drifting timing with rpm
4 Attachment(s)
I have a 99 turbo runing MS3 Basic stock triger and GM coils, i see an increasing timing discrepancy as rpm rise when i look at timing on the engine with a timing light, i could see it on the dyno when i set all the timing numbers to 12 deg on timing table 1 and did a run and i can see it reving the engine after setting timing to 10deg fixed timing (not the timing table) the timing shift is about 14 deg at 6000RPM and aroud 16 at 7000rpm.
I will test a different cranck sensor tomorrow to see if that changes anything. Any ideas of what can cause this? |
Here is a utube video of what i see with a timing light when reving the engine with fixed timing set at 19 deg
|
What do you have the spark hardware latency set to?
--Ian |
2 Attachment(s)
it is set to 0, can i put a negative number there?
|
Start with 100, see what happens. That's what rev used for a 32-1 wheel I've used.
|
The exact number varies from car to car, but it will always be a positive number (representing electrical delay from crankshaft to ECU) and will generally be in the tens to lower hundreds.
It's a trial-and-error setting. |
Thanks, i will test it tonight
|
I progresevelly changed and burned with each change the spark latancy from 0 to 255 (the maximum) with fixed timing set to 10 BTDC, the timing still advances with rpm from 10 at idle to arownd 20 BTDC at 6000rpm, there did not seem to be a difference as i changed the latency number, my tuner said that some cars have this issue and is caused by sensor variation, he dos say that this is the only Miata he has seen with this issue.
|
Chiming in, as it's definitely the first time I see it as well.
I would start by updating to the latest stable firmware, which is 1.4.0. The latency is only used if the timing retards, not if it advances. Also, this might seem silly, but do try with a different timing light as well. |
Originally Posted by Reverant
(Post 1283842)
The latency is only used if the timing retards, not if it advances.
Strange happenings afoot. Prediction? |
Thanks everybody, i will try a diferent timing light, pickup sensor, and the latest firmwere, i am still runing 1.4 alpha 9 that Scott Clark from DYI installed before tuning my car earlier this year.
|
2 Attachment(s)
Originally Posted by Reverant
(Post 1283842)
Chiming in, as it's definitely the first time I see it as well.
Also, this might seem silly, but do try with a different timing light as well. We were going to tune with the flex fuel sensor and E85 but desided to wait until i can figure out what is going on with the timing. |
The spark latency setting is to help control timing retard induced by the vehicle's electronics. It will not correct self advancing ignition in relation to RPM.
You may be triggering off the wrong edge of the tooth. I have seen cases where an unstable edge will cause timing advance with increased RPM. |
Ben, i am runing OEM triger wheel and sensor, i see that you reccomend 2mm tall teeth and thicker steel with your sensors.
Do you reccoment i install one of your wheels? I dont see a bolt on wheel on your website but would like to know if you have one that you reccomend. Thanks |
Why?
Just try different input capture edges and see if you find a combo that works better than current. |
All my units work with rising edge and going high. Setting to falling edge will cause the edges to overlap and the car won't start. These units use optocouplers, and I've never seen this issue before.
Which firmware are you on right now? |
I am runing 1.4 alpha 9 that Scott installed before tuning the car earlier this year
|
Alpha firmware should NEVER be installed if you don't have a reason to try it. Scott should know that. Upgrade to 1.4.0 or downgrade to 1.3.4.
|
The reason for loading it before the tuning was that we wanted to run the flex fuel sensor and E85 blending maps during that session and 1.4 had some new features in this area.
I will load 1.4.0 and see if that changes anything |
Originally Posted by Reverant
(Post 1284162)
All my units work with rising edge and going high. Setting to falling edge will cause the edges to overlap and the car won't start. These units use optocouplers, and I've never seen this issue before.
Which firmware are you on right now? Mine is showing falling edge and going high (inverted) and I've had issues cranking forever but once it starts it runs ok. That is what it was on default when I got it. |
drifting timing
2 Attachment(s)
I wonder if my ecu happens to have the wrong or malfunctioning clock crystal, I see that I am never getting to my 7200rpm redline on any of my dyno plots when that is the hard revlimit and I have a soft revlimit at 7000rpm, the wrong clock cystal may change timing on the MS3?
I remember changing cyistals on my stock ecu when running STS to get a few hundred more rpm's and that changed fueling, in my case to lean the engine a little bit that was a good thing since it was too rich to begin with. |
Originally Posted by Dlaitini
(Post 1285573)
Is this true with your MS2's as well?
Mine is showing falling edge and going high (inverted) and I've had issues cranking forever but once it starts it runs ok. That is what it was on default when I got it. |
Rev,
Can you send me crystal P/N is on your MS3basic builds? |
This is definitely not a crystal issue. If it was, you would not have communications to the ECU, as serial is time-based, and if the crystal was not working right, you wouldn't be able to communicate with the ECU.
On top of that, the crystal used on the MS3 is "as-is" on the MS3 daughterboard, I don't have a P/N as I buy the daughterboards preassembled, just like everyone else. Have you tried the stable firmware yet? |
I downloaded 1.4.0 from DYI's site and was looking for instructions on installing the firmware since i dont remember if i need to install a jumper on for this or not and it has been over a year since i did this, hope to have time to to it this evening to load it and change the crank triger sensor.
|
No, you don't need to install a jumper.
|
2 Attachment(s)
Well i atempted to install the latest firmware but ms3loader_win32 could not find the port, i re-started the computer to make sure that tuner studio was compleatley off and moved the usb comunications cable to different ports 3 times but was not sucsesful.
Could it be because this is a windows 64bit computer, the port check program can see it and tuner studio can as well.:vash: |
Originally Posted by HHammerly
(Post 1285776)
Well i atempted to install the latest firmware but ms3loader_win32 could not find the port, i re-started the computer to make sure that tuner studio was compleatley off and moved the usb comunications cable to different ports 3 times but was not sucsesful.
Could it be because this is a windows 64bit computer, the port check program can see it and tuner studio can as well.:vash: |
Thanks Ben!!! I will look for it on your website.
Coming to PRI this year? |
|
Originally Posted by HHammerly
(Post 1285793)
Thanks Ben!!! I will look for it on your website.
Coming to PRI this year? I'm not optimistic that replacing your crank sensor will solve your issue. I still suspect the ECU is looking at an unstable tooth edge. |
I will try different sesnor distance from the trigger wheel (it is about 0.03" right now) if that does not fix it ill try the other sensor and other sesnor settings...
|
After adjusting sensor gap, using another sensor, changing sensor settings (it runs going high and going low BTW) and not finding a solution i changed dwell from 5ms to 4ms and BINGO!!! That was the culprit timing actually retards with rpm by 2 deg at 7000rpm.
Thanks everybody!!!! Time to schedule another tuning appointment �� |
Ben, I noticed on the 1.4.0 release notes that the table blending is not working any estimate on when that may be fixed?
Thanks! |
Megasquirt Support Forum (MSEXTRA) ? MS3 pre-1.4.1 beta 2 released (View topic)
Beta 2 fixes a Blend curve 3 issue. But I didn't see anything in the 1.4.0 release notes MS3 1.4 firmware Key changes and Gotchas - from 1.3.x and earlier ======================= 1. Baro Previously the code used to divide by the barometer when calculating the fuel pulsewidth. If using an old tune you need to enable the "old style" baro calculation to enable the old behaviour. 2. CLT rev limiter The TPS bypass rev limit setting has been removed, the standard hard limit is now used in bypass mode to reduce confusion. 3. Closed-loop boost The initial values table is now a bias table. This needs to be tuned as an open-loop table using "setup mode" before enabling full closed-loop. See the tooltips and manual. 4. On/Off outputs using bitwise AND The meaning of threshold and hysteresis has changed, see the manual. 5. Other changes are covered in the Setting Up manual. Key changes and Gotchas - from 1.2.x and earlier ======================= 1. Idle control settings re-arranged Need to reset: Stepper vs. PWM idle valve Open-loop vs. Closed-loop Output pin for PWM. 2. Closed-loop idle control The settings are re-arranged to simplify setup, but existing users will need to retune. 3. GM/TFI ignition settings changed. 4. MAF configuration changed. 5. VSS input configuration changed. 6. Generic PWM output type and frequency selection changed. 7. On/Off outputs enabling changed. 8. Various tables now use "kPa" instead of "%", the numbers should be ok. 9. VVT adds min/max duty settings, you may want to use them. 10. Two TunerStudio dash indicators have been renamed: Not Synced -> Not RPM Synced Half-sync -> Half-RPM sync You will need to re-load you dash (right-click on a blank area of the dash select Load/Save, then Load dashboard, then Accept) or update manually. 11. Rev limiting is changed. Check your settings. 12. Some "trigger wheel" users may need to change their tooth#1 angle if it was close to 360deg or 720deg. (A previous bug was fixed.) Key changes and Gotchas - from 1.1.x and earlier ======================= 1. MAF There is a new MAF calibration implementation like MS3. The old method is still valid. 2. Baro The default settings for baro were on the MAP sensor calibration page as 147, -47 with a tweak curve on top. The new method sets these two numbers to zero and exclusively uses the adjustment curve. 100% means un-altered fuelling. 3. Air-density In previous versions, there was an internal calculation for air density with a tweak curve on top. The new method exclusively uses the adjustment curve. 100% means un-altered fuelling. Note: MAT does change air density and the speed-density equation relies on this to estimate intake charge. 4. Spark output polarity The name has been simplified "Going High" - was called "Going High (Inverted)" "Going Low" - was called "Going Low (Normal)" DOUBLE CHECK YOUR SETTING BEFORE CONNECTING COILS. 5. Boost Control To make things more intuitive we have now changed the firmware so that - "Normal" is the most typical output polarity setting - larger boost duty% numbers mean more boost. If you are a new user starting out from scratch, then nothing needs to be changed, just use the default settings. If you are upgrading from a previous firmware version, then the Boost Output polarity setting will be the opposite. For most users it will now be "Normal". This is true whether using open- or closed-loop boost control. If you're using closed-loop, nothing else should need to change. Open loop boost users most likely need to set their table so that: new_cell_value = 100 - old_cell_value Remembering that more duty = more boost. 6. Serial protocol The firmware now uses the "newserial" protocol. This requires compatible tuning software and firmware loaders. (Note, it you want to revert to an older firmware version you will need to use the new firmware loader, or use the boot jumper.) |
Originally Posted by HHammerly
(Post 1286021)
After adjusting sensor gap, using another sensor, changing sensor settings (it runs going high and going low BTW) and not finding a solution i changed dwell from 5ms to 4ms and BINGO!!! That was the culprit timing actually retards with rpm by 2 deg at 7000rpm.
Thanks everybody!!!! Time to schedule another tuning appointment �� |
3 Attachment(s)
I have LS truck coils, i'm not sure what you are asking regarding setting dwell with a scope, i changed dwell in the ignition settings
https://www.miataturbo.net/attachmen...ine=1448330927 |
2 Attachment(s)
Originally Posted by HHammerly
(Post 1286119)
i'm not sure what you are asking regarding setting dwell with a scope,
For any given combination of ignition coil and driver, the time required to achieve maximum current differs. Thus, the most optimal method for determining the ideal dwell-time for any given coil is to attach a DC current probe to its primary side and look at it with an oscilloscope. Here are some scope traces I took years ago of the stock coils on an '02. First, I dwell the coil for about 4.7ms: https://www.miataturbo.net/attachmen...ine=1448333045 (pardon the noise- lazy wiring on my part.) I don't remember the scale of the probe I was using at the time. Let's arbitrarily say that it's 1V / A. You can clearly see the coil current start at around 1.5 A and then starts rising more or less linearly. After about 3.1 ms, the coil primary reaches saturation and the current levels off at 2.2A. Dwelling the coil beyond this point produces no further rise in primary current, so in this example I'm just needlessly creating excess heat. Here, I decrease the dwell time to about 3ms: https://www.miataturbo.net/attachmen...ine=1448333045 This is just about right. The primary cuts off at pretty much the exact point that primary current reaches its maximum. I could have dwelled the coil for another few tenths of a ms just to be safe, but this is in the ballpark. |
Thanks Joe, i will set the dwell to 3.1 ms and get the car re-tuned, i would have expected that very long dwell timing would retard timing as rpm increases rather than advance it...
I set the dwell to the 5MS time because i found in this in MegaManual. "The LS1 built in coil igniters (the amplifier that drives the coil's primary current based on the sequencer signal) will follow the sequencer signal pulse width. When the signal from the sequencer is high (3 to 5+ Volts - with very little current from the controller, a few dozen milliAmps), the coil current will be building. When the signal from the sequencer is pulled low (shut off), the coil will spark. The duration of the signal from the sequencer determines the dwell (though the coil igniter limits this to no more than ~8 milliseconds). As the dwell time increases, the peak coil charge current increases. The coil output current is not as linear. The LS1 coils do not fully saturate until around 8 milliseconds, but the spark energy does not increase much when the dwell exceeds 6 milliseconds. So stick with a 'running' dwell setting near the design value of 5.6 to 5.8 milliseconds. This will keep the coils cooler and extend their life. Note that MegaSquirt-II uses a nominal dwell setting, called the 'maximum dwell', at 12.0 Volts and adjusts this for the continuously measured running voltage (since higher voltages charge the coil more quickly, the dwell is shortened, and conversely, low cranking voltages need extended dwells to properly charge the coil for easy starting). To get 5.6 milliseconds of running dwell, the nominal dwell parameters should be set to: " |
Originally Posted by HHammerly
(Post 1286131)
Thanks Joe, i will set the dwell to 3.1 ms and get the car re-tuned,
i would have expected that very long dwell timing would retard timing as rpm increases rather than advance it... I don't think dwell time is the root cause of whatever problem you're having. I'm not 100% certain it isn't but I can't see how it would be. "The LS1 built in coil igniters (the amplifier that drives the coil's primary current based on the sequencer signal) will follow the sequencer signal pulse width. When the signal from the sequencer is high (3 to 5+ Volts - with very little current from the controller, a few dozen milliAmps), the coil current will be building. When the signal from the sequencer is pulled low (shut off), the coil will spark. The duration of the signal from the sequencer determines the dwell (though the coil igniter limits this to no more than ~8 milliseconds). As the dwell time increases, the peak coil charge current increases. The coil output current is not as linear. The LS1 coils do not fully saturate until around 8 milliseconds, but the spark energy does not increase much when the dwell exceeds 6 milliseconds. So stick with a 'running' dwell setting near the design value of 5.6 to 5.8 milliseconds. This will keep the coils cooler and extend their life. Note that MegaSquirt-II uses a nominal dwell setting, called the 'maximum dwell', at 12.0 Volts and adjusts this for the continuously measured running voltage (since higher voltages charge the coil more quickly, the dwell is shortened, and conversely, low cranking voltages need extended dwells to properly charge the coil for easy starting). In the absence of scope data, I'd tend to trust the manual. Also, ideal dwell time varies with voltage. There's no one single number which is ideal for all cars. |
@codrus recently scoped the GM coils. He has data.
|
I believe the LS coils will fire if they think they're being over-dwelled. This would lead to advance at high rpm.
|
1 Attachment(s)
<Stupid double post>
|
1 Attachment(s)
I scoped the LS3 coils, not the truck coils. They're probably different, but:
https://www.miataturbo.net/attachmen...ine=1448338880 (There's an obvious typo on the second line in the chart). The first three columns are voltage, because saturation dwell varies depending upon the input voltage to the coils. This is why the MS3 has a curve. I measured it at various points because I was curious how much it varied, but it's the ECU-observed one that really matters because that's what's used as an input to the MS3 dwell curve. The delta between ECU-observed voltage and coil-observed voltage may vary by car, though. Note that you don't necessarily want to saturate the coils all the time. For maximum coil life you want something around the minimum dwell that still produces a strong enough spark not to misfire, and the saturation level merely indicates the limit beyond which you're not going to get any more spark strength. --Ian |
Originally Posted by stefanst
(Post 1286141)
I believe the LS coils will fire if they think they're being over-dwelled. This would lead to advance at high rpm.
--Ian |
How did you alter the voltage?
|
Originally Posted by Joe Perez
(Post 1286133)
In the absence of scope data, I'd tend to trust the manual. Also, ideal dwell time varies with voltage. There's no one single number which is ideal for all cars.
--Ian |
1 Attachment(s)
Originally Posted by aidandj
(Post 1286148)
How did you alter the voltage?
https://www.miataturbo.net/attachmen...ine=1448339393 --Ian |
And the MS3 only has a 2d table correct?
|
Ah, didn't even think about running it off something besides the battery. Duh.
I had a 30amp variable power supply at school. I miss that. |
Originally Posted by aidandj
(Post 1286151)
And the MS3 only has a 2d table correct?
--Ian |
4 Attachment(s)
Here is the current dwell - voltage correction table.
Timing change must be hapenoing outside of the MS control or it would apear in the logs? attached is the log from the video on an earlier post, it shows a steady 10 deg when the actual timing was 10btdc deg at idle to over 20btdc at 6k rpm |
OK, this is starting to make sense.
Your coils were automatically discharging due to over dwell. This is something that some variants of (not all) LS coils do. The coil manufacturer felt it best to discharge the coil, and save its life, and fire the charge in advance of the ECU control. This saves your coil and destroys your engine. Good stuff. I would replace all 4 coils and set dwell with a scope. But based on your observation with a timing light, 4ms is a better number than 5ms. |
Originally Posted by Ben
(Post 1286297)
OK, this is starting to make sense.
Your coils were automatically discharging due to over dwell. This is something that some variants of (not all) LS coils do. The coil manufacturer felt it best to discharge the coil, and save its life, and fire the charge in advance of the ECU control. This saves your coil and destroys your engine. Good stuff. I would replace all 4 coils and set dwell with a scope. But based on your observation with a timing light, 4ms is a better number than 5ms. FWIW I run the IGN1A Race coils, and they will not auto-discharge on you, and are actually a bit stronger than the GM coils. I run mine at 3ms dwell, .030" gap, and no misfires at 28 PSI. With 4.5ms I could run .040" gap with no misfire at 28 PSI. Currently at 26 PSI, 3ms, .030, works great. |
I really appreciate everyone's help on this!!
I have ordered a scope and will try to confirm Ben's theory by using the coil test setting with > 5ms dwell to see if it is firing before the 5ms, i will post the coils p/n if i can find it to hopefully help others from having the problems i had. |
Originally Posted by HHammerly
(Post 1286328)
I have ordered a scope and will try to confirm Ben's theory by using the coil test
If you don't already have one, you will also need a current probe capable of measuring steady-state DC. (Most cheap current probes are AC only). If you don't have one, you can construct a cheap substitute by just wiring a very small-value resistor in series with the +12 to the coil primary, and using the scope to read the voltage which develops across the resistor. I'd suggest something like 0.1 ohm. It doesn't have to be precise, as we're interested not in knowing the exact current but merely the shape of the curve. Just be sure it's rated for at least 1w, as it's going to develop some heat. Example; 0.1 Ohm 5W Resistor Wire Wound 5% Tolerance Also, be aware that the act of inserting the resistor will change the dwell curve slightly, by reducing the effective voltage at the coil. This will skew your observations by a few tenths of a ms. |
I need a current probe. Time to go to surplus.
|
You can get reasonably decent ones off of eBay for under $100. This is the one that I have: CP 06 AC DC Current Probe 10kHz 40A 6 5mm Jaw Size | eBay
The build quality is very poor, but the actual performance is pretty decent. As compared to the $5,000 Tek probe we had at Harris, it was accurate to within a few %. More than good enough for automotive use. The ones made by Hantek also seem reasonably well-reviewed, no first-hand experience: http://www.ebay.com/itm/NEW-Hantek-C...AAAOSweW5VPgRX |
I had one at school. I'm just going to check the salvage carts until I find one. I'm not in dire need.
|
We use this one:
https://www.picoauto.com/products/cu...-current-clamp No comparisons against a $5k clamp, but it's held up over 8 years of shop use, so I think it's pretty good. |
All times are GMT -4. The time now is 04:53 PM. |
© 2024 MH Sub I, LLC dba Internet Brands