Microsquirt and 60-2 VR Sensor - What value shunt resistor?
 

I'm using a Microsquirt v3 with a Porsche 993 etc VR crank pickup and my flywheel cut to 993 etc 60-2 tooth pattern.

I got a great start and picture perfect composite logs at low RPMs but ran into signal breakup and the "fake rev limiter" behavior at around 2,600 rpm.

I added a 10k ohm shunt across the pins in-harness for the crank pickup and that raised my "rev limit" to about 4,600 rpm, but its still there.

If you've gone down this path, what value shunt resistor did you end up with to eliminate this problem?

Cheers!

Did you enable the internal jumper that connects the high tooth count internal shunt? You can see it at the bottom of this pic. The place that says "JMP HERE" VR sensor. VR2 is for cam input. Leave it alone

http://forums.pelicanparts.com/uploa...1557796080.jpg


The 10K is likely too high. It will change the internal voltage resistor divider. Usually you should start with a R value that is close to your VR sensor Resistance value.

I am going to guess your VR sensor is in the 800Ohm range. Try a 500 or 1K resistor if the internal shunt doesn't work.

Thanks Jamie

I chose 10k because that seems to be a widely recommended value on the msextra forums, but I couldn't get a clear understanding of whether I should go higher or lower given what I'm observing.

My logic says I should go lower, because that will bleed more voltage off to ground, reducing the voltage saturation issue at high RPMs. But, there's all sorts of contradictory recommendations on the forums :)

That there is...

The only real way to set this is to look at what voltage is coming off the VR at your max rpm, add some margin and then calculate a resistor value that makes the divider limit at this voltage.

BTW, Have you enabled the FW tach period rejection filter in the microsquirt? I use this on my Hall Sensor installs. Basically, you have a tooth every 6 degrees. Calculate a time period for your max rpm plus 1000. then enable the curve to reject short pulses as a function of rpm. It works well.

Thanks for the tips Jamie.

Jumpered the board, left the 10k in parallel with the sensor, configured noise filtering.

May possibly have raised my "fake rev limiter" a couple hundred rpm, but its still in the high 4,000's close to 5,000 rpm.

Frustrating.

try putting a 1K in series (in line) with the positive side of your VR sensor. This will limit the voltage.

Do you see signal loss or jumps to high rpm when you hit your 5k rev limit? The composite or tooth logger should show. Do you still see the -1 missing tooth?

Are you running a second VR sensor for sequential

So this weekend I pulled out the MS and added the jumper on the board. Negligible if any difference.

So, my next step was to remove the 10k resistor I had put across the pins on the sensor connector. I installed a 2pin molex connector here so I can adjust the resistance easily while testing. It also makes a nice location for oscope connections.

Without any additional resistor other than the on-board shunt I'm back to a 2,600 rpm "rev limiter". So, I'm not really sure what the on-board shunt does. Didn't seem to have any effect for me.

I measured the resistance across the two pins on the CPS while disconnected from the Microsquirt - 1098 ohms.

I also captured some oscope logs while idling, not the image I expected.

http://forums.pelicanparts.com/uploa...1558363000.png

I did note that I got loss of sync as the voltage from the sensor approached 100v.

I'm not running sequential, no second sensor.

Jamie, to confirm, you are suggesting in-line NOT in parallel with the sensor? Ie this is NOT shunting voltage to the ground side. Or are you suggesting that IN ADDITION to the 10k shunt across the pins?

There is a TON of misleading and conflicting information about curing this issue.

Super helpful information here.

I do not know what the Vin (max) voltage limit for the microsquirt VR circuit is. The fact that you hit a hard limit at 100V could be at the limit - especially when common component choice could be close to the Vrms of this 100Vpp signal.

So, we need to drop the voltage of the signal. There are 2 ways to do this:

1) Move the VR sensor further away from the toothed wheel. The VR signal is non-linear with distance following a magnetic power law. Basically is close to a square function. Double the distance and you cut the signal. What is your sensor clearance? Try adding 1 or 2mm and this should help.

2) Build a voltage divider into the VR wiring input before it gets to the conditioning circuit. The VR sensor signal is very clean in your scope trace. There is a slight slope change on the up-slope but I don't think that is enough to confuse the ECU. You added part of the voltage divider circuit. This is why I suggested to add a 1K in series with the signal.

Consider three connections to the circuit. Vin = VR sensor output. Vout = Signal going to the ECU. Ground. You added half of a resistor divider network. Your 10K resistor is basically R2 in the following circuit where R1 is just a wire connection (R1 = 0). You need to move this to the R1 position and add R2. Connect them as follows.

http://forums.pelicanparts.com/uploa...1558367234.png

A 1K resistor will cut the voltage by 1/11 according to

Vout = ]R2 /(R1+R2)] * Vin

This will give you ~10V signal at your 2600 rpm cutout point and should give plenty of voltage headroom at your presumed 7k-ish redline.

Ah, NOW I see what you're suggesting. Excellent, I'll give that a try. Before I do I'll take a couple of scope readings at different RPMs and see if I can predict what the uncorrected voltage would be at 7k rpm, and adjust the resistor values to try to keep it below 90v.

Yes, I noticed the slight difference on the upslope. I want the cleanest zero crossing possibly and it's easy enough to swap pins while I'm doing this change, so I'll do that.

I'm using the stock Porsche sensor mounted on the stock boss so moving the sensor would require modifying the sensor and probably not easy to do now the motor is back in the car.

I'll post back with updated data next weekend. In San Fran all week.

Composite tooth logs look fine, but I don't have a log for when it breaks up. Must have lost it. Interestingly I just noticed I think the onboard shunt makes the tooth logger look different.

Before shunt:
http://forums.pelicanparts.com/uploa...1558371400.png

After shunt:
http://forums.pelicanparts.com/uploa...1558371420.png

The signal looks inverted with the shunt.

Yes, but it doesn't change the running behavior. And it's a PITA to take the MS out from where its mounted, so I'll probably just leave it.

I've just been fooling with this, at assumed you meant that the GND in the diagram was the VR- pin, and the Vin was VR+.

I've tried a few different dividers, .5, .3, .2 without much change. In rethinking, I may have made an erroneous assumption.

Are you taking the GND in the diagram above to actual engine ground, and not to the VR- pin? Such that, VR+ off the sensor is modified by the voltage divider, and VR- runs straight from the sensor to the Microsquirt?

Getting quite frustrated by my lack of solution to this, and the level of disinformation, and general dearth of documentation on the Microsquirt product has me very frustrated.

To date I've tried:
-nothing other than connecting sensor - sync loss 2,600 rpm
-10k across the sensor pins - sync loss 4,000+ rpm
-bridge the shunt on miscrosquirt board - I'm buggered if I can tell any different with this thing
-voltage dividers across the pins of the sensor
-voltage dividers going to engine ground

Whatever I do, I can't get anything to work reliably. My next option is to
-remove the useless bridge on the MS board
-remove the current 1/3 voltage divider that bleeds to engine ground
-try a very low resistance shunt across the sensor pins (like 1k)
-come up with a conditioning circuit of my own, possibly utilizing optoisolators, varistors, transistors to generate a clean 5v or 12v tooth pattern for the MS

IDK why this is so difficult, this isn't some esoteric setup I'm running.

Gaaaaaaaaaahhh!

Decided to quit screwing around and build a conditioner using a MAX9926 https://www.digikey.com.au/product-detail/en/MAX9926UAEE+T/MAX9926UAEE+TCT-ND/3879279

Are you sure you are keeping the VR away from high EMI emitting components like the coils?

I've used ferrite beads on the VR sensor wiring, faraday cages around my coilpack, and low pass filter on the VR sensor wiring. Nothing ever seemed to change in my setup, so I figured I was wasting time with it and left it alone.

I was having a massive explosion when coming on boost like NOS that I thought was VR related. It probably is simply in my tune.

oh yes, totally isolated harness separated by at least 6 inches and only really gets close at the terminating end.

boy I get a massive backfire when I loose rpm signal, made worse as I don't have the exhaust on right now. First time it happened my gf came running in, thought there was an explosion. The cops came over and said they had reports of shots fired :D

I'm not going to get involved any specifics of your ECU or advice there, however for VR signal conditioning I could recommend this as worth looking at :

https://sirhclabs.com/product/cortex-ebc-speed-sensor-adapter/

We had a similar requirement in needing to convert a VR signal into a consistent voltage square wave for a high speed cylinder pressure instrumentation system. Used the above device - have viewed and logged in high speed scope at over 10,000rpm - and it worked very well for what I thought was an extremely reasonably cost and no hassle factor.

Yes that uses the same chip I plan on using. Its a very simple circuit, needs only 2 additional resistors.

:D

BTDT, when I had the Ford EDIS coilpack wired wrong....

The MicroSquirt v3 circuit is based on a max9926
http://forums.pelicanparts.com/uploa...1559260174.jpg

full document is here microsquirt_v3_0_1.pdf

the internal 5k shunt on the MicroSquirt is inside the current limiting series resistors (10K each) meaning it does very little to provide an adequate load for the Bosch VR
The max9926 has a high impedance 100K+ Ohm input internally and the internal shunt drops the impedance the VR loads against down to 25K

The Bosch VR is quite strong and much happier with a decent load to play with
an external shunt of 1K @ 2 watts should do the trick and give the VR sensor in the distributor some load to work against

The purpose of the shunt is to change what would be a high impedance input in to a low impedance input. One that the relatively low output impedance VR sensor can still easily drive, but that induced noise cannot significantly drive.

Hi,
Did you ever get to the bottom of this issue? I am having the same issue at higher RPM's, only I have been messing with the resistors to raise the breakup point. Now at around 6200 rpm. using 3K series and 1k shunt.

Sort of. I purchased a VR conditioner from the speeduino project. Pretty cheap. Connected that up inline. Works like a charm.

There is SO much misinformation about this I had to leave a reply to save others from going around in circles.

Firstly, the correct place for a shunt resistor is where indicated in the Megasquirt schematic, rather than directly across the VR sensor. That is, after the input resistors, which in the Megasquirt example are 2 x 5k resistors in series.

The reason there are 2 x 5k resistors instead of 1 x 10k is to do with the voltage rating of the resistors. Generally, resistors will have a voltage rating of about 200V. Using 2 in series allows 400V input.

One of the reasons the resistors are there is to provide protection for the inputs of the MAX9926. The inputs of the MAX962x series are protected with diodes connected to the supply voltage and ground. Thus any input that tries to head above supply or below ground will be clamped by these diodes. However the diodes can only withstand +/-40mA absolute maximum. (Safe design would allow for half that). Poking a VR sensor straight up the inputs of the chip will destroy these protection diodes causing peaks of almost 300mA. The result will then likely be erratic output (I traced a "will not start" fault in a Volvo to this exact issue after the car was due to go to the scrap yard because the dealers could not find the problem).

So a 10k input resistance allows for 200V nominal (400V worst case) on the input with no shunt present.

Secondly, the 10k input resistors and the 1nF capacitor act as a differential input filter. These values provide a cut off for frequencies about about 8kHz which is designed to get rid of high frequency noise. Similarly the 10nF capacitors act as a common mode filter. All designed to get rid of external noise.

This circuit works by keeping the signal path of both the VR+ and VR- identical. The MAX992x subtracts one from the other (difference amplifier) and compares the result to an internal reference (2.5V). The idea is that noise will be induced equally into both lines, therefore the difference amplifier will cancel out the noise leaving the true signal to be amplified.

This is also why the wires from the sensor should be twisted together (called a "twisted pair") and posisbly also screened.

This circuit arrangement is almost exactly the same used in Bosch Motronic ECU's used on millions of cars. They don't use the MAX992x, rather a special Bosch ASIC, but the input arrangement is very similar. (Using a VR on the crank and Hall on the cam)

If you place a shunt directly at the input, it will need to be about 5W to 10W, depending upon your expected input voltage. Your VR sensor will also dissipate heat from its internal winding resistance. None of that is a good thing.

If you place the shunt directly at the chip input, the power rating of the resistor only needs to be 0.008W (ie a 1/4W resistor is very comfortable). It will achieve about a 4x voltage reduction at the input, although this is only really relevant at lower speeds where the VR output is less than about 30V (because at higher voltages the signal gets clamped by the protection diodes inside the MAX9926.

If you are going to believe some of the internet wisdom and fiddle with resistor dividers etc, then try to make them balanced (i.e. equal in each leg) that is, a series resistor in each leg (10k?) and a shunt resistor (4k7?) between the two. This is effectively exactly the same as just using the shunt on the input to the MAX9926.

I am sorry this is so long. Can I put it another way? If you have noise problems, then you likely have a noise being induced into the input that is [I]different[I] in each of the VR+ and VR- wires. Or very large in both. Either way, look at the wiring path, look at current paths in the engine block and bell housing around the starter motor. Make sure both wires are the same lenght(!) and twisted around each other. This is important at any connectors making sure NOT to have a big loop of wire in one leg and a short wire in the other. Shield if necessary and ONLY earth at the ECU end. Make sure the engine and transmission are well earthed. Make sure you haven't damaged the MAX9926 inputs with inadvertent overcurrent stress.

If this is a starting related issue note that the starter often finds earth in the bell housing and this is then bolted back to the block to make its way back to the battery. Dirt under the bell housing is known to cause crank sensor noise problems in some cars. Manufacturers have even issued instructions to clean and rebolt the bellhousing faces to cure misfire and starting issues. An earth strap directly to the starter might help.

Kram, thanks for the insight.

I did fix my issue with adjusting the noise filtering in Tuner Studio. It was set just enough to mess with me at the higher RPM's.
Thanks, herb