Faking a proportional valve

[Tinine]

Anyone ever PWMd the coils of a bang-bang valve? I tried it recently so that I could reduce machine setup-time by eliminating the manual flow controls. Seemed to show potential. Now I'm wondering about modifying the spool to reduce the deadband. I would've thought that others had experimented with this sort of thing. Huge cost difference between a bang-bang and a proportional valve.




Rgds,




Craig

 

[Steve Bailey]

The cost difference just might be offset by the performance difference. You would expect a proportional valve to be linearized so that a 12 mA signal would result in the valve being 50% open. I doubt if that would be the case with 50% on time in a PWM signal, so plan on a lot of tinkering to get it to operate the way you want over the full range. I would also expect significantly reduced solenoid life when you tried to operate it from a PWM signal.

 

[Peter Nachtwey]

A big problem is that you don't have spool position feedback like the servo quality proportional valves do.
In simulations it didn't work. I couldn't get the valve to switch fast enough. If it won't work under ideal situations, then it won't work with all the uncertainties in a real system. The PWM frequency needs to be high. What you are trying to do is shift mechanical complexity to electronic.

 

[PreLC]

Apart from the response of the solenoid itself(Over time your 50% duty cycle PWM might go from 100 units/s to 10 units/s, might need an inline flowmeter to keep it closed loop), it's also worth noting the fluid response: If you are opening-closing a flow every few milliseconds, you are also inducing a lot of vibrations most upstream systems are likely not designed for.

Proportional control valves usually maintain flow by changing the orifice size(Gate valve, needle valve etc.)

 

[TheWaterboy]

This is done in Pro Mod transmissions to provide a smooth(er) creep to the starting line. REALLY hard on the spool. And wow what a difference a couple milliseconds makes.

 

[Tinine]

Very cool that we have hydraulic discussion.


Who cares about spool feedback? The above was to eliminate a manual flow control to save time. The manual flow control doesn't compensate for viscosity(temp) changes, right.


If you need closed loop, screw the valve. You need feedback from the actuator that you are driving. Close the loop and all will be cool.



Here's my real world:


Do you have any idea of how many machines are being produced, utilising a Vickers controller-card + Proportional Valve?


I have one client who ships 300 machines/year.


Now, what happens: They need the greatest possible speed but they also need repeatability. So what happens?


Well their PLC has a quadrature ($$$) encoder input.
They drive a digital output to tell the proportional valve to go in the positive direction at high flow-rate. When a certain trip-point is achieved, they switch to a lower velocity which involves the Vickers built-in ramp generator. They arrive at target position and they switch the valve off before rapidly returning. Minimum $2,500 for this valve + card.


Now, RPi Pico...$7
LS7366 Quadrature decoder/counter (SPI)$5
LM358 to convert PWM to analog +/-10v $1.00

Suitable PSU $20.00

24vDC bang-bang valve $150


The components listed above could easily generate a closed-loop trapezoidal profile and drive the arm MUCH more smoothly than the Vickers dumb solution.


But hey, let's just do what we've always done and then we can go watch football or some soap operas.


It's effed-up




Craig

 

[Peter Nachtwey]

If you are opening-closing a flow every few milliseconds, you are also inducing a lot of vibrations most upstream systems are likely not designed for.

YES! In my simulation I tried a PWM with 100KHz frequency and the pressure shocks or swings were too big. My conclusion is that there would need to be a small accumulator to absorb the shocks but that ruins the response.

 

[Tinine]

YES! In my simulation I tried a PWM with 100KHz frequency and the pressure shocks or swings were too big. My conclusion is that there would need to be a small accumulator to absorb the shocks but that ruins the response.

How so? That is only the carrier frequency. What were you doing with the important bit...Aka: Duty Cycle?




Craig

 

[Tinine]

Try 1KHz or lower. It's impressive what can be achieved.


Yeah, I had the PhD guy from Vickers visit me to inform me that my highly repeatable and reliable machines were "just wrong" because I wouldn't spend double the amount on their zero-overlap servo valve. Propaganda Horse Doo-Doo that all too many fall for.



"Yeah I'm a controls engineer, just got back from BeckHoff training"
"Oh cool so what did you learn?"
"Oh well like EtherCAT is awesome and stuff"
"Cool but like how?"
"Oh it just is and stuff"


That's the way it is.

 

[Peter Nachtwey]

Very cool that we have hydraulic discussion.
.....

Your example where a company is doing the same thing over and over is a good example. Good enough is good. Sometimes ignorance is bliss.

There is a huge difference between servo quality proportional valves and other so called proportional valves that have no spool feed back.

You mentioned Vickers, what controller and valves are you using? I am somewhat familiar with ancient Vickers controllers.

Trapezoidal ramps were old in last half of the 1990s. The problem with trapezoidal ramps is that they generate too many high frequencies that hydraulic systems can't respond to. Back in 1993 I started the first of our second generation controllers that used sinusoidal ramps that avoid or reduce the high frequency components of the target generator. For instance, ramping from 0 to 100 mm/s in 0.1 of a second is only 1 m/s^2. That is easy but if frequency of acceleration is 5 HZ. This is because it takes 0.1 ms to ramp up and 0.1 ms toramp down so the total ramp time is 0.2 seconds or 5 Hz. This means the hydraulic actuator should have a natural frequency 4 times 5 Hz or about 20 Hz.
A trapezoidal ramp will have higher frequency component that will require the natural frequency of the actuator and load to be even higher and that costs money.

Now we use 5th order ramps. 5th order ramps are generally better than 3rd order ramps. In theory 7th order ramps should be better than 5th order ramps but the problem is that 7th order ramps require a much higher peak acceleration and deceleration that requies either extra torque or bigger cylinders so 7th order ramps are not good for quick short moves that are required for fast produciton in industry. 7th orde ramps are good for trains so people's drinks don't fall over.

 

[Tinine]

If you are opening-closing a flow every few milliseconds,

This is not what is happening and furthermore, the spool simply can't respond.


The solenoid is opposed by a spring. Through PWM you modulate the current in the coil and therefore the flux.
The valve becomes a flow control.


Did you ever push one of these spools (de-energized coils) with something like a small Allen wrench? The flow is proportional to how far you push the spool.


Rgds,



Craig

 

[drbitboy]

The solenoid is opposed by a spring. Through PWM you modulate the current in the coil and therefore the flux.
The valve becomes a flow control.

Isn't this kind of like the DC analogue of a VFD for an AC motor?

 

[Tinine]

Your example where a company is doing the same thing over and over is a good example. Good enough is good. Sometimes ignorance is bliss.

There is a huge difference between servo quality proportional valves and other so called proportional valves that have no spool feed back.

You mentioned Vickers, what controller and valves are you using? I am somewhat familiar with ancient Vickers controllers.

Trapezoidal ramps were old in last half of the 1990s. The problem with trapezoidal ramps is that they generate too many high frequencies that hydraulic systems can't respond to. Back in 1993 I started the first of our second generation controllers that used sinusoidal ramps that avoid or reduce the high frequency components of the target generator. For instance, ramping from 0 to 100 mm/s in 0.1 of a second is only 1 m/s^2. That is easy but if frequency of acceleration is 5 HZ. This is because it takes 0.1 ms to ramp up and 0.1 ms toramp down so the total ramp time is 0.2 seconds or 5 Hz. This means the hydraulic actuator should have a natural frequency 4 times 5 Hz or about 20 Hz.
A trapezoidal ramp will have higher frequency component that will require the natural frequency of the actuator and load to be even higher and that costs money.

Now we use 5th order ramps. 5th order ramps are generally better than 3rd order ramps. In theory 7th order ramps should be better than 5th order ramps but the problem is that 7th order ramps require a much higher peak acceleration and deceleration that requies either extra torque or bigger cylinders so 7th order ramps are not good for quick short moves that are required for fast produciton in industry. 7th orde ramps are good for trains so people's drinks don't fall over.

Just to emphasise the ridiculousness of this application, its all about approaching the final encoder count at a rate where it stands a chance of switching off the command to the valve to ensure repeatabilty. They don't care about smooth/abrubt.



There are soooo many examples out there of some sucker being sold on this overpriced valve (I can get a 2KW AC SERVO MOTOR complete with drive and 17bit absolute feedback for 30% of the cost of these ********e valves). There is nothing to them. Absolute ripoff.


Rgds,


Craig

 

[Tinine]

Isn't this kind of like the DC analogue of a VFD for an AC motor?

It's the same as an LED dimmer. All that's happening is that it's being switched on and off at a high (carrier) frequency.


If the LED was being switched on and of at 1KHz and spent 50% of the time in the off state and 50% in the on state (50% duty cycle), to the human eye, it's simply half brightness.


This is why our servo motors emit high frequencies. The motor currents are PWM regulated and the carrier frequency should be beyond human hearing.


Craig

 

[drbitboy]

PWM duty cycle, % ≠ spool position, %

And lets not even get into valve characteristic, though of course that can be modeled.

If the speeds are linear, the frequency probably contributes more than the duty cycle. The only way this works is either with spool position feedback or non-linear actuator-vs.-spring characteristic (model/feed-forward).