[Diy_efi] Re: uego ground (nernst and pump)

[Bernd Felsche]

On Thu, Feb 27, 2003 at 07:26:52PM -0800, Stewart Prince wrote:
> Yes, I see this now.  If the reference for both cells are tied
> together internally, the best way do to it would be to reference
> ground signal to 2.5 volts.  This makes an H-bridge unnecessary as
> any buffer with 0-5 volt DC output can drive 5mamp, easily.  Thus,
> a PWM signal with a 50% duty cycle would equate to Ipump=0 (on
> average) and with the proper selection of a series resistor, a
> 100% duty cycle would equate to whatever positive pumping current
> is needed, depending on what Air-fuel ratio you want to calibrate
> to, and a 0% duty cycle will equate to a negative pumping current.

> One could size the op amp gain and the in-line resistor to yield
> the proper max pump current at 100% duty cycle, but how does one
> know if the pump current has really driven the nernst cell voltage
> back to .45 volts?  I haven't modeled this system, but I'm
> assuming that it is a type "0" system in terms of transient
> response, meaning there will be steady state error.  If you close
> the loop (the purpose of the op amp) you will need a high gain to
> minimize the error, and this might create "overshoot" and
> "ringing" of the circuit response.  Perhaps what you really need
> is to take the nernst cell signal, process it using a PID
> algorithm, then send it out to the pump cell using PWM.  This
> sounds like a perfect fit for a microprocessor.  I believe this is
> the path we will take.

If you are driving the sensor with a micro-controller, you may find
that the micro-controller has an analogue comparator built in. Such
a comparator can typically set an interrupt when the comparator
crosses the threshhold level. If the threshhold level floats at
0.45V above the virtual ground of the Nernst cell, then the
interrupt can be used to change the pulse direction for driving the
pump. The Nernst voltage determines which direction you should be
pumping - useful for starting. 

It's not PWM per se. The cycle period is not fixed. Response will be
*fast*. The comparators can interrupt in under a microsecond.
Hysteresis will have to be built in by software; perhaps the
comparator's interrupt disabled just prior to and until just after
pump reversal, to allow the comparator voltage reference to
stabilise - it'll be "riding" on the pump voltage.

Once again, the ratio of the reverse and forward pump cycles will
correspond to lambda. The "amount" of pumping you do will be the
V.I product in each direction. Driving constant-voltage or
constant-current is an advantage... voltage is easier to drive, but
the current more difficult to measure. Cell impedance is NOT
constant with respect to temperature.

-- 
/"\ Bernd Felsche - Innovative Reckoning, Perth, Western Australia
\ /  ASCII ribbon campaign | I'm a .signature virus!
 X   against HTML mail     | Copy me into your ~/.signature
/ \  and postings          | to help me spread!

_______________________________________________
Diy_efi mailing list
Diy_efi at diy-efi.org
http://www.diy-efi.org/mailman/listinfo/diy_efi