VSS VR circuit improvements

[Bernard Fife]

One issue that has come up for some beta testers is the VSS input triggering many times at low speeds, sending the indicated speed high and shifting the trans to 4th gear. The root of this is some false triggering in the VR circuit, in which the signal drops out very briefly creating multiple pulses (rising or falling edges) at low frequencies and low amplitudes.

small_40Hz_VR1_and_Q13.jpg

I have made some improvements to the code (especially the masking boundary calculations, which had a serious bug in them). However, the hardware could also be part of the solution. I have been playing around with the circuit in LTspice, and it looks like changing both R37 and R45 to 680K Ohms helps a lot with the low amplitude triggers. The circuit will then trigger reliably down to 50 millivolts and 20 Hetrz (in LTspice - real world result yet to be confirmed!).

These changes shift the trigger point with the frequency though, and wouldn't be good for a positional input (like a crank wheel input for MS-II). However, in the case of the VSS, we aren't looking at position, only speed. In any case, switching C29 for a 47pF capacitor (from a 470 pF capacitor) alleviates most of the phase shift. I suspect fine tuning of the value for C29 could result in good performance (especially on a crank signal where the total speed range is relatively small).

These changes will also simplify the BOM, so they may become the 'official' build once they have been proven in a number of installations. Any beta testers that are seeing the issue describe above might try these changes.

Lance.

[Bernard Fife]

For those that are interested in 'virtual' experimenting with the VR circuit, attached is the LTspice model of the VR1 circuit (as a zip file). You can get LTspice from http://ltspice.linear.com/software/LTspiceIV.exe - it is free.

The reason that increasing R37 and R45 (Rb and Rd below) seems to work is that they increase the bias (steady voltage) on the base of the transistor, turning it on with less signal voltage. Without this bias voltage the transistor wouldn't turn on until 0.7 volts or so was reached. In the original design, the 2 megaohm and 300K ohm resistor acted as a voltage divider on the 5 volt source to provide a 5.0*(300K/(2M+300K)) = 0.65 Volt bias.

VR circuit

vr.gif (8.7 KiB) Viewed 8426 times

For very low voltage signals, however, the voltage is also applied back through the VR sensor, and the 300K resistor is essentially in parallel to ground with the 220K (Rb) and 10K (Ra) resistors (which provide a path to ground through the VR sensor). In this case, the effective resistance for the original circuit is 1/(1/300K+1/(220K+10K))) = 130K Ohms and the bias voltage is thus only 0.06 Volts.

With the 680K resistors in R37 and R45 , the effective resistance is 1/(1/680K+1/690K) = 342K, so the bias voltage is 0.73 Volts and the circuit triggers on very small signals.

Lance.

[dch]

The reason that increasing R37 and R45 (Rb and Rd below),( the 2 megaohm and 300K ohm resistor)
With the 680K resistors in R37 and R45 ,

Lance.

Looking at the build guide R37 is a 220K ohm resistor and R44 is a 2.0M ohm resistor, do we replacew R37 or R44? Thanks Dave

[Bernard Fife]

Dave,

Sorry about thhe numbering mix-up - my testing was done on beta boards which had different component numbering, and I messed up my description above too...

R44 remains at 2M Ohms, you replace both R37 (220K) and R45 (300K) with 680K Ohm resistors.

These three form a voltage divider (R44 and R45 when the input voltage is high, R44 and R45+R37 when the input is near ground), and the 680K resistors bias the transistor base up so the signal triggers the transistor more easily, making it much more sensitive to small signals (i.e., at very low vehicle speeds).

Lance.