quote:

That is why I don't consider calculating the natural frequency of a motor to be relevant to computing the acceleration time or velocity.
http://forums.hydraulicspneuma...=484101371#484101371

Unless I misunderstood it, and I don't think I did, the discussion in that thread was calculating the natural frequency, and then finding the minimum allowable acceleration or deceleration time to avoid instability. I like this idea and think it's a good thing to double check. One of the things that I never got around to doing is to try and find out how much internal leakage would offset any results. Would it be probable that in most cases the amount of internal leakage/damping would take us away from this danger zone anyway, making this check a waste of time?

quote:

Originally posted by Doug Hanson:
Unless I misunderstood it, and I don't think I did, the discussion in that thread was calculating the natural frequency, and then finding the minimum allowable acceleration or deceleration time to avoid instability. I like this idea and think it's a good thing to double check.

The method in the book is optimistic. The true rise time is much longer than 0.2 seconds but it depends on the valve coefficient and the supply pressure.

quote:

One of the things that I never got around to doing is to try and find out how much internal leakage would offset any results.

The leakage makes a significant difference. When the valve is open to get to full speed the damping is very high because energy is lost across the valve.

quote:

Would it be probable that in most cases the amount of internal leakage/damping would take us away from this danger zone anyway, making this check a waste of time?

I agree when accelerating but not when decelerating. As the valve shuts there is little energy lost across the valve so the system will tend to oscillate without motor, valve or cylinder leakage and friction.

It is strange that the books never consider what happens when an actuator stops.

Updated spreadsheet.

Natural_FreqC.xls (108 KB, 22 downloads)

Well, this thread is just an example and applied to "proportional DCV" valve. More things have to be considered like valve pressure drops, pump pressure(constant for all phases of actuator movement), etc. These other things look simple but greater understanding is needed to really be able to design(properly)...a lot of considerations...different actual applications have different approach...maybe we need to learn more if necessary...

Humnnn, after the NF calculations...I will just resort to some rule of thumbs for undamped natural frequency and acceleration/deceleration...I am sure these are proven...



Goodnight...

I don't know whether it is funny, stupidity or there really is a basis of using "two" servo valves on "one" actuator moving more faster in this set-up than using a single servo. I have seen quite a number of test fixtures in our company using two servos in one actuator.

Today I saw a system with a equal area double rod end supposed to be accelerating fast in approximately 1" stroke per half second. The cylinder is around 5" bore with a rod about 2" dia. There are two servo to which I guess using one to extend(A port and plugged B). The retract is B port and plugged A on the other servo. This is the only possible circuit that I could imagine for the cylinder to run. Meaning activate one servo to extend and activating the other one to exhaust and vise versa. I can see that the servo size is sufficient enough to run the system(probably 3/4 to one inch port, Moog, pilot operated and "mechanical feedback). The system is also equipped with an LVDT and a load cell. There are two accumulators also(around 1.5 gallon each). I don't want to comment more on the capability of our engineering people but I just can't imagine why as a hydraulic component manufacturer our people are very newbie on hydraulics. Since the main hydraulic PU is broken(they say it takes 3-5 months to fix/replace the pump... I am sure this is stupidity also), they try to run it using a pump between 20 to 25 gpm at 4100 psi pump. I was informed that to test our product there must be 40 GPM to maintain the 4100 psi pressure. I did not try to approximately calculate, but it seem like using a conventional calculation/approximation based on the size of the cylinder/rod a 20gpm can do it. 1 liter per .5 sec approximately.

I just gave the approximate size but my main point is the use of two servos... Since the spool had no electrical position feedback, there will be an error that may result to another rims delay or cavitation because the synchronocity is quite hard to do...IMO.

Please don't qoute this with my name because I am going to delete this in a couple of days after reading some response/comment.

I have no idea also about the accumulator precharge pressure but tha system pressure decays to around 2000 psi in a few seconds. I guess it's around 2000 psi and the load can move at 2000 psi too if the main pressure decays to 2000 psi. The test has to have data at a period of around 12 seconds for a desired cycle. I am not

I mean 3 stage servo...

quote:

Originally posted by maglub:
I don't know whether it is funny, stupidity or there really is a basis of using "two" servo valves on "one" actuator moving more faster in this set-up than using a single servo.

Yes, I had a customer that couldn't get a 100 to 120 gpm valve in the required time so he used two Moog 60 gpm valves in parallel to effectively make a 120 gpm valve. It worked well.

quote:

I have seen quite a number of test fixtures in our company using two servos in one actuator.

It doesn't make sense to use one on each side of the piston from a control stand point. It would be much harder to control than sending the same signal to two parallel valves.

quote:

Today I saw a system with a equal area double rod end supposed to be accelerating fast in approximately 1" stroke per half second.

That is not fast. The maximum velocity is only about 3 inches per second and the maximum acceleration is only about 18 inches per second squared. That should be easy.

quote:

The cylinder is around 5" bore with a rod about 2" dia. There are two servo to which I guess using one to extend(A port and plugged B). The retract is B port and plugged A on the other servo. This is the only possible circuit that I could imagine for the cylinder to run. Meaning activate one servo to extend and activating the other one to exhaust and vise versa. I can see that the servo size is sufficient enough to run the system(probably 3/4 to one inch port, Moog, pilot operated and "mechanical feedback). The system is also equipped with an LVDT and a load cell. There are two accumulators also(around 1.5 gallon each). I don't want to comment more on the capability of our engineering people but I just can't imagine why as a hydraulic component manufacturer our people are very newbie on hydraulics.

Before being too critical I would examine the circuit more closely.

quote:

Since the main hydraulic PU is broken(they say it takes 3-5 months to fix/replace the pump... I am sure this is stupidity also), they try to run it using a pump between 20 to 25 gpm at 4100 psi pump. I was informed that to test our product there must be 40 GPM to maintain the 4100 psi pressure. I did not try to approximately calculate, but it seem like using a conventional calculation/approximation based on the size of the cylinder/rod a 20gpm can do it. 1 liter per .5 sec approximately.

I calculated about 4.3 gpm

quote:

I just gave the approximate size but my main point is the use of two servos... Since the spool had no electrical position feedback,

Hydraulic servos don't have electronic spool feedback. The feedback is mechanical and internal to the valve. It would be good to know the valve part number.

quote:

I have no idea also about the accumulator precharge pressure but tha system pressure decays to around 2000 psi in a few seconds.

A lot depends on the pressure at which the pump start stroking.

Why such a big cylinder?
Did you see my video from last week posted in the position control thread? The sinewave progam moves exactly 1 inch peak to peak and at 2 Hz. That is four times out and back. I can even get that up to 5 Hz but not for long because the pump can keep up. My cylinder is only 2 inches in diameter but my pump is only 5 gpm at 1500 psi.

The system you describe should be able to move the 1 inch easily. Your system must be moving a lot of mass.

BTW, you have mentioned helicopter rotor testing. Our controllers are used to do that. We have about 3 or 4 customers that test rotors.

Thanks Peter, if your calculation is 4.3 gpm and my estimate is 15 to 20 they could have used just one valve. Your customers application required 120 gpm with two valves in parralel and supplying to one port to accomodate such big flow then that is logical and understandable.

I am also thinking why they used such a big cylinder! The unit being tested is a small damper(I guess for landing vibration). The size to which is approximately between 2.5 to 3 inch bore with a rod of about less than an inch. The unit has a(small) spring loaded accumulator with around 2 to 2.5 inch bore.

I will try to get the valve part number. Btw I can't open that PDF, maybe my pc is not compatible, i will try again later.

It's quite funny, but the unit being tested has an opposing load of about 159/190 to 300 lbs. The result is in lbs/inch/sec^2?! the unit is to be cycled in 60 sec at 8.5 Hz. It looks like the cycling cylinder has 6" bore too. There's no part# for the moog servo(erased) but it's big and the tube is around 25 mm which I could guess that the servo size is between 3/4 to 1". The test circuit has a closed center servo which could indicate a possitive overlap too.

I just smile when I saw the displayed circuit of the servo just drawn as a conventional closed center DCV and not indicating a proportional diagram...

I can't help on being critical with our technical people but they are just a waste of money trying to fix a lot of problems. This company is classified as small scale but we have around 300 employees here. I can even bet that among the so called technicians and engineers only probably 5% can read a hydraulic digram properly. In my deparment alone(repair station), not a single tech/engr can read a circuit or use the circuit, except me. I was even told that the circuit is not necessary...very funny whenever I heard it.

quote:

Originally posted by maglub:
It's quite funny, but the unit being tested has an opposing load of about 159/190 to 300 lbs. The result is in lbs/inch/sec^2?! the unit is to be cycled in 60 sec at 8.5 Hz.

Cycling at 8.5 hz is much more difficult than it seems at first and many under estimate the difficulty. The formula for calculating acceleration is
acceleration=Amplitude*(2*π*Hz)^2
If the amplitude=1 inch and Hz=8.5 then
acceleration=1.0*(2*π*8.5)^2=2852.3 in/sec^2 or 7.4 times gravity so the mass would take 7.4*300 =2220 lbf. The cylinder is MUCH bigger than what is required to apply the necessary force. The large diameter cylinders are necessary to get the natural frequency up. Large diameter cylinders make the system control stiffer.
So how are you controlling the system? A PLC will not be fast enough.

This message has been edited. Last edited by: Peter Nachtwey,

Ever heard about 1282A Schenck Pegasus? It's the servo used... I thought it was moog. The rating is around 50 gpm. I supposed it's rated at 1000 pd. Analyzing the piping looks like one servo on one cylinder port to one servo each.

Anyhow, this fixture/testing can only work with sufficient pump flow(above 40 gpm as they say) to maintain the input pressure of 4100, otherwise the controllability is very small percentage(spool openning).

Later...

valve spec from a web search.

OUTPUT FLOW 80 GPM MAX,CROSSOVERPRESSURE 300-550 PSI AT 1000 PSI SUPPLY,4 PORTS

Cheers Woody

Thanks Woody and Peter. Based on that specs the pressure drop is around 1000 to 1100 psi. If this valve have a different spool sizes (example 50 or 80 gpm) and the specs on the nameplate is 50 gpm, I might try 50 then 80 if I would like to calculate the critical cylinder mass natural frequency in the absence of valve frequency response curve. The only unknown that I need is the bulk modulus of mil-6083 fluid. Honestly, with a flapper type servo, i am having difficulty understanding the frequency at f(-90) and with respect to the valve pilot pressure(?!).

Peter, it looks like PLC(small circuit board) was used, and no pc at all. The electronic controls are rag-tag. I cant even assume that there is a valve manufacturer amplifier card. I cant really get closed enough to their testing/troubleshooting because my leadman is like a freakin' he-who .

Btw since the unit under test is a damper(with a spring loaded accumulator), there should be a negative load/force on the opposite stroke(actuating cylinder retracting, damper extending).

Actually the testing was discontinued because the other engineering he-whos suspected that the servo has a problem(they said the servo has too much leakage ). I can only guess that there is a big percentage of negative overlap, whereby the limited pump discharge is just being diverted to tank at around 2000 psi(two servos), and cant increase the input pressure anymore. I just hope that I could examine the oscilloscope reading and find out some more data...it might take days again...

The card they are using for the servo is moog. The input to the torque motor are connected in parralel but in reverse polarity. I guess so that when one position of servo is energized(to extend P-A), the other servo is energized(to exhaust B-T).

There is an unacceptable flow to return when both valves are neutral. Around 12 gpm. The signal is also unstable, and lagging too much.

This message has been edited. Last edited by: maglub,

Finally, the problem was temporarily solved and the units were tested(last week). They changed the size of the servo(smaller). I realized that I made a bad post regarding the null setting. The unacceptable amount of leakage IMO was because of two big servos having too much negative overlap.

I had the feeling that because of very small adjustability range, it is very difficult to control the system. IMO, the system could have performed much better if using just one servo which the amount of leakage due to negative overlap can be reduced by another half. With a limited pump flow/system pressure, the pilot pressure will also be affected and the servo NF too(?!).

Also, another accumulator was added...