Page 1 of 2

VR board

Posted: Fri Mar 04, 2016 2:25 am
by AndreyB

Re: VR board

Posted: Fri Mar 04, 2016 5:08 am
by stefanst
Just ordered the board. Will report back.

Re: VR board

Posted: Fri Mar 04, 2016 11:35 am
by kb1gtt
Take note that these VR sensors can generate high voltages at times around 300V. See this page for additional details http://rusefi.com/wiki/index.php?title=Manual:Hardware_Trigger#VR_lower_level_details.2C_formulas.2C_app_notes.2C_etc

You may want to add some additional insulation on your VR wires to prevent these high voltages from getting over to the discovery board. As well you may want to check that spread sheet to verify you'll have enough voltage at low RPM to detect a signal.

Re: VR board

Posted: Fri Mar 04, 2016 10:44 pm
by Horizenjob
I've been wondering on this VR sensor stuff a bit now too. I'm using the MAX9926 on my 5634 board and I'm concerned about how to handle the inputs and wether or not I should provide protection. On my prototype board I provided for Bussmann SURGX ESD protectionTVS diodes. They are 0603 and easy to use, but I think not appropriate by themselves on these inputs. They would handle ESD well, but they are not rated for inductive sources. We're talking 300V into a 10K load so not a powerful signal though.

When you look at the functional diagram for the MAX part you can see it clamps it's inputs to the power and ground rails. The inputs are then fed thru 100k resistors to an internal comparator. So it should be OK to clamp the signal externally to the MAX chip.

For the clamp you could either use a fancy diode like Cree C3D1P7060, or just a fast rectifier like COMCHIP acurn102 http://www.digikey.com/product-detail/en/comchip-technology/ACURN102-HF/641-1643-6-ND/5013971. This would limit the input signal swing to 1.7V. I think they would be fast enough such that the VR signal would not spike and then the SURGX parts can do their job. If one of the signal lines shorts to ground or power it won't be a problem because this just clamps the common mode signals...

Re: VR board

Posted: Sat Mar 05, 2016 12:42 am
by kb1gtt
ESD diodes absorb very small amounts of energy, but at high short term energies. A repetitive signal would likely put those short spike at a rate that exceeds there thermal capabilities of the package.

Also your working voltage of an 0805 is commonly about 150V. Which depends allot on which creepage spec you use (pollution degree, etc), the specs vary quite allot. I use the 150V as it's a middle of the road low humidity spec. That's why I use 2X of 0805 package, to get a combined 300V. Technically this could allow for a 600Vp-p signal with 300V isolation to your GND or 5V rail. The concern is that creepage currents can ionize and dump excessive current to your rails. See below PDF for a general working voltage spec.

https://www.bourns.com/pdfs/chpreztr.pdf

Also when thermally cycling a PCB, in an industrial range, you commonly have about a 50% failure rate with 1206 components, and higher failure rates with larger components. The differences in thermal expansion from the PCB to component, commonly causes the copper adhesive to shear off PCB, fatigue cracks, etc. I suggest avoiding 1206 and larger packages because of thermal cycling issues.

Rail clamping diodes don't necessarily start to conduct at .7V from the rail. Those diodes could start clamping at like 15V or more if MAX so choose. I'm not sure what MAX has done. I see this note in the datasheet, which seems to indicate you should run a 300V signal, then measure if your current is above or below 40mA.
Current into IN+, IN-, IN_+, IN_-.......................................±40mA
I'm not sure how this works when it's powered with a 5V rail. Even if that's +/- 2V for 4Vp-p, you still aren't at 5V. I must be interpreting this datasheet incorrectly. Or perhaps they have some magic inside.
Overdrive = 2V to 3V, zero-crossing
Keep in mind you want your 10k resistors to be large enough that you get enough energy at low RPM to accurately detect the signal, but don't have to much energy at high high temperature and high RPM's. If you limit your detectable voltages with external clamping diodes you'll limit the range you can detect, such that it will either be harder at low RPM, or harder at high RPM.

I think the MAX chip is clamping to 5V, perhaps as high as 5.1V. I would use the internal clamping diodes, instead of adding external ones. I could see how an external diode could have a slow recover time which causes it to skew the signal. I could also see how a fast switching diode might create high frequencies that can propagate past the other circuitry on the chip. Fast transients can cause latch up issues, etc. I would suggest that people use the Honeywell datasheet, and Honeywell product line to figure out what kinds of voltages they might expect.

Re: VR board

Posted: Sat Mar 05, 2016 5:50 am
by Horizenjob
Keep in mind you want your 10k resistors to be large enough that you get enough energy at low RPM to accurately detect the signal, but don't have to much energy at high high temperature and high RPM's. If you limit your detectable voltages with external clamping diodes you'll limit the range you can detect, such that it will either be harder at low RPM, or harder at high RPM.
The 10k is not an influence on the low signal voltage, they provide a current limit for the clamps and also are part of the low pass filter. I just say this because they are followed by 100k resistors in the chip.

If we put a clamp diode between the signal lines, that controls the signal voltage to a set value, for instance 1.5V. The MAX chip absolute ratings for the pins are -0.3V to (VCC + 0.3V). So that's what they clamp to. I'm just thinking of clamping the differential signal.

They show a graph of the over voltage on the inputs on page 6 I think of the datasheet. I think the only way they could make that scope photo is measuring the voltage before and after the 10k input resistor.

Re: VR board

Posted: Sat Mar 05, 2016 5:55 pm
by kb1gtt
The 100k resistors internal to the MAX chip are only for setting the gain of the op-amp. The op-amp's inputs pass almost no current, typically the impedance between the + and - inputs is greater than 1 meg ohm.

When your VR signal is between the 0V to 5V rails, the impedance seen by the VR is greater than 1Mohm. Such that when the VR makes small energy, it results in a larger voltage swing. You'll be able to detect the signal until you get down to some some smallish level of mV's, which is resulting from some small uJ perhaps nJ of energy.

When ever the VR signal has enough energy to exceed the 0V to 5V rails, the impedance seen by the VR will change based on how much the diodes are conducting. The impedance could drop to 10kohm, but probably the real impedance is above that.

If you are concerned it might not be rugged enough on the inputs, you could install a clamping diode for R111, or R112. I have that as an optional 5kohm resistor which is not typically populated. I suggest a resistor as it prevents the potential dynamic issues of a diode noted a couple posts before this post. As well resistors have a better high temperature tolerance. This resistor would decrease the impedance seen by the VR, and would limit the energy dumped to the rails. Using a diode instead of a resistor for this component, would likely make the circuit maintain it's sensitivity at low RPM, but who knows what kind of issues could arise from the diodes dynamic characteristics. So I put it in the schematic as a resistor, and figured it's not normally needed. If someone happens to need something they can put on a resistor or diode when that time comes. For now pretty much any OEM VR will not need this.

Re: VR board

Posted: Mon Mar 07, 2016 4:46 pm
by Rhinoman
kb1gtt wrote: Also when thermally cycling a PCB, in an industrial range, you commonly have about a 50% failure rate with 1206 components, and higher failure rates with larger components. The differences in thermal expansion from the PCB to component, commonly causes the copper adhesive to shear off PCB, fatigue cracks, etc. I suggest avoiding 1206 and larger packages because of thermal cycling issues.

I think the MAX chip is clamping to 5V, perhaps as high as 5.1V. I would use the internal clamping diodes, instead of adding external ones. I could see how an external diode could have a slow recover time which causes it to skew the signal. I could also see how a fast switching diode might create high frequencies that can propagate past the other circuitry on the chip. Fast transients can cause latch up issues, etc. I would suggest that people use the Honeywell datasheet, and Honeywell product line to figure out what kinds of voltages they might expect.
I haven't seen anything like that severe an issue with 1206 components, even in more severe automotive testing. I would think that's a problem with the footprint or the process. We had some issues with thermal cycling over an extended temperature range but resolved that with a change of solder composition.
The MAX chip clamps to rail, I have used external clamping diodes because a relatively large current through the device can cause a shift in the reference voltage level. If I remember correctly the input current is specced at 2mA maximum which is easily exceeded unless you use large value series resistors which start to slew the input waveform.

Re: VR board

Posted: Wed Mar 23, 2016 7:15 pm
by Horizenjob
I read the datasheet for the Comchip 1206 diode more carefully. The package is an FRP substrate with a glass passivated diode embedded. So it may be a reasonable match for the thermal coefficient of the board.

It was interesting to read the STMicro datasheet for the support chip and see the example VR and Hall circuits.

I'm going to setup a timing wheel in my drill press and scope a couple of sensors at slow speed.

Re: VR board

Posted: Wed Sep 07, 2016 8:04 pm
by stefanst
I've had those boards sitting on my desk since April, when my life took a sudden detour. I'm back- at least for a little while. Is there still any interest in me populating these and giving them a whirl?

What chip do I use? Digikey has the 9927 at $5.54 and the 9926 at $7.19. Doesn't look like we will ever need the added capabilities of the 9926- correct?
Also I have a lot of 0603 resistors in my parts bin, but no 0804s. I assume nobody will complain if I use the 0603s.

Re: VR board

Posted: Wed Sep 07, 2016 8:53 pm
by AndreyB
stefanst wrote:What chip do I use? Digikey has the 9927 at $5.54 and the 9926 at $7.19. Doesn't look like we will ever need the added capabilities of the 9926- correct?
http://rusefi.com/forum/viewtopic.php?f=4&t=268&p=14614&hilit=9927#p14614 says NO to 9927

We still need to validate this little board, never validated it.

Re: VR board

Posted: Wed Sep 07, 2016 9:05 pm
by kb1gtt
Use 0603's at your own risk. The 0805's have a working voltage of about 150V. Per a Honeywell app note, if you have certain VR sensors with certain geometry in your teeth, you can get voltage as high as 300V. This is why I used 2X 0805's in series, as it creates a working voltage of about 300V, and decreases the possibility of erroneous energy getting across the resistor. In reality, most VR sensors are around 50V or less, which is OK for 0603 components. Other resistors and caps not directly connected to the VR wires can be 0603 if your eyes and patience allow for them.

Re: VR board

Posted: Fri Sep 09, 2016 1:40 am
by stefanst
Hopefully the last question:
5k resistors run at almost $2/piece on Digikey. I assume using 4k7 at a price of $0.01/piece instead will not be a problem. Correct?

Re: VR board

Posted: Fri Sep 09, 2016 9:19 am
by kb1gtt
Correct 4.7 kohm is close enough. The resistor used on Frankenso is 4.99k and costs $0.10 in qty 1. See link below.
http://www.digikey.com/product-detail/en/RNCP0805FTD4K99/RNCP0805FTD4K99CT-ND/2240591?curr=usd&WT.z_cid=ref_octopart_dkc_buynow&site=us

Re: VR board

Posted: Fri Sep 09, 2016 4:05 pm
by AndreyB
stefanst wrote:5k resistors run at almost $2/piece on Digikey.
I know, it must be the 5k resistor cartel. Fight it, use 4.99k :)

Re: VR board

Posted: Fri Sep 09, 2016 4:08 pm
by stefanst
4k99s are in the basket. Any input on the 9926 vs. 9927 question?

Re: VR board

Posted: Fri Sep 09, 2016 4:26 pm
by AndreyB

Re: VR board

Posted: Thu Oct 20, 2016 2:45 am
by frdstang93
Newbie here, I just ordered this board, going to try my hand at this. Good thing is my wife is certified in soldering and does it for a living! Is there a Bill of material for this particular board?

Re: VR board

Posted: Thu Oct 20, 2016 10:08 am
by kb1gtt
BOM links added to the first post. These are the BOM's one has components grouped together, the other has every item on it's own line.

Cool to have a solder expert at hands length. Keep us posted and good luck.

Re: VR board

Posted: Thu Oct 20, 2016 3:47 pm
by frdstang93
Perfect, thank you!! I will post up progress and some testing with it as soon as possible.

-Josh

Re: VR board

Posted: Thu Oct 20, 2016 5:25 pm
by stefanst
I got three of these boards from Oshpark - still unsoldered sitting here. Also have the components. Would you like me to send you the whole shebang, so you can hand it to your wife? I just didn't have the time to deal with it and don't have any real current need for the boards anyway....

Re: VR board

Posted: Mon Oct 31, 2016 3:53 pm
by frdstang93
That would be great, what do you want for everything?

Re: VR board

Posted: Mon Oct 31, 2016 5:38 pm
by stefanst
I paid $4 on Oshpark for three boards and $21 at Digikey for the components (I only ordered two of the 9926 chips). So how about $25 shipped to you?
Mind if I keep one of the boards and a few resistors/caps, so I can build my own if the need arises? I think the board could also be used to decode the ABS signal for some traction control maybe....

PM me your email if you're interested and I get things moving.
@, @ and I will definitely want a report when you'r done though :)

Re: VR board

Posted: Mon Oct 31, 2016 10:57 pm
by kb1gtt
This might be a problem, I'm not sure I've ever finished anything, so that report might have a delay :)

Yes these should be OK if used for wheel speed sensors.

Re: VR board

Posted: Tue Nov 01, 2016 3:53 am
by frdstang93
stefanst wrote:I paid $4 on Oshpark for three boards and $21 at Digikey for the components (I only ordered two of the 9926 chips). So how about $25 shipped to you?
Mind if I keep one of the boards and a few resistors/caps, so I can build my own if the need arises? I think the board could also be used to decode the ABS signal for some traction control maybe....

PM me your email if you're interested and I get things moving.
@, @ and I will definitely want a report when you'r done though :)
Ok Ill send you a pm, decoding an abs signal is exactly what i want to do with it. I have a jbperf board doing it currently so it will be easy to use my car for testing.

Re: VR board

Posted: Mon Aug 07, 2017 5:21 pm
by AndreyB
Some content in Russian regarding radiator fan interfering with VR sensor and driving MAX9926 crazy see http://rusefi.com/forum/viewtopic.php?f=15&t=1233&p=26888#p26888

Re: VR board

Posted: Tue Aug 08, 2017 1:58 am
by kb1gtt
Is the wire twisted in that wire loom? I guess they are showing the magnetic is getting directly into the sensor tip. Seems odd the motor is that close to the sensor.

Re: VR board

Posted: Tue Aug 08, 2017 2:00 am
by AndreyB
kb1gtt wrote:Seems odd the motor is that close to the sensor.
Not OEM as you can see.

Re: VR board

Posted: Mon May 27, 2019 10:21 am
by AndreyB
If thedd input signal voltage remains lower than the adaptive peak threshold for more than 85ms
MAX9926 watchdog again. Why am I not affected with my test mules - is that because only VR sensors are affected and my Hall sensors are happy? Is that because I have enough tooth on my trigger to not be affected?

Re: VR board

Posted: Mon May 27, 2019 10:57 am
by puff
just healthier trigger disk with more teeth...