[cjjonesarmory]

I just had a brainstorm thinking about a recent earthquake that happened in my area. Then I began thinking about something I saw on one of the science channels where they were testing construction techniques. They used what appeared to be a hydraulic cylinder to shake a platform. I imagine that for their test to be valid it would have to be a very precise system. So I began wondering if hydraulic systems could be used as precision positioning systems, and if such a system could also be adapted to something such as CNC or robotics. I have considered electro-mechanical linear actuators, but the size of the actuators is too big for the amount of force I need. So a hydraulic cylinder with a pump that can be mounted out of the way seems to be the answer.

This must have been done before, I can not be the first person to consider something like this. The application in mind requires compact size and large amounts of force, but also requires a certain amount of accuracy and precision to be controlled by a computer, at least in part, so using standard manual hydraulic valves would not work.

I'm wondering if the piston can be positioned similar to the way a stepper motor is positioned, meaning simply metering the amount of fluid coming and going to and from the cylinder. My current use for such a system would have an accuracy and repeatability of about .100" to .250" so it doesn't need a high degree of accuracy. Coming from an automotive background, I know that a lot of Dodge/Chrysler vehicles use an automatic transmission that uses PWM solenoids to meter fluid into the various hydraulic circuits to change gears. This is also how fuel injectors meter pressurized fuel into the engine. The application I have in mind is not quite the same, but similar so I think the concept is valid, just not sure if this would work with hydraulic cylinders.

Any thoughts or information on the subject would be greatly appreciated.

[underthetire]

Ya, it's been done. Back in the 60's and 70's, even a few earlier. Servo valves, thermal changes in the oil, run-aways, leaks, noisy, positioning problems. It used to be the old stand by for a lot of NC machines. Some press brakes still use something similar for the rams. I'm glad its all gone myself.

[cjjonesarmory]

As a follow up, I'm wondering what would make a good linear position sensor for such a system. The precision machinery in my shop uses glass scales for the DROS, but my system will be exposed to dirt, impact, and a lot worse I'm sure. I had considered using something like a linear digital scale, but again, not too confident about the ruggedness.

So, to summarize, I am curious what would make a rugged linear position sensor that has an accuracy of around 0.100" to 0.250"? I suppose it would have to be some type of fully enclosed sensor, perhaps a linear resistor where voltages could be mapped/calculated to specific positions.

[underthetire]

Newall makes the best scale for this type of application..
Newall Electronics - Digital Readout (DRO) and Linear Encoders

[Al_The_Man]

Back in the '80s Bosch had a CNC machine actuated by Hydraulics and Proportional Servo valves.
Typically using scales for position.
Today hydraulics can be found in Tube bending machines where the rotary forming is done with hydraulics, servo valve and rotary encoder.
Also Elox used them on their EDM machines and the servo valve advanced and retracted the electrode by feedback from the spark gap.
Al.

[cjjonesarmory]

Thanks for the input. When "rotary encoder" was mentioned for the tubing bender it sparked another thought, again from my automotive background: I remembered that Throttle Position Sensors use a type of rotary encoder I guess you might call it. It uses a 5V reference signal. As the throttle is opened the voltage coming back from the sensor varies proportionally with the amount of movement. I'm thinking I could use a rotary encoder on the hinge/joint instead of a linear sensor that measures the amount of extension of the cylinder. The hydraulic cylinder would be used to move a component in a similar fashion to a tubing bender.

[Al_The_Man]

I remembered that Throttle Position Sensors use a type of rotary encoder I guess you might call it.

I think you will find they are proportional controls, IOW it most likely was a variable resistance pot that changed on the rotary sensor.
It depends on how complex you want to make it, there are various ways to roughly position a Hyd cylinder.
Al.

[cjjonesarmory]

Al the Man: I think you are right about being proportional controls. I'd have to calibrate the sensor and correlate a given voltage to an angular position in degrees.

TPS (Throttle Position Sensors) can't be the only thing that functions this way, but it seemed to me like a good starting point. From my automotive background I know that the inputs to a car's computer are getting more and more precise. In fact, on a lot of new cars, including Corvettes (I believe), they have coined the phrase "Drive By Wire", meaning that the gas pedal is no longer even connected directly to the throttle plate. The gas pedal only moves a sensor which in turn causes a servo (or some type of motor) to actuate the throttle plate. I would imagine that there would be a great deal of sensitivity in such a sensor for a system like this.

Now, I don't necessarily believe it would be cost effective for me to the Chevy dealer and buy a couple dozen "gas pedal sensors" or whatever they call it, but clearly a system that has been put into widespread use must have a lot of engineering expertise behind it. I just need to find out the details of what exactly is being used in that type of system and duplicate it.

[wildwestpat]

I have built self levelling systems for radar survey equipment using hydraulics as the motive power to an accuracy of better than one minute of arc from true horizontal in any direction. The radar head weighed in at 2Tons plus when mounted on the self levelling turntable which in turn was mounter on the survey vehicle.

The problem is that all the pipe work after the control valve/s must be in ridged tube not flexible hose. The small expansion in the flexible hose gives a lot of spring to the system that is difficult (very) to compensate for. The control valves are a problem as you need to reconcile the large flow of fluid for rapid traverse with the need for micro control to get precision positioning. Variable rate valves IMO are a bundle of problems in themselves and three pre-set flows are preferable using high pressure drops across small knife edge orifices to get a reasonably constant flow independent of temperature. I used the jewelled sapphires used in watch making as a source of hard edged holes that are resistant to wear. To state the obvious the system has to be clean on the inside and small flows of fluid are easily obstructed. Knife edge restrictors are preferable to the screw thread sort on temperature viscosity grounds and are easily incorporated with poppet valve spools. Dual activating cylinders are attractive but the control becomes a bit of a problem and best avoided.

Apart form tube / hose expansion there is a tendency for the piston seals to cause stick-slip as the seals have to resist high pressures as well as permit linear motion of both piston and ram. Temperature changes cause the oil volume to alter and hence the need for feedback from sensors to keep checking and correcting.

There is something very satisfying about the primeval motive power of hydraulics but the power can cause problems in the event of an obstruction.

You can get a good feel for hydraulics first hand if you can get to have a go at a JCB or similar with hand controlled valves on the motion. Hydraulics would make ideal motive power for a shaping machine but not one IMHO as axis motive power for CNC making watch parts!

PS What I am saying I think is hydraulics could be OK for rough motion even at small sizes but not easy for precision use unless you need the brute force that hydraulics can bring to the design

Nice idea though - so good luck if you build one. Regards - Pat

[Al_The_Man]

I would think you could make a simple comparator from a couple of LM311's or similar.
IOW, one input would be a set-pot and the other a feedback pot.
The output would switch off at coincidence.
This would be if a simple positioning motion that once in position and coincidence was reached the hydraulic cylinder solenoid would turn off and rely on 'instant' stop of the hydraulically blocked cyclinder, if this is accurate enough for your needs.
If large pressure is not needed, a air over oil system is cheaper and quieter.
A similar LM311 could be used on the retract if required.
If you were able to use something like a PicMicro processor these have all the functions you would need to do it simply and cheaply, as it may be done in the automotive application you mention.
This method would not be practical for CNC or robotics, however.
Al.

[wildwestpat]

As with all machine designs start with the requirement. Power to move the tool including any reactions from the work - precision necessary on placement of tool - time to react - total movement required.

This will fix most of the parameters for the design of the motive power pack. Air or any other gas is going to have problems with a proportional placement on two counts -expansion of the gas with temperature changes - compressibility of the gas will give rise to a very springy actuator which will make for imprecise actuation and low frequency oscillations which will have to be filtered out. Fluids being largely incompressible should rapidly repay the slight extra complication.

If you want to run open loop then consider using use of heat as the actuating force. The heat causing a fluid or gas to expand rapidly actuating a piston linked to some form of rotary or ratchet device. One blip of heat increments one notch. This would be simple to interface to the electronics and would also allow open loop control. The action would be one actuator impulsing a ratchet in the plus direction and another for going in reverse. Actuators have been built on semiconductor wafers that will pass through the eye of an ordinary sewing needle so a bigger one should be easy to fabricate. The choice of fluid would be easily vaporised liquid or a gas. The problem being to cool the gas after each stroke so Stirling Engine design buffs may have some useful advice - I am suggesting a ratchet action not rotary.

Regards _ Pat

[cjjonesarmory]

I wasn't sure I wanted to mention the specific application before, and I still don't want to give away too about the exact application because it needs to stay under wraps for now, but I'll divulge a little bit more info.

The system I am working on would be primarily outdoors and subjected to lots of dirt, vibration, and who knows what else. The best example that I can think of without giving anything away is to imagine a computer controlled back-hoe. I believe that is what they are called anyway, the big scoops on the end of an arm that has multiple joints. It's a little more complex than this, but similar.

So basically what I would need to do is equivalent to adding sensors and computer controls to a back-hoe arm. Hope that helps.

I appreciate all of the suggestions so far.

[Al_The_Man]

If you want to employ continuous positioning, i.e. servo PID loop, you are going to need a proportional servo valve and amplifier, not cheap.
I have built a similar application using a rotary encoder and used a simple National Semi IC LM759 to drive the valve directly.
For the PID loop controller you could look at a Galil single axis motion card such as the legacy DMC-1500 and set it up to run off of 12v or 24v DC.
If you want to use the cheaper proportional method which can give you continuous postioning, there should be many Application notes in any of the IC manuf literature, such as Nat. semi., Freescale, Texas Instruments, Allegro to name a few.
Any sensing device used in that type of environmental is going to require rugged protection and be able to take the vibration etc.
Al.

[wildwestpat]

Some where out there I have seen a hydraulic ram with an inbuilt linear sensor. Might have been at Dowty Mining or Lockheeds. The thing was used to manipulate loads of many tons in the quarry / mining industry. I was interested until I discovered an ultra sensitive electro level which allowed the system to run closed loop on the parameter (absolute level) that was important. In that application ram extension was only needed to determine that the end stops were not approached and that was only necessary as the system was powered by compressed gas bottles not a pump as noise was a problem.

Good luck - regards - Pat

PS you can use a differential poppet valve arrangement to change the speed of extension and retraction which may well be very different as the volumes to be shifted are considerably less on the shaft side. Careful arrangement of the poppet seating will cushion the snap change in oil flow but for the accuracy posted should work well. Cushion the poppet's valve seat using a spring loaded piston so that the fluid that is behind the piston is enclosed and completely separated from the main high pressure fluid paths. Use a small one way bleed to control the escaping fluid with a differential bypass of the restrictor to permit rapid return of the valve seat support piston when the pressure is reduced due to the high speed poppet solenoid valve closing. Use either a bleed and one way valve from the unpressurised side of the system or a small air reservoir. The steps in oil flow only require extra electronic comparators in the feed back loop to drive the solenoids operating the by pass poppet when the error is large and closing when the error is small and the desired position is being approached. Two steps may be sufficient or you may need more depending on the ratio of fast to slow movement that is required.The big advantage is that the fine restrictors are only used for fine movements saving wear on the orifice (this is why I used watch jewelled holes). I suggest you try changing the pressure manually on a ram similar to the ones you intend to use. Some stick-slip like mad and you may well find that a solenoid operated system will give finer control than a variable rate valve due to the inherent jerks induced by the solenoid operated valves. IMO solenoid valves are cheap to make and easy to incorporate into the pipe matrix block with the restrictors non return valves and other things hydraulic.