A couple quick pics of some stuff around the garage. First is of the installed piston oilers. You can see them if you look close. Second is the current status of the Walbro 400 in tank. Both wires to the pump are 10 gage with gold plated terminals. I'm in process of making my own hotwire kit for the pump and will post pics when done.

Get'm Piston Oil Jets Installed

Walbro 400 installed into stock bucket

 

im interested in measuring size, matching and maybe even characterizing low pulse width. the first two are easy but the low pw stuff im assuming you just run a test at steady pw and measure the volume of liquid and divide by number of pulses, then iterate through until you have enough data points to make a curve?
im just not sure what all it should be able to do. im writing sorta complicated code so i should be able to make it do anything needed.

 

Ok, sweet. I'll break it down a little bit for people not super electrically inclined. I'm still learning as well, so some things might not be perfect. Also, if I'm going to be honest, I didn't do nearly enough characterization at low pulse widths. I needed way more data to be accurate, but it's good enough for now. It's a really time consuming process and I can honestly say that you're probably not going to want to do low pulse width characterization unless you're running 'yuge injectors and want a lean idle and pop-free decel.


To electrically drive an injector, you have to have a few things: 1) A "switch" 2) a controller 3) a voltage source. The voltage source is just the voltage across your battery terminals. The next two, the switch and the controller, get a little more complicated.
Let's start with the controller. In technical terms, the controller is called a "microcontroller." The microcontroller is the device that stores and executes code. (I'll follow up a little on the code after I've finished explaining the hardware.) The code that runs in the microcontroller and controls its outputs has no direct control over the injector itself; the micro outputs a super weak "off" (0 volts) or "on" (5 volts) on it's digital output pins. By itself, the microcontroller doesn't have enough jam to turn on the injector. By jam, I mean that it can't supply enough current to move the pintle (injector controller) without burning itself out. Think of the microcontroller like a person flipping on a lightswitch. The person has to put in very little effort to control the light but can control it whenever they want to.

In order to turn the injector on, you have to feed it a bunch of current (compared to what the microcontroller can supply). To do so, the injector is connected so that one side sits at battery voltage and the other side sits at ground/0 volts. By Ohm's law (V = I * R, or Voltage drop = Current * Resistance), a 12 ohm injector will "drop" all the battery voltage (let's assume 12 volts for simplicity) while drawing 1 amp. Current here = 1 amp = 12 Volts / 12 Ohms.

This is where its going to start getting a little complicated. Electrical current through the injector coils (inductor) is kind of like a river with a dam controlling it. Much like current through an inductor/injector coil, if you take the dam out from in front of the river (switch closes), the water (current) starts flowing. The flow can't change quickly since the water has mass, but over time it stabilizes to a single flow rate. This is your electrical current Now, imagine slamming the dam closed when the water is quickly running. The water will push against the dam and rise higher and higher until it comes to a stop. Think of the water height as your voltage.
So, as your switch slams closed the injector current can't instantly stop, so it creates a large voltage to keep driving the same current across the switch. These are called "inductive kicks" and can be damaging if not handled correctly. In order to handle them correctly, the voltage is "clamped" to an upper threshold using a zener diode. In our dam/water example, the clamp is like a trench dug just before the top of the dam that allows the water to go elsewhere instead of over the top of the dam.
Based on some information I was able to find on Bosch Semiconductor's website (many major OEM's use Bosch components and ECUs; VW, Audi, Ford, etc), the low side drivers that they provide are clamped at 55V. Altering the clamping voltage of the low side switch will alter the closing/seating characteristics of the injector, so it's best to find the clamping voltage of the components used in the Delphi ECU's.

The switch can either be a dedicated low side driver (hooked up on the ground side of the injector) or a MOSFET (metal oxide semiconductor field effect transistor) with a 55v clamping diode and FET driver IC to help with the FET turn on characteristics if you want to get complex.
MOSFETS are the internals for solid state relays, if you didn't know already.

The rest of the mechanical setup is fairly straightforward. Get a fuel pump capable of your desired injector static flow rate plus a good margin, variable fuel pressure regulator (if you wish, not necessary for vehicle setups with a boost referenced FPR), a precise scale, Stoddard solvent (you'll have to correct for this later, but you're less likely to blow yourself up while you're testing), a variable voltage power supply, and an enormous fuel rail or one that has a damper built in. The last thing you want is the fuel in your tester to resonate. Keep local rail pressure stable over the top of the injector and the downstream injector pressure equally as stable.

*Edit* I forgot about the pressure and temperature sensors for your test fluid. If you want to get super accurate and consistent results, you have to monitor your test fluid temperature and pressure so that you always test at the same fluid conditions. The pump itself will heat your test fluid either through recirculation and/or conduction if it's an in-tank setup. If you have a fairly large barrel of test fluid, you can run a few tests before the temperature starts to drift. If it's a small volume, your temperatures can start to creep in the middle of the test. How you control test fluid temperature is up to you and your budget.

Test using the average value of as many shots as your measurement system is capable of holding for the most accurate results. Fuel injectors aren't as consistent as they may seem on a shot to shot basis. I took the average of 1000 shots when I did mine.

Set up your microcontroller code (I'm assuming you're running an Arduino or similar) to loop as fast as possible, check the elapsed time at the beginning of each loop, and use conditional statements to turn on/off the injectors at specific intervals. Don't rely on loop times or iterations to solve for injector opening time; that's too variable.


TL:dr: Use either a combination of a MOSFET with a 55V freewheeling diode and a dedicated gate driver or a dedicated low side driver with an internal 55V clamp to drive your injector. Structure your micro algorithms to take loop time instead of number of loops. Average mg/pulse over as many shots as your scale and container will allow to minimize shot volume variance in your final characterization.

I'll come back and proof read this in a day or so. Spent a couple hrs wiring this up and don't feel like reading through it all.
-Cal

 

i know all that, what i meant to ask was the logic of how to time things and calculate the injector characterization at low pw. is it as simple as run at x microsecond pulse widths until you get to a certain fluid volume then divide the number of pulses by the volume? or is there something more complicated?

 

It's simple but time consuming. If you can measure volume precisely, use volume. Otherwise, use a scale and back calculate the volume vs pulsewidth using density and weight.

 

good info, thank you.

 

Well, the frame's fixed. Took a month, but she's right as rain. Ended up taking the bed off, wire wheeling the rear half of the frame, and repainting it while I was under there.
Since last update, here's what I've done.

• Installed Walbro 400 and DIY hotwire kit. I haven't checked the fuel pressure since install, but based on the lambda values running the old VE map, its very close to what it was. I'm running the flex fuel crossover and reg.
• Installed and routed catch can with inline check valves. Seems like it's working well so far. Interestingly, it seems like its affected part load charge biasing between banks quite significantly. Up to 10% between banks at times.
• Punched a hole in the firewall and installed the 3d printed grommet. De-pinned and repinned connectors through the grommet for the lambda, thermocouple, and DC power hookups.
• Installed the desktop into the slot where the the CD changer used to be. Seems like it fits pretty well. I need to solidify mounting in the next few days.
• Ordered up a few pure sine power inverters for the instrumentation and PC/Monitor along with a monitor stand.
• Re routed wastegate line and installed a smaller spring. I finally have some control over boost. Before, I saw around 22psi (boost, not absolute) for a second...I think I had previously pinched a line or wasn't flowing enough to open the gate in time.

I'll be back with more pictures this week.
-Cal

• 
  Top of populated injector driver board
  
  Bottom of driver board...first row of fets weren't great until I found the right soldering temp.
  
  Populated lambda board, finally.
  

 

Well, it's not the interior pics I promised. But, I moved all of the water meth nozzles to the underside of the manifold. I only got it about half done but wanted to keep this up to date, ish. Next steps are to run the rear 4 cylinder hard lines and design a manifold to connect them all together.

The carb jets threaded into tapped AN to NPT adapters should allow me to dial in things a bit better for each cylinder. I decided to stray away from the individual check valves and put the check valve elsewhere.

50% completed water injection installation

 

Half way to being one of the fanciest stock truck manifolds on the planet...Still have to drill and install the two remaining thermocouples when they arrive and build and connect a water-meth manifold to the lines. I might also try to get runner pressure taps on the underside of the manifold, I'm just not entirely sure how useful it will be at this point as there are widebands on each cylinder for charge bias detection. I'll think on it a little bit more, but in the meantime, here's two more pictures.

Thermocouple (front) and methanol injection nozzle (rear)

Manifold underside with WM and thermocouples installed. Didn't have enough thermocouples to complete it, but ordered more now

 

This is the cockpit as I've got it currently set up. Still trying to figure out if I want the computer to sit a little bit recessed in the dash or flush behind the bezel I printed out. All of the wiring and power supplies are routed through the center console. Inverter is in the center storage area and has holes cut for the wires in the bucket to keep the space semi-usable. I have to get rid of that steering wheel cover too...

This ended up being a safe location for the firewall grommet to go through. No surprise electical cuts and it ended up right below the blower motor and cover so it hides very nicely. I'll have to redo a couple thermocouple extensions to get them to the data box inside.