My MegaSquirt box was now almost completely hooked into my DeLorean, to the point that I could now do some testing.
Battery and Basic Tests
The first step reconnecting the the battery to the car's electrical system and seeing if the basics worked. The door lights came right on, and turning the key lit up the dashboard and center console. The vent blower motor worked fine, as did the headlights, although the high beams wouldn't come on.
The most obvious thing , though, was how the cooling fans came on immediately. This made sense; MegaSquirt's FIDLE is now used to switch the fans, and apparently it is on by default.
I fixed the headlights first. Turns out I forgot to install the high beam relay. After putting that in, flipping on them on didn't turn on the lights -- it actually shut off every accessory in the car. I spent some time tinkering before I found that simply tapping the battery jump post with a multimeter turned the accessory relay back on. It seems I simply forgot to tighten the bolt on the battery jump post inside the car (next to the bulkhead connectors), so there wasn't enough of a connection to guarantee adequate current to all the components. Once I fixed that, the lights worked reliably. To be honest after replacing the fuse box and installing the EFI wiring, I expected more things to not work, so I'm sort of surprised everything is functioning properly.
To configure MegaSquirt, you need to use a laptop that connects to its DB9 serial port. Since most modern laptops don't have serial ports, I used a compatible USB to Serial Adaptor from DIYAutotune.com and a USB extension cable. I'd recommend getting a longer one than DIYAutotune provides (or using a 6' male to female DB9 cable, as DIYAutotune suggests, as it seems they have less of a chance of RF interference as compared to USB cables) if you want to be able to put the laptop in the passenger seat, for example. Running on the laptop is a copy of TunerStudio. TunerStudio is free for basic configuration.
I'm not going to do a full TunerStudio tutorial here, since I'm pretty new to it myself, but I'll cover what I learned. I bought a cheap Core2Duo-based Windows laptop for this purpose, mostly because I didn't want to keep dragging my main laptop around. You can get one for under $150 pretty easily, making this one of the less expensive parts of an EFI conversion.
I created a new project and had TunerStudio auto-detect the COM port. Remember to turn the ignition to at least the accessory position, or MegaSquirt won't have any power and won't be detected. Mine was found easily. After hitting Ok, I got a screen full of gauges. Most of them were 0, as you'd expect with the car being off, but the engine temperature was reading 62 degrees F, which matched the current temperature in my garage. I also played with the throttle position sensor, which seemed to run from 15% to 85%; there's a calibration in TunerStudio, but I'll get to that later.
Cooling Fans and FIDLE
The first thing I looked into was how to adjust FIDLE to control the fans. This can be found under the Boost/Advanced button as Programmable On/Off Outputs. Enable the FIDLE output, with the Power On Value set to Off and the Active Value set to On. Then change the Active Conditions Output Channel to "coolant", leave the test to ">" (greater than), and enter 206 for the Threshold and 10 for the Hysteresis.
What that does is start the car with the fans off. When the temperature rises above 206 degrees (the threshold value), FIDLE will turn on, which trips the relay to power the fans. When the temperature falls to by 10 degrees (the hysteresis value) to 196, the fans will turn back off again. The 206 and 196 degree values are the normal values used by an unmodified DeLorean to turn on and off the cooling fans.
Burn Failure, Battery Issues and a Loose Wire
To upload changes to MegaSquirt, you hit the Burn button. At this point I had a problem -- MegaSquirt disconnected and failed to upload. A few power cycles later it reconnected, but TunerStudio said the firmware was corrupt. I downloaded new firmware MS2/extra, upgrading from 3.3.3 to 3.4.2 in the process, but it wouldn't properly install.
At this point I decided to bring the MegaSquirt home and try it there. I plugged it into a JimStim I bought for future testing, but more importantly here to provide power to the board. I had no problem at all flashing the firmware.
I was starting to think I might have a power problem. My Optima Yellow Top battery is about nine years old, and the car's battery meter was only reading about 8v, so I'm fairly certain that was the culprit. The Yellow Top is better for heavy loads and deep cycling, but I'm not sure I was really using it that way so I decided to go with the less expensive Optima Red Top this time. Some people like Optimas, some don't, but one notable feature is that they don't gas out like conventional batteries, which is pretty nice when your battery is in the cabin with you.
The Red Top showed about 11v on the car's battery meter, and 12.5v with a multimeter (even the Yellow Top was still showing good voltage), but I still had the same problems -- trying to burn settings to MegaSquirt failed and I had to re-flash the firmware.
I finally noticed that TunerStudio had a battery gauge, so I flipped that on and saw that MegaSquirt was getting only 6.1v from the car's electrical system. That was probably the culprit right there. When plugged into the JimStim it reported 11v, which was probably fine for the stimulator and implied that MegaSquirt itself was working fine -- something was wrong with the wiring in the DeLorean.
To start diagnosing the wiring problems, I took the case off the DB39 connector at the end of the wiring harness, and immediately spotted the problem -- the red main power wire wasn't actually connected to its pin, and the wire was pulled a good two inches back. I must have tugged on it by mistake when wiring everything up in the car and broken it solder connection. MegaSquirt must have been leaching 6v from one of the other lines.
A quick bit of soldering fixed the problem, and the TunerStudio battery meter properly reported 12v.
Now that I had full power, I could go back to setting up MegaSquirt's state through TunerStudio. I mostly just followed this page in the MegaManual, but it did take some time to find some of the settings, as TunerStudio's interface seemed to have things in different locations than the MegaManual suggested. Here are some of the basics, as of TunerStudio MS 2.6:
In Ignition Settings section, under Ignition Options/Wheel Decoder, choose EDIS. Everything else can be left at defaults. The details of this can be found in the Mega Manual under Settings/MegaTune.
Idle Air Controller
Under Startup/Idle, Idle Control, choose Stepper (4 or 6 wire). I left everything as at the defaults.
Under Fuel Settings, Injector Dead Time/PWM, I made best guesses based on the Ford FOTE-9F593-D9B injectors I have, which are 22 lb/hr high-impedance injectors.
The only thing I did was set the Injector Dead Time to 0.900 as per the tooltip. I left everything else at the defaults, which seem to be correct for high-impedance injectors.
We also have to generate an initial volumetric efficiency (VE) table. The generation tools require you to buy a license, so you're free ride is over. The license isn't expensive ($60 at the time of this writing) and considering all the other bits I bought to get this running, it's not a big expense.
Under Fuel Settings, choose Fuel VE Table 1. This opens a VE Table window. From its Tools menu, choose VE Table Generator. I entered the following settings based on page 2 of the 3.0L Engine Overhaul Manual:
Engine Type is Naturally Aspirated
Idle at 790 RPM
MAP at 35 kPa (which is the default; I couldn't find any real values for this)
Redline at 6500 RPM (I just used the redline on the DeLorean's tachometer)
Peak Power at 150 HP at 5000 RPM, 100 MAP (again, I couldn't find any MAP values, so I used the default)
Peak Torque at 171 lbs-ft at 2750 RPM, 100 MAP
Displacement of 2975 cc
I believe a MAP of 100 is a special value to TunerStudio that tells it that you don't have a specific MAP value to use. 100 kPa is atmospheric pressure at sea level, and can be considered the ultimate wide-open throttle.
Hit the Generate button to create the new VE table, and the click Apply to keep it.
The AFR (air/fuel ratio) table is set up the same way. Under Fuel Settings, choose AFR Table 1, and follow the steps above to generate and apply the table. The engine specifications should already be set from when you did the VE table; if not, just re-enter them here as well.
Also in Fuel Settings, under EGO Controls, I changed the Sensor Type to Single Wide Band to match my sensor.
As per MegaManual, the AFR must be updated for the sensor. Go to the Tools menu, choose Calibrate AFR (you might have to first choose Un/Lock Calibrations first to enable it) and from the dialog choose Innovate LC-1/LC-2 Default from the dropdown, and then click Write to Controller.
There area few things to do here to indicate the basics of the engine for MegaSquirt. This is under Basic Settings, Engine and Sequential Settings.
Required Fuel was set to 14.4. This value was generated by clicking the Required Fuel button, then entering:
2975cc displacement. This specific value came from the 3.0L Premier/Monaco Engine Overhaul Manual.
22 lb/hr injector flow
14.7 Air-Fuel Ratio. This is just the idea stoichiometric value as described in the MegaManual Principles of Tuning Programmable EFI Systems.
Back in the Engine and Sequential Settings dialog, the Number of Injectors are set to 6. Everything else left at defaults.
I also had to calibrate my TPS. This is pretty easy from the Tools, Calibrate TPS menu. I had a long enough DB9 cable that I could sit the laptop on the engine and pull the throttle arm by hand.
Initial Spark Advance Table
While MegaManual claims it doesn't have a Spark Advance Table Generator, it turns out the docs are just a little out of date and one does exist here. I used the following settings from page 2 of the 3.0L Engine Overhaul Manual:
Cylinder Bore of 3.66 inches
Combustion Chamber Type of 2-valve closed chamber w/ optimized quench
Fuel as regular (87 octane)
Compression Ratio of 9.1:1 to 10.0:1; the actual ratio is 9.3:1
Idle Vacuum of 15 in-Hg. This is the default value; I wasn't able to find any information about the engine vacuum in the manuals or forums. I expect this is a little low (Googling suggested 17-21 was normal for most engines), but it's just an initial table and will have to be tuned anyway, so it should be good enough as a starting point. (Update: I later found this table to be about 2 degrees higher in all cells than a known good table from another DeLorean, so I just used that table. I’d suggest using idle vacuum values in the 17-22 range, aiming for 14 degrees at 40 kPg in the first and second columns).
Maximum RPM of 6500
Idle RPM of 790
Spark Advance Table Dimensions of 12x12
The generator then creates an XML file containing the table for TunerStudio. Go to Ignition Settings, then Ignition Table 1. Choosing Load Dialog Settings from the window's File menu first asks if you want to include additional dependencies, which I did do. After choosing my table file, I got a "signatures do not match" warning got a mismatch warning -- in fact, the signature was "null", which is the value that programmers use to mean "not present". I'm fairly sure that this has to do with table being saved through the generator tool instead of manually through TunerStudio, so I simply ignored the warning and loaded it anyway.
Spark Plug Dwell Setting
I spent way too much time trying to figure this one out before I discovered that EDIS handles this setting automatically. There is no need to manually figure out the dwell yourself, so you can ignore that part of the MegaManual entirely.
Connecting the Throttle Cables
Before I could go much further I really needed to connect my throttle cable. The Jeep throttle Josh had provided included a clip that the throttle cable runs through. I bought a cheap set of brass throttle cable stops (aka "cable locks") from Amazon to secure the cable. The idea is simple enough: run the cable through the clip, then through the hole in the cylindrical cable stop, and tighten the screw on the top of the cable stop to lock the cable into place.
Connecting the Transmission Cable -- The Wrong Way
The problem was that I didn't need to connect just the throttle cable -- my automatic transmission has a separate governor cable that also needs to be hooked up here. I drilled a second hole through the throttle cable bracket that Josh had made, using M8 x 1.0 nuts and washers on each side to hold the cables in place. The only problem was the length of the transmission cable. Or so I thought.
Replacing the Transmission Throttle Cable
While the sheath was long enough, the cable inside was just a touch too short, so I decided to replace it. I first cut the barrel off the end of the cable. Under the car, I unscrewed the philips screw holding the governor arm in place and took it off the governor, being careful not to lose the spring. After working the rusted cable end back and forth a bit to break it free, I was able to remove it from the arm.
The old cable measured about 44" long, so I ordered a new 60" long cable with barrel ends from Amazon for under $10 (it came with the sheath, but I only needed the cable part) so that I'd have a lot of extra slack. I cut the barrel off of one end and pulled the cable out of the sheath.
A small snag was that the barrel end was a tiny bit bigger than the one on the original transmission cable, and wouldn't fit in the governor arm. This was fixed by carefully widening the hole on the governor arm with a drill until the new barrel just fit. I ran the cable back up the transmission sheath from under the car. Even though the cable was thicker than stock, it had no problem running up the sheath.
Re-attaching the governor arm is a bit of a pain, but not as bad as some other things I've had to do on the car. The pivot is keyed, so first you need to line it up properly with the shaft on the governor. But you also need to attach the spring to the arm and the body, while pressing it onto the shaft and over the guide pin. If you have trouble with the spring, flip it over (along the spring's axis) so that the hook is against the arm instead of across from it, since it'll be less likely to pop off and be much easier to attach. Once everything is together, put on the screw, but don't over-tighten it -- I'm pretty sure the threads are plastic, and it'll be easy to strip if you're not careful.
Attaching the Cables
The Jeep throttle had a small clip that one cable could run through. I widened this with a drill so that I could run both cables through. I also cut off the end of the clip so that the cables would have as much spread as possible as the throttle was opened. Both cables fit through one of the cable stops. Since the transmission cable has a spring, I pulled it as far as it would go before tightening down the screw on both cables..
And this is when I realized my mistake. I noticed that the governor's spring would keep the throttle from closing properly, so I started looking at adding another spring when I decided to check the DeLorean Workshop Manual one more time. There, on page G:05:02, it clearly shows the governor arm being in the forwards position when the throttle is open all the way. This is the opposite of how I had it set up. I checked the old DeLorean throttle spool to confirm this, and sure enough the transmission cable pulls when the throttle is closing, not when it's opening.
CONNECTING THE CABLES -- THE RIGHT WAY
Luckily, the Jeep throttle had a long arm with attachments for both opening and closing cables. I reconfigured one that pulls the throttle open to have only the throttle cable on it (which means I mangled the clip for nothing).
For the transmission cable, I used a second clip I bought from Amazon (called a "detent cable end"). It slightly loose, but I was pretty sure the spring tension from the governor arm would keep it on the throttle arm.
Josh's throttle cable mount (the one I'd modified previously by drilling a hole for the transmission cable) wasn't positioned correctly to reach the other end of the arm, so I built a second one out of steel stock, welding an "L" piece to a flat piece. In retrospect, I could have saved my self some trouble and just got an aluminum or steel "L" and cut it with an angle grinder into the same shape, but I needed more practice welding, so I don't regret the extra work, even if using aluminum would have avoided the need to to prime and paint it. This new bracket mounts on top of Josh's bracket using the same bolt holes, and aligns the cable correctly with the other end of the throttle arm.
...and a mouse nest
A small aside, but when I replaced the fuel pump at the start of this conversion, I left off the access cover between the trunk and the pump. A mouse snuck in over the winter and made a little nest in the trunk on top of a canvas shopping bag, using some unimportant papers that were in there. No damage was done to the car, and it was easy to vacuum out. It's a good thing that I had been keeping the doors closed whenever I wasn't working on the car, since I definitely didn't want him in the cabin. The electrical seems good, so I'm thinking the mouse just wanted someplace to hide, not to eat my wiring.
Fuel System Test
Before doing anything else, I wanted to test the fuel pump and lines for leaks. For the first time in three years, I put five gallons of gas of 89 octane unleaded in the tank.
I'd already heard the fuel pump running when I turned the key to the "run" position. MegaSquirt only runs the pump for a couple of seconds when the engine isn't actually turning, which isn't really that helpful for testing for leaks. I pulled the fuel pump relay and used my PowerProbe to run 12v to the pump for about ten seconds.
Forgot to Tighten a Fuel Hose
The pump ran, and I heard a hissing sound coming from the engine bay. Some fuel was visible under the car, below the relocated fuel filter. Annoyingly, the puddle was grey, because the gas was removing the paint on my newly-refurbished frame. I ran a longer wire to the pump's relay socket so that i could run it from the bay, and listened around while it hissed. It turns out this leak was quite obvious -- it seems I'd forgotten to tighten the fuel line on the bottom of the filter, so it was leaking fuel down the line. The noise was much louder than the visible fuel, and I had to find it by touch. Luckily it was easy to clamp down. Running the pump again revealed no leaks; I ran my fingers over all of the connections just to be sure -- all dry.
Next I tested the fuel pressure. I had permanently installed a gauge on the line, but when I ran the pump it only moved a small amount, and turning off the pump immediately dropped the pressure back to zero. I first thought I misunderstood how these worked, and that they required manifold vacuum to operate -- there's a vacuum line running to it -- but that didn't seem right; at high engine speeds there is very little vacuum, so if it relied on vacuum to function at all then there wouldn't be any fuel pressure and the car wouldn't run.
It turns out the problem was my own incompetence -- I had put the gauge on the return line, after the pressure regulator. At that location it would only showing the pressure the excess fuel after the regulator was done with it. Once I moved it to the proper location on the feed line just before the fuel rail, everything worked fine -- the pressure read about 39 PSI when the pump was energized , and 36 PSI when off. Pressure held, which suggests that the regulator and the rest of the fuel system doesn't have any leaks.
One interesting note is that the Moaco/Premier Service Manual page 14-4 suggests that the pressure should be 28-30 PSI. However, some Googling suggests that 22 lb/hr fuel injectors should run closer to 40 PSI, so I'm fine with these results.
Before I could start the engine, I needed to install the serpentine belt to run the water pump and alternator off of the main crank pulley. While my idler pulleys were installed, the A/C compressor was not; the old 2.8L R12 unit doesn't mount to the valve covers without modification, and I decided to wait on my planned R134 conversion until after I actually had the car running. While serpentine belts are usually tightened with an automatic tensioner, DeLorean engine swaps seem to more commonly use the same mechanism as the original 2.8L engine -- an adjustable bar between the alternator and the engine block.
I used a tailor's measuring tape, wrapping it around the pulleys. Tailor's measures are cloth, so they're perfect for finding the size of a belt (as compared to a metal tape measure, say). The catch is that the tape is much thinner than a belt, so it'll get you in the ballpark but not the perfect size. I measured between 49 and 51 inches, depending on how far I moved the alternator adjustment arm. Don't forget that the belt has to be long enough to slip over the rim of the pulleys, too. I tried three before settling on the CarQuest 5060495, which had the perfect width (0.82") and circumference (49.5").
At this point I realized I'd installed the alternator bracket and adjustment bar incorrectly. The reason you have to cut the 2.8L bracket and get a new bent adjustment bar is because the 2.8L position is too far back on the 3.0L engine. The belt needs to be in a single plane once installed, and in the 2.8L position it sits at an angle.
After fixing the bracket and flipping the bar, the belt aligned perfectly. I tightened it with by wedging a pry bar between the mounting bracket and the alternator, wedging it outward as far as I could and holding it with my foot while I used two sockets to tighten the adjustment bar against the engine block, and then the other end attached to the alternator.
Adding Fluids, Solving a Coolant Leak
I'd already added oil to the engine for my compression test, so that left refilling the transmission with Dexron III transmission fluid, which is how I found that I hadn't installed the transmission dipstick seal correctly (I've covered that in another post).
Next was adding the three gallons of 50/50 coolant/water mix to needed for the coolant system. I had transferred the Wings-B-Cool self-bleeder from the 2.8L engine from the 3.0L engine, but I wouldn't be able to bleed the radiator until I ran the engine.
After adding about two gallons of coolant, I noticed a puddle forming on the floor. It seems that I hadn't tightened the hose clamps nearly enough, and a few joints were leaking. I went through and retightened them all, but one still leaked quite a bit. Of course, it's the one hardest to reach, connecting to the T just before the hot water valve. It's not just a few drips either -- it's just about pouring out of the hose in a steady stream.
No matter how much I tightened the clamp, I couldn't get it to stop leaking. I loosened the clamp and shifted it as close to the "T" joint as possible, but still it leaked. I finally just pulled off the hose and added a second clamp. This seemed to do the trick, as the leak was no more.
UPDATE (August 2018): The above is not correct -- it still leaked. I eventually discovered that the engine has a 3/4" connection, bu my "T" has a 5/8" connection. The 3/4" hose I was using was too big to seal on the 5/8" "T". My solution was to buy a Dayco 87613 hose meant for a Mitsubishi. It's a 1' long 3/4" to 5/8" hose, perfect for this application.
At this point I decided to actually turn the key and see what happened. The headers were on (sees you should never run an engine without at least the headers), but without the exhaust the car would be incredibly loud. I turned the key only for a few seconds, and after some cranking it did try to catch. It may have actually started if I held the key longer, but I'm not sure.
I did find some more coolant leaks due to loose hose clamps, this time at the water pump. They were easy enough to tighten further, which fixed the leaks.
Coil Hold Downs
Along with coolant leaks at the water pump hoses, cranking the engine revealed another problem. I always thought that the spark plug boots were a little loose, and this was proven true as the rocking of the engine actually popped some of them loose from the cylinder head. It's interesting, since if you put a loose spark plug coil in the boot it held in place quite firmly, but it seems that the boots don't slide down as far over the plugs once they're both installed in the heads.
The solution to this was to build some hold downs brackets. The plastic engine cover piece (which contains the air filter in the 2.8L engine but is just cosmetic here) mounts through bolt holes on the edges of the fuel rails. I bought some longer bolts that would extend all the way down to where the coils are. The hold-down bracket is simply a piece of aluminum cut to fit over the tops of the coils, and ends of the longer bolts would push down on it to keep them in place.
I created a bracket template using a piece of cardboard that zig zagged around the coil connectors and the fuel injectors. I cut this into flat aluminum stock using an angle grinder for the perpendicular and diagonal cuts, and a Dremel and a bunch of cut-off wheels for the horizontal cuts. A reciprocating saw with metal blade would probably have worked well too. I had to do this twice, as the first piece of aluminum wasn't wide enough to reach from the coils where the cover bolts would go. Once bolted down the coils were firmly locked in place.
I also had concerns that the spring conductors inside the boots were too short. I was going to buy new ones, but it seems you have to buy them with the coils. The guy at NAPA suggested something much simpler -- stretch the spring. That worked perfectly, and the springs now reached all the way to the end of the boots.
Transmission Kickdown Microswitch
The automatic transmission has a kick-down microswitch that is actuated when the throttle is all the way open. This was attached to the throttle spool on the 2.8L engine, but there was no such facility for this on the 3.0L engine.
I built a bracket out of a piece of steel angle stock and mounted it to the back of the passenger side's cylinder head, which has two convenient holes in the back. An angle grinder cut easily cut out the extra bits of the stock that were in the way. I bought a microswitch with a roller ball arm from Adafruit and mounted that to the underside of the bracket.
Rather than dealing with tiny screws and nuts for the switch, I just used zip ties -- they held the switch in place very firmly, and saved me a lot of trouble. The holes in the back of the cylinder head are M7 and not very deep, and I had trouble finding short bolts, so I wound up cutting down two longer ones with an angle grinder and a cut-off wheel and cleaning up the edges with my bench grinder. To adjust the position of the bracket, I drilled holes that were more like slots into the bracket so that I could slide and tilt it up and down a bit. To adjust the depth of the switch, I just added some more washers between the bracket and the cylinder head (wound up using three). Nice and simple.
For wiring, I ran a ground wire from the engine block to the a female blade connector and plugged that into the "common" pin of the switch, and plugged the green/white wire from the harness into the "normally open" pin of the switch.
One small problem I hit was that both of the holes in the back of the valve cover were stripped. I have an M7 x 1.0 TimeCert kit, but there isn't a huge amount of space between the head and the firewall, and not enough space to fit a proper wrench with the throttle cable and wires there. I wound up starting the bits by hand, then wither using pliers or an approximately-sized socket wrench to turn the bit. I found that a 6 point sockets might not fit the squared end of the bit, but an identically-sized and 12 point would. Still, this was tedious, and I only fixed the upper bolt hole; the threads on the lower one were good enough to hold the bolt, but not quite good enough to properly tighten it, but the top bolt held firm and the switch didn't slip, so I was good.
I already covered a lot of this in another post, so at this point all I had to do was install it. The joints and pipes are commonly available as exhaust and intake parts, with exhaust hose clamps firmly attaching the joints to the clamps. I was short a couple of T-bolt clamps, so I ordered some 2.25" ones from SiliconeIntakes.com. They carry all the other parts needed for this as well.
The only problem I had was that the fuel pressure gauge is directly under where the air intake runs, so you can't actually see it very well. I was able to turn so the you can shine a light in and lean over to see it peer at, which is good enough for how this is going to be used. Still a little annoying; I sort of wonder if I should have just mounted it at the fuel filter, but I'm just leaving it as is for now. (UPDATE: I later moved the gauge to the firewall so I could still see it.)
Later I mounted the A/C compressor, which made the routing a little less clean, as the compressor is now in the way just a little. The result is that the air filter is tilted slightly downward, as is the last bend in the pipe. Everything still clamps tightly together, so it should all be fine..
Draining the Fuel Tank
At this point there was about a year and a half break as I waited for a new exhaust to arrive, gave up, and just built my own. This process is documented in another post. I hadn't expected this wait, so I had gas sitting in the car for a full year and a half. Some quick Googling noted that gas is only good for a year at most, so I had to drain the tank and put fresh gas in.
In the past, I'd used a hand pump to remove the gas. It was tedious and annoying, and seemed leaky. The is time I used a hand siphon. If I had a way to get the gas can lower than the tank, I could have used the siphon to pull the fuel out with gravity, but instead I had to pump it by hand the entire time. I still found this much easier than the old pump, and I was able to do it one handed. Rather than removing the entire fuel pump, I removed the cap that replaces the original fuel sender and snaked through there, putting the end with the brass connector in the tank to weigh it down.
Once I had the majority of the fuel out, I used a rag tapped to a stick to try to wipe out the tank as best I could. I could only do so much through the sender hole, but since I'd cleaned the tank thoroughly before, there wasn't much that needed to be done.
The next step is to turn the key and see if it actually starts.