Thank you kindly. I never would have asked that before, but I have a "spare" MAP sensor to show for that effort now :bigtu:

OK, VVT is wired. One wire to +12v, one wire to INJ9 on the AEM.


On to the settings, and understanding WTF they do:

Attachment 199932

Starting lower left-hand corner, I assume this is the 2D table dictating what degree of advance/retard correlates to what percentage on the solenoid. The hard part about this is as Jason said - it's not a straight correlation, and we have to increase the duty to get the VVT to advance and then bring it back to stop advancing it. I think this map is basically going to get set so that the solenoid sees ~.54v, which is steady cam timing, across the board, and then use the error table to actually do the adjustment.

Upper left is the timing map. Load on Y, RPM on X. The cells are either duty cycle percentage or cam advance. Because you can't put negative numbers into these cells, I am thinking this is the duty cycle percentage, but then there's no negative numbers in the 2D table either, so maybe it's based on total degrees of advance from the max retard point (i.e. +12 would be 0deg adv/retard on the Miata, since there is a max of 12deg of retard).

Lower right is the money table, the timing error vs. duty cycle table. Because there is no direct correlation between the DC% and the advance, this is the table that will do the adjustment. As the RPMs increase and the desired retard increases, the error will get larger and larger, which will trigger duty in one direction or the other. As it gets closer to the target, the error decreases, brings the voltage back closer to .54, etc.

I have absolutely no idea what any of the stuff in the options box does. I think Cam Range would be the 360/number of teeth on the gear to get degrees per tooth, but I'm lost on the rest of it.

The VVT Feedback options, I figure the VVC FB min/max are the limits for the error table, but I don't know what the VVC#1 FB I/P settings do.

Tell me how wrong I am now, please :hustler:

Keep VVT disconnected.

See if your setup can read cam phase to being with.
Rev it to ~3000 RPM. Apply 12V to solenoid.
See if it reads 47* less retard or more advance.
Double check too that in the cam/crank trigger advanced screen, that advancing the cam doesn't cause loss of sync.

Then let's discuss closing the loop.

Cam range I think is the # of crank teeth time that the cam can move around. With the factory 99 trigger wheel, you have 1 tooth per 180* of crank rotation. So 47* would be 0.261 teeth of range.

Later.

Correct. You will have to find the duty cycle such that it holds advance. I posted the estimated duty cycle in another thread. BTW this system will probably work better if you use ~100~200 Hz and you placed an SB530 diode backwards across the solenoid (cathode or stripe connects to the 12V side of the solenoid)

Later when you actually start tuning this, lets start with advance not changing with MAP but only changing with RPM. It will make tuning easier.

Wrong. This is your advance target vs RPM and load. It's just that "advance" is expressed as a non-zero number. So max retard would be 0 or close to 0, and max possible advance would be 47. WARNING - I may have it backwards, and the AEM may interpret this as RETARD - so that max advance may be near 0, and max retard will be 47. Do a search on the AEM electronics forum - you'll probably have most luck in the K20 section.

Correct.

VVC FB P and VVC FB I are the gain terms of the P and the I in the PI loop, respectively.

In AEM convention, you set these to be negative numbers, unless your system gain is inverted, such that more duty cycle decreases the thing you're measuring/targeting.

Because you're usually better off using the error table, set the P gain to 0.
The I gain should be very small, and it's only used to find the exact value of "hold the advance" duty cycle.

The FB max and min are the max allowed change in duty cycle as driven by the P and the I terms .... e.g. you put in +10% and -10%. As for the latter, I forget if the convention is positive or negative... if positive, the min will be -10.

Can you provide a layman's explanantion of what the SB530 diode will do or would it just confuse me? :giggle:

I agree. Once I have the 2D version of that map down (no MAP changes), I am interested to see what kind of response gains can be had by advancing the cam at low load (if any).

Cool, that's kind of what I meant. Timing as in relative cam timing target. I will check on the AEM forum. Just a linear line from full advance at idle to full retard at redline should get us going, once I know what values correlate to full advance and full retard. Even if the K20 section turns up nothing, one test drive will reveal whether the slope of that graph needs to be positive or negative (advance at 0 or 47).

It's for the protection of the output pin and the driving transistor on the AEM processor. When the output pin switches to low level, the current tries to continue flowing through the coil and it is harmlessly diverted through the diode. Without the diode no current could flow and the coil would produce a damaging high voltage 'spike' in its attempt to keep the current flowing.

Jim

The standard output driver in the AEM has a voltage clamp, a 60V zener diode (built into the MOSFET). When the current to an inductor or solenoid is interrupted during the off time of the PWM signal, the inductor's voltage "flies" very high and is clamped to 60V. The current then decays rapidly. The SB340 clamps the voltage and prevents the current from decaying very rapidly. The net result is that the average current (over the duty cycle) follows the PWM duty cycle more closely and the solenoid will require less duty cycle to reach the target current. The main desired effect is that the current will be a more linear function of duty cycle, making it more predictable.

Jason, SB340 or SB520?

It didn't read anything different. I assume I am looking here:

Attachment 199889

The motor ran very rough, but it did maintain sync with +12v on the solenoid.

The lowest it will allow me to set that number is 0.39.

sb520.

Can you show your advance cam/crank sensor screen? There may be a setting there related to VVC.
If you get your settings to read, it should read some value under "Cam#1 ADV" and "Cam#1 Angle" even when running at full retard.
Do they read anything while idling?

Finally got round to installing the AEM s/w on my recently crashed laptop...

Cam range looks like simply a multiplier that scales teeth to displayed degrees.

I will assume it's crank teeth. So with the factory 2 tooth wheel, it's 180* per tooth.

The K20A calibration that comes with AEMPro uses 37.11 in that cell.

It may have been that I had the VVT control off - I'll turn it on in the software and try again today.

Looks like the AEM won't accept 180° in °/tooth. Put in 90 and remember that the numbers displayed are in CAM degrees and not crank degrees.

Where is the god damn fan control page? I am getting tired of the fans kicking on and off a bajillion times a second because there's no hysteresis because I can't find the page to add some.

3000rpm, Valve control on, */teeth set to ~90, no change. The car usually idles around 29-30kpa, at 300rpm as soon as I hit the solenoid with 12v, the RPMs drop to 1800, 60kpa, runs rough, and the cam angle drops from ~1.9 to ~1.6.

Attachment 199689

Open AEM Tuner. Go to: "Tools". Next, hit: "Configure Outputs". There, search for Radiator Fan. You're done.

If you're using AEM PRO. Go to: "Options", hit "List Selection" and then unlock: "Radiator Fan". Set temp Off and On. Done.

Aem tuner no workie correctly with Gen 1 boxes.

Is your gutted thermostat in the back of the head? Where's the temp sensor? Just wondering.

Both in the back of the head.

This is promising. 1.9 to 1.6 teeth is delta of about 54* (crank) which is close to the 47* that the intake cam moves. Looks like it measures the distance from the last significant crank tooth, to the index cam tooth. So more advance = shorter time, fewer teeth. If you then multiply this by 90 (deg per tooth), it will read retard and not advance.

What I don't understand is why the parameter "Cam #1 ADV" is reading 0.
Try setting the option "Cam #1 Start" to 1.7 teeth, just to see if it will start reading something.

You may want to browse the K20 section on the AEM forums.

Thanks Rafa. It was really wreaking havoc with my idle. Getting these 1000cc injectors to idle steady at anything except 11.5:1 AFR is a little tough.

Back of the head, temp sensor is just past it. It would be in front of it but there's something funkly in the casting of my head and my temp sensor won't thread into the stock hole (it's plugged on 01 heads normally but its still there). Eventually I will get rid of it entirely and pull my coolant temp readings from the AEM via an IQ3.

You're welcome. You're doing a great job.

Please keep this thread going.

I have a reading. 0.31 cam#1 ADV. It took plugging in the solenoid to make it work. It seems you can't just trigger it with a +12v and get it to read. I'm going to see if I can advance/retard it at idle using the duty cycle table, leaving the target and error tables alone.

Attachment 199670

Can't seem to adjust it, though. No adjustment to any of the tables appears to do anything.

When all else fails check the manual.
Page 166 of AEMPro User Guide V2.0.pdf in C:\Program Files\AEM\AEMPro\Instructions describes VVC.

I was right about range. The exact minimum value of the 1.6 number that you typed, minus a small tolerance, should be the "start" value. This will be the position that the EMS will call "zero".

Range has a funky 0.8 factor:
Cam Range = (360 Degrees/# of Fuel Teeth per rev)/0.8
And it doesn't accept the actual range of the cam - I guess the software doesn't really care how far the cam can go, it just uses this number as a scaling factor. Again because it doesn't accept the 112.5 number which the formula would yield, enter half of that, or 56.25, and all your degree units will be in cam degrees.

It does appear too that a bigger number is more *retard*, so think of full advance at 0*, and full retard as 23*.

Because the numbers represent retard and not advance, you need the error table to increase as error increases, and you need the 'I' factor to also be positive. I would first try something like 10% at 5* error, -10% at -5* error, 20% at 10% error, and flat outside that.

As discussed earlier keep the entire duty table flat at the duty cycle that holds the current phase. Except that at max retard, bring it down to zero. So flat all the way to 25*, then zero from 30* onwards.

BTW do this: Configure-Unit-AFR Units, select AFR Gasoline so your AFR reads normally.

Did you install the SB520 diode?

Your VVC out max should be what I posted in the other thread, the max duty needed to get maximum rate of advance.

I would set 'I' to 1, and set VVC FB max and min to +10 and -10 respectively.

Like this.
In this example I set the targets to full retard everywhere.

Will do, next week. My number 2 injector locked wide open while I was fiddling with all of this. Sending them all back along with a very nasty phone call on Monday - these are about 40 miles old.

OK, folks, I need a bit of help with this. I switched injectors this evening to isolate the issue and it appears the AEM is locking injector number 2 open as soon as I crank the car. We're talking enough fuel to hydrolock the cylinder, and it blows puddles out the exhaust. I'm reloading my calibration to see if I somehow managed to screw something up there, but what's my next recourse? Send the box back to AEM?

It appears to be fixed. Reloaded my calibration from the dyno a few weeks back and it appears to have stopped. Going to give it 12 hours to clear the fuel from the exhaust and try to start it back up.

Jason, do you know what the period of the solenoid is (the first box)? I can scope my friend's 2001 tomorrow but wondering if you've already done it.

Under crank settings there's options for "inject all" or run batch fire during cranking, amount of time to inject extra fuel before starting, etc. I think. One of those may be way off for your huge injectors and making it seem like the injector(s) are locked open?

But glad to hear it's fixed.

Sav, between 100 and 200 Hz should work for that solenoid. i.e. between 5,000 and 10,000 uS. Someone here scoped it, may hvae been Reverant.

Guys, we have a problem. Something in the VVT system is locking the 2nd injector (the one that fires on 0 tooth) open as soon as the motor cranks over. I just switched the #2 injector to INJ10 because I thought it was damaged. Fired up fine. Went to plug VVT in and immediately had the same problem. I am wondering whether something in the injector phasing is screwing things up.

I am going to turn off PWM#9 and the VVT control in the software once the car cools down and I can empty out the cylinder yet again - if the car runs normally with VVT control turned off we need to solve this before further testing can continue :(

Did you have the same problem with inj output #2?

Inj 10 is used for VVC when it is enabled; you can't use 9 and 10 as injector outputs when you use VVC.

Jason, pretty sure you have to enable each one in VVC#1/2 options to control solenoids vs. injectors. I had the exact same issue with injector driver 2, thought I had blown it out inside the box so I switched.

Also, there's no fuse in the VVT slot, so the VVT solenoid was never getting +12v power. Durr. Once the car is clear of fumes in an hour or so, I'll go back out and try again with the VVT solenoid unplugged. It ran fine, and then as soon as we plugged the VVT solenoid in, it immediately flooded.

I do not know whether the flooding is isolated to software settings or having the solenoid plugged in - I will check that in a bit.

It is a software issue.

My best idea is that it is something in the injector phasing. Injector 2 is triggered on the "0" tooth.

No the phasing just changes when the injector fires.
Just a random check, try inj output #6.

The only other thing I can think of is that there is a short in your harness between the wire going to the injector and the VVT solenoid.

Looks like hardware issue to me, not software ...

After Savington mentioned flooding cylinder #2 I did some digging in the software and seems to me to make the '90-'92 Miatas work (the 1710 box covers all the '90-'95 Miatas), on the daughter board AEM have connected the injector output 9 & 10 to injector output 1 & 2. This way they could implement the batch injection (software and the ECU doing sequential and the daughterboard faking the batch by shortening the driver pairs). Just take a look at the '90-'92 Miata calibration ... 9 & 10 injectors are enabled ...

Are you guys sure, outputs 9 and 10 are really on the extra connector 3O and 3P at all?

I remember AEM Engineer telling me in the past about the 1710 Miata box needing hardware tweaking to support VVT (because there are no injector outputs 9 and 10 routed to the connector) and I mentioned it here some months ago . But someone argued about my statement saying 3O and 3P are injector 9 and 10 (I have no idea whether he made any tests at all).

So ... my suggestion is test the injector outputs manually to see if you have 9 and 10 in the connector at all or whether those two are connected to pins for 1 and 2 and you either have to move some jumpers on the daughter board to separate them or you may even have to cut some traces on the daughter circuit board and route them to the proper pins.

To do a test:

1. Open AEMPro, go to Options -> Injector screen
and turn off all injector outputs there (so the software temporary doesn't treat them as injectors).

2. Disconnect the plugs from the injectors.

3. Since all our injector outputs are not treated as injectors anymore but as general LS drivers.
In AEMPro, open Options -> Configure Outputs and there manually toggle Fuel 1, 2, 9 and 10 to ON one by one.
Monitor their status in a parameter window (Ctrl-P) and check with a multimeter the pins 2U, 2V, 3O and 3P.

My prediction is that when Fuel 9 is ON, pin 2U will be grounded instead of 3O and when Fuel 10 is ON, pin 2V will be grounded instead of 3P.
So, to make 9 go to 3O and 10 to 3P, you'll either have to move some jumper or do some trace cutting and/or soldering on the daughter board

This is just for Savington's AEM EMS. Jason uses one with a Honda daughterboard so I guess his is fine.

This statement most likely is true, but after my theory in the previous post -> the short is by design and is most likely on the daughter board, so he has to fix it there.

This makes perfect sense. Only with PW#9 turned on does it ground the injector out. I tested the harness, there's no continuity between the wire for injector 2 and the wire for the VVT.

Julian, if I opened up the box and snapped a couple photos would you be able to tell me what I need to cut/move?

Or instead of taking apart the board/moving jumpers/cutting leads/resoldering/whatever, just use move (physically and software wise) injectors 1 & 2 to 5 & 6, after you've done this check to see if it's the problem.

Ah, that's the better idea! I guess my brain was half off that late last nite :D

What Tim suggests also won't break the box warranty if it is brand new.

Move the OEM injector harness wires from inj 1 & 2 to something like 5 & 6. In the software in "Options->Injector" disable 1 & 2 (disable = uncheck "Active") and enable the two new ones (set them as primary, etc. options there just like 1 & 2 were).

Then open the "Fuel->Advanced Fuel->Injector Phasing" and put the values you have from Inject Tooth #1 & #2 into the two new ones you used.

After this start the engine to make sure you did the switch correct.

Then connect the VVT solenoid to injector 2 (or 1 - whichever the 9 is shorted to)

Jason is the hardware guy - he could give you a much better guidance :) But I think Tim's idea is much much better because you won't have to mod the ECU at all and all you have to do in hardware is move the 1 & 2 injectors wires to two other pins.

It wasn't my idea. Jason posted it first and I thought it made more sense. I just retyped it to make sure Sav tries it.

OK, I moved injectors 1 and 2 over to outputs 7 and 8. Still not totally sure why activating injector 9 as VVT control is grounding out 2/10 but whatever. Hopefully it's solved now. Changed the active injectors and injector phasing and the car fired up fine. I will play around more with VVT tomorrow and see if I can get it to run the circuit without locking an injector open, and then power the solenoid and start zeroing in on some settings.

It appears to be solved. I can activate PWM9 in the software and the injector stays closed as it should. I'll move on to settings today/tomorrow.

One quick question about knock sensing - I hooked my Bosch sensor up to the AEM and stuck it on the side of the block. I placed the sensor in the OEM location using the FM adapter. I can tap the sensor and get ~1v out of it, but when the car is idling and revving it won't go above .02v. Any ideas?

You may find the answers to this question in this thread: https://www.miataturbo.net/forum/t33781/

J_man's answer to that question. Here you go:

Quote:
Originally Posted by stinkonamonkey
You can use a saturn knock sensor for the EMS. Search the AEM forums for details. You have to run a shielded wire to the sensor and mount it to the block using a modified bolt. I've got one installed but I haven't got it working real well yet.


+1. The resonant Saturn sensor is the one he needs.
Flat response sensors won't work.

I posted a bandpass filter and amp for the AEM to be used with the 99 knock sensor.
There was a thread about using said circuit for a Bosch type sensor. I answered there but I don't remember my answer.

The GM resonant sensor center frequency is a bit off from the miata's knock resonant frequency. Those resonant sensors have extremely narrow bands so I don't know how well they would work compared to a flat sensor with a proper bandpass filter.

BTW the Link doesn't have a bandpass filter either based on a schematic I was shown. (Unsure ifi I saw the FM version)

I may have a '99 sensor laying around but I know I have a 2002 sensor. Mazdacomp's site is cocked up so I can't check part numbers, but are the 2001+ sensors similar to the 99/00 sensors?

I don't see any reason for the 99 sensors being different from 01.

Flat response/bosch/wide band knock sensors do work, I use one. It just doesn't work as well as something setup specifically for the miata, or something that outputs up to 5volts like a dedicated filter & amp circuit would.

I can check when I get home Andrew, but I don't think the knock sensor volt moves much when I rev my engine either (using FM supplied bosch sensor from my Link ECU days). But it definitely "spikes" when I get knock. The spike is less than 1 volt but it's still obvious when there's normally a constant 0.x volts. You have my settings to see how much timing it pulls/fuel it adds per volt but obviously there's a bunch of resolution not used when using a bosch sensor wired directly to the box.

EDIT: Sooo a dedicated black box tuned/amplified for use with a Bosch knock sensor will be included in the crank trigger/vvt kit you guys are putting together for us lazy AEM users, right?

I had just a Bosch flat response sensor in the past connected to the AEM - it is kinda unusable. It picks the entire frequency range and often there are high level sounds which are not knock frequncy but could fool the ECU.

I currently have the Saturn resonant sensor hooked to the AEM.

Jason says Miata knock frequency is abnormally high - if that's true I'll switch back to the Bosch sensor with his bandpass filter / amp (you have to filter out the proper frequencies and amplify the signal before feeding it to the AEM if using a flat response sensor).

Hey Jason, was your frequency pick based on just that single wav file you posted back then (with a couple of knocks in it)?
Or did you test with a bunch of induced knocking at low, mid and high rpm to confirm that the frequiency is the same over the rpm range at both mild and strong knocking - but you never posted those?

Both should be the same since it is a simple flat response sensor - the filtering and amplifying in the OEM case are done in the OEM ecu.

I only tested off-idle knock and found 13 kHz to be the best center frequency.
I would appreciate some knock recordings - I can analyze them, and then I can re-adjust the center frequency of my circuit.

13 kHz is not "abnormally high". I tested some GM Saturn sensors and IIRC they were at 12 kHz IIRC.

Yeah, I've tested your wav file too and saw the freqiencies there.

I should've said compared to that GM paper often posted about how to calculate the knock frequency knowing the cylinder bore. Following that formula IIRC the Miata engine knock was somewhere around 8Khz.

With that other thread about the electronic det cans group buy, I hope people post knock recordings too sometime soon :)


Hmm. John @ J&S posted in the past that the Saturn sensor has lower than 12Khz resonant frequency. No idea whether he tested but IIRC the sensor part info came from him.

These settings are what I started with. I can run the duty up to 90% and there's no effect on the solenoid. I raised the output limit to 100% since it was being limited at 48 and it still doesn't want to respond. Idling at a steady 11.8-12.0 cam degrees.

Post a screenshot of the cam #1 template.

Did you look at the parameter screen to see what duty cycle it's actually outputting?

And, if your "cam start" number is correct, it should idle at ~23* cam degrees if you're outputting zero duty cycle.

J_Man, a super easy way to find the resonant frequency of a piezo sensor is to use a signal generator with 1V or so of output on it. It will act as a loudspeaker. there is a very narrow range of frequency where it gets loud. I posted this a long time ago on MF.

Here I found my thread

http://forum.miata.net/vb/showthread...ight=resonance

Two piezo sensors with the same part number show 11.6 kHz and 8.5 kHz. I have a feeling the 8.5 kHz one got over-tightened and damaged.

If you listen to the recording, you get a descending note, which I would think is because the piston is moving down the cylinder. Which means that bore diameter is only part of the picture.

Also, 13 kHz is exactly one octave above 6.5 kHz ......