DIY: Build your own computer-based compression tester for $68 (or $100).
 

CAVEAT LECTOR: I haven't tested this against real numbers yet. I LITERALLY just finished building it today. If somebody has a TR-01 or Rotary Diagnostics ReCT-04 for me to compare against, I'd love to borrow it. I'll even pay shipping both ways.

This DIY assumes you have access to either a "Dremel"-type rotary tool or more common sense than me. Also, you'll need basic soldering skills. I do mean "basic". I've done a bit of wire soldering in my time but this was the first time using prototyping board and I managed to get it right the first time. Any cheapo $15 pencil style with 25W power is more than enough. Some heatshrink tubing is best but electrical tape will also work short term. If you have any questions or concerns about soldering, consult YouTube; tons of instructional videos there.

I got this idea from this post on RX-7 Club:
http://www.rx7club.com/3rd-generation-specific-1993-2002-16/$150-pc-based-compression-tester-1021454/

Theory:
As near as I can tell from the specs, the Rotary Diagnostics ReCT-04 compression tester uses the same DataQ DI-145 I specify here. The unit is powered via USB cable. The DI-145 will work with transducers that output voltage or current signals. If you go with a transducer that outputs current, you'll need to add a 250 Ω resistor (see DataQ's website for this). The DI-145 also comes with software (the same software as the ReCT-04) so you can follow that manual for software setup for now (though I plan to write my own, eventually).

The 9milelake transducer has 3 wires:
Red +5V
Black Ground
Yellow Signal

It requires 5V power (not 12V like some of the more expensive transducers use). Fortunately, USB 2.0 ports provide 5VDC at 500 mA. The DataQ box requires < 100 mA and the transducer draws < 4 mA so you can tap into the power/ground lines on the USB cable for the DataQ. The pigtail on the transducer is fairly short so you'll need to splice in some wire to connect it to the DataQ box. The transducer outputs 0.5 V (at 0 psi) to 4.5 V (at 200 psi) but it only has one signal output. The DataQ box wants + and - signal. When I connected the sole signal wire from the transducer to the + terminal and nothing to the - terminal, the 0 psi reading was -1V. To correct this, simply connect the - terminal to ground.

I chose the 9milelake transducer because it was 1/4 the price of the next 0-200psi transducer AND because it runs on 5VDC. You could go with a Dwyer unit for ≈ $87 (and should also work with 5VDC). You will also be able to eliminate the 1/8 to 1/4 adapter. I may end up with one of those if the one I bought turns out to be crap.

Basic ugly tester:

Here's a way to make a quick'n'dirty compression tester tool. I don't have any photos of it because I took one look at it after testing assembly and couldn't leave it that way. However, if you just want something simple and ugly, this is the bare minimum.

Electronics:
1) Take an old Ethernet, USB or any other kind of cable at least 6 feet long and strip out 4 wires.
2) Strip the jacket off the USB cable supplied with the DI-145
3) Peel back the shielding
4) Cut the black and red wires.
5) Re-solder the black and red wires with the matching black and red wires that you obtained in step 1. Make sure when you tape everything up that you tape the wires individually so they don't short out.
6) Take the 3rd wire and connect it from the - terminal on the DI-145 to the black (ground) wire on the USB cable.
7) Solder the new red and black wires to the red and black wires on the transducer pigtail.
8) Take the 4th wire from step 1 and connect from the yellow wire on the transducer pigtail to the + terminal on the DI-145.

You're done with the electronics.

Hardware:
1) wrap the threads on the transducer with teflon tape and fasten it tightly to the 1/8 inch NPT F to 1/4 inch NPT M adapter.
2) Repeat step 1 for connecting the adapter to the smaller of the KD Tools 901 Air Hold Fitting Set.


Parts:
KD Tools 901 Air Hold Fitting Set
Robot Check
$7.44

9milelake Stainless Steel 200psi Pressure Transducer
Robot Check
$19.50

Speedhut Adapter, 1/8 inch NPT Female to 1/4 inch NPT Male
Robot Check
$9.15

Black 10 ft Hi-Speed USB 2.0 Printer Scanner Cable Type A Male to Type B Male
http://amzn.com/B00YNZ8P50
$2.69 (this was my source of wire for connecting everything)

DataQ Model DI-145
DI-145 USB Data Acquisition Starter Kit
$29.00

Total: $67.78

Random Laptop capable of running WinXP
I'm using an ancient Dell Mini 1012 (single-core Atom CPU) that's been gathering dust for a while.

Better Implementation:
Alright, so the previous setup is, uh, well... It's crap. I couldn't live with it. The electronic setup is identical to above but connected in a much less janky fashion.

I bought an aluminum enclosure but forgot to account for the fact that the screw supports run the entire height of the box. So, the DI-145 doesn't fit as shipped. WARNING: THIS WILL PROBABLY VOID THE WARRANTY.
Here's how I fixed that:

1) Pull the 4 rubber feet off the bottom of the DI-145 and remove the 4 screws hiding beneath them
2) The top will come off complete with terminal connectors, USB port, LED's and circuit board.
3) Use a rotary tool to trim back one of the screw posts near the bottom so that the top+PCB will fit flush on the bottom.
4) Cover the bottom of the PCB in the electrical tape.
5) fasten the PCB to the bottom of the box with foam double-sided tape.

OR...
BUY THE RIGHT BOX TO BEGIN WITH!!!! :cussing:
(i.e. not the one I list below, though it will work with some effort)

Either way, you'll need to cut holes in the box to fit the USB connectors. I soldered the connectors to the board and traced them with an ultra-fine point permanent marker. Then I used a 1/2" drill bit to drill them out and a rotary tool to turn the circular holes into squares. It isn't pretty but it's better than the janky setup above.

The prototyping breadboard I list below is nice because you'll get 8 usable boards... so plenty of extras to screw up with. Same goes for the 10 USB sockets. The breadboard has a bus strip on along each of the long sides. Use these for your + and - (ground) lines. Mount your devices on the side with the silkscreening on it and solder on the opposite side. The solder pads are connected in pairs so that you can solder a device to one and then solder a jumper wire to another area.

The USB ports have 4 pins (not including the two support fingers).
1 = 5V
2 = D-
3 = D+
4 = Ground

Electronic Connections:
1) Solder the pins for the first USB socket such that each one ends up on a different solder pad pair (see photo). (This one will be used for connecting to the PC)
2) Connect Pin 1 to the bus bar along one edge of the board using a wire.
3) Connect pin 4 to the bus bar along the opposite edge from step 2.
4) Repeat steps 1-3 one more time for the second socket. (This one will be used for connecting the transducer).
4) Take the 6-inch Mini USB cable and cut off the USB-A connector (the bigger one).
5) strip back the jacket and wires.
6) Solder the red wire to the bus bar from step 2
7) Solder the black wire to the bus bar from step 3
8) Solder the green wire to the solder pad connected to the D+ pin on USB socket 1.
9) Solder the white wire to the solder pad connected to the D- pin on USB socket 1.
10) Cannibalize two 6-inch wires from some other cable you don't care about.
11) Solder one (preferably black) to the bus bar in step 2 and connect it to the - screw terminal for Ch1 on the DI-145.
12) Solder one (whatever color you like) from the D+ pin on USB socket 2 and connect it to the + screw terminal for Ch1 on the DI-145.
13) take one of the 10-foot USB cables and cut off the flat end (the one that DOESN'T MATCH the sockets you just soldered).
14) strip back the jacket and wires.
15) Connect the red wire on the USB cable to the red wire on the transducer pigtail.
16) Repeat 15 for the black wires.
17) Verify which wire corresponds to the D+ wire by connecting the cable to USB socket 2 and using a voltmeter with conductivity (or resistance) mode. The colors don't always match up between manufacturers so don't assume that green on the StarTech cable is the same as green on a YumCha cable. In my case, the D+ wire was white (opposite from the StarTech cable).
18) I used cyanoacrylate glue (super glue) to hold the ports and circuit board in place. A bead of glue along the edge of the board where it contacts the box as well as around the USB ports will hold it in place. For extra support, I grabbed a piece of plastic from a long-forgotten project and glued it in as a post to hold up the rear section of the board.

That's pretty much it. This will power both the DI-145 and transducer off the 5V USB power and connect the - signal wire to ground. If you buy a transducer with a separate - signal wire, modify step 11 to connect to that pin on USB socket 2 instead of ground.



Parts:
ALL OF THE ABOVE PLUS...

uxcell 10 pcs USB Female Type-B Port 4-pin Right Angle PCB DIP Jack Socket
http://amzn.com/B00EZK7E28
$4.15

Black 10 ft Hi-Speed USB 2.0 Printer Scanner Cable Type A Male to Type B Male
Robot Check
$2.69 (in addition to the one above)

BUD Industries CU-476 Aluminum Econobox, 6" Length x 3-1/4" Width x 2" Height
Robot Check
$9.70

SB4 (Two Pack) Snappable PC BreadBoard, Scored PCB
Robot Check
$10.00

StarTech.com USB2HABM6IN 6-Inch Mini USB 2.0 Cable - A to Mini B
Robot Check
$2.99

Total
$97.31

(Placeholder for software setup and basic usage instructions.)

DRAFT:

By default, the WinDAQ software will show you all the channels on the device but we really only care about one of them. We'll need to tell the WinDAQ software which channel to show and how to interpret the data on that channel. The DI-145 maxes out at 240 Hz sampling rate and this is divided among all of the sampled channels. If you have 4 channels then you would get 60 Hz per channel. A rotary engine moving at 300 RPM will have rotors turning at 100 RPM. 100 RPM is only about 1.67 Hz so 60 Hz will be enough but it won't be very pretty.

Incidentally, this means that you could probably read from two pressure transducers at once so you could get both rotors at the same time if you wanted. (I will probably do this once I'm satisfied that this setup works well enough to invest another $40 into).

1) Find a computer with USB 2.0 ports running Windows XP or later.
2) Install the WinDAQ software on it.
3) Connect the DI-145 to the computer and fire up the "DATAQ Instruments Hardware Manager"
4) If everything is working correctly, your DI-145 will show up after a few moments. Highlight it and select "Start WinDaq"
5) Select Edit --> Channels...
5a) The cursor will be shown and you should type "-2" followed by to eliminate ("subtract channel 2")
5b) Repeat 5a for channels 3 and 4.

Now we need to tell the software about the data it will be recording on that channel(s):
6) Select View --> Format Screen --> 1 Waveform
7) Select Edit --> Engineering Unit Settings...
.... Upper Level Volts = 4.5
.... Upper Level EU = 200
.... Lower Level Volts = 0.5
.... Lower Level EU = 0
.... EU Tag = psi
.... Click "OK"
7a) If you chose a different transducer, change these numbers to reflect the values of the unit you went with.
8) Select Edit --> Engineering Unit Settings... (It's like Déjà vu all over again)
.... Click on "Set Offset"
.... Desired Reading = 0
.... Click "OK"
9) Select Scaling --> Limits...
.... Top Limit = 200
.... Bottom Limit = 0
10) Select Edit --> Sample Rate...
.... Sample Rate = 240
11) Select File --> Save default setup

Alright, you're ready to start collecting data and I'm ready to start going to bed.
This will be continued tomorrow.

Any updates on this? Would love to see the completed product.

Yah, sorry, been busy, uh... not doing this.

Realistically, the finished product is a combination of the first three images in my first post (except there's a cover not shown in the 1st image). I'll take another photo of everything hooked up tonight or tomorrow. I'd upload screenshots of the software but they look identical to the ReCT-04 manual.

I hooked it up and followed the Rotary Diagnostics ReCT-04 manual to set everything up. I got low compression numbers that but that was expected (I'm pulling the engine for rebuild tomorrow).

I'm thinking about redoing the PCB with a homemade custom board and a socket for another second transducer. I've never made a custom board out of copper-clad but it's just chemistry and I have a BSc in that subject. How hard can it be?

Can you draw out a schematic or make a wiring diagram? I'm a visual learner. If not I'll try and follow the directions. Looks like s fun project too!! I ask for a schematic so I can fabricate my own pcb.

I'll see what I can cook up over the weekend. I'll also be getting my car, laptop and compression tester back so I can take more photos (they're all at a friend's place awaiting the re-installation of a rebuilt Renny)

I've been playing with some PCB prototyping software and I think I can export an image or take a screenshot.

My lousy soldering skill aside, it was a pretty simple project. I'm really surprised nobody has posted something like this before.

oooo - sub'ed... The comp tester subject itself is interesting, but also I've been interested in learning about USB comm...

Here's the schematic that I just mocked up in Osmond:

Attachment 286902

This includes the hypothetical second transducer so you can measure both rotors at the same time. If you want to build one with a single transducer, just eliminate U1 or U2. The schematic is of the bottom side of the board (the side you'll put solder on).

The circuit board is just a pass-through for electrons; there's no resistors, capacitors, IC's or anything modifying them (aside from what DataQ put inside their box).

The only true USB communications happen between the computer and the DataQ box. I used a USB port and cables for the rest because it's a good, cheap, durable connector technology with exactly the number of conductors that I was looking for. Realistically, you could use RJ-11, RJ-45, DB-9, HD-15, MagSafe, screw terminals... Whatever is conductive to get the transducer signal(s) into the DataQ. Actually, now that I've though about it, if I make another I'm going to look into the feasibility of using MagSafe connectors (connectors have 5 pins but I don't know how they're wired).

The 1+/1- and 2+/2- holes get wires soldered to the board and then connected to the screw terminals on the DataQ. I cut the USB-A end off of a mini USB cable and soldered the appropriate wires to the GND (black wire), D+ (green wire), D- (white wire) and 5V (red wire) holes. The mini-side of the USB cable gets connected to the mini USB port on the DataQ.

I built one. I used the DATAQ box as the enclosure. I had to use a different transducer

It claims Output: 0.5V - 4.5V linear voltage output. 0 psi outputs 0.5V, 75 psi outputs 2.5V, 150 psi outputs 4.5V. * Works for oil, fuel or air pressure. Can be used in oil tank, gas tank, etc

Still had to set the

Upper Level Volts = 4.5
Upper Level EU = 200
Lower Level Volts = 0.5

When I set the Upper Level EU to 150, my readings were SO low, my engine technically should be able to run.

My readings were

Front
7.19, 7.32, 7.12 kg/cm2

Rear
7.26, 7.64, 7.17 kg/cm2

All normalized to 250rpm

If thats indeed true, then thats going to be a problem. My thread, post 16 details the start of my hot starting difficulty.

https://www.rx8club.com/rx-8-discuss...rx-8-a-261105/

The problem with the dealer is when I went to have my airbag changed under the recall, I asked for a compression test. They didn't take my car for another hour, and so I cancelled it because I don't want a test on a cold engine. Perhaps they would know and take it for a drive to warm in back up, but this is a dealer, so my expectations are lowered accordingly. I also am nervous that when they jack it up they could damage the pinch welds like what happened on the black RX8 I have (See the above thread for photos)

I really wish I could borrow a Mazda compression tester or a Twisted Rotors unit.

Yah, until we can validate the readings, they are only useful for watching trends.

Time to post in my local regional forum to see if I can find somebody with a ReCT-04 or TR-01.

Thinking about it, the two adapters create some dead volume. That would artificially reduce the dynamic pressure readings... I need to find the time to validate the results but my new job keeps getting in the way.

So I built and used the el cheapo version from your instructions, thanks for this! Works great!
About compensating for dead volume, did you ever figure this out?

I measured the adapter and sensor internal volume to be exactly 3mL, using a syringe and oil. Just need to know the compression chamber volume at TDC.

I think I have it:

Renny single face Vmax: 654cc
Compression ratio 10:1
Therefore Vmin: 65.4cc
Sensor volume: 3cc
Vmin with sensor installed: 68.4cc
CR with sensor: 9.56:1

By PV=nRT, nRT stays constant for this so
P(sensor)*68.4cc = P(adjusted[/I])*65.4cc
So P(adjusted)=P(sensor)*1.0459

So if you have a 120 psi peak, it should be a 125.5 psi peak. This sensor will read about 5% low.

Wow! That's awesome! :worship:

I haven't worked with this much for a while because that laptop I was using died and I haven't had the time to deal with it until recently.

In the mean time, I picked up an RCT-V5 and plan on comparing the two.

Cool, I would be curious to see how well they match. Don't really have a good way to check sensor calibration.

I noticed something else too: the peaks are a bit inconsistent. The same face will give +/- 5-7 psi, so ai'm thinking there is a sampling rate issue. If the sample doesn't hit exactly on TDC, you don't get the exact compression peak. Need to play with it to see the sample distribution per face. If you want I can share my data while your laptop is dead.

The DI-145 has a sampling rate of 240 Hz (for a single sensor, 120 Hz each for 2 sensors).
300 RPM at the e-shaft is 100 RPM at the rotor which is 1.6667 Hz. So, you have 144 data points per RPM of the rotor or 48 data points per face.

One strategy would be to take a large-ish sample (obviously limited by how much you like or dislike your car battery and starter). Then just take the highest value for each face (taking advantage of the fact that the rotors always pulse in the same order).

My laptop is back up and running now and I plan on getting more data soon (this weekend at the latest).

That's a good point, 144 points per revolution is, let's call it a tick every 3 degrees. My starter probably doesn't do full 300 rpm ( to validate with the collected dataset). So it should be landing within at most 1.5 degree of TDC, but the sensor itself has noise (at atmospheric it bumps about 0.05V which is a few psi?). So you probably get 1 tick close to TDC, but that tick can be +/- noise. If all this is true, we should see a normal distribution of peak values for each face. I'll play with this at lunch tomorrow.

Also I just had a track day, so I'll recheck compression when I change back to everyday tires. Wonder if a day of italian tuneup is worth a few psi ;)

Some more news: with this tool I can test compression much more often than before (or than is healthy), but this is interesting. Since September, my front rotor has fallen from 7.4-7.9-7.7 to 7.2-7.12-7.14. This includes 1 track day and about 1000km of mostly highway driving. So I'm not sure which result to believe, and I'm impressed how much measured compression can vary between tests.

FWIW the previous test my car had, with a TR-01, 2 years ago, showed 6.3-6.5 across the board. And the one before that in 2012 showed 7.8-8 with the Mazda tool. All normalized to 250rpm and sea level.

So it's all over the place!

Did you change premix type or concentration?

I don't premix regularly, and just so happens I premixed with Lucas Semi-Synth the tank of this last test. Other than that no engine changes since 2012.

Did you measure atmospheric pressure at the time of the test? That's the only thing I can think of.

Alright, time to resurrect this...

A project at work has me confident that I can do this with an Arduino so that I won't need to have it connected to a computer. The end goal is to replicate the functionality of the RCT-V5.
One thing I learned from my work Arduino project is this: Holy living crap is 2k RAM a tiny amount to do anything with. Ugh.

For that reason, I chose an Adafruit ItsyBitsy M0. Instead of the 8-bit 16Mhz ATmega328, it uses a 32-bit 48Mhz SAMD21 CPU which has 32k RAM. If this isn't enough, the ItsyBitsy M4 runs at 120Mhz and has 192k RAM.


I also found (what I hope is) a better 1/8th MNPT -> spark plug hole adapter. https://www.thehungrydiesel.com/ADAP...7966-R7966.htm

The SAMD21 chips are 3.3V logic chips so the 5V pressure transducers have to be converted down so you don't fry the ItsyBitsy. I have each going through a voltage divider (R1 = 240 ohm, R2 = 510 ohm).

I've also added a barometric pressure sensor (for altitude/weather correction) and a 20x4 LCD display. The LCD display and it's I2C backpack are 5V units so I purchased a serial-friendly line-level converter.

I just got done calibrating it with a bike tire pump. Not exactly a NIST traceable standard but it's what I've got handy right now.

That was all the easy part.

The hard part is figuring out the programming.
I'm not a very good programmer. In fact, I'm a fairly crap programmer.

Lets say the motor spins at 300 RPM. The rotors spin at 1/3 of the eccentric shaft but there are 3 of them so that cancels out. We should get 300 pulses/minute or 5 pulses/second. However, we need to capture many samples per pulse to ensure we record the peak. This math is a bit above my current brain level. Hopefully, appropriately sampling two 5 Hz signals should be do-able for a 48 MHz processor.

Okay, so, we think we can sample fast enough. Now we need to figure out how to...

1. Wait for cranking to start.
2. Save only the peaks into an array.
3. Select three consecutive peaks from the array.
4. Calculate the time elapsed between the first and last data point, convert to RPM
5. Adjust the results for current barometric pressure
6. Normalize the results to 250 RPM
7. Display the results
8. Tell the owner to start weeping (or celebrating).

Really, it's number 2 that's going to kick my ass.

To quote Steve from Monster House, "All that's left is everything."

Hah yeah Arduino memory is hilarious. I was going to say take a raspberry pi, if you have no power constraints. Then you can run whatever you want. You can even put it in a gameboy case. Or an Arduino Méga, that should give you the 0-5v inputs you want and 256k?

As far a sample frequency, I think you need to sample far faster. 5hz works if you catch the peak exactly. The peak duration is tiny, like 5ms, so you need to sample at 200hz+ to confidently catch the max of each one.

You can use the old derivative method to find peaks. Anywhere where the slope is 0 is either a min or a max. Throw away the mins and you're set.

As for when to start recording, record the last x samples continuously, so the problem becomes detecting stops. This should be easier, you're just looking where X time has passed without peaks. The first few peaks are garbage anyway, the starter doesn't reach full speed until after a few revolutions.

This has taught me one thing - boy do I need to brush up on my electronics skills. Not since Radio Shack days....