I feel kinda responsible for this mess, now
Unfortunately, I broke down my setup to use the solenoid for something else. It was a much simpler rig than DelSolid's. Just a single gas ( CO2 in N2 ) With a regulator (1lpm) and solenoid. Then, a tube running into one of those Innovate heatsink bungs, fitted with #10 barb fittings. I would let the sensor settle in at ambient air, then, trigger the solenoid with the CO2 mix. Times from the solenoid pick to about lambda 1.2 were around 30ms. Then, lambda 1.2 to 1.0 was about another 6ms ( on the DAC). I do not know how much latency was built into the solenoid and tubing. I had always assumed about 30ms. But, obviously, it is less than that. So, I was getting ~35ms
On the trace that I posted, there is a bit more info than the slew rate of the DAC. The DAC slew rate would be the changes between the 2ms update steps. What was impressive was what happened with the steps as the value approached lambda 1.0 Usually, with that steep of a slope, one would expect overshoot/ringing. Or, to have it "round off" and taper towards its final value, from over-damping.
for #1. The 61.9ohm resistor is just a current shunt. Taking a differential voltage reading across it gives the normalized pump current ( it is in series with the pump cell ). The resistor is not limiting the current. The current would be the same without it. You just wouldn't know what it was.
#2. The max current values are sustained values, with a couple of caveats. Mostly they are worried about the voltage across the pump cell. If the potential is greater than about 1.2v ( IIRC ) the pump cell will start to eat itself ( binding potential of the ZrO2 or catalyst material ). That happens long before any self-heating damage. For the same reason, the test gas MUST have an oxygen component that will disassociate at a lower potential than the ZrO2 or the pump cell will eat itself to get at that oxygen. i.e. you can't run it on straight Nitrogen, for example. That said, even repeated peaks will do cumulative damage to the pump cell. The limits are not symmetrical because it's a cell. i.e. it has its own voltage/polarity that must be accounted for.
For all the consternation, it really does seem to do what it says it does. 35ms from free-air to Lambda 1.0 is better than 20ms t63. And, that includes the test rig latency. A real world test would be to set it up on an engine and pull an injector wire. Then, look for the lean spots at an RPM/2 repeating rate.
Poking around the signals, it looks like they ARE using a controlled current source. But, not in any normal way. The red (pump) wire is always at 2.5v. The orange wire has a 1kHz square-wave signal on it. So does the black wire. The green wire seems to not be connected to anything except at the sensor connector. And, like I mentioned before, the heater wires have a DC voltage of around 8v.
Usually, the orange wire is 2.5v. The pump current is applied to the green wire. And the 61.9ohm resistor goes across the green and red wires to measure the current.
If they are using a constant current source, they already know what the current is. So, no shunt resistor. fine. PWM driving the pump cell? Ok. But, what is to be gained by driving it from the "wrong" end ?
In truth, the only reason I registered on this forum was because I saw your post and your lament that the CAN bus update rate was only 100hz. and I figured I would offer to up that for you on a 1-off basis. That message speed was chosen because many of our beta testers use one per cylinder and any more than 100hz would kill the bus. If a unit were being used in isolation you can demand a lot more of the bus capacity.