Date: Tue, 3 Dec 2013 19:07:14 -0800
Reply-To: Brett Ne <brettn777@GMAIL.COM>
Sender: Vanagon Mailing List <vanagon@gerry.vanagon.com>
From: Brett Ne <brettn777@GMAIL.COM>
Subject: Re: Arduino and Vanagons
In-Reply-To: <CABx3TkVCD2gAMaSM9cA=V5_jgv0iyP2b3pbk05kWG9LvK0y1wA@mail.gmail.com>
Content-Type: text/plain; charset=ISO-8859-1
I'll keep that in mind as I continue to play with it. There is an internal
pullup resistor built into each input pin that is about 25k. And I do like
to keep a series resistor on the input pins for an extra measure of
safety.
Brett
On Tue, Dec 3, 2013 at 4:59 PM, Ryan Press <ryan@presslab.us> wrote:
> Most hall sensors are open collector (or drain) so it's not the hall
> sensor's fault, it's the ECU that isn't providing much of a pull-up.
> Try adding a pull-up resistor, I'd bet that gives you some more oomph.
>
> You could tie directly to the Ardunio input as well, like you did with
> the current limiting resistor. This will cause the body diode inside
> the uc to conduct, clamping the voltage. A small RC with a 1 nF cap
> or something would be a good idea too.
>
> Check the datasheet, but most body diodes can handle 5 mA or so, make
> sure you size the limiting resistor to have a good margin there. You
> could tie your pull-up resistor to 5V of course as well, but in any
> case put resistance in series with the input pin to protect against
> over-voltage stress.
>
> Ryan
>
>
> On Tue, Dec 3, 2013 at 4:46 PM, Brett Ne <brettn777@gmail.com> wrote:
> > You need an input buffer followed by a voltage clamp.
> >
> > Perhaps...I'll know more once I get a good shot of the pulse shape on a
> > scope. And I'm assuming that the 89 Fox Hall sensor behaves closely to
> the
> > Vanagon Hall sensor. I'm still surprised by how weak that signal is;
> you'd
> > think they'd need a shielded wire, especially being so close to the high
> > tension ignition wires.
> >
> > Brett
> >
> >
> > On Tue, Dec 3, 2013 at 3:08 PM, Thomas Hargrave <tphjr@hiwaay.net>
> wrote:
> >
> >> You need an input buffer followed by a voltage clamp.
> >>
> >> Tom
> >> ------------------------------
> >> From: Brett Ne <brettn777@GMAIL.COM>
> >> Sent: 12/3/2013 4:18 PM
> >> To: vanagon@GERRY.VANAGON.COM
> >> Subject: Re: Arduino and Vanagons
> >>
> >> Just a quick update.
> >>
> >> I bludgeoned together an Arduino program to calculate and display the
> >> engine speed based on the Hall sensor output(my programming "style"
> makes
> >> hackers' code look elegant). To deal with the presupposed high
> voltage(to
> >> a microcontroller, anything over 5v is high voltage) from the Hall
> sensor ,
> >> I put a 10k and a 3.3k resistor in series with one end connected to the
> >> Hall output and the other to ground. The Arduino input pin was
> connected
> >> between the two resistors, which should give a voltage output of 1/4
> >> whatever the voltage is coming from the Hall sensor. I tried hooking
> it up
> >> a couple of times with the engine running and it stopped it cold
> >> immediately each time. That is one *weak* signal! A 13.3kohm path to
> >> ground shuts it down. Just to live life on the edge, I removed the 3.3k
> >> resistor going to ground and tried again. This means that the full
> voltage
> >> coming from the Hall sensor will be going to the Arduino input pin(it's
> >> only $2-3 for a new chip if I burn out the pin). This time, the engine
> >> continued running without complaint and I was getting rpm readings on
> the
> >> LCD. WooHoo! The readout was too high and very unstable, so there is
> >> still much work to be done in this first step. I'll have to recheck my
> >> code; my formula was based on timing 8 pulses, but I may have set the
> >> counter wrong and it's counting only to 7 pulses, giving a higher than
> true
> >> reading. I think the the instability has to do with the handling of a
> very
> >> weak signal.
> >>
> >> Sometime in the next day or two I'll drag my laptop out and hook up my
> USB
> >> oscilloscope to it and get a better idea of what this Hall sensor output
> >> looks like. Because the Hall signal is so easily disrupted, It looks
> like
> >> we'll need to add an op-amp voltage buffer in order to be able to use
> this
> >> signal without affecting the ignition system.
> >>
> >> This gives yet another option for those thinking of installing a hidden
> >> vehicle disabling system for security. Tie in a wire to the Hall
> output,
> >> run it to a hidden toggle switch that switches it to ground via a 5k
> >> resistor. Less obvious than a fuel relay disabling system. It'd take a
> >> good mechanic at least a half hour to find it. An average mechanic
> would
> >> take several hours and would have replaced the engine computer, coil,
> cap,
> >> rotor...
> >>
> >> I've started looking into the circuitry required, but am not to a stage
> yet
> >> where I think it would be fruitful to report.
> >>
> >>
> >> Brett
> >>
> >>
> >> On Wed, Nov 27, 2013 at 3:34 PM, Brett Ne <brettn777@gmail.com> wrote:
> >>
> >> > At this point, I don't know if it's worthwhile spending much more time
> >> > studying duty cycle outputs under various conditions. David's
> provided
> >> > enough data to have a pretty clear picture of what the ICU is doing
> under
> >> > various idle conditions. The purpose of gathering the duty cycles
> isn't
> >> to
> >> > provide target values for the microcontroller to strive for, but
> rather
> >> to
> >> > get a feel for how much the duty cycle needs to change to compensate
> for
> >> > various loads. In fact, I'm no longer convinced that the duty cycle
> >> values
> >> > above closed throttle position have much use for us. Our version of
> the
> >> > ICU would quickly drop the duty cycle to zero as the throttle is
> opened &
> >> > the rpm's climb and keep it at zero until the throttle is closed to
> >> > decelerate whereupon the duty cycle will start to increase when the
> >> engine
> >> > speed drops below the idle speed target. We may get away without
> needing
> >> > the throttle switch input. But then, there may be driveability issues
> >> with
> >> > that approach; there may need to be some baseline idle valve air flow
> >> level
> >> > needed.
> >> >
> >> > We will definitely need coolant sensor voltage readings at a known
> cold
> >> > value and a known hot value so we can calculate the temperature from
> the
> >> > sensor voltage. If someone has an infrared thermometer and a decent
> >> > digital multimeter, that would be the easiest way to get those
> readings.
> >> >
> >> > Brett
> >> >
> >> >
> >> > On Wed, Nov 27, 2013 at 1:22 PM, JRodgers <jrodgers113@gmail.com>
> wrote:
> >> >
> >> >> David, glad to hear you are crawling back into the light!
> >> >>
> >> >> \John
> >> >>
> >> >>
> >> >> On 11/27/2013 1:45 PM, David Beierl wrote:
> >> >>
> >> >>> At 09:33 AM 11/27/2013, Brett Ne wrote:
> >> >>>
> >> >>>> A member pmailed results of some testing last night:
> >> >>>>
> >> >>>
> >> >>> Ok, I'll 'fess up. I'm the guy who's been supplying numbers for
> >> >>> this. I've been going through a very rough patch over the summer
> and
> >> >>> not wanting to talk to anybody, but at least for the moment this and
> >> >>> some other things happening here are bringing me out of my hole to
> some
> >> >>> degree.
> >> >>>
> >> >>>
> >> >>> "...just checked a spare ISV using DC, operation is proportional to
> >> >>>> applied
> >> >>>> voltage. Starts opening somewhere around 3 VDC, fully open
> somewhere
> >> >>>> around 8 VDC."
> >> >>>>
> >> >>>> So we can treat it as if it's a dc motor operating against spring
> >> >>>> pressure. Keep in mind that we are feeding it a modulated signal,
> so
> >> >>>> even
> >> >>>> at 11.5 v and a 50% duty cycle, the valve will operate as if it's
> >> >>>> receiving
> >> >>>> a constant 5.75 v steady signal(11.5 x 50%) and will not be fully
> >> open.
> >> >>>>
> >> >>>
> >> >>> Here are some better numbers for the spare valve, which is a bit
> >> >>> sticky at the extremes (maybe they all are, I don't know):
> >> >>>
> >> >>> Resistance 4.0 ohms.
> >>
> >> >>>
> >> >>> Impedance at 150 Hz (sine wave) using a signal generator with 50-ohm
> >> >>> output: 1.0 Vrms --> ~25 mA so ~40 ohms.
> >> >>> Impedance at 1000 Hz ditto: 1.0 Vrms --> ~9 mA so ~110 ohms.
> >> >>> My generator couldn't maintain a square wave while driving the valve
> >> >>> because of its higher output impedance (result was a set of spikes
> >> >>> with smoothly declining tails) but the numbers at 150 Hz were
> similar.
> >> >>>
> >> >>> I may be able to hook up an amplifier to the generator and actually
> >> >>> drive the valve at operating levels, but in the meantime, using a DC
> >> >>> supply with max 1300 mA output:
> >> >>> On increasing voltage valve jumps open slightly at about 2.4V / 500
> mA.
> >> >>> On decreasing voltage valve closes at about 1.5V / 325 mA.
> >> >>> On suddenly applying 6.5V / ~1300 mA valve snaps fully open.**
> >> >>> On suddenly applying 5.8V / ~1100 mA valve does not fully open.
> >> >>>
> >> >>> **The open-circuit voltage is probably somewhat higher since I'm
> >> >>> driving the supply at its current limit here.
> >> >>>
> >> >>> So lets crunch some numbers to see what's going on during cranking.
> >> I'm
> >> >>>> going to calculate the power (watts) going to the idle valve during
> >> cold
> >> >>>> idle and during cold cranking and compare.
> >> >>>>
> >> >>>> First, let's calculate the resistance that the idle valve provides
> so
> >> >>>> that
> >> >>>> we can determine current flow for different voltages. We know
> that at
> >> >>>> 9v
> >> >>>> and 29.4% duty cycle the valve draws .157 A, so we can get
> amperage @
> >> >>>> 100%
> >> >>>> duty by dividing .157/29.4% = .534 A.
> >> >>>> Resistance = Voltage/Current = 9v/.534A = 16.85 ohms of resistance
> >> >>>>
> >> >>>
> >> >>> Actually you can't extrapolate to DC this way because the valve is
> >> >>> highly inductive. As your duty cycle approaches 100% the valve
> >> >>> effective impedance will get closer and closer to its four ohm
> >> >>> resistance. If I can get the amplifier hooked up and delivering
> some
> >> >>> approximation of a square wave I'll be able to vary the duty cycle
> >> >>> directly and provide more numbers.
> >> >>>
> >> >>> Also bear in mind that the 6V/9V/10-11V numbers I supplied for peak
> >> >>> voltages in the system while operating a) are very approximate,
> >> >>> eyeballed off a tiny pocket oscilloscope screen at two volts per
> >> >>> division and b) need to be increased by about a volt each because
> the
> >> >>> baseline was about a volt negative. The ~10-~11 range was
> >> >>> immediately after starting vs after running a few minutes; probably
> >> >>> because B+ was recovering after extended cranking with the ignition
> >> >>> disabled.
> >> >>>
> >> >>> Vrms and duty-cycle numbers are from a fancy 4-1/2 digit Fluke meter
> >> >>> with flat response well above our frequencies of interest and IIRC
> >> >>> four measurement cycles per second. If I indicate a ~n.whatever on
> >> >>> one of those readings it's because it's fluctuating. You shouldn't
> >> >>> use any more decimal places in calculation than the least that I
> >> >>> provide (i.e. if I give you ~9 for an input, a calculation from that
> >> >>> resulting in n.nn has to be trimmed to ~n to be meaningful).
> >> >>>
> >> >>> The white wire on the ISV is less than an ohm to the alternator
> case,
> >> >>> so you're right that it's grounded.
> >> >>>
> >> >>> Duty cycle at a cold start at 60F ambient was down to around 23% by
> >> >>> the time I could leave the front of the van, walk through the house
> >> >>> to avoid the soaking wet cedar tree and reach the back. Quickly got
> >> >>> down to 21% or less. I don't think I've yet seen it below 20%.
> >> >>>
> >> >>> I've not yet been able to catch the duty cycle for the short
> >> >>> excursion above 1000 rpm that happens a few seconds after a cold
> >> >>> start. But I think a more important number is the rpm reached
> >> >>> (guessing 1100) and duration and how this is affected by
> >> >>> ambient/engine initial temp. And is this normal system behavior or
> >> >>> an oddity of mine? Like the short period above 1000 when I idle
> down
> >> >>> while driving, it's consistent among different ECUs both 022D and
> >> >>> 022F, but unknown if consistent among different ICUs since I have no
> >> >>> spare.
> >> >>>
> >> >>> Playing with the throttle plate as it was beginning to warm up it
> >> >>> seemed to vary between about 22-24% without any particular
> systematic
> >> >>> behavior. It's hissing down with rain so I'm not going to wait for
> >> >>> full warmup at this point. I think the changes are some artifact
> of
> >> >>> system operation and don't matter.
> >> >>>
> >> >>> Stabbing the throttle produced slightly wider variations up to maybe
> >> >>> 20-25% but I have trouble seeing how it can matter given the input
> >> >>> from the driver's foot swamping everything else.
> >> >>>
> >> >>> This was quite surprising to me; the power going to the idle valve
> is
> >> >>>> less
> >> >>>> during cold cranking than during fast idle. I was thinking that it
> >> >>>> would
> >> >>>> receive more power during cranking to get a good supply of air &
> fuel
> >> >>>> into
> >> >>>> the engine to really get things going. Then I thought maybe it's
> >> >>>> trying to
> >> >>>> restrict the air a bit to richen the mixture, but the ECU knows
> when
> >> the
> >> >>>> engine is cold cranking and fuel mixture is one of its main jobs.
> >> >>>> Maybe by
> >> >>>> restricting airflow, it effectively lessens the compression ratio
> to
> >> >>>> make
> >> >>>> starting easier. Or maybe I'm just overthinking this and it's
> trying
> >> to
> >> >>>> put out the same airflow as cold idle but isn't able to accurately
> >> >>>> make up
> >> >>>> for voltage drop during cranking.
> >> >>>>
> >> >>>
> >> >>> I suspect the last. It would be answered by doing cold cranking
> with
> >> >>> a booster to keep the voltage up, but I'm not equipped to do that
> >> >>> (though I'm probably going to have to replace the tired battery soon
> >> >>> if I want to drive at all this winter). This engine has always
> >> >>> seemed to crank hard since I've had the van, and replacing the
> >> >>> starter didn't change it. Sometimes it cranks fine and sometimes it
> >> >>> goes to its knees on the first revolution, then recovers.
> >> >>>
> >> >>>
> >> >>> I think that it would be best to aim for the same airflow for both
> >> cold
> >> >>>> idle and cold cranking and adjust as real world results suggest.
> >> >>>>
> >> >>>
> >> >>> I think that's sensible.
> >> >>>
> >> >>> Yrs,
> >> >>> d
> >> >>>
> >> >>>
> >> >
> >> >
> >> > --
> >> > Brett in Portland, OR
> >> > "Albert" '82 VanaFox I4 Riviera
> >> >
> >>
> >>
> >>
> >> --
> >> Brett in Portland, OR
> >> "Albert" '82 VanaFox I4 Riviera
> >>
> >
> >
> >
> > --
> > Brett in Portland, OR
> > "Albert" '82 VanaFox I4 Riviera
>
--
Brett in Portland, OR
"Albert" '82 VanaFox I4 Riviera
| 
