because it measures the cause of airflow change, throttle angle) tells the ECU to tip in some more fuel to 'fill the gap' until the AFM catches up.
It also does the same but opposite on decel, it cuts fuel when the throttles snap shut even if the AFM is still registering a higher reading.

I'm not sure if the earlier ECUs do this, but i can tell with OBD that:
There is a long term fuel trim in place, that can adjust a 'cell' (rpm vs load) up or down by 20%.

If you've got the AFM mashed open, then it will be adjusting this high load load cell only for long term fuel trim, under all conditions. So when you are cruising along, and it's running too rich, it will be trying to lean it out.
Then when you go full throttle, it might still be trying to pull this 20% fuel out when there's actually high load...
Which could be making it run dangerously lean. Or more to the point, running leaner could be what accounts for the power increase, rather than the airbox being removed.
I guess one way you could isolate this further is by holding the airflow meter flap open, while the airbox etc is still attached. If the car is still faster, you know it's fuel/ignition related rather than airbox related. In which case money spent on an ECU is well spent.


If you're able to afford trackdays but not engine mods it's probably best to either:
-Leave it alone/standard, for maximum reliability
-Skip a trackday or two and spend the money on an ECU.

If you want a more scientific method for quantifying your accelleration results apart from a stopwatch, there are apps you can download to a phone or tablet that can use the GPS to calculate accelleration rates etc for comparison.
Or at the very least, conduct the test in a 3rd gear pull or similar that puts it under higher load for longer.
If you've got a 0.1 second margin for error, (or whatever) spreading this out over a longer time period reduces the error percentage.

[Stu-]

Fck thats an awesome post Roman

[RomanV]

Cheers!

While we're somewhat on the topic (a bit off topic from first post)

I've had a few different scenarios with my car, that I've been able to analyse with OBD2 to see what the ECU thinks which has been interesting.

The first problem - a persistent knocking under light load.
It seemed as though the airflow meter was not getting enough of a reading, so I was thinking it might be running slightly lean and advancing the ignition too much.
I had an 'Aha' moment... I would use a piggy back air fuel computer to increase the AFM signal. Which adds fuel, and removes advance.
So this added fuel... which the ECU took back out via fuel trim. But then it must mean the ignition would retard for the extra load? Nope!
It was still wanting to run right to the limit of knock. The actual problem turned out to be that I had the wrong knock sensor fitted (engine with different bore size/resonant frequency) so since the ECU wasnt seeing any knock it was all like SWEET AS BRO and chucked in heaps of timing.
Fitting the correct knock sensor resolved the issue.
But the point here is, if you ran some magic space fuel that you could advance the bejeezus out of the timing for, you'd make more power and get better economy automatically.

Second problem...
Car started spluttering and sounding like it was running lean or not getting any fuel.
I check the OBD2, and find that it's trying to add 20% fuel across the board... Hmmm trying to add fuel but none is getting there, sounds like a fuel filter problem. Sure enough, check the fuel filter and it was completely blocked. Correct diagnosis without even opening the engine bay!

third 'problem'...
One my injectors packed up, and the only other injectors I had were 13% bigger.
Fitted them up, drove the car for an hour or so and the ECU showed that it had pulled about 13% fuel out of the load cells I'd been driving in. Awesome! It self 'diagnosed' bigger injectors and adjusted to suit.

Fourth problem...
An exhaust leak ahead of the oxygen sensor.
I knew there was a leak, because there was an annoying noise. (it kept blowing a gasket at trackdays)
But one thing I noticed when there was a leak, was that my fuel trim went from -13% (to suit injector size) down to -2% or so.
The leak made the engine think that it wasnt running rich anymore. My fuel economy on a trip around Coromandel was 10-11l per 100km.
When I remade the exhaust to fix this leak, while cruising the car now gets a consistent 7-8l per 100km with no other changes, and ran a crapload better.
Oxygen sensor is super important to getting the fuel right! It's pretty amazing how well a factory ECU can tune with a narrowband sensor.

Thinking about it though, a factory ECU needs to work with any combination of slight manufacturing /wear and tear variations in: fuel pump > fuel filter > injectors > fuel pressure regulator > oxy sensor.

If each of those had a 5% variation then it makes sense that a factory ECU works pretty hard to try figure out the state of affairs on a given engine.
As just one inflexible tune would turn to custard on a certain percentage of cars, straight off the factory floor unless they were conservative with timing and fuel in a way which would hinder the rest in terms of emissions/performance/economy.

[Lith]

So this added fuel... which the ECU took back out via fuel trim. But then it must mean the ignition would retard for the extra load? Nope!
It was still wanting to run right to the limit of knock. The actual problem turned out to be that I had the wrong knock sensor fitted (engine with different bore size/resonant frequency) so since the ECU wasnt seeing any knock it was all like SWEET AS BRO and chucked in heaps of timing.
Fitting the correct knock sensor resolved the issue.

Very nice posts, sir - love seeing good analysis and sensible conclusions being done and shared, adds some quality science to this thread and improved it's value substantially

One thing which raises my eyebrows a little bit about what your findings in this quoted bit (not a reflection on you, what you've seen and concluded seems very sensible) is that it implies the stock map is basically over-advanced and the ECU relies on knock detection to dial things back until it's "OK", running on the assumption that ECUs don't tend to add timing automatically so much as have a starting point for fuel and ignition and trim timing back as needed (there isn't a sensible way of determining if more should be added) and do some closed-loop and long term fuel trim tweaking as things truck along. Makes it seem either the base tune is either ruthless for our fuel (or fullstop), not matched to the engine, or the base timing isn't matching what the ECU is expecting?

I do feel like I have heard that BT 20vs have partial throttle drive-ability issues when in pretty much stock form on factory ECU, and a lot of tunes (in general) which I've picked up that previously had some "mystery" hesitations are often over advanced in light load throttle areas, though I'd actually assumed that with BTs it was to do with how they'd dealt with calculating load with an unreliable MAP signal.

[RomanV]

When I fitted the correct knock sensor, the ignition advance at low load pulled back by 5-10 degrees and audible light knocking was gone. And idle became stable when previously the car would stall.

It fixed an idle issue, hesitation issue, and improved fuel economy.

I was more than surprised by how much of a change it had!

It might act more conservatively at high rpm/load though, most of my observations have just been from cruising speed... As that's where I've had the problems, but it's also where it's easiest to monitor because the engine conditions are stable.

Every engine has an rpm range which is a dead zone for the knock sensor, the frequency of the motors vibrations matches exactly that of knock.

On 3S engines its 6000rpm or so from memory.

So it cant rely entirely on the knock sensor for establishing ignition advance in that rpm range. BUT

It could possibly draw conclusions from low rpm/load ignition advance about the fuel being used. Use it to effectively find out an octane value of the fuel and then spam it's predictions across the rest of the map. (Pure speculation here, its also possible that it just reverts to known values when it's above a certain load/rpm)

Unfortunately my OBD2 program doesnt give any indication of the knock sensor activity, or VVTI advance so there's not much more to be learned about its logic compared to the injectors.

[RomanV]

Hmmmm so I wouldnt want to test this on my beams motor because it seems to run like crap on anything but 98 anyway.

Bbbuuutttt my Toyota Echo with the 2nz engine runs a 10.5:1 comp ratio, incredibly happily on 91 octane.

I guess I could datalog some load/rpm points and check the ignition advance. (It's a lot easier to do full load/high rpm engine testing with a car that only makes 80hp haha)
Then fill with a tank of 98 octane and see if the ignition advances further under the same load and rpm conditions. (My guess would be yes)

[Lith]

What I meant is I don't like this idea or theory that an ECU would try and find a timing curve which is near the limit of detonation for that engine on that fuel - I can't see why a manufacturer (or anyone) would employ such a strategy... if I am interpreting what you are saying correctly.

[RomanV]

Hmmm yeah, you can make observations but theres no way to really know any more about it unless you did some pretty extensive datalogging.

But some casual observation is interesting for me, because I want to get an aftermarket ECU at some stage but I dont want to go to anything worse than the factory one.

So there's benefit to understanding its logic, and makes it super easy to put together a base map in terms of ignition advance when you know what the factory computer does at various rpm/load combinations.

[Lith]

Hmmm yeah, you can make observations but theres no way to really know any more about it unless you did some pretty extensive datalogging.

But some casual observation is interesting for me, because I want to get an aftermarket ECU at some stage but I dont want to go to anything worse than the factory one.

So there's benefit to understanding its logic, and makes it super easy to put together a base map in terms of ignition advance when you know what the factory computer does at various rpm/load combinations.

Yeah I definitely get where you are coming from - I've definitely learnt a bit, confirmed ideas I had and also come up with a few new ideas (or actually in some cases, reasons) for things by monitoring what factory ECUs do and how they have dealt with things and it has been quite interesting... often if nothing else, giving me a question to investigate I never realised I needed to find an answer for and end up finding whole new cans of worms

Definitely good to keep your eyes open and watch what OEM do to get answers, but do bare in mind that there will be multiple reasons for some of what they are doing and not all the actions you see are necessarily going to be clear or relevant to technology these days.

[RomanV]

Yep I know what you mean... Especially when OEM is generally concerned with a larger number of variables than just making power.

Either way, it's something interesting to nerd over that doenst cost me any money

Slightly relevant to to thread. But since remaking exhaust and replacing dodgy fuel pump, did some GPS datalogging.

Over a specific speed range compared to GPS data from a previous trackday the car is now consistently 6% quicker.

Which equates to, that I was losing ~10hp because of either my crappy exhaust, running lean, or both.

Looking forward to seeing whether this translates to an improvements in laptimes or not haha.