fujioko wrote: ... This week I'm building the servo throttle controller and ordering the parts for the servo load controller. I should have the dyno controls automated fairly soon. Stay tuned!
Mind the servo loop damping factor, and watch out for phase change instability.
erohslc wrote:fujioko wrote: ... This week I'm building the servo throttle controller and ordering the parts for the servo load controller. I should have the dyno controls automated fairly soon. Stay tuned!Mind the servo loop damping factor, and watch out for phase change instability.
Also, with that much charging capacity, your dilithium crystal realignment defibrillator might optimize the reversed helix quadrant, so youll want to compensate with a collapsed nutrino-field cascade.
4cylndrfury wrote:erohslc wrote:Also, with that much charging capacity, your dilithium crystal realignment defibrillator might optimize the reversed helix quadrant, so youll want to compensate with a collapsed nutrino-field cascade.fujioko wrote: ... This week I'm building the servo throttle controller and ordering the parts for the servo load controller. I should have the dyno controls automated fairly soon. Stay tuned!Mind the servo loop damping factor, and watch out for phase change instability.
Next build thread?
@ erohsic, Good point. I think the control loop is going to be too slow to realistically duplicate fast changing loads such as gear changes and whatnot. I'm thinking of sorting the real world data into bins and running a profile based on bin data. I'm not sure that makes sense but that is sort of what I want to try.
Stay tuned
Next build thread: fujioko builds a low buck interositor ...
Progress....
Building the interocitor.
The hydraulic system needs to be converted from manual to electronic control. I picked a heavy duty stepper motor plus controller for $60.00 on ebay. The pulleys and belts were from McMaster Carr.
Drilled four holes and assembled some threaded rods to hold the motor plate into position.
The bottom pulley wasn't flanged so A few fiberglass flanges had to be fabricated.
The pulley locks onto the shaft with a set screw. If you stare at the pulley long enough weird things start happening.
A stub shaft is installed to give the multi-turn potentiometer something to grab on to. The feedback pot is for position reference and for manual control. Eventually the load cell will be used to determine the position of the shaft.
All mounted and wired up
20 lines of code and it woks like a charm.
[/URL] The interocitor so far. It features an Arduino UNO, MEGA, load cell amplifier plus two Stepper motor controllers
Stay tuned!
In reply to fujioko:
Ha, GREAT stuff, the spiral on the pulley is Genius!
Thanks!
Forgot to mention the software controlling the stepper motor mimics the classic analog feedback servo loop for manual control of the load. Turing a dial on the control panel will spin the load valve the appropriate amount. I have to write some more code for fully automatic load control but for now the stepper servo loop works well enough to inspire confidence.
The throttle stepper servo is almost done and should be the next update.
Stay tuned
This thread is off the chain. Very cool.
I really like this thread
I think i should start a kickstarter just to pay you to build this full time and to update this thread faster....
....'ya know, all joking aside, I have thought about that.... I love doing this stuff, but for now the day job keeps a roof over my head.
for some reason the quotes don't work on this forum....
Mezzanine asks....
What are your plans for after you get the 1.3 all tuned? Rent the thing out to your GRM buddies?
Good question,
I want to explore the limits of custom tuning, everything from spark plug indexing to deep down the rabbit hole stuff. Throw in a few Smokey Yunic myths and some internet folk lore and I want to give it a try.
This is basically a big toy that Santa clause overlooked, The closer I get to finishing it the more excited I am.
While I have some basic research to do on behalf of the B3 Miata, I'm wide open to suggestions from the hive mind. ummm.... twin turbo intercooled alcohol injected fire breathing B3 with NOS and zoomies would be a little over the top on a shoestring budget but its not out of the question.... just ain't going to happen for a while.
Dyno update...
Massive failure on the throttle servo build. The servo was finished last Sunday and is way to complicated and incredibly slow. It went into the scrap bin.
The parts arrived for the new servo and it should be finished within 48 hrs.
Stay tuned!
The throttle servo was a cast iron bi*ch to sort out. The trouble was due to three different technological hurdles to jump. Physical hardware, electrical hardware and virtual software. .... plus add safety, speed and durability.
The first-generation of the throttle servo started by harvesting a stepper motor and gearbox from a laser printer. A wooden baseball bat was the primary harvesting tool. A complex gizmo consisting of gears pulleys and cables was constructed and but was ultimately scraped. The total cost was $0.00
The second-generation servo uses a powerful stepper motor directly coupled to the throttle shaft. For safety, the stepper motor can be deactivated and the throttle return spring will immediately close the throttle. The associated software takes up 18 lines of code for manual mode. A few more lines of code will be necessary for automatic mode. Total cost was a budget busting $45.00. I know it seems cheap but free is always better when possible.
Load cell and amplifier
The load cell is actually unremarkable in the sense it works right out of the box. The $40.00 load cell and the $23.00 amplifier are wired together and the output is wired into the ardureno. A few lines of code converts the voltage into pounds.
The load arm on the PAU measures exactly 14 inches.. I built a second 14 inch calibration arm on the opposite side of the PAU. The idea is to apply a known weight to the calibration arm and tweak the amplifier (if necessary).
The loadcell is rated for one kilo Newton ...that works out to 224 pounds in dog years or something like that. Metric weights don't make any sense.
Just for giggles, I weighed myself on a UPS scale then climbed a ladder and stood on the calibration arm. Close enough. I figure I can eat 100 more Big Macs and I will weigh enough to full range the load cell or I need to buy a calibration scale. Decisions...
Anyway, the real calibration will be done at some point but for now I'm happy the load cell and amplifier appear to work. Due to the 14 inch load arm, the actual torque generated by the PAU has to be multiplied by .86 to convert to foot pounds. This is just one line of code and never has to be an issue.
At this point I had enough stuff cobbled together to do some basic testing. Getting all the ducks in a row is about as complex as launching the space shuttle but I managed. Long story short, it all works!
With the engine running I can dial in a load and advance the throttle electronically. So far I'm at 65 lines of bullet proof code.
The next step is getting the shaft speed sensor sorted out. I have all the parts in stock however this last hurdle is massive and I'll explain when this thread gets updated again.
Let's look at some pictures...
Getting the parts to build the first generation throttle servo. I had to beat the E36 M3 out of a laser printer to harvest the stepper motor and some gears. If you seen Office Space, you know the song....
Gears and motor in hand
The first generation throttle servo was insanely complex and eventually ended up in the scrap bin.
The second generation servo is simple, fast and safe.
Stepper motor directly coupled to throttle shaft makes life a lot simpler. Flipping a bit in the software will disable the motor controller and the throttle will snap shut in the blink of an eye. Nice!
Meh, no drama no surprises. The load cell and amplifier work perfectly out of the box. Five lines of code and it was time to move on.
This is the nearly complete dyno controller. The knob on the left is for manual load control and knob on the right is for throttle control. The manual controls are for trouble shooting and setting up the dyno. Flipping a switch will disable the manual controls and a PC will automatically take over.
I'm going to try to shoot some video tomorrow and let you folks get a look at this thing in action.
Stay tuned!
So.Dang.Cool!
Thanks!
Had a chance to shoot a quick video of the dyno in action. The environment was a bit smokey and somewhat loud so the video ain't the greatest.
Have a look.
Does the dyno have a fluid cooler or do you risk boiling the fluid during long testing?
how does the dyno account for vicosity change due to fluid temperature if it does not have a cooler? do you have to run it for an amount of time before it is up to temp?
Mad_Ratel wrote: Does the dyno have a fluid cooler or do you risk boiling the fluid during long testing?
Hi 'ya,
The low pressure side has a liquid to liquid heat exchanger that connects to a garden hose. Because Michigan, the garden hose is put away for another four or five months. I reckon a cooling system could be fixed up on the cheap if need be.
The dyno can do 15 to 20 min of run time before the fluid gets close to 200F. I think the danger zone is in the 250F range or something like that... have to check.
Ninja edit..
The viscosity change isn't an issue because the power absorber uses a strain gauge type load cell to measure torque.
The dyno project has made great progress up until the point where driveshaft speed needed to be measured. Shaft/pump speed is measured in frequency and is required for horsepower calculations.
The simple solution is to pick up the engine RPM and divide it by the transmission gear ratio and ta'da shaft speed. The cost would be next to nothing because it's just wire and software... ..After about a month of trying other methods, I finally said berkeley it and did it the simple way.
The problem with measuring the driveshaft speed directly was a slight wobble in the pump adapter. The wobble made the hall effect sensor method unreliable. I next tried optical and it got too complicated.... so then I installed an electric VSS sensor from a '99 Miata. The electronic VSS sensor didn't provide the resolution I wanted so I finally decided to go with engine RPM and a little math.
The reason I avoided the easy button was I wanted to keep the dyno controls independent from the engine controls. My concern is mostly due to noise that is typical in automotive electrical systems. Fortunately the tach signal was extremely clean and for extra insurance I routed the signal through a buffer. The buffer provides a firewall of sorts and keeps the two electrical systems isolated.
A side project was to build a permanent degree wheel that would mount to the crankshaft pulley. The face of the degree wheel was found on the internet and all I had to do was print it out. The printout was then glued to posterboard and run through a lamination machine to make it water/oil proof. The complete laminated face was then glued to some fiberglass sheets and cut out to the right size.
Pictures are more interesting...
Hall effect sensor and magnets was the preferred method of generating a frequency. Unfortunately a slight wobble in the pump adapter made this unreliable. Other methods were tried to get actual shaft speed and ultimately I had to grab the engine RPM signal. A little math in the software is used to correct the transmission gear ratio. The dyno appears to work best when the transmission is in 3rd gear.
VSS sensor provided a great signal but I wasn't happy with the relatively low frequency I was seeing. The VSS generates an AC signal that had to be converted to DC square wave with a few transistors and resistors. This idea was scrapped as well.
VSS slides into place.
A side project was to make a degree wheel
fiberglass, paper and clear plastic makes this wheel indestructible
Some studs were added to the crankshaft pulley to support the degree wheel.
Screwdriver indicates where new pointer will be mounted
for E36 M3s and giggles I took a picture of the spinning degree wheel. The timing light and the camera works great together.
So... we now have shaft speed and torque.....the next picture is the money shot.
^^^^THIS^^^^
The engine was spun up to speed, a load was applied and data was acquired. This is berkeleying awesome!
Ten pounds of E36 M3 in a five pound bag. The electronic junk will need to be re-packaged into a larger enclosure.
Stay tuned!
Awesome work!
If you do any forced induction, you should try using DX cooling for intercooling.
This is the best thing on the Internet.
Thanks!
All that is left to do is some cleanup work and some more code. Pounding out code in an unheated environment sucks but it will mostly be debug on the automatic controls.
The elephant in the room is the indicated 21HP being absorbed. I'm as shocked as the rest of you but this is just the beginning.
Instant quantitative feedback on power and fuel consumption will takes the B3 Miata to the next level. I'm thinking sometime in March I'll have the Miata re-configured for a bump in power and the engine management totally sorted out.
The goal is to achieve 50MPG in the Miata, however I also want as much "free" power as I can get. Sort of like the old saying "have my cake and eat it too" . When the car goes ball to the walls, I want warp speed factor 9. On the other side, when car is cruising, I don't want to waist a drop of fuel. Now, I ain't going to get LS2 type acceleration out of a 1.3L engine, but I think I can improve on the factory Festiva rating.
The B3 Miata has the cold unsympathetic heart of an appliance but is sole is still MX5. As a daily driver I get the zoom zoom of RWD, compliant suspension, and real brakes. The power is adequate but a little moar wouldn't hurt.
Stay tuned!
With the majority of the dyno electronics more or less sorted out, it was time to reorganize and repackage the controller (interocitor). fortunately I have been hording industrial scrap and the expense of this upgrade was next to nothing.
Due to the ever changing schedule during the holidays, it was important to do most of the teardown and rebuild in less than 24hrs. I don't have any schematics and didn't feel like making one. My memory is good enough to hold it all together for a day or so, but after that I start forgetting stuff.
So... I hauled the controller into the house and setup a makeshift workbench.... the rest is a blurrrr.
Pictures tell a better story...
queue up the Mission Impossible theme music
The old enclosure was too small for all the crap. Time to rebuild it.
OMG... what have I done...
This is the new enclosure... well sort of new. I rescued this box from the scrap pile and gave it a fresh paint job.
Parts laid out in a semi logical order. This new controller will have two Arduino 3560 micro computers plus a bunch of other stuff.
Once the lay out was complete, it was time to put it all together. Here is a MacGyver quick tip. You can make a awesome centerpunch from a stainless steel rod salvaged from any CDROM/DVD drive. Before you toss your busted drive in the trash, open it up and grab the rod. You will need to chuck the rod into a drill and dress it on a bench grinder. The point will stay sharp for hundreds of uses. Try it!
This is a great way to bust a tap. However, if you set the clutch just right and fiddle berkeley the forward/reverse switch... you can tap a bazillion holes in record time. Takes some practice but this is how I roll.
This is where I was after five hours. The basic system powered up with no smoke. yippee ki yay.
At the ten hour mark, 90% of the wiring was done.
Day two, the box was mounted to the hydraulic tank. The dyno harness was modified to plug into the box and all that was left was to throw the power switch. No smoke, but no workey..Oh, that's right I need to flash the new Arduino with some code. A little while latter the upload was done and to my amazement it all worked.
The cover looks like this. It will need some paint to look pretty, but for now it keeps the junk covered.
.... And now for something completely different. This thing was formally known as Mr. Fusion. I took the crazy out of it and now it is just an injector that squirts into a mason jar. This is going to be used to calibrate how much fuel is actually being used. It is electronically coupled to one of the real injectors and through a convoluted process I will be able to measure exactly how much fuel is being used. We'll revisit this soon.
The device formally known as Mr. Fusion is mounted and is awaiting some plumbing.
Ho Ho Ho.... Santa brings gifts to the dyno. This is a 100 pound digital scale that will be used to verify the load cell is reading correctly.
The scale hangs on the calibration arm and is pulled down to put weight on the load cell. According the the scale the load cell is a few pounds off on anything less than 15 lbs, but the loadcell appears to be measuring the correct load all the way up to 80LBS. I wasn't able to get past 80lbs because this is really hard to do and I need to MacGyver something up. The good news is the loadcell is putting out reliable numbers.
That's all for now,
Merry Christmas everyone!
This is amazing.
I Love this. You've got skills!
so you can work with fiberglass and electronics and you understand data and shizz.
I bet your next project is going to involve rebuilding an automagic transmission or something too.
Keep up the good work!
My Heros:
You'll need to log in to post.
