"Cat whiskers" sensor as Z-probe?

oliof

@o_lampe build it! The big question, as always, is repeatability.

o_lampe

@oliof
Repeatability is pretty stable when using a Schmitt Trigger. The analouge value would depend on the silicone bulb.
I could build it, but I have no test-equipment (scope at least) nor budget for PCB production.
And I'd need help with implementing such sensor in RRF.

oliof

@o_lampe maybe @PCR would be able to collaborate, time permitting.

o_lampe

I just ordered two sensorboards from China. Will take 3 weeks to arrive.
There is a Cpp library written for Arduino which uses the I²C port. I guess, it's not too difficult to rewrite it for SPI if necessary.
Would any of the two options have an advantage for RRF implementation?

Now working on the mold. The chip is so tiny, I'm afraid a printed mold wouldn't be accurate enough for massproduction
I have enough time to figure out...

o_lampe

I just found a possible candidate for the silicone bulb:
Lawn mower carburetor primer

They are quite cheap, easy to get and might also work as underbed-sensor: Put a few of these under the bed and connect them with PTFE-tubes. Guide the tube to an airtight housing with the GY-63 inside. Voilá, underbed Z-probe.

dc42

One issue with any kind of bed height sensor is the response time. The datasheet for that barometric sensor gives the maximum output data rate as 100Hz, implying a response time of 10ms. That's on the slow side for a height sensor. For example, if you probe at 20mm/sec (F1200 in the M558 command), then a variation in the reading time of 10ms implies a position uncertainty of 0.2mm.

o_lampe

@dc42
Good point, I found a response time of 0.5-8ms. Depending on the sample rate (256 times oversampling will do for our usecase)

Question is, can we reconfigure it?
The other point is the interface, which would cause latency too. (*)
Maybe a simple pressure sensor would be better? We don't want a weatherforecast from Duet boards, right?

*) The TMC drivers use SPI, what's the latency/loop time there? Can we squeeze another SPI client in queue?

o_lampe

Another way of hotend probe came to my mind, where a silicone tube around the heatsink neck is used to feel the nozzle touch the bed. Like 'Precision Piezo', but without the need for speed.

arhi

last time I was looking at any decent (resolution and speed) gas pressure sensor for a project I could not find anything worth looking below 80 US$ in volume so I gave up. I see this one is 8US$ in volume so that's acceptable but the speed is serious problem here. Also you would have to make differential measurements non stop as this is heated environment so air pressure varies a lot with temperature ... also the housing of this sensor is not very friendly for what you'd wanna do.. something from SDP3x series like SDP31-500PA-TR-1500PCS would be much better (3x more expensive, 20x faster with 2kHz update rate )

o_lampe

@arhi said in "Cat whiskers" sensor as Z-probe?:

but the speed is serious problem here

As I wrote above the response time can be as low as 0.5ms = 2kHz if we find a way to change oversampling rate.
But I'm open for your pressure sensor, it's also well suited. But regarding integration we'd start from zero. ( the analog version would be the simplest, just treat it as a switch)
Let's see, if we can get some development samples.

The GY-63 MS5611 has been used in Arduino&Multicopter world and comes with a basic Cpp-library.

o_lampe

Ouch, those SDP3x sensors are expensive. ~25€ a piece...
The analog sensors are quite slow (3-9ms response time), while the I²C version in 2kHz continuous mode also offers averaging, smoothing and mass flow data. That's quite useful, we just have to read the direction of flow, and it doesn't matter if it's hot or cold.

arhi

@o_lampe yes, ~25E at volume, 3x more expensive than 5611 that's ~9E at volume ... still super cheap comparing to sensors I checked few years ago, looks like price dropped significantly (these types were around $100 and more) ... one should investigate more there might be other ones better suited... imho 5611 has not only issue with speed but serious problem with packaging, how will you attach a tubing to it? you would have to put the whole pcb inside a balloon... not an easy design decision..

arhi

btw I have in my "garbage box" NBPLANN150PAUNV no clue why I purchased it but I guess I found it useful at the time. This type of sensors are used in medical apps (blood pressure monitor for e.g.) so might be interesting for this application

https://sensing.honeywell.com/honeywell-sensing-board-mount-pressure-tbp-nbp-series-datasheet-3008328.pdf

.. they are analog output $8 at volume, $10 in onzies .. Wheatstone Bridge configuration so should be super fast and precise (IIRC 0.2% accuracy) .. of course for this sensor going with 0-1psi version (available) would be better than 0-150psi as you would want to detect minute changes so a very small range is optimal ... proper barbed port so you can attach a tube..

o_lampe

@arhi
Thanks for all the input, it helps not starting from zero or work on something, others have abandoned already.

It would be pretty easy to set up a wheatstone bridge with our beloved 100k NTC thermistors and use them as mass flow sensor.
THAT would be cheap, but requires an old-school analog circuit expert. (me: looking in the mirror)
But at first I want to have a working sensor and go on from there.

If anyone else wants to work on this concept, I don't mind and I have no financial interest.

arhi

@o_lampe well the NBP series are imho great for this

lot of choices for range so you can select best one for application
good package so you can easily attach a tube
$10 for single piece is not a dealbreaker for prototype design
they come both smd and through hole version so soldering is not a problem even for beginners
wide range of power option (1.8 to 12V)
wide range of temperature (compensated 0-85C for TBP version)
analog output, bridge, should be fast and sensitive, even the lowest output of mV/V of 1.5 will produce 0-7.5mV if powered by 5V that's decent output spam to not be drowned by noise and not to have to use expensive opamps

So, IMHO, I think it is the ideal candidate for a sensor like that. I think the mechanic design and manufacture is going to be a much bigger problem.

o_lampe

@arhi
It only makes sense to spend more money than a BLTouch would cost, when the result is equal or better. I've hoped the footprint would be small, but the silicone bulbs I linked earlier are 20mm diameter. Good for underbed probes, but a real winner would be the heatsink-necklace as it has zero offset and requires less force than piezos.
We'll see. I have plenty of projects ATM. Maybe someone beats me with this one?

arhi

@o_lampe said in "Cat whiskers" sensor as Z-probe?:

@arhi
It only makes sense to spend more money than a BLTouch would cost, when the result is equal or better.

To me it would make sense investing time if it would solve a problem that exist and is not solved. TBH I don't see what unsolved problem this could solve except to work as underbed probe. ATTM I don't know of reliable underbed probe. Probing from the top, I doubt it can have smaller footprint than some piezo/bltouch/easytouch/.. sensor .. and as for precission and repeatability a simple microswitch is "good enough" so

but a real winner would be the heatsink-necklace as it has zero offset and requires less force than piezos.

I can't visualize how would this work