In cylinder pressure sensor device.

If you have ideas for a new board, for new features for existing boards, or want to suggest new applications for the I/O Extender, this is the place to post.

In cylinder pressure sensor device.

Postby badbug » Sat Nov 07, 2015 7:04 pm

Would like to see an in cylinder pressure sensor device made for tuning.
badbug
 
Posts: 20
Joined: Sat Apr 17, 2010 3:19 pm

Re: In cylinder pressure sensor device.

Postby jbelanger » Sat Nov 07, 2015 7:15 pm

So would I. Actually they do exist but they are very expensive systems.

The problem is that the sensors that can be used for this are very expensive and require very precise (and expensive) engine positioning for the pressure data to be used. So this is not something I will be able to provide because I don't have the funds to do the research and development.

Jean
Image
jbelanger
 
Posts: 3633
Joined: Sat Oct 03, 2009 12:24 pm
Location: Quebec, Canada

Re: In cylinder pressure sensor device.

Postby Yves » Sun Nov 29, 2015 11:56 am

assuming you have access to a sensor, could the IOX be setup for it ?

I'm asking because, assuming 8000 rpm would equate to ca. 50 kHz in sampling rate necessary.

I know that the sensors are about 1500 USD each. The daq costs another 4000 USD and then some.
Yves
 
Posts: 93
Joined: Sat Feb 28, 2015 9:44 am

Re: In cylinder pressure sensor device.

Postby jbelanger » Sun Nov 29, 2015 8:42 pm

The sampling rate is one thing but the other one is what to do with the data. There isn't enough resources on the IOx to process or store that much data and I'm not sure there would be enough to sample the data and send it out and keep everything synchronized.

So at best it would have to be a dedicated task with external storage and processing. It would not be an additional feature to the current IOx feature set.

There are usually good reasons why something is that expensive. While there are certainly ways to find significant savings, it would be surprising to get a satisfactory solution at less than 1/20th of the cost.

Jean
Image
jbelanger
 
Posts: 3633
Joined: Sat Oct 03, 2009 12:24 pm
Location: Quebec, Canada

Re: In cylinder pressure sensor device.

Postby dontz125 » Sun Nov 29, 2015 9:37 pm

Yves wrote:I'm asking because, assuming 8000 rpm would equate to ca. 50 kHz in sampling rate necessary.

I freely admit I have no knowledge of this - but why do you need to sample 375 times per crank rotation?
dontz125
 
Posts: 199
Joined: Mon Feb 22, 2010 10:27 pm

Re: In cylinder pressure sensor device.

Postby Yves » Mon Nov 30, 2015 5:53 pm

It is actually 48 kHz, but I rounded it off to 50 kHz. :)

Jean,

Look at it from another side, a lot of people are looking for something like this, so there is a sizeable market for it if price is doable. 6k USD for a system is not doable.

I'm no electronics expert, so which is why I'm asking. As far as storage capacity is concerned. I do not think this needs to be stored on the IOX card. You could just as well have a laptop connected to process the data and store it. The IOX could deal with the signal conditioning and sending the raw data to a laptop or so to be further processed. Preferrably TS as this provides other useful info such as load, TPS, temperature and crank angle.

As for the sensor : if you look at the way a spark plug is built, it will allow for some strain gage to be attached to the ring just above the gasket. The porcelain is basically pushed out of the metal jacket with pressure applied on the chamber side. This will stress the metal jacket. So when measuring the stress in the metal jacket after you have calibrated it, I think with a wheatstone bridge you could translate the stress to pressure. Strain gages that you can glue to a spark plug are a dime a dozen. No idea about spark plug temps though.

The difficult aspect is tieing the data to crank degrees. Now remember you only need pressure from about 90° BTDC to around 90° ATDC so in total 180° for every 2 revolutions in order to analyse peak pressure point and detonation/pre-ignition.. This makes for a completely different sample rate. In this case at 8000 rpm, we are looking at 12 kHz.

Anyway, my non-expert .02$
Yves
 
Posts: 93
Joined: Sat Feb 28, 2015 9:44 am

Re: In cylinder pressure sensor device.

Postby jbelanger » Mon Nov 30, 2015 7:01 pm

Yves,

My main problem is that even if I could come up with a complete system that would cost me a very small fraction of the $6k, it would require significant R&D to make sure that the measurements are actually valid. And that would mean a lot of expenses because I would need to have the $6k setup (at a minimum) to validate my own in addition to an engine test bench.

Which means that I would need to spend tens of thousands of $ up front. And I would have to get that back in addition to my time spent on this and some profit margin. So it might end up with a system costing not that far from what the available setups cost.

So if the goal is to have a system that will give reliable data and have rugged components that will survive the conditions, then there is no way for me to do that.

If the goal is to have the IOx sample data from some sort of strain gauge at a rate appropriate for pressure measurement and correlate them with an engine position, that's another thing and might be doable.

But that would mean that the user would need to make their own sensor and calibrate them. They would also need to deal with all the noise and heat issues. And there would also be a need to have a standard way to get a precise engine position which will get you the needed precision with a minimum of processing. That way you can actually have the number of sample points needed at high RPM and correlate them with an engine position. That still is likely to be a stretch for the resources of the IOx CPU.

And there's the need of transferring the data (correlated engine position and sampled pressure) at the needed rate. A 12kHz rate would mean 48kbytes/s assuming 2 bytes each for the engine position and for pressure. That's just for raw data without any overhead for the communication protocol or any additional data or control transfer. That is more that the serial link can handle and would fill up the CAN bus. That means it would need a dedicated link to a dedicated application on the PC. And it also assumes that this transfer rate does not overwhelm the IOx CPU with the numerous interrupts or affect the data sampling and engine position measurements.

So even the basic and minimal solution is not trivial and definitely not just the addition of one more IOx feature. It would need to be dedicated code on the IOx and PC with a dedicated comm link working with specific hardware but it would also require a lot of user involvement on the sensor, wiring and calibration side.

Jean
Image
jbelanger
 
Posts: 3633
Joined: Sat Oct 03, 2009 12:24 pm
Location: Quebec, Canada

Re: In cylinder pressure sensor device.

Postby dontz125 » Mon Nov 30, 2015 7:07 pm

One sample per engine cycle at 8000 rpm is 67Hz per cylinder. What am I missing?
dontz125
 
Posts: 199
Joined: Mon Feb 22, 2010 10:27 pm

Re: In cylinder pressure sensor device.

Postby Yves » Mon Nov 30, 2015 7:39 pm

dontz125 wrote:One sample per engine cycle at 8000 rpm is 67Hz per cylinder. What am I missing?


8000 rpm in total, with 2 revolutions to complete a 4 cycle event, means that every 2 events you need to sample during 4000 rpm/60 = 66.66 times a pass at TDC during power stroke. 66.66 X180° = 12000
Yves
 
Posts: 93
Joined: Sat Feb 28, 2015 9:44 am

Re: In cylinder pressure sensor device.

Postby Yves » Mon Nov 30, 2015 7:49 pm

jbelanger wrote:Yves,

My main problem is that even if I could come up with a complete system that would cost me a very small fraction of the $6k, it would require significant R&D to make sure that the measurements are actually valid. And that would mean a lot of expenses because I would need to have the $6k setup (at a minimum) to validate my own in addition to an engine test bench.

Which means that I would need to spend tens of thousands of $ up front. And I would have to get that back in addition to my time spent on this and some profit margin. So it might end up with a system costing not that far from what the available setups cost.

So if the goal is to have a system that will give reliable data and have rugged components that will survive the conditions, then there is no way for me to do that.

If the goal is to have the IOx sample data from some sort of strain gauge at a rate appropriate for pressure measurement and correlate them with an engine position, that's another thing and might be doable.

But that would mean that the user would need to make their own sensor and calibrate them. They would also need to deal with all the noise and heat issues. And there would also be a need to have a standard way to get a precise engine position which will get you the needed precision with a minimum of processing. That way you can actually have the number of sample points needed at high RPM and correlate them with an engine position. That still is likely to be a stretch for the resources of the IOx CPU.

And there's the need of transferring the data (correlated engine position and sampled pressure) at the needed rate. A 12kHz rate would mean 48kbytes/s assuming 2 bytes each for the engine position and for pressure. That's just for raw data without any overhead for the communication protocol or any additional data or control transfer. That is more that the serial link can handle and would fill up the CAN bus. That means it would need a dedicated link to a dedicated application on the PC. And it also assumes that this transfer rate does not overwhelm the IOx CPU with the numerous interrupts or affect the data sampling and engine position measurements.

So even the basic and minimal solution is not trivial and definitely not just the addition of one more IOx feature. It would need to be dedicated code on the IOx and PC with a dedicated comm link working with specific hardware but it would also require a lot of user involvement on the sensor, wiring and calibration side.

Jean


Ok, understand what you mean (more or less).

As far as the calibration is required : I'm not sure how hot the metal jacket of a spark plug gets, but in view of some comments I've seen on CHT's, they shouldn't be to far of the vicinity of 100-125°C. Vishay has sensors that live up to 350°C.

As for calibration : since the strain gage is not in contact with the combustion itself (indirect measurement) I think you could calibrate it with some sort of metal cylinder in which you fit the spark plug and that you put under pressurized air. If need you could heat up the cylinder to 125°C to have working temps or so. Maybe I'm looking at this in a much too simple way, but that's how I would do it.

If the data rate is an issue there is a variation on the theme that one could do and that is to only detect the exact time at which maximum pressure is present in ° ATDC.
It depends on the engine characteristics as an engine can either be knock limited or mbt limited, meaning it can either hit knock before it reaches mbt or it can hit mbt before knock. Obviously the second would be the most ideal situation.
Assuming the goal of setting the ignition timing would be to reach mbt and therefor maximum power, you would only have to register the strain gage for maximum output and relate that to the number of degrees ATDC. For all intends and purposes and on a normal engine this would have to be somewhere around 15-20° ATDC.

Of course that would leave knock detection and/or pre-ignition detection on the table.

You also have me wondering about those NI DAQ's I have lying around (unfortunately in PCI format)
Yves
 
Posts: 93
Joined: Sat Feb 28, 2015 9:44 am

Re: In cylinder pressure sensor device.

Postby jbelanger » Mon Nov 30, 2015 8:42 pm

That is also how I would do the calibration. But I have no idea if the static result would still be valid under dynamic conditions. And there are not many ways to validate that other than doing back to back comparisons with a known calibrated sensor. On the other hand, you don't need absolute values so it might be enough to simply get sample data and then fit a curve to compute the actual peak pressure position (ppp).

But that's also why you don't want a simple maximum sample. Since you will have some noise and the maximum won't necessarily fall exactly on a sample position, you need at least enough data points to fit a curve to find the ppp. Of course, if you have knock you won't have a curve but I'm not sure how you can analyse the data to determine that.

One way to reduce the amount of data needed is to limit the sampling to a small window around TDC so that you can detect knock and get the ppp. This way you have some time to transfer data even while you're not sampling as long as you finish before the next event.

If you can do some test with the NI cards in some way and with the simple calibration setup, it would be a good initial test to see if you can get something meaningful out of a simple strain gauge. And if the NI setup is fast enough, you might even be able to see pressure variations from a (tankless) compressor.

Jean
Image
jbelanger
 
Posts: 3633
Joined: Sat Oct 03, 2009 12:24 pm
Location: Quebec, Canada

Re: In cylinder pressure sensor device.

Postby Yves » Tue Dec 01, 2015 3:07 am

I'm not sure however if you can transfer anything that is programmed to work with the NI setup and programmed in labview can be used later on to be integrated into TS. The thing is that those programming languages are not compatible. So sharing sensors would be difficult I think.

Personally I lack the programming skills to make this work.

Anyway and fwiw, on another forum we were discussing ignition timing, knock and other aspects. The guy from TFX Engine (nitro) chimed in just a couple of threads ago. Might be of interest.

http://speedtalk.com/forum/viewtopic.php?f=1&t=44918
Yves
 
Posts: 93
Joined: Sat Feb 28, 2015 9:44 am

Re: In cylinder pressure sensor device.

Postby Yves » Tue Dec 08, 2015 4:48 am

A bit further down my research route. I have NI pci-6024 cards (2 of them available). Seems that with a pci to laptop adapter unit I could make them work on the car.
In the past I studied a bit labview but haven't continued this. However, it seems that the use of DAQ and programming it in labview is doable (at least it seems like that :D )

I have contacted Vishay to see if they have a suitable stain gage for mounting on the spark plug and that would hold up in use.

The only thing I still need to figure out is how to go from the saw tooth crank and cam signal to degrees of crank rotation.
Yves
 
Posts: 93
Joined: Sat Feb 28, 2015 9:44 am

Re: In cylinder pressure sensor device.

Postby Yves » Thu Dec 10, 2015 7:10 pm

A representative of Vishay locally can equip the spark plugs with the strain gages. So far so good.
Yves
 
Posts: 93
Joined: Sat Feb 28, 2015 9:44 am


Return to Wish List & Requests

Who is online

Users browsing this forum: No registered users and 2 guests

cron