P/Q valves?
Jiri Toman said:
Of course you do that is what Arburg uses. Talk to Bosch.
I have, I visited Bosch, in Raccine, many years ago before it assimilated Rexroth. At this time Bosch had its own valve similar to the one Rexroth valve in the link. I have always preferred the Rexroth version though. This the valve that was recommended for injection molding injection applications. Now that Bosch and Rexroth are one I don't know of any other manufacturer that makes a valve like this.
ftp://ftp.deltacompsys.com/public/PDF/re29067_2003-02.pdf
Go to page 8. This is NOT a PQ valve. This valve does not have an analog input for pressure, yet is does an excellent job of injection control with the proper controller. It uses the Q2 ( injection molding ) spool. Notice that the B port is almost always ported to the tank. This is VERY import because it insures that the pressure on the rod side of the piston is 0. This allows one to use only one pressure sensor on the blank side of the piston to monitor the applied pressure/force. One can also see that the flow is only ported to the B port when the control signal is trying to make the piston retract with a signal of –5.8 volts ( on the graph ). The range from about –5.8 volts to –10 volts should only be used when there is a need to retract the screw without using pressure from the plastic. In normal operation the rod side of the piston is at tank pressure, ideally 0 PSIG.
To ignore the pressure on the rod side of the pistons is an error because the pressure on the rod side of the piston can subtract substantially from the net applied force. This often happens when people use P/Q or normal servo valves to monitor the pressure of the hydraulic oil. If one really wants to use a servo valve with a servo spool, a pressure sensor on the rod side must be attached so the differential force can be calculated based on the different surface areas of the piston.
A P/Q valve would be OK if you want to control nozzle or cavity pressure of the plastic only as in the application below. If one wants to control or limit the hydraulic pressure, nozzle pressure and mold pressure, the P/Q valve would need to have a way of switching between the these pressure sensors and have different PID controls for each pressure sensor. I don’t know of any P/Q valves that can do that.
With a valve with the Q2 spool, one can monitor either the oil injection pressure, nozzle pressure or cavity pressure independently with separate analog inputs on the controller. The controller should a separate PID for each analog input one doesn’t need to worry about switching gains on the fly.
The link below is a graph of a dry shot and a HPM 500 tom pressure taken in 1998. It took me only a few hours to get this setup.
ftp://ftp.deltacompsys.com/public/jpg/H500INJ.JPG
Expand the .jpg to full size to see the detail.
(Cyan) target position
(Red) actual position
(Magenta) target velocity
(Blue) actual velocity
(While) target pressure
(Gray) actual pressure
(Green) Control to valve. 0 volts is approximately in the middle between the top and bottom of the graph.
Notice the first 5 seconds that there is a sizeable error between the target position ( cyan ) and the actual position ( red ). As I pointed out before, the valves with the Q2 spools, are not meant to be used for PID position control. The controller was trying to move the actual to the target position, but it couldn’t. In this case the valve’s input polarity was reverse such that negative voltages made the screw extend or the pressure increase. You can see from page 8 from the valve above that the valve will not port oil from the pressure to the rod side to make the injector retract. Normally this is not a problem because the plastic pushes the screw back while plasticizing. However, this was a dry shot. At 5 seconds the injection cycle begins and you can see the actual position/velocity follow the target position/velocity. This is using only a 5 steps out of 10 step motion profile and only 6 steps out of the pressure profile. Notice how well the pressure can be controlled even when ramping up or down. Normally, you don’t ramp the pressure back up as in step 5 but the OEM wanted to see how the controller would do.
One can see that even though the flow characteristics of a valve with a Q2 spool look odd, it can be used to control pressure very accurated.
Jiri Toman said:
You did not mention velocity , you mentioned position!
Yes I mentioned position, that is what most injection machines use for feedback. How else would they know when to change velocity when injecting and how far to let the screw be pushed back when reloading for another shot? Velocity is the derivative the position. I haven’t found the secret to time control. I haven’t found the secret to controlling position with out controlling the velocity or controlling the velocity without controlling the position or at least the change in position.
Good info. When I walk around the NPE shows, I see many machines with analog MDT rods. These have got to be the worst things you can use. First there is noise. Second, what good is a position if you don’t know when it is valid? These analog MDT sample asynchronously to the controllers so there is a lot of jitter in the time between samples which leads to a lot of fudging ( filtering ) the velocities on many controllers. At least with an analog slide pot the sampling is under the control of the controllers A to D circuitry which usually gathers data synchronously to its own scan so the controller knows when the data is valid. The best feedback to use is the SSI MDT rods that sample position synchronously to the motion controller.
Jiri Toman said:
Profiling in IMM terminology is pretty much the same across all the machines, HPM's, Arburgs, Sodicks, Engels, New Britains, Huskys, Demags etc.
Profiling is pretty much shown by Norm in his pictures.
You set up 10 step profiles Pressure vs Time or Pressure vs positon etc.
You mean pressure LIMIT vs position. One can’t control pressure and position. I don’t want anyone to get the idea that they can. I think it is wrong, or maybe just ignorance, that people are mislead to thinking they can control position and pressure at the same time.
Jiri Toman said:
If you are profiling velocity you will have pressure limits
and you will eventually switch over to pack/hold based
on one of 4 different triggers (if you have them enabled).
In the case of the above graph the screw hit the end of the cylinder because this was a dry shot and entered a pack and hold phase bacause the pressure started to increase. This transition from position/velocity mode to pressure mode is very tricky and in this case made much more difficult because the metal on metal contact makes the pressure rate much higher than when pushing on plastic. It is difficult ( impossible ) to bring that pressure rate to 0 instantly. Real plastic would have a slow rate of change in the pressure because it compresses much more than steel. See the “Under pressure” thread on Ron’s hydraulic power forum.
Notice that both position/velocity and pressure have s-curve profiles. This is not common on most injection machines. Most machines just change velocities or pressures without regard to the rate of change. One should be able to specify the acceleration rates or rate of change in pressures with s–curves. This is important because one can’t change the pressure in oil or plastic instantly.
Jiri Toman said:
The cavity pressure is something you are obviously not familiar with in addition to many other aspects of an IMM.
I don’t claim to know what the cavity pressure should look like as a function of time as shown in the graphs. What I do know is that there is a lag between the pressures seen on the hydraulics and the nozzle pressure and the nozzle pressure to the mold. This will lag will vary depending on the machines and molds. I haven’t seen all the machines and molds yet. I do know is that if the customer knows how he wants the pressure in the mold controlled, we can do it. I don’t see where one can say they know all about injection molding with experience on just one or even a handful of machines. I don’t see how one can get a P/Q valve to monitor three different pressure sensors with one analog pressure input. Even if an there is and analog switching device, I still can’t see how one changes the PID ( is there a PID? ) gains in the P/Q valve as it switches inputs.
