I have been chosen to replace Jack Johnson as technical contributor for H&P magazine 1

Good luck with your endeavor Peter and try to keep a sense of humor.

Keith Cress
kcress -

What is a sense of humor?
If I seem like a sour puss much of the time it is because we have seen too many screwed up designs. Somehow people to learn a least a little about mechanics, hydraulics, electric motors and control theory.

Seriously, any ideas for topics?


Peter Nachtwey
Delta Computer Systems

Seems like you already have one flow makes it go -- NOT ;-)

Have you considered extending the depth of some of Mr. Johnson's original articles, particularly on modeling?

TTFN (ta ta for now)
I can do absolutely anything. I'm an expert! faq731-376 forum1529 Entire Forum list

Yes, hearing "flow makes it go" winds up my integrator.
I have been trying to make "flow makes it go" go away for 20 years on various forums.
I take delight in rubbing in the fact that Newton did not include flow in his 3 laws of motion.

Have you considered extending the depth of some of Mr. Johnson's original articles, particularly on modeling?

Click to expand...

Very good question.
Jack Johnson is leaving his post because he wants to spend more time he want to spend more time on modeling pumps, motors and valves. Jack has written many articles where he lays out a model of a motor or pump that looks like an electrical circuit. This makes sense, Jack's background is electrical. However, what is the reader going to do with that article? There is a recent JJ articles on modeling motors. Was it useful? This is the problem.

In the past Jack has wanted me to get involved with a committee to standardize models. I think it is a good idea but the reality is that if there was a good model of a motor, what would you do with it? I think, what would I do with it? Do you need to know about the case drain and how much oil leaks across port A to B etc? People have been "getting by" for decades now. I saw an article that defined 5 levels of details for pumps. I would prefer to keep the models relatively simple. I don't want perfection to get in the way of the good.

Here is what could be done. If I had flow meters for the inlet, outlet and case drain, pressure sensors on the A and B ports of the motor and encoder to measure the speed and finally a dynamometer, I can model the motor very thoroughly. It requires solving many differential equations simultaneously and finding the coefficients to those differential equations. I know how to do that. I just don't have the equipment but if I did, then what? It is clear that few understand Laplace transforms and fewer yet would understand systems of non-linear differential equations. Who would supply the data for modeling the pumps?

What to do?

Another thing I would do is define some python standard classes ( actual I have ) for valves, cylinders, pumps. I chose python because it is free and very capable like Matlab. Would the big companies follow my lead or do their own in some other language and sell their simulator? BTW, don't be surprised if some python code shows up in the articles. How many people know python?

I know that Jack Johnson would really like to make progress on this topic which is why he is letting me take his place. I wish him luck but I don't see how one can herd cats ( big hydraulic companies ).

Since I know IRstuff knows Mathcad I will provide this link.
I use this file as a starting point for doing simulations for customers.
How many people would understand what is going on there.







Peter Nachtwey
Delta Computer Systems

Congrats, Peter! Looking forward to seeing your columns, curmudgeonly or not...

Curt, why is it that you haven't written many articles?
I am sure that while DTD was still independent, there was still the freedom to do so.
There must be hundreds of applications that you could write about.
I was always aware of what DTD did and how they did it. I read the manuals.
I think DTD and Delta had very little overlap in the market place so there really wasn't much competition.
BTW, I did get a change to meet Dmitri Dmitri, DTD founder, and Jacob Tal, Galil founder, back in the late '80s.

I have turned in my first paper as technical writer.
I will always be looking for topics that people want to talk about.

My style will be much different from Jack Johnson's because I come from a manufacturer's point of view instead of a college professor's point of view.
Jack did a lot of fine work that most people don't know about because the data is in his books. I will bring up some of these topics too.
However, mu priority is not to sell books, it is to sell motion controllers.
Training/knowledge is everything.




Peter Nachtwey
Delta Computer Systems

Peter:

Most of the publications I wrote for and had established relations with the editors are now gone. I had started focusing on writing for groups like ASPE (American Society for Precision Engineering).

Lately I have been very busy managing the transition. I do hope to get back to writing more articles as things settle down.

My first article is out. It more of a hydraulic through my eyes.
Allen Hitchcock edited it a little.

I need more suggestions for articles.
I might write one about doing hydraulic motion control in a PLC. Yes, we get asked use a motion controller when one can use a PLC or Arduino.

I think Curt will back me up on this. It isn't the PID that is difficult. It is the target generator.






Peter Nachtwey
Delta Computer Systems

I will suggest that as you write articles you define the TLAs (three letter acronyms) when you first use them in an article. Yes, every article. It will help educate first time and remind old time readers and may even reduce apprehension of new designers you are trying to attract.

This first article gives you a good introduction into your new role. The education will move us ahead and perhaps dispel some more myths about the world of hydraulic power.

Ted

I think I rarely use HPU. Hopefully people know what that means. I use VCCM ( valve control of cylinder motion ). I would rather have called it VCAM. ( valve control of actuator motion ).
I usually define symbols I use.
Where:
CLTF(s) is the closed loop transfer function expressed as a Laplace transform
K is the open loop gain with units of (mm/s)/(%control)
s is the Laplace operator. A frequency.
ζ is the damping factor. Unitless
ω_n is the natural frequency in radians per second.

I use that equation a lot in my work but not in articles. This is because few understand Laplace transforms.
I feel this is my biggest problem
Also, I wish this forum supported LaTex or MathML.
We use it on our website.

H&P has used some of my formulas in articles that aren't my own. I don't mind but there any reference to where they came from or how they are generated. The stuff I don't isn't found in text books. The question I have is will people just accept it as truth? I wouldn't. I have seen so much bogus information printed.
There is another topic on my forum about acceleration times. The formula he quoted was
This is definitely wrong and extremely conservative. I have seen this in print in a few different places and people repeat it.

I would like to point out some of the formulas people use are wrong but the math may get to deep.







Peter Nachtwey
Delta Computer Systems

I could see 3.5 maybe up to 5, depending on the level of accuracy, 35 seems awfully long.

TTFN (ta ta for now)
I can do absolutely anything. I'm an expert! faq731-376 forum1529 Entire Forum list

In your articles it may be wise to put some notes about safety. 10,000 psi in hydraulic system is not unusual and that can be very dangerous when components are improperly installed. I remember a case where an ironworker was severely injured when one quick disconnect on the retraction port of a single acting jack was not connected to a hose and the little ball inside the quick disconnect shot out and cause the ironworker to almost loose his life.

Peter:

I like the article! I fully agree that the target (aka trajectory or setpoint) generator is often much more important than the PID loop.

One of my pet peeves about academic controls is the almost total absence of trajectory considerations. And with that, a great underemphasis on feedforward techniques (which can only be used with good trajectory control). I realize that some control fields do not (or cannot) use good trajectory generation, but many can.

Some years ago, I went to a presentation by grad students in controls at MIT. One student had done some nifty work in sliding-mode control feedback. I asked him what was next for him, and he said he was presently stuck trying to generate a trapezoidal profile.

I do think that it would make a good article to describe when a dedicated motion controller should be used over a PLC or low-cost embedded computer. I've seen quite a few "roll your own" projects come to grief because they completely underestimated the complexity of implementing a ROBUST hardware/software solution.

And at the risk of starting a "religious war", I would suggest an article on when to choose electrohydraulic versus electromechanical control. (In the motor control world, stepper vs servo has long been one of those "religious wars".) The boundary has definitely moved since we got into the controls field, but there still is a boundary.

Curt Wilson
Omron Delta Tau

I could see 3.5 maybe up to 5, depending on the level of accuracy, 35 seems awfully long.

Click to expand...

The problem is that there is no justification given for 35/ω_n yet people keep repeating it. Even professors and valve companies repeat this. I knew this was wrong back in the late 1980's or early 1990's when we made our first hydraulic system. It had a natural frequency of about 28-30 Hz. The acceleration time should have been 0.2 seconds but we were easily doing it half that time. The 35/ω_n would cause hydraulics designers to be too conservative and design systems with cylinders that were twice the diameter they needed to be. This means 4 times the flow and for times he HPU size.

There is a recent article I wrote in H&P about the natural frequency and frequency of acceleration. There is a relationship between the two but much depends on the control algorithm. For a standard controller with a PID with velocity and acceleration feed forward the ratio of the natural frequency to the frequency of acceleration should be about 4 to one but that can be reduced significantly if better algorithms are used. If using a proportional only control the ratio of natural frequency to frequency of acceleration must be much higher than 4 to 1.

In your articles it may be wise to put some notes about safety. 10,000 psi in hydraulic system is not unusual and that can be very dangerous when components are improperly installed. I remember a case where an ironworker was severely injured when one quick disconnect on the retraction port of a single acting jack was not connected to a hose and the little ball inside the quick disconnect shot out and cause the ironworker to almost loose his life.

Click to expand...

I have installed hydraulic controls for steel mills in the US, Turkey and India. All of these systems used system pressures in the range of 2000-3000psi. The only time I have come across such high pressures are for hydro forming but that used hydraulics in the 3000 psi range and used intensifiers to boost the water pressure higher. I also did some work on a diesel fuel injection system. The hydraulic cylinders were in the normal range of about 2000-3000 psi but the fuel was pressurized to 43,000 psi using diaphragm pumps. My point is that I haven't seen any systems where the hydraulic system pressure is 10K PSI. Maybe 5000 psi at the most.

I do think that it would make a good article to describe when a dedicated motion controller should be used over a PLC or low-cost embedded computer. I've seen quite a few "roll your own" projects come to grief because they completely underestimated the complexity of implementing a ROBUST hardware/software solution.

Click to expand...

This is in the works already. The is a common question we get. Also, we are trying to sell into China. Most people try to roll their own controls with a S7-200 Siemens PLC.
At first I was going to make it a PLC vs motion controller article. Then I decided against that because obviously I am biased. However, we have many different PLCs here for compatibility testing. One of our engineers is tasked with making a S7-1500 PLC control our basic hydraulic system. The S7-1500 is much more powerful than the old s7-200 but the S7-1500 is what we have got.

It is funny/sad that college students can't work out a simple trapezoidal move. That is trivial. A third order target generator is much more interesting. The real challenge comes when the short move is requested where the velocity never reaches the command velocity and the acceleration never reaches the command acceleration. So what is the acceleration and velocity that the target generator should ramp too? Few people solve that problem. A third order target generator is beyond the capability of all but a few. This means the are limited to trapezoidal ramps where the acceleration changes abruptly and there is no jerk gain.
There is NO WAY a third order target generator can be done in a PLC.










Peter Nachtwey
Delta Computer Systems

My second article has been published..... finally

The next article will continue where this article leaves off.
In the end there is no way a PLC can do hydraulic servo control as well as a hydraulic servo controller.
The software to generate the target position, velocity and acceleration is too complex to run on a PLC.
There is a thread about this on





Peter Nachtwey
Delta Computer Systems

Below are the links to more Hydraulic and Pneumatic articles.

Anybody designing servo hydraulic systems should read this one

The next article will be about the VCCM equation.

The article after that will be about using the VCCM equation in its different forms to do cylinder and valve sizing.
This article draws on Jack Johnson's work but adds a few twists to take into account reality.

I need more ideas.





Peter Nachtwey
Delta Computer Systems