Experimental Instrumentation

[fredrosse]

Electronic Pressure Transducers work well with recording engine cylinder pressure. Their response time is on the order of fractions of mili-seconds, so they are fast enough to record an "indicator diagram" for steam engines. As far as temperature ratings, it is a simple matter to put a water leg into the installation, having perhaps 2 inches depth, to protect the transducer from steam temperature. This water will not interfere with the readings at engine speeds of several hundred RPM.

Transducers are available for higher temperatures. Standard Omega transducers can operate at a maximum rated temperature of 185F, the 32B model is rated for 335F environment, and the PX1004 model is rated for 600F.

These transducers would operate at temperatures much lower than inlet steam temperature, unless they were screwed directly into the main steam supply pipe and insulated. Even if screwed directly into the steam cylinder, they would approach an operating temperature somewhat less than the average of inlet and exhaust temperature.

There is more information on a previous thread in this forum, look up the topic "How Much Water"

[87gn@tahoe]

It is a design based on the Tverskoy engine. I haven't had a chance to boot into Windows, but as soon as I do, I'll make some diagrams from the SolidWorks files. By SBA you mean the folks over here: http://www.steamboat.org.uk/ ?

Yep! That's them

[Dhutch]

I think there were/are some members of the SBA working on a digitized engine indicator

A blast from the past, but if you still use the site, any idea is this came to anything?

Given we have transducers at work that will do 500bar (yes bar, 7000psi) and 150C/300F there must be transducers that can manage 185c (wet steam at 200psi) else as you say, a short syphon would take the temperature down significantly.


Daniel

[fredrosse]

"there must be transducers that can manage 185c (wet steam at 200psi) ..........."

Just look at the previous post that is 2 positions previous to yours, fast acting pressure transducers that are rated for the temperatures and pressures of interest, and just stock catalogue items are available.

[johngriffiths]

I seem to recall the late Robin Wallace Sims producing a unit that worked quite well, the difficulty being the attachment of the piston position sensor which varied so much, engine to engine, it had to be workshop modified for each situation. His software was written for a somewhat elderly portable computer and after damaging one, he had difficulty in replacing it. I remember thinking the sampling rate was a little low for high speed engines.

Does anyone know where this unit is now?

JohnG

[Lopez Mike]

I've been puzzling over this for a year or two. The transducers aren't the big problem. Less the $50 each for fast and low volume units.

The issue I have not solved yet is a decent way to measure the piston position.

I'm beginning to think that the way to do it is to measure the crankshaft position and hand enter the stroke and rod length into the algorithm and let the stinking lap top solve for piston position.

The computer is going to spend almost all of it's time dozing away between data points so it shouldn't be any problem to do a simple trig problem every degree or so. One TDC pulse from the crank sensor per revolution and let the data system do an average an interpolate for crank angle.

Hey Bart! This sounds like a great Java project so as to make it cross platform.

[fredrosse]

Picking up crankshaft angle position for every revolution, and having the computer find the piston position for any crank angle is a good idea, and fundamentally works well for determining the "x" axis of an indicator problem. Of course the "head end stroke" is not the same relationship to the crankshaft angle as the "crank end stroke" with a connecting rod of finite length.

However there is a problem with just picking up a single crank pulse for every revolution. Our launch engines can seriously change speed during a single engine revolution, so don't tell the computer to divide each revolution into 360 equal spaces representing degrees of crank rotation. Depending on the size of the flywheel, around 10 to 20% RPM variation occurs with a single cylinder double acting engine that only runs at a few hundred RPM. A pulse system with at least 36 pulses per revolution (every 10 degrees) would do OK, and show the actual crank position adequately.

High speed, multi-cylinder engines do not have this problem, but on our steamers it is pronounced.

[Mike Rometer]

Automotive ignition systems have been doing just this for years, to determine timing changes. The early ones (at least) used a 36 division plate with a reluctor coil. They were able to work down to points of a degree. The pulse generation is remarkably simple. The counting and mathematical manipulation of the results is probably another matter.

[barts]

A slotted disk and optical pickup (LED & phototransistor) will work here easily... a separate pickup for TDC will get the whole thing going after the first revolution after startup.

A lookup table to convert crank position to piston position (filled out using the stroke and crank length) and a small microcontroller could easily drive a small LCD display w/ a real-time display of cylinder pressure vs piston position. It may not be practical to update the screen at the engine frequency; averaging results to update the display less often may make more sense. Linear interposition of crank position between 10 degree updates from the optical sensor should work fine.

Many of the controllers have analog input that would be suitable for reading the pressure values from the transducer; remember to low-pass filter the pressure transducer signal; filter cut-off should be at half the sampling frequency.

- Bart

[Mike Rometer]

Optical sensors were abandoned for automotive use because of dirt getting on the sensor and obscuring it. Not a problem with a magnetic reluctor. T.D.C. was sensed with a slightly longer pulse.