Another MAX9926 VR board

[ivan141]

Since getting my hands on a brickrpm board soon is not an option at present I decided to do my own so I'd have something to during
my vacation (and with my Maxim samples). I know this must be the latest in a long line of similar boards, but since at least 2 other
forum members here have done their own boards I'd appreciate any feedback I can get before sending the Gerbers to China.

Schematic:

Part was a modded item from an Element14 Maxim library.. awkward crossing of the input lines is a result.

Top of board:

Not much going on.. as you can see I created a power & a ground plane, and the resistors to the VR sensor are THT so I can
spare myself the trouble of ordering 1206 1/4w smt resistors.. and the basic through hole stuff doesn't take up all that much space
anyway, none that wouldnt be empty with the mounting holes I put on there.
It's neatly snapped to an 0.1" grid so it can be fitted to prototype board with pin headers, or just screw it on somewhere and use
wire to board method.

Bottom of board:

I went through a lot of trouble trying to keep things 1 layer, and then considered the madness of it, since it doesn't really make
a dent production price wise. The mandatory too-much-free-time-on-my-hands logo is there, and the traces to the sensors, nothing else.

Once I order these there is a good chance I will end up with many more than I have use for.. I'll put them up for sale here @cost if anyone's interested.

[Fred]

Apologies for not selling you a BrickRPM board!

You should include three high-power resistors per channel. 2x 0.5 Watt, and one 2 Watt shunt.

Otherwise it's fairly straight forward. The only thing that you have to be careful with is ensuring no coupling between pre resistor traces and post resistor traces.

Fred.

[ivan141]

No worries about the BrickRPM...
I'm not too worried about coupling of traces since they run on different sides of the board, and do not run in parallel.

But I do get the idea that few other MAX992x boards have gone to the trouble of adding support for a shunt resistor.
After reading up I understand it has to do with the output voltage of the VR sensor being internally clamped to Vcc,
causing a loss of resolution when using high output voltage VR sensors (because the noise will also have a higher amplitude
and thus possibly cause false triggers).

I'll see if I can cram another THT resistor on there (for optional use), but I dont understand your proposed power rating of 2W.
Since I'm using tht resistors, there will be the regular 2 1W 10k resistors per channel (couldn't find 0,5W, and the 1W package is virtually
identical to the regular 1/4W fare). As I understand it, the shunt resistor varies somewhere depending on the output voltage of the VR
sensor applied. Why in the world would it need a higher power rating than the 2 10k resistors in series with it?

Some quick (and likely dirty) calculations:

I will be assuming the high impedance inputs of the MAX9926 can be left out of the equation to keep things simpler.

Assuming a 1k shunt, my 10k 1W resistors are useful up to sqrt(1/10000) = 0.01A
Highest acceptable voltage across the channel inputs would be 0.01*21000 = 210V (I know, it would not be necessarily smart to go that far)
The shunt resistor would in this case only have to deal with 1/10 W correct?
Note: voltage across shunt would now be 10Vpp, I would go with a lower value shunt to get it closer to 5Vpp, but I'm just focusing on the power ratings here.

Now bump the shunt up to 5k for a smaller voltage division, and all that has changed is
that we can now handle 250Vpp, and the shunt has about 1/2W dissipation.

A 2W shunt would only become interesting from 20k and upward, but I fail to see the point of such a shunt value.
The higher the VR sensor output voltage, the lower a shunt should be to get the output range closer to Vcc, which is the
whole point of the shunt, so one would only see a larger shunt with lower VR output voltages.

In reality the 10k resistors would need to be considerably beefed up to be of any use at the more extreme VR voltage ranges (say 300Vpp).
To bring 300Vpp down to a 5Vpp signal at the sensor you'd need a 340r shunt, at which point the 10k resistors need to dissipate 2.175W,
and the 340r shunt would still only dissipate a meager 0.075W.

f*ck me I'm long winded... anyhoo, fire away, I expect to have made some incorrect assumptions somewhere.

[Ethan]

Hi ivan141,

I did the calculations in a very simple manner, and it all seems to check out.

Is there a chance for noise at lower voltage input, when there is too much attenuation?
from what I read, voltage is related to the speed at which the teeth of the wheel pass the pick-up,
meaning low voltage at low rpm?

[Ethan]

Also the max voltage through the shunt will be vcc + diodebreakdown*2.

so 5V + 0.3*2 = 5.6V.

The negative cycle of the wave will be clamped to -0.3V.

so current from the VR sensor will be the two series resistors (10k + 10k) and the shunt (1k),
making 21k.

so with 200V in, 200V/21K = 0.0095A
with 5.6V/1k = 0.0056A through the shunt,
the other 0.0039A going through the clamp diodes.

0.0056A*5.6V = 0.03W Shunt
0.0095A*95V = 0.9025W Series
0.0039A*0.6V = 0.00234W Clamp Diodes

[Fred]

Apologies for not replying here sooner!

@Ethan: The shunt does not go inside the current limit resistors, it goes outside it, absorbing and attenuating full VR output voltage, hence high power rating required. Additionally, the total voltage between + and - input pins never exceeds 5 or 6 V, but the direction of that voltage differential varies.

Ivan, when you posted that, I read through it, and it seems SNAFU to me. Sorry for not responding sooner, but I'm short on hours at the moment...

Let me do some math!

V=IR
P=VI
V/R=I
P=(V^2)/R

V = 115V RMS
R = 20k Ohms (forget the 5v in the middle for now)

Thus

115*115/20000 = 13225/20000 = 0.66125W total
0.66125 / 2 = 0.330625

0.330625W minimum power dissipation capacity required per 10k resistor, hence my recommendation for 0.5W parts and my use of them on BrickRPM boards as shipped out to various countries around the world.

Now that that's settled, let's play with the shunt:

115*115/5000 = 2.645W < realistic with a high tooth count at a high rpm with a fine tolerance between sensor and wheel.
115*115/340 = 38.897W < unrealistic due to low resistance killing ability of VR sensor to generate voltage.

Resistors being thermal devices, we can derate them a bit as no one is going to hold their engine at 19000 RPM continuously with a 60 tooth wheel except maybe ghost rider, and he's apparently quite dead now. If Jewish, use 1/4w parts for the limit resistors and 1w for the 5k shunt.

Fred.

[Ethan]

Ahh I see, ivan141's calculations assume the shunt is after the 10k resistors. he just rearranged P=(V^2)/R, and solved for V with a known P and R.
Since with the shunt before the 10ks it no longer creates a voltage the divider, does the attenuation occur because of the pickup coils VA? By creating a larger current draw it pulls down the voltage?

cheers,
Ethan

[ivan141]

Good thing I havent sent out any gerbers to china yet.
I second Ethans question about the way the shunt would work if it's not a voltage divider. My knowledge
on inductive circuits has dwindled a fair bit in the 10 years that have passed since I finished my studies.
Sadly I dont have a VR sensor and toothed wheel for testing, but it would be an interesting experiment.

I understood the main function of the shunt to be to lower the input voltage across the MAX9926 inputs
in order to avoid large amplitude noise causing false triggers, so I naturally assumed a voltage divider.
Is there any reason a shunt inside the current limiting resistors would not work just as well as one outside
them?

[Fred]

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.

The drop in voltage is because although that low output impedance is low, it's not zero.

Fred.

[Ethan]

Has there been much investigating into different values for the input capacitor? Looking at the data sheet for the NCV1124, there are a bunch of equations to calculate the input resistors and capacitors, for a given sensor impedance, peak voltage and frequency.