Harlan's Impossible Turbo Build 2
10-18-2016, 07:10 PM
#26
Here are a couple snips from the drive home today. A 4th gear and a 6th gear pull to see where boost is coming up.
While it's not a kick in the kidneys like the old turbo, but it's a very subtle added power and it follows the engine up as VE starts getting better. Got a low end boost spike around 2500 rpm that is showing up on both Mani and turbo outlet... dunno maybe SSV opening? Odd...
Need to do some more analysis and tweak the boost controller so it will modulate better instead of flying open and closed.
What do you guys think? I think it's a keeper.
While it's not a kick in the kidneys like the old turbo, but it's a very subtle added power and it follows the engine up as VE starts getting better. Got a low end boost spike around 2500 rpm that is showing up on both Mani and turbo outlet... dunno maybe SSV opening? Odd...
Need to do some more analysis and tweak the boost controller so it will modulate better instead of flying open and closed.
What do you guys think? I think it's a keeper.
10-20-2016, 07:11 PM
#27
Here's a video of me testing the piggyback timing control on the car with an oscilloscope. I have the new coils installed now but not connected, hopefully tomorrow I'll get the new coils connected and test them for real. Then it will be time for ion sensing. I also have fuel pump pressure control tested at 55psi in preps for a permanent control system and plumbing modifications, and that video should be out next week.
Currently working 12 hour shifts and only getting 1-2 days off per week, so free time is tight, but I'm getting things accomplished so it's motivating me to do more.
Currently working 12 hour shifts and only getting 1-2 days off per week, so free time is tight, but I'm getting things accomplished so it's motivating me to do more.
10-21-2016, 05:54 PM
#28
Spent my one day off trying out my new ignition coil mod and testing ion sensing. Went great other than shocking the crap out of myself with an ignition coil 
, and spending 4 hours chasing a problem that turned out to be a wire on the wrong terminal.
I was able to adjust leading and trailing timing from my phone, and I could see the peak pressure point advance and retard with timing. It's just a rough test setup right now, so no driving yet, but a couple more days off and I should have something use-able on the road.
This was my last big hurdle before designing/printing a permanent PCB, so that's coming soon too.
I'll post some pictures when I have time, and I should have a video of the test out in about a month. I have several other videos to release first.
, and spending 4 hours chasing a problem that turned out to be a wire on the wrong terminal.I was able to adjust leading and trailing timing from my phone, and I could see the peak pressure point advance and retard with timing. It's just a rough test setup right now, so no driving yet, but a couple more days off and I should have something use-able on the road.
This was my last big hurdle before designing/printing a permanent PCB, so that's coming soon too.
I'll post some pictures when I have time, and I should have a video of the test out in about a month. I have several other videos to release first.
11-04-2016, 08:55 PM
#32
Just got this edited and uploaded. Here is my first test of the new coils and coil driver with the piggyback ECU. Spoiler it works but I got shocked.
Still need to get my ion sensing video up, and even more important I need to get work done on the car.
Still need to get my ion sensing video up, and even more important I need to get work done on the car.
11-05-2016, 08:33 PM
#34
It's all fun and games until someone gets hurt then it's hilarious. I guess I did know the engine would run without a single spark plug lead, but I'd never really thought about it.
Got the next video edited up, showing ion sensing and how advancing/retarding timing can move the peak pressure point in idle. I'll set it up to release next Thursday. Hopefully by then I'll have more accomplished.
Got the next video edited up, showing ion sensing and how advancing/retarding timing can move the peak pressure point in idle. I'll set it up to release next Thursday. Hopefully by then I'll have more accomplished.
11-06-2016, 09:08 AM
#35
Love it Harlan! I'm glad you didn't edit that out 
I love your videos, and not just for the humor, please keep it up! The tech is a lot of fun. I've been digging into some more techie things lately and really enjoying it.
I love your videos, and not just for the humor, please keep it up! The tech is a lot of fun. I've been digging into some more techie things lately and really enjoying it.
11-10-2016, 05:43 PM
#36
Got my first successful attempt at ion sensing on youtube. It reads the peak pressure point, but it averages it slowly. I worked on it some more today, but I'm having trouble getting it to respond quickly and smoothly. More work required before I try closed loop timing. In other news tested the piggyback ignition all the way up to redline and no misfire/surprise sparks.
11-13-2016, 06:55 AM
#38
I tweaked around with the amplitude by changing resistors, and now I have PPP indication throughout the rpm range, and got my first successful set of logs while driving today. Looks like stock timing is pretty close.
It's also tracking quick enough to try closed loop timing on my next day off, so that's coming along quickly. And I'm going to transfer my knock detection from the knock sensor to the ion sensing output.
The setup is behaving nicely at 6psi. Because of the size of the turbo it's a little gentle to build boost, but it does not feel choked at the top end, and it maintains boost in the power band. I think 8psi will be the real go/no go for this turbo. It may still be oversized for my HP goals...
Also I need to restore the cruise/economy mode because it is seems to be a couple MPG hit having the WG closed vs open in cruise.
It's also tracking quick enough to try closed loop timing on my next day off, so that's coming along quickly. And I'm going to transfer my knock detection from the knock sensor to the ion sensing output.
The setup is behaving nicely at 6psi. Because of the size of the turbo it's a little gentle to build boost, but it does not feel choked at the top end, and it maintains boost in the power band. I think 8psi will be the real go/no go for this turbo. It may still be oversized for my HP goals...
Also I need to restore the cruise/economy mode because it is seems to be a couple MPG hit having the WG closed vs open in cruise.
11-14-2016, 10:02 AM
#41
I'm about halfway to having an ECU replacement. Pondering if I want to go the other half, I think lean burn will make that decision for me. I really wish there was a better platform to start from. I may start playing with custom code on a megasquirt.
Coding for closed loop timing this morning. I'm setting up in code to control front and rear housings separately and have separate timing trim on all 6 rotor faces. It may be overkill, but coding it is straightforward and mostly cut and paste.
After I have it idle tested and road tested (I plan on a 5deg adv/ret limit for initial testing). I'm going to set up a lean burn map and see what happens to EGT and fuel economy.
I'm not revealing my timing strategy yet, but I doubt anyone has tried what I'm about to push. I'm still searching for the 20% lost HP of a rotary vs a piston engine. I'll let you guys know if I find it.
Still need to work on fuel pump pressure control. But my time is still limited right now.
Coding for closed loop timing this morning. I'm setting up in code to control front and rear housings separately and have separate timing trim on all 6 rotor faces. It may be overkill, but coding it is straightforward and mostly cut and paste.
After I have it idle tested and road tested (I plan on a 5deg adv/ret limit for initial testing). I'm going to set up a lean burn map and see what happens to EGT and fuel economy.
I'm not revealing my timing strategy yet, but I doubt anyone has tried what I'm about to push. I'm still searching for the 20% lost HP of a rotary vs a piston engine. I'll let you guys know if I find it.
Still need to work on fuel pump pressure control. But my time is still limited right now.
11-14-2016, 05:01 PM
#43
It now idles on closed loop timing. It's not perfect, but it works. The biggest problem is I don't have fuel trims dialed in so the idle is loopy to begin with. Meh. Won't matter when I go to open loop fueling.
Wasn't feeling good enough to record a video, but eventually I'll get there. Still working on the ion sensing signal strength.
Next big step is to take it on the road with closed loop ignition. And then I need to install my fuel pump pressure control mod.
Wasn't feeling good enough to record a video, but eventually I'll get there. Still working on the ion sensing signal strength.
Next big step is to take it on the road with closed loop ignition. And then I need to install my fuel pump pressure control mod.
11-14-2016, 07:45 PM
#44
Oh and another note, I implemented closed loop timing on both front and rear housings and it's bizarre. They both kinda chase STFT which is to be expected, but they also oscillate independently. I need to get individual rotor face timing implemented, because this is just odd.
It almost looks like the shudder cycle from a misfire. Like one housing/one face is loaded up with more fuel than the others, and then it burns off and another takes its place. Weird.
Also it doesn't seem to like to idle with the PPP at 45ATDC. Might have to reduce it for idle closer to the stock 40ATDC, or maybe the stock is more for emissions and I just don't have STFT stable enough. Oh well, when I put fuel in Open Loop we will find out.
Added: Just went through several iterations of closed loop. And had nothing but trouble. Idle was hunting, and it just kept getting worse. The PPP kept getting more advanced even with 15deg of retard. I just threw my hands up in the air and removed the closed loop code. And it kept having the same trouble. That's when I discovered in all my testing at idle, and start/stops of the engine I had flooded it and I was chasing my own tail.... More testing later after I've cleared all the gasoline from my housings.
It almost looks like the shudder cycle from a misfire. Like one housing/one face is loaded up with more fuel than the others, and then it burns off and another takes its place. Weird.
Also it doesn't seem to like to idle with the PPP at 45ATDC. Might have to reduce it for idle closer to the stock 40ATDC, or maybe the stock is more for emissions and I just don't have STFT stable enough. Oh well, when I put fuel in Open Loop we will find out.
Added: Just went through several iterations of closed loop. And had nothing but trouble. Idle was hunting, and it just kept getting worse. The PPP kept getting more advanced even with 15deg of retard. I just threw my hands up in the air and removed the closed loop code. And it kept having the same trouble. That's when I discovered in all my testing at idle, and start/stops of the engine I had flooded it and I was chasing my own tail.... More testing later after I've cleared all the gasoline from my housings.
Last edited by Harlan; 11-14-2016 at 09:38 PM.
11-16-2016, 03:56 PM
#46
Still not getting closed loop timing to work well. I think the issue is caused by my choice of reading the leading plugs, and how late the leading plug fires/flame propagation at idle.
Since I want this to work with both positive and negative split in all conditions, I'm kinda at a loss. Next step will be to try reading both leading and trailing plugs at the same time and hoping for a clearer result.
The ideal is that the spark plug fires, the flame front develops, and you read the ion signal after everything has calmed down. This should give a simple curve that's easy to read. So I either need to get to that point, or develop more advanced signal processing.
Since I want this to work with both positive and negative split in all conditions, I'm kinda at a loss. Next step will be to try reading both leading and trailing plugs at the same time and hoping for a clearer result.
The ideal is that the spark plug fires, the flame front develops, and you read the ion signal after everything has calmed down. This should give a simple curve that's easy to read. So I either need to get to that point, or develop more advanced signal processing.
11-18-2016, 05:42 PM
#47
Decided to do some logging and graph out what was going on. Glad I did! The signal is pretty clear when averaged out over many cycles. Here are three plots, first is stock, followed by Trailing +11 and Leading -5, and then Trailing -11 Leading +5.
It looks very pretty in this form, and it's easy to see where the PPP is. Stock is about 45deg.
The Leading -11 shows it far to the left but masked by the coil firing.
The Leading +11 shows the actual event, and it's far too advanced because of the extra trailing timing.
Now I just need to tweak around with how it averages and try to get data quick enough to try closed loop again.
It looks very pretty in this form, and it's easy to see where the PPP is. Stock is about 45deg.
The Leading -11 shows it far to the left but masked by the coil firing.
The Leading +11 shows the actual event, and it's far too advanced because of the extra trailing timing.
Now I just need to tweak around with how it averages and try to get data quick enough to try closed loop again.
11-18-2016, 06:20 PM
#48
Decided to do some logging and graph out what was going on. Glad I did! The signal is pretty clear when averaged out over many cycles. Here are three plots, first is stock, followed by Trailing +11 and Leading -5, and then Trailing -11 Leading +5.
It looks very pretty in this form, and it's easy to see where the PPP is. Stock is about 45deg.
The Leading -11 shows it far to the left but masked by the coil firing.
The Leading +11 shows the actual event, and it's far too advanced because of the extra trailing timing.
Now I just need to tweak around with how it averages and try to get data quick enough to try closed loop again.
It looks very pretty in this form, and it's easy to see where the PPP is. Stock is about 45deg.
The Leading -11 shows it far to the left but masked by the coil firing.
The Leading +11 shows the actual event, and it's far too advanced because of the extra trailing timing.
Now I just need to tweak around with how it averages and try to get data quick enough to try closed loop again.
11-18-2016, 06:28 PM
#49
Yes stock is closest to correct, but I kinda expected that in the first place. These graphs are all at idle. Guess I failed to mention that!
I was trying to find out how clean the curve was, and what changing timing would do to the unreadable part of the curve (after the coil fires, but before it makes a nice curve.)
The more advanced the leading timing, the better the curve looks and the more readable it gets.
Also advancing either timing moves the PPP to the left as expected.
Next I need to fix the code that finds the PPP in degrees so I can use it to drive closed loop timing. Hopefully I can get it to read quickly and accurately.
I was trying to find out how clean the curve was, and what changing timing would do to the unreadable part of the curve (after the coil fires, but before it makes a nice curve.)
The more advanced the leading timing, the better the curve looks and the more readable it gets.
Also advancing either timing moves the PPP to the left as expected.
Next I need to fix the code that finds the PPP in degrees so I can use it to drive closed loop timing. Hopefully I can get it to read quickly and accurately.