1988 Turbo 'Vert Project

[toplessFC3Sman]

So.... project creep set in on the arduino-based controller that was supposed to turn on the water-to-air intercooler pump & switch on the radiator fan if necessary... Now it has an LCD display that I will be integrating into the idiot light cluster where the clock used to be, it will talk with the megasquirt via CAN to display its sensor inputs or outputs, it has a set of conditional statements to turn on or off various outputs and it has a few more inputs that can be used for other analog devices. It also grew a rotary encoder/button combo so you can change the display and adjust various options like the switching point for those conditional statements, forcing the outputs on, and changing the display backlight color & brightness for both daytime (headlights off) and nighttime (headlights on).

I still need to get the CAN part of it working - thats the last step since I need to dig the ECU out of the car, update the megasquirt firmware version, and bring out the CAN signal. Anyway, here are a few pictures... most of what you see are jumper wires which will all disappear once I actually get it onto a circuit board, the 5V regulator that is much bigger than it needs to be, and a pair of pots (variable resistors) that are standing in for the temperature sensors for debugging purposes.

[ITSWILL]

Looking good. I love how inexpensive yet powerful these micro controllers are.

Couldn't you vary brightness based on the dimmer in the cluster. Write a code based on the dimmer voltage output and correlate that to screen brightness?

[toplessFC3Sman]

Yes, thats the plan - basically I have 2 sets of color & brightness variables. one of them will be used when the dimmer wire is grounded (headlights off), and the other when its at +12V (headlights on). This just makes it configurable so that you can specify how bright, what color etc once its already installed in the car and can compare to the other interior lights. The wire that runs to the idiot light cluster seems to only be on/off though - it didn't seem to vary with the dimmer switch. Does the clock vary in brightness with the dimmer? Mine never worked reliably and hasn't been in the cluster in a number of years now.

[fidelity101]

good lord that is sweet! very clever idea too. I'm excited to see the final outcome.

[toplessFC3Sman]

I got a little bit of time in the garage yesterday to make some noise, and finished the brackets that hold the intercooler over the engine in roughly the same spot as the OEM TMIC. They aren't very complicated, its just difficult getting time in the garage when I'm not worried about using angle grinders and things that would wake Paul up when napping/sleeping.
In front, it is basically just a flat strip with a chunk cut out to allow the alternator adjustment. This sticks down and bolts onto a threaded ear on the front iron that is almost directly behind the alternator adjustment bracket.


The rear one comes off of one of the TMIC mounts on the UIM, uses some carved up L-channel to come across and bolt to the intercooler, then it bends down and twists slightly to bolt to one of the studs/nuts that hold the throttle body inlet piece on. On this one, the aluminum had fatigued and broken from all the bending and adjustment, so I added a few small bolts to attach the two pieces.

[murz]

Oh man this is awesome! I like how you kept the original shape too.

[toplessFC3Sman]

The pump got mounted last night now too - not a huge thing but I have to keep chipping away at it.

The bracket itself is really just a piece of aluminum bar-stock with two tapped holes for the mounting bolts (they're threaded through from the backside and loc-tited up against the bar stock to act as studs), and two double-holes filed down to pass hose clamps through. I used some adhesive-backed felt and insulation foam for doors and windows between the hose clamps and the pump, and between the clamps, bracket, and the car to cut down on vibration and rubbing.


This lines up the outlet at the bottom of the heat exchanger nicely with the pump inlet at the lowest point of the system, and has the pump outlet shooting out between the two heat exchanger fittings to head back to the cooler.

[toplessFC3Sman]

That pump right there ^^^ was the cause of a few headaches yesterday. I got the system plumbed and everything connected last night and... it wouldn't push any water. After confirming this by having it suck out of a gallon jug, through the intercooler and heat exchanger, and pour into a bucket, there was no flow. Even when I unbolted it, put longer hoses on to allow it to hang down below the car it only pushed a trickle out - maybe a gallon a minute at the rate it was draining the bucket, nowhere near the 500 gallons/hr (~8.3 gal/min) it is advertised as having. Looking through the spec sheet further, it does say that the flow drops off to about 340 gal/hr (5.7 gal/min) at 3 1/2 feet of water (i'm assuming thats pump head pressure), but when doing my research I figured that would be fine since there is less than that total difference in where the components are mounted in the engine bay. Whelp, I guess I didn't account for the flow losses through the intercooler and heat exchanger or something too, because the actual flow rate was pretty pathetic. Turns out that 3 1/2 feet of water is only about 1.5 psi, which means that on the chart below, this pump's flow rate drops like a rock in comparison to most of the pumps specifically designed to be used as automotive intercooler or waterpumps.  (8.3 gal/min @ 0 psi to 5.7 gal/min @ 1.5 psi) If its flow curve looks anything like the other ones on that chart, and it likely does since it is still a centrifugal pump, it is all out of flow by 3 psi.



So, instead I bought one of the Bosch CTS-V pumps - these are pretty well known and have been around for many years now in other applications like as coolant pumps, intercooler pump for the supercharged Mustang Cobra from the 2000s, etc. If it can flow enough coolant for those engines, it should be able to handle my needs since I'm nowhere near those power levels.

[toplessFC3Sman]

Just to drive the point home and compare a couple other Bosch part numbers that I came up with, here is that flow vs pressure plot again with a few dashed lines overlaid.


All the dotted lines are based on measured flow numbers from RULE or Bosch for their pumps - they are both 12V pumps, but I don't know whether these tests were done at 12V, 13.5V, or what. Anyhow, the RULE bait-pump looks pretty terrible despite claiming 500 gallons/hr, which it can only do with no pressure. Add some pressure and it drops like a rock. As far as I can tell, theYou are not allowed to view links. Register or Login pump is the Cobra/CTS-V intercooler pump, and the flow curves line up pretty nicely between Lingenfelter's testing and the spec sheet. One other pump I stumbled across that had higher flow/pressure capability was the You are not allowed to view links. Register or Login. Its a 3-wire design with a +, -, and then one labelled "S"... is it PWM-speed controlled? I don't really know, but it looks very capable based on the specs.

[toplessFC3Sman]

After the last troubles with the pump, I went back and looked closely at all of the potential problems with the coolant loop - anything that could hinder flow or cause the pump flow to stall. Basically, this came down to hard turns that could kink the hose, places that air could get trapped, and making sure that the pump was as low in the system as possible so that it always would be getting fed water since it cannot create any suction.

The first one was simple - basically just looking at the hose runs and adjusting things and adding brackets or tie-downs as necessary.

The second - removing all air when filling - was a bit more complicated. Its possible to get a vacuum pump to suck all the air out, then pull water or coolant into the system, but knowing how likely it is that I may need to add water or disconnect things for tweaking etc, I decided to try to go a different route and not require a special vacuum pump to fill things. This meant that I needed to put air purge valves at all the local high spots to allow me to force out all the air when filling normally, which I could then close once they started showing water trickling out. The two biggest local maxima were in the top of the heat exchanger under the radiator support, and in the top of the intercooler itself since the fittings attached in the center of each of these. A look through mcmaster turned up these very nice, low-profile purge valves meant for radiators, but they would work great for my purposes since they are resistant to corrosion, pressure, and require you to manually open a valve (as opposed to many which require pushing a button, which I could easily see doing accidentally, or see it opening due to vibration). Its a very small thing, and just requires an 1.8" NPT thread be tapped or welded into where-ever it is to be installed.


I had thought that I would need an Al bung welded on the top of the heat exchanger to thread this in, but when I went to drill the hole that the bung would sit in, the end plate was thick enough for 7-8 threads, completely sufficient to thread the purge valve into.

These require a special square key to open and close, but due to the position under the radiator support, the supplied key was too short so another had to be made. Fortunately, an old piece of brake tubing was just about the right circumference, and I could bend it over the square end of a thread tap to make an adequate tool. A bit of drilling through the top of the radiator support, and now there is access to the top of the valve!


The same valve was installed in the upper-most corner of the intercooler as well to get as much air as possible out with a regular gravity-fill.


This meant that the hoses could be routed under the OEM battery location, since I don't really want to tackle the project of moving the battery - I have enough wiring ahead of me before I can drive Savannah again!


The last piece of the puzzle is dropping the Bosch pump a little lower, putting it below the bumper bar off on the side of the air channel heading to the radiator and other heat exchangers. This way it is always pulling water from the oversized IC heat exchanger, and will always be pushing fluid since all the air is out of the system. Some quick testing shows that it seems to be flowing a lot more than previously - I'll need to run the hoses back into the milk jug and bucket to get an estimate, but it seems to be a substantial improvement.


Also, looking back through this, I'll need to get some pictures of the bracket I made to hold the fill cap, and where that is mounted too.

[toplessFC3Sman]

I've been slowly plugging away at the various things left to do - reinstalled the radiator & hoses, filled up the radiator &  IC system with water & coolant mix, minor plumbing things like running the overflow hoses, attaching a vacuum/pressure source pre-throttle to draw the idle-control air from, etc.

The biggest thing has actually been taking all those circuit diagrams and turning them into reality. At this point, the power-relay side of things is done, with the relays from the Lexus spliced in and hooked up so that they trigger correctly when each signal wire is grounded. I haven't run the engine long enough to see whether the thermo-switch is working as-intended, but it tested fine in a pot of boiling water, so I don't see why it wouldn't.



The biggest things left to do are:

• Update MS to a later version of code with CAN support
• Reinstall & confirm its working, set static timing with timing light
• Confirm Arduino controller is reading CAN signals correctly
• Package controller, screen, CAN module & 5V regulator in idiot light cluster
• If there is room, put the switching transistors in cluster, otherwise put them down with the manual switches and control knob in ash tray
• Finish gauge wiring
• Plastic block-off panels for holes around IC heat exchanger
• Reinstall underbody brace, wheels, bumper, hood (may need to trim one of the braces)
• Typical shake-down & new engine build break-in driving

[toplessFC3Sman]

• Update MS to a later version of code with CAN support
• Reinstall & confirm its working, set static timing with timing light
• Confirm Arduino controller is reading CAN signals correctly
• Package controller, screen, CAN module & 5V regulator in idiot light cluster
• If there is room, put the switching transistors in cluster, otherwise put them down with the manual switches and control knob in ash tray
• Finish gauge wiring
• Sort the battery mounting - I'd like to use the stock mount but will probably need to run the IC water hoses slightly differently
• Plastic block-off panels for holes around IC heat exchanger
• Reinstall underbody brace, wheels, bumper, hood (may need to trim one of the braces)
• Typical shake-down & new engine build break-in driving

Rather painless updating the MS, but there have been a few changes to different settings since I last updated the firmware that may change the way it runs, so a little bit of tuning may be in order again.

[fidelity101]

I'm surprised that updating the software is enough to trigger the need for a re-cal.

[toplessFC3Sman]

It shouldn't be much of a re-cal, but there were some changes to the way the closed-loop idle control is done, so I anticipate having to go in and do some tweaking there to get a more steady idle without having it hunt around too much - this was where I spent an inordinate amount of time before, getting the balance just right between having the proper "valley" in the fuel map that would allow the idle air controller (using the BAC) to respond well and catch the idle under all sorts of conditions.

[toplessFC3Sman]

A lot of progress was made over this past weekend, although it doesn't look like much in context of this list...

• Update MS to a later version of code with CAN support
• Reinstall & confirm its working, set static timing with timing light
• Confirm Arduino controller is reading CAN signals correctly
• Package controller, screen, CAN module & 5V regulator in idiot light cluster
• If there is room, put the switching transistors in cluster, otherwise put them down with the manual switches and control knob in ash tray
• Finish gauge wiring
• Sort the battery mounting - I'd like to use the stock mount but will probably need to run the IC water hoses slightly differently
• Plastic block-off panels for holes around IC heat exchanger
• Reinstall underbody brace, wheels, bumper, hood (may need to trim one of the hood braces)
• Typical shake-down & new engine build break-in driving

The first big step was getting CAN communications working between the Megasquirt & Teensy 3.2 controller using a Waveshare CAN transceiver, You are not allowed to view links. Register or Login. For a while, I was getting no communication whatsoever... dead silence on the line. It turns out that all the coding & wiring I had done was correct, but the library of CAN commands that it was referencing was out-of-date and while all the commands themselves were there, the order that arguments were passed to them had changed, which meant that my code was initializing to the wrong speed and looking through a filter which removed all the MS's messages. A lot of troubleshooting later, and everything was working as intended. Since that was the last bit of the project that was un-tested, I could start the miniaturization process to try to fit two breadboards worth of prototyping stuff into the warning light cluster.

First was the display - I had already done some minor filing of the edges to make it short enough to fit, but it was still as large as physically possible while still fitting where the clock had been, so everything else needed to fit within it's footprint behind it. Everything else consisted of:
- Teensy 3.2 controller
- Waveshare CAN tranciever
- 5V voltage regulator
- 3 TIP120 NPN power transistors (the Lexus relays were drawing too much current for the much smaller 2N3906 transistors to comfortably handle? - they were rated for 200mA but did not work in practice)
- Various resistors & capacitors for setting screen contrast, thermistor input ranges, output current limiting for the TIP120 transistors
- Wiring connections for stranded-wires to run to external harnesses for the control knob, outputs, inputs, power & ground, etc

So... lots to pack in. I started with a general Radioshack IC prototyping board that I had laying around since it was laid out with two "lines" running the length of the board for regulated 5V and ground, as well as pads coming outward with 3 holes for each pin of the Teensy controller chip. It also had a row of 2-hole pads along the top and bottom which would be useful for the header to attach the screen. However, this board needed to be trimmed down a bit to fit within the cluster and behind the extents of the screen - this is it shortly after starting after soldering down the connections for the screen along the top (power, ground, various data lines, back-light LED colors, etc), then the voltage regulator on the right side, one of the TIP120 transistors on the left, the screw-connections from the CAN transceiver board on the lower left (the board itself is mounted underneath this proto board), and a bunch of resistors over capacitors for the inputs in the lower right. The second transistor would be going directly to the right of the first transistor, but wasn't soldered down yet to give better access for soldering more components.


The third transistor is on the back-side of the proto board, as is the Teensy controller itself and the Waveshare CAN transceiver. The Teensy was on the back so that I could solder on the opposite side of the board directly on the copper pads so I would have decent access since it would be mounted with a set of headers (so the controller can be removed without soldering), and to keep it out-of-the-way from most of the other soldering. Here it is just stuck in place for reference without the headers. The height of and access to the screw-terminals and size of the CAN transceiver dictated that it be on the back as well, and I just ran out of room on the front for the last of the transistors. The three wires here are the outputs to the relays for fan speed and IC water pump. This really highlights how much forethought really needed to go into placing all the components and trying to ensure access to get everything soldered down.


And the side view, without the headers for the Teensy controller or the screen in place. The screen would go on the bottom of this stack the way I'm holding it, and the Teensy would be lower than it is currently positioned with the headers. The micro-usb programming cable is currently attached and sticking off to the left - I want to keep this accessible for programming, but it won't be installed with this in place.


In these pictures its a bit further along - the headers are installed (I had to trim down the ones for the Teensy so it would sit comfortably between the screen and the prototype board), the screen is on and trimmed a bit, and the proto board is more complete. Filtered power from the cluster has been wired to it, but sensor input wiring and control knob wiring hasn't been attached. Also, you can see the power filtering circuitry to give a more steady 12V and switched 12V for the gauges and this project, now reconfigured and wedged into the spot for the left-most lights. The convertible doesn't have a hatch warning light, so these were empty already; I just had to hollow out the area to wedge these components into.






Since the rotary control knob & button combo that I am using to control the screen couldn't be mounted up next to it (no space, plus I don't want to be reaching all the way up there for it!), I needed to find a new place to put it. The ashtray seemed like a good candidate, especially since it gets to hot to stick a phone in anyway. Since I would be taking up that space, I decided to throw a few manual switches for the outputs in there, and some LEDs to show if it was on (whether by the manual switch, the controller output, or the thermoswitch for the fan speeds)


And now, with the proto-board finished, and all the various wires added. I still need to carefully consider how these are all going to be routed out, since they bulk up the entire package a little bit more and prevent it from sitting nicely.


The switch still needs to be mounted in there with a cover panel, but all the connectors fit through the existing hole in the side. It will also have the calibrate & status LED for the LC-1 wideband controller mounted on that panel.

Here it is, powered up on the bench. The inputs and outputs aren't connected, nor is CAN since those are all in the car, but this at least gives a taste of what it'll be like.