Running a little behind this week. We are kind of swamping with orders on the online store and Brain got a call from his mother that things were not good on the left coast. So he drove out to Californy to the Hills of Beverlee. I’m stepping and fetching like Rochester and kind of ran out of the week of the 13th, Jack Rickard, airspeed, and altitude all at the same time.
Many thanks to a special visitor this week, Royce Wood of K&R Performance Auto of Westerville Ohio. K&R stand for “Karen and Royce” of course and he did bring his lovely bride along. He also brought his Genesis Scanner along to plug into the Elescalade.
[jwplayer file=”news071312-iPhone.m4v” hd.file=”news071312-1280.mov” image=”http://media3.ev-tv.me/news071312.jpg” streamer=”rtmp://s3einxnpkaij93.cloudfront.net/cfx/st/” provider=”rtmp” html5_file=”http://media3.ev-tv.me/news071312-1280.mov”]
The results were mixed, but getting better. We were able to “see” our MAF and MAP inputs as something other than a voltage (grams/second for MAF, kP for MAP).
I actually have ANOTHER Cadillac Escalade. ALthough an ESV, it has the same basic engine and layout as the EXT. So I wired up a little 10v voltmeter on the MAP input of both and then drove them both. What I found by comparing this is basically what several of our viewers have been saying all along – Manifold Pressure is more akin to current or power in the electric motor than it is to RPM.
The voltage on the ESV would be about 1.5 to 1.8volts at idle. If I put it in gear, it immediately jumped up to 2.0 or 2.1volts. And if I took off it would go to a higher value as a function of how briskly I accelerated. Three volts for moderate and as high as 4.0v for brisk acceleration. But what I noticed was at 30 mph and high rpm, if I took my foot off the accelerator, the voltage plummeted to 0.20 or 0.30, much lower than the book would imply, and then recovered gradually as the engine came under load. So the “feel” is one of “load” on the engine – NOT RPM.
So I redesigned the circuit.
I went from a pulse counter chip to a different chip entirely – the TI912IN CMOS Operational Amplifier. And we made some serious changes to Jeffery Jenkins RC filter input. I tied a 2.7k resistor from the signal to ground and a 240 ohm current limiting resistor in series with the input. I also increased the value of the capacitor to 33 uF and did away with the pulse clipper zener diode.
In this way, I can use the full 12v PWM signal into the RC filter. By using a larger capacitor I kind of increase the integration function of the filter into the input. This should give me a higher dc level at the lower pulse widths we have during idle.
The operational amplifier can use any power from 5v to 18v. I powered it using the 12v supply from the Soliton rather than the 5v supply from the ECU. Again, this would naturally result in a higher output from a lower input. The ECU expects something 0-5v but the problem I was having was a low voltage at idle. So we put a 5.1v Zener diode to clip the output to no more than 5v or so.
The heart of the circuit is a 5k potentiometer and a 470 ohm resistor forming a feedback network to the inverting input of the opamp This lets me adjust the gain using the formula GAIN = 1 + POT/470. So I can have some amplification of the signal, but adjust that from 1:1 to anything up to the 12v supply. The idea is to get more definition at lower power levels. The higher power levels kind of taking care of themselves and limited by the 5.1v zener.
What Royce saw on his scanner was a pretty good MAP simulation at that point. But the MAF simulation was a disappointing 2 gm/sec that just didn’t change.
So we changed the input to the Voltage Controlled Oscillator to look pretty much like what I did with the MAP sensor, albeit with a smaller 22uF capacitor. We took the series 2.7k out and put it instead to ground. We removed the clipper zener. ANd we changed the cap to 22uF. I wold probably like to replace the 240 ohm resistor with a 500 ohm pot frankly but I didn’t’ get this done – no 500 ohm pot laying around.
With the truck up on jackstands, Royce seemed happy with the MAF and MAP sensor inputs, but the ECU was not. He was also having troubles keeping his Genesis talking to the ECU. I suspect noise on the 12v supply.
We played around and a gained determined that it drove the wheels better with the brakes on, implying that more load was a GOOD thing for one or both of these signals. Since I have MAF tied to the POWER output of one Soliton1, and the MAP sensor tied to the CURRENT output of the other Soliton, this kinda/sorta makes sense.
But at some point, the ECU would slam the throttle plate shut and that would be that.
We also played around with a couple of values in the Soliton. The pedal deadband value was set at 5%. Deadband is the percentage of throttle position that the Soliton ignores. In this case, the first 5% of the voltage calibrated as min to max just doesn’t do anything. The ECU clearly did NOT like the die that throttle could be applied at all without a corresponding increase in RPM. So we set that to zero.
The other setting is the 50% current setting. This allows you to set the percentage of total current the SOliton will apply to the motor when the throttle is at 50%. By setting this to 40% or 30%, for example, you can “spread” the sensitivity of the lower end of the throttle where half the throttle throw distance controls the first 30% of the current available. The remaining 20%, by implication, goes to the second half of the pedal.
This makes the car much more controllable at low power levels, i.e. parking, than it does at high power levels where you’re basically squashing the hog with your foot to make him squeal.
So we played with that setting a bit.
All of this DOES change the reaction of the ECU. Sometimes better. Sometimes worse.
After Royce left to return to Westerville, I worked on a couple of things.
First, I took it off the jackstands to see how it drove (of course). Since it liked a load from the brakes, it out to be ecstatic to have a load from the ground.
Actually this was the closest we have come to driving. Not only will it idle around, but you could actually drive off with a little accelerator. I got it up to about 30 and even shifted into second gear, but it did pretty quickly shut down.
Then I added a 100 uH inductor to the input of both power supplies. As i said, Royce seemed to find the connection to the ECU very sporadic and alluded to power supply difficulties. I thought actually installing Jeffery’s ripple suppression choke on the input to the DC-DC supplies was pretty much in order.
I have some blue LED lights in the battery box we haven’t had on much recently. So I connected those while I was back there. This caused the 12v supply to dive and in playing with THAT I managed to kill the MAP sensor circuit. Suddenly it had one output – 5v and it was NOT adjustable. So I had to solder up another little circuit board and replace that.
Drove the car. It was marginally better behaved but just marginally. Still not there.
So basically we got our ass kicked on this one and went home losers this week. I don’t like that. I just take losing badly. With the convention coming up soon, and some other projects pressing in on us, Im feeling the clock failing on this Elescalade.
I do have an ace up the sleeve though. Six months or a year ago I ordered a product called VCM Suite from HPTuners. I really don’t like these guys. They have their product all locked up and you have to buy “credits” to use their software after you buy their software. The product support is terrible but the worst part is just the attitude of the whole company. It is completely run through with negativity and they are just SURE you are going to steal their hard work and boy have they got ways to stop you. Authentication codes and serial numbers and so forth.
I grow weary. I started with Bill Gates diatribe about software piracy in the 1970s when he found people were making pirated copies of his BASIC interpreter that he was selling in baggies on paper tape. And I’ve had to watch this play out with company after company after company for over 30 years now.
Each new generation of software programmers is willing to do battle with their customer base in a heroic wrestling match over who owns what and whether they paid for it.
The outcome is ALWAYS the same. Companies that engage in this, rarely accomplish anything with the software pirates. But they DO annoy the shit out of their paying customers.
And so what happens is, kids and people just playing with things STILL manage to purloin the software. But the anti-piracy tactics drive away the paying customers. The irony is the pirates were never going to buy the software or pay for anything anyway. And so the company pays the penalty, but never gains a benefit. I have not seen a single company survive this although Adobe comes perilously close.
The bottom line is that this is not the way to succeed in software.
But I did pull the program out and install it on a new Vista laptop. And I did begin the long tortuous learning curve of yet another hopelessly detailed software application.
This one is hopelessly detailed not only from the usual comical attempt at an innovative interface, but the ECU truly is a kind of a little puzzle box.
But I learned a couple of things without even hooking it up to the truck. First, we having somehow, by dumb luck or animal cunning, instinctively homed in on precisely the right two signals. This ECU is ALL ABOUT MAP and MAF. And I mean ALL ABOUT. All remaining signals get a passing mention.
The program is actually enormously capable. It can read not only the data tables out of the ECU, but a flash image of the entire EEPROM to a file on disk. You can then access the data tables, but you can also essentially rewrite the operating system if you are sufficiently familiar with it.
You can reflash the VIN, turn on or off the antitheft function, we will EASILY be able to turn off both the check engine light and the text display advisory for any DTC. You can basically recode the whole ECU and write it all right back into the ECU – operating system and all if you like. Or just the data tables.
We can change the shift points in the transmission controller, the torque converter lockup, and even the speed/hardness of the shifting for any particular gear at any particular speed. And you can do that separately for WOT (wide open throttle) or other modes of the throttle.
Best of all, it APPEARS that the values reported by the MAF and MAP sensor are actually done with a lookup table, sometimes TWO of them for each. So you can redefine how many kP of pressure is represented by how many volts. Or how many gm/sec is represented by what frequency. By having the data in a table, you basically paint a “curve” of response across the sensor output.
This is really what I needed. With SOME output VAGUELY representing motor power or current, I can modify that curve ENDLESSLY to whatever we need it to be.
The problem of course is, unlimited power to change it, presents unlimited opportunity to f***k it all up. I could spend a lifetime tuning this thing. And of course everything is interactive.
It’s kind of like our motor controllers, with 1500 crucial variables.
All we have to do is make some changes, test drive, make some changes, test drive, and continue that process reiteratively perhaps 1500×1500 times.
Nonetheless, I now have some confidence we can get the vehicle running and shifting the transmission. We can even control the lights on the dash and the text advisories. We can probably even CHANGE the text advisories.
Time and patience are a problem here. I wish I had someone knowledgeable to hand this off to. It’s very doable. But very time consuming.
Worse, we’ve wandered off into an area that doesn’t make good video, and is so deeply technical I fear no one will care. This is beyond watching paint dry. It’s just technotrivia of the deepest sort at some level.
The bottom line is, this CAN be done with modern cars using available tools, and probably anyone can do it. Big learning curve on the tools. But the ability to actually entirely reprogram the ECU on almost any car is there. And I’m kind of obligated at this point to success with the Elescalade.
But we face a very real problem at EVTV that reminds me of the old saying: “THE HURRIER I GO THE BEHINDER I GET”
I would love to do the lawnmower right now. Simple project. Sufficiently inexpensive anyone could do it. It mows your lawn.
The THING. Simple project. This may be the BEST platform for an electric car in the world. Room for batteries, standard VW rear end, seats four adults. Etc.
B&B Manufacturing is pregnant to deliver us a CARBON FIBER molytube wide bodied Speedster that will weigh less than a thousand pounds rolling. We can probably do a 1600 lb Speedster Light with a 50 mile range and an affordable price.
I’ll reward our handful of blog readers with a glimpse into coming weeks. We have entertained a group from Tokyo. They have purchased Speedster Duh and put down a deposit on a second unit. They want to order five more by year’s end. And they are hell bent on selling electric Speedsters in Tokyo and hoping to make big noise at some car show there in December.
So with Brain out of town this week, you can seem I’m feeling a little pressure here.
John Hardy is helping the cause with some pretty impressive battery testing. We introduce him in this episode. He alludes to some graphs that are publicly updated or streamed from his shed to the Internet that anyone can follow. More on that HERE.