i am working on the best setup to do this testing, including the tar content before it enters the engine, lambda, EGT…
I have done some testing with not so perfect charcoal, have to extract the data to put in in picture wise…
It looks that higher tar content has a higher SO2 reading in the gas but that reduces in the exhaust gas.
Second is the leaner burning brings down the NOX ( see above picture for the O2 % )
Did testing also on my CRF250, if i restrict the exhaust, the NOX gets really low…
Working and studying on a lot of things at same time…
9 Likes
making some advancements on the controls. The GUI looks the way it does as there are other data displays with in it that are blacked out. These are data points I had planned to implement but needed to learn how first. So rather than rebuild this GUI later I built it with all these displays from the get go. So I have learned how to do all this now but will be working on one thing at a time.
So one tough one was eliminating the engine start button and implementing this into the display. Unfortunately there is no way to create momentary button in the display. So my solution was to integrate the starter engagement with the mixer valve adjustment from the GUI, as you generally are adjusting this as you start the engine. I made the slave controller able to detect the movement of the slider adjuster values. If its moving the engine starter will engage and then there is a short lag after releasing or stopping movement. So you can idle for a very short interval but you do need to keep it moving while trying to start the engine. Only bench tested for now, I may try to get this loaded up for an evening run after the sun goes away. 
The next thing that sort of goes hand in hand with engine starting is the AFR Read out. The AEM guage is inside the enclosure so you can not use this with the doors closed. The plan was to integrate this into the GUI, so I have finally achieved this. The Arduino slave basically compresses this signal and sends it over I2C to front end master. Then this data stream is mapped to accordingly for display.
Yeah was up till 5:30 this morning back up at 10 and just wrapped it up. Going to take a nap and play later 
10 Likes
Next on my plate is a TAR sensor. Never ever again will I or a client deal with tar getting inside the engine. Ive been planning this all along but have just now got around to testing this. It works!!
8 Likes
Interesting! So the tar is electrically conductive? I’m not getting this.
Rindert
2 Likes
Indeed, it has a slight conductivity that I can detect. There will be a probe inside a sheave inserted just before the carburetor inlet. The sheave will be milled so it will be sort of like a spoon with the probe extending over top of it. If tar is produce it will hopefully collect here and bridge the probe and sheave.
I’m using a pull up resistor in the control and this probe is wired to ground. So when we detect tar it will pull the voltage down to a preset parameter. I have more to test yet as I need to test water and see if there is a difference. Im hoping the tar has more conductivity than the water and also hope it will not collect in the sheave. There will be a drain hole so it can drain off. but tar should still collect.
The voltage drop is around 1.5 volts and this is plenty enough for detection via the controller.
4 Likes
Isn’t there a lot of moisture in wood gas? What are you actually sensing that can distinguish the humidity of wood gas vs tar?
5 Likes
I still need to test the condensate vs tar yet. So far there is a difference between tap water and the test sample. However I think in process it wont work like that as the two will be mixed. But I still am expecting different readings between high tar content moisture vs lower content.
Unfortunately the tap water has a higher conductivity. But this may change with another test sample that has a better mix of condensate mixed in it. With the cleaner condensate having a higher conductivity this makes the code a little more complex. As the condensate will drive the values past the tar values setting off the code in process. So will need to use sampling. So the code will perform multiple test and count them. If it is condensate then the values will be driven past the dirty parameters and once past these values will reset the the sample count back to zero. When the gas is dirty it should fall in this dirty parameter set and after multiple readings and the count expires it will fault out. I will make this work
3 Likes
What range of resistance values are you seeing for tar, for tap water, and for the mix?
I’d have thought that the gooey tar in your recent video would have a pretty low resistance compared to even “dirty” tap water. Sounds like a lot of room for experimentation with a wide range of voltages, frequencies, DC pulses, etc…
Pete Stanaitis
3 Likes
I have not tested resistance yet. as this sample was a scraping from a drain pan. I need to get some better samples to try and get more accurate testing. But you are on par with what I have found so far.
Yup thats what Im hoping for is a range of resistance clean vs dirty
2 Likes
What I have found so far… the voltage on my input pin is 4 volts (pullup) when I have the probe in the sample and is grounded I see a voltage of about 2.5 volts. I tested a damp sponge with tap water and I see a voltage of around 1.5 volts. So about 2.5 volts drop water and 1.5 volts drop for the tar sample.
2 Likes
I also have to wonder, if the steam carried in the gas is alkaline? Is it becoming alkaline only after condensing and mixing with the ash?
If so, than I think what we are going to find is the cleaner moisture will be less conductive than the dirty gas.
3 Likes
But the tar will be carried as tiny particles in the wood gas, right?
3 Likes
Yes, however, at this point is where we generally see tar rapidly condense at the carburetor inlet. So this sensor probe will be designed to catch it. It will have drains as well. so the thinner condensate can drain off vs the tar will accumulate. That being said if tar is detected than this probe will need to be cleaned. But Id much rather clean this than deal with this stuff gunking up an engine.
2 Likes
I would think it would be acidic. H2S is produced during pyrolysis. Then it combines with water to form H2SO4 (sulfuric acid).
Rindert
3 Likes
ah ha thank you 
2 Likes
So i did some more experimentation today. I collected another tar sample from a drain pan thats been in the sun. So it is just a gooey glob of tar. I can not detect resistance with my meter and can only see a very small amount of voltage drop via the controls. Not enough to even detect a change.
So check this out, what if the sensor can detect alkaline moisture all the time? Now if we are making tar then what should happen is it will coat the probe and insulate it!! I think in real world practice this is how it will actually work. I still think there will be a variance in the moisture mixtures as well, but this will be the precursor to the probe becoming insulated.
I took a sample of pure condensate from the machine this I have plenty. With the probe submerged there is around a 3+ volt drop from the original 4 volts starting. So more than tap water. This is good as this can work too our advantage. In any case we just need to see a change when tar is produced.
I implemented a read out on the touch screen so the tar content can be monitored. 10 is bad and zero is best. If at zero that means the condensate is fully pulling down the input to zero. As it rises this would mean there is a higher concentration of tar or nothing has accumulated yet. So I think if tar is produced the cut off point will be long before reading the upper limits: The higher end of the scale will be ignored or at least at first start up.
I ended up completely re building the GUI over the weekend. I finally found a code snippet with an example of a working momentary push button. Unfortunately I could not get it to work into my existing code. The formatting just would not work. So I made a copy and erased the core of the code leaving a clean canvas. Started with this button and then added in my code piece by piece testing every time I would add something. Long a tedious process. But this allowed me to really clean it up at the same time. Much cleaner and organized and now I have a push button for the engine starter it is soooo cool.
To the mid right you can see TCS (tar content sensor) So this will be a cool means to monitor this sensor. Hopefully I will be back up and running later this week!!
4 Likes
So you are seeing something higher than 30 megohms?
Are you using a 10K ohm pullup? That may be way to small. Try 100K ohms, or even one meg.
Maybe time to read some schlolarly papers on this subject?
ie:
One idea that I saw was that they fed a 4 ampere current through one sample and read the voltage drop at 3 or 4 places in between the input probes. Sounds like even the “experts” find this to be a tough job.
—Be patient, the article takes a while to load.
I think you planned to use this sensing as sort of a “fire alarm” to shut down in case of tar, right? If so–
Maybe some kind of a filter that plugs easily, causing a serious drop in gas flow that is easy to detect. —Or— even an easily plugged filter that is physically dislodged, tripping a sensor as it moves?
Have you seen Alpowerlabs tar testing system from several years ago? They used cloth filters in series with the gas and determined “tariness” by the amount of color change that the darkened cloth imparted to a jar of solvent in a certain period of time. I wonder if a person could cycle the cloth in and out of the gas stream, constantly testing for color or density change.
Pete Stanaitis
4 Likes
I’d vote for the static filter idea in some form.
Another way it could work would be as a final safety filter, then observe pressure drop across the filter. Though by the sounds of it, specific conditions to assure condensation would have to be met.
If a bubbler type filter was used, a “change oil” commercial automotive system could be used, I assume they look at the colour / clarity of the liquid.
I agree this is going to be quite challenging. But it’s worth exploring.
3 Likes
I’m with the understanding one can’t filter tar out from the gas yes the tar can gum up an engine. Is there a certain metal that the tar would be attracted to over another type of metal? If so, if one had wide flat probes of that type of metal in the gas flow stream that were spaced just far enough apart would not produce a reading but close enough where if the tar starts building up on the probes it would start bridging and then produce a reading.
5 Likes
Im using the internal pullup resistors built inside the Arduino’s .
pinmode (Pin#, INPUT_PULLUP);
You can direct short this internal resistor and read an analog sig as well. it is already in perfect range for detection.
The meter I have will not detect any resistance. The batteries maybe shot they are pretty old. The meter still works, but I should be able to at least get an ohms reading on the condensate. But Im not even seeing a reading there. This would be good info to know; however, I already have a usable reading for detection. Just need to learn how to build the sensor and implement it.
Hopefully this first probe design will work.
I dont follow APL and am trying to get away from any type of filters. This probe will be fairly easy to clean compared to filters and then the need to replace them.
This sensor should never go off. It is there to make sure tar is never ever an issue. If it goes off then corrective action is required somewhere. Something broke, bad fuel. clog some where etc. Fix the issue, clean the sensor and your back in business.
2 Likes