DIY-EGOR ideas (was: No Subject)

[Bernd Felsche]

Jörgen Karlsson tapped away at the keyboard with:

> > > We cannot pwm anything across the pump cell, it has to be a
> > > constant flow there, I think that the way to go is to filter the
> > > pwm output to get a stable current. If the sensor ground is put at
> >
> > I haven't been able to find a reference about not being able to PWM
> > the pump current. Are you able to tell us where you've found that
> > information? What sort of load (RLC) does the cell present?
> 
> Yup, I guessed...;)
> 
> The addition of a rc network wouldn't hurt, would it. As I see it
> the filter makes it harder to damage the cell. We can also use a
> higher maximum current then if we only limit current by using a
> resistor. We can of course use other types of current limiters but
> they are more complex.

In this case, a 1k resistor from the regulated 5V rail limits
maximum current in both directions to 5mA (without accounting for
Vdrops through the FETs).

> I know that these sensors can measure afr's in the 9:1 range,
> maybe lower.  IF WE ASSUME, that anything above 6mA kills the pump
> cell we have to limit the maximum current to 6mA (THIS NUMBER IS
> TAKEN OUT OF THE AIR!!!!) if we don't have a filter we will pulse
> the cell at it maximum current all the time! If we use the filter
> the cell only see the needed current...

The maximum current is limited by the current which can be switched
as there is no capacitance on the input to the H-bridge after the
current-limiting resistor.

> > > 2.5v above the circuits ground the processor can drive the pump
> > > cell and filter circuit directly.
> >
> > Depends if the cell is floating or tied to ground somehow. If it's
> > floating, then a bridge circuit can be used to reverse current flow.
> > It's quite common in DC motor drives.
> 
> The pump cell and the sensor cell has a common ground, but this
> ground is not common with the heater ground... At least that is
> what i find when I look at the schematic that Bruce posted. The
> sensor Ground is clamped to 2.5 volts there.

That might be applicable to the cell in question. I haven't found a
block-diagram of the NTK/NGK sensor used by Bruce, etc.

> > What sort of filtering is necessary?
> 
> I would use a rc network.

> > > We know that Bruce and the guys used up towards 1.1A heater
> > > current, lets assume that is max. Let someone else do the
> > > damage ;) First of all we filter the pwm output from the mcu
> > > and then we use the filtered voltage to drive the power
> > > transistor.
> >
> > > We also need to measure the current across the sensor, we only
> > > put a series resistor and measure the voltage drop. By using a
> >
> > That's why there's a current-sensing resistor in the ground-lead
> > of the H-bridge in the circuit described earlier.
> 
> I haven't looked much on that schematic, I haven't printed it and
> it is not fun to watch it when it is verticaly oriented on the
> screen...

Sorry about that; Eagle outputs the Postscript in that direction and
Distiller, etc won't rotate automatically. Instead of the Acrobat
reader, see if you can find "Ghostview"; it rotates automatically.

I've updated the web page with revised schematic and notes, adding
more image formats to the schematic. Please don't ask me to do GIF.

> > > We also need the voltage across the heater element it self, if
> > > we have the current and the voltage across the heater we know
> > > the resistance, that we can translate to temperature. I think
> > > that aiming for a set resistance here is the way to control
> > > the heater correctly, If the resistance goes up we must lower
> > > the heater
> >
> > Why not measure the voltage drop "directly" using an ADC channel
> > in the AVR? As in my discussion circuit where there's a current
> > sensing resistor, and a voltage-divider to scale the nominal
> > voltage to below AREF. That reduces the number of external
> > components.
> 
> That's a good one, I feel stupid...

We all have our "moments". :-)

> > > current and the other way around. I think that we can be
> > > pretty lay back while doing this, I think that we will use one
> > > voltage across the sensor for initial fast heating, then at
> > > some resistance we drop the voltage to prevent cracking, the
> > > correct values can probably be found in the base shematic.
> > > After that we let the raising resistance of the sensor take
> > > care of the adjustment, we just put two more regulation
> > > points, one that lowers the voltage a bit when the sensor is
> > > close to maximum temp and one that almost completely shuts of
> > > the current when it gets even hotter.
> >
> > Sounds very complicated. If you're controlling the current
> > through the heater, then you're controlling its power output.

> I am sure that many of the sensors will see a lot higher exhaust
> temperature then 600ºC, when that happens the heater will have to
> be put on idle until the temp drops within operating temp of the
> sensor again. Someone mentioned that the sensor must be heated
> slowly at high temp. I don't think that we can use any simpler
> temperature correction, at least not if the sensor will se hot
> exhaust gases.

Indeed. 600 is the minimum operating temp for the Bosch sensor.
Another Bosch patent mentions up to 850C.

> As I see it we can manage without the H bridge for the pump cell
> on your circuit and use your current sensing on the heater
> element.
> 
> I think that the next problem to address is if we need to measure
> the current across the pump cell or if we can use the voltage
> across it instead?

You'd need at most 3 ADC channels to do it. One is already in the
schematic shown. With the H-bridge, you'd measure the voltage
(scaled) across it's outputs (i.e. as applied to the cell). As long
as the pulse-width is longer than the 3 (!!) AD conversion times, 
you can easily determine both current and voltage drop.

Alternatively; a measurement of each value on successive pulses
might be just the go... as the schematic I've drawn uses the PWM
output pin to trigger an interrupt, the decision of which you want to
measure can be done in that handler, with appropriate timing to
allow the current to ramp-up.

With PWM, the pump cell voltage is however nominally "constant"
(regulated rails - drop through limiting resistor - gate drop) so
there's little need to measure it if you're measuring the voltage
drop through a current-sensing resistor. OK; there'll be some
variation due to temperature changes in the components but as the
power dissipation is minimal, most variation will be due to sources
outside the circuit. An on-board temperature-sensor could be added
for "soft" calibration, etc.

-- 
 /"\  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!            |
----------------------------------------------------------------------------
To unsubscribe from diy_efi, send "unsubscribe diy_efi" (without the quotes)
in the body of a message (not the subject) to majordomo at lists.diy-efi.org