Analysis of Transducer Accuracy on the Repeatability of the Injection Molding Process
Ernest Griesser and Kenneth Pinkham
Abstract
Injection machine users are demanding improved performance with respect to shot-to-shot uniformity, particularly in areas of medium to high volume molding runs of critical parts, where the need for "zero defects" is demanded by customers and management. This is why molders are bypassing machine controls to use mold and nozzle pressure transducer control over hydraulic control for better results. This paper analyses the injection molding process with respect to precision molding requirements versus transducer type, location and performance on control system repeatability. The results show that a new type of nozzle transducer having a Silicon on Sapphire diaphragm capable of measuring both pressure and temperature in contact with the plastic melt offers a magnitude of ten times better shot weight control. This gives freedom from routine daily adjustments of the pressure fill/flash ratio over all other pressure transducers and load cells currently in use.
Introduction
In a previous paper [1], the development of Silicon-on-Sapphire pressure transducer technology was traced from the aerospace and defense industries to its application in the plastics industry. Figure 1 shows a photo of a silicon on sapphire diaphragm transducer. This technology has recently been applied in the injection molding industry with great success. The unique characteristics of the silicon on sapphire diaphragm transducers are their ability to sense both temperature and pressure in a single machine penetration, more accurately than any other transducer. In direct contact with plastic melts, they measure pressures linearly from zero to as high as 40,000 psi and temperatures up to 800oF. Their advantages include:
- Fastest response times; < 100-500 Micro-sec.
- Best accuracy.
- Linear response without low pressure dead bands.
- No hysteresis.
- Ability to zero temperature effect on pressure readings.
- On-diaphragm SRTD for measuring temperature.
- Fatigue-proof diaphragm.
- Abrasion proof diaphragm.
- Corrosion resistant diaphragm
- Best cost/performance ratio
- Lowest lifetime cost.
Injection Molding Process
Figure 2 is a diagram of a typical hydraulic injection molding machine.
A. The drive at A rotates the screw which deposits the shot behind the nozzle and into the front portion of the barrel, in front of D. As the screw rotates, it moves backwards and becomes shorter, typically generating less and less adiabatic or frictional heat in the molten plastic the more it moves back.
B. When the mold is ready, the extrusion section advances to kiss the nozzle to the mold sprue bushing and inject the plastic. In this step, a check valve at D prevents backflow of the melt and injection pressure is intensified by a hydraulic piston. This multiplies the hydraulic pressure from ten to twenty times, to to provide sufficient pressure to inject the molten plastic into the mold.
C. The hydraulic pressure transducer at B is typically rated at 0-3,000 psi with an error of +20 psi and a response time of <5 milli-sec. The piston behind the screw at C multiplies this error at B by 15 times, (actually 10-20) providing a nominal error at the nozzle of +300 psi. This variation of melt pressure at the nozzle is what causes less than precise shot weight control.
D. The friction of the piston at C plus the screw in the barrel can consume from 10 to 30% of the overall pressure theoretically applied to the melt. This reduces the pressure at the nozzle by that amount and contributes to additional pressure fluctuations at the nozzle, typicallly from +1% to 3% or more of maximum theoretical nozzle pressure, (Ppeak/H x Piston area multiplier.).
E. When a single part mold, or a several piece family mold is run with a conventional sprue gate, there won't be much difference in pressure between the nozzle and sprue locations. Thus, a silicon on sapphire nozzle transducer used to control or transfer Ppeak/N and Ppack/N at the nozzle, E, will provide an accuracy ten times greater than the hydraulic pressure transducer used to control Ppeak/H and Ppack/H. This is by virtue of +30 to 40 psi melt pressure control at the nozzle versus +300 to 400 psi or more variation of melt pressure at the nozzle using hydraulic control only.
This is confirmed in Example 1.
Pressure Transducers
Figure 3 shows a typical pressure-time profile for a single cavity mold being filled. As the mold is filled, the pressure inside initially rises, levels and then peaks the instant it is filled. Then, shrinkage takes over as the plastic is cooled, decreasing the volume slightly, as the part weight remains the same. To avoid shrinkage sink marks, the mold is packed at an elevated pressure below Ppeak at Ppack. Packing pushes some more molten plastic into the mold until the gate is frozen. Thus part weight is directly and totally controlled by pressure, which is best controlled in the mold or as close as possible, such as in a hot runner manifold or the nozzle. Hydraulic pressure transducers are too remote and indirect in function to control mold fill with any precision.
Variations of peak melt pressure and pressure duration at the nozzle will affect part weight. Therefore, any improvement in pressure uniformity at the nozzle will improve shot weight uniformity and quality. The best type of control for this situation would be sapphire mold, manifold or nozzle pressure control transducer which is capable of catching the Ppeak and Ppack in the same place every shot.
It must be recognized that a mold pressure transducer can only control peak mold pressure with transfer to pack pressure, while a Silicon on Sapphire nozzle melt pressure transducer can control the complete pressure process, including low back pressure control. In most cases, a Silicon on Sapphire nozzle transducer used in the machine control system will eliminate the need for external cavity transducers. In fact, the process control improvement achieved with silicon on sapphire nozzle control was found to be superior to all the cavity types except silicon on sapphire itself.
Injection Molding Transducer Comparisons
Table I compares the technical characteristics of the general classes of pressure transducers examined. Each will be discussed individually.
Any pressure transducer used to control peak injection pressure having a response time >5 milli-sec is not adequate for the job. This eliminates mercury units and IR temperature units when they are used to follow and control pressure with crystalline polymers which exhibit momentary temperature rise under compression.
Pressure Transducer Error Calculations
Pressure Transducer manufacturers specify their factory specifications as produced at room temperature with laboratory testing apparatus and temperature gradients not relating to actual working conditions on machines. It is the "real world" conditions therefore, which govern real world variations as opposed to factory spec variations. This study investigates the effect of real world, conservative operating machine and plant parameters on transducer accuracy.
In order to better understand the source of variability of various types of pressure transducers used in injection molding, the authors made calculations of mold and nozzle pressure variations based on manufacturers data and conservative machine operating conditions. Calculations were made for absolute error, the maximum deviation from true pressure, and for repeatability, the common error that can be expected from shot-to-shot. Several examples are given that confirm the calculated results.
The first step was to consult manufacturers specification data and to make some basic operating parameter assumptions concerning temperature effects and operating conditions. In each case, the authors were able to calculate good estimates of absolute and repeatability errors that are highly enlightening in explaining variations observed in field examples and production runs.
The probability that all errors are strictly additive is nil, so the square root of the sum of the squares method was used. For the comparisons to be as uniform as possible, all the calcultions assumed that theoretical peak pressure is 30,000 psi while Ppeak/N is 24,000 psi and Peak/M is 20,000psi. This assumes a nominal 20% pressure drop from the intensified hydraulic hydraulic side to the nozzle melt pressure. The pressure drop from nozzle to mold is assumed to be 4,000 psi.
Hydraulic Pressure Transducers
Hydraulic pressure transducers are mainly bonded-foil strain gauge types, typically made in 0-3,000 psi ranges. Compared to silicon on sapphire, their combined error and repeatability specs are typically the same. However, due to their location and the error multiplying effect of the pressure intensifying piston, they need to be at least ten times more accurate than they are at present. This means, to equal the precision provided by a silicon on sapphire transducer located in the nozzle or mold, a hydraulic transducer would need to have a combined error of 0.025% to 0.05% plus a repeatability of 0.01% or better, along with a lower temperature effect on pressure. The actual probability of developing such a unit is extremely remote, while if it could be accomplished, the cost would be very high. The use of a silicon on sapphire hydraulic pressure transducer would provide no advantage over existing hydraulic types, except durability. However silicon on sapphire nozzle or direct cavity types provide the greatest advantage for precision molding control.
Calculations for Case 1 assume the maximum variation from starting a cold machine until the hydraulic oil reaches a temperature of 120 deg F. Case 2 assumes a small oil temperature variation during a run of 10 deg F. The calculated values of hydraulic pressure variations are then multiplied by an average pressure intensifying piston ratio of 15. The resultant melt pressure variations were thus calculated as +819 psi, or +3.4% of the absolute error at the nozzle and +292 psi or 1.46% for the repeatability error at the nozzle. This confirms the reason for the repeatability variation of +338 PSI found in Example 1, and explains why hydraulic control is a poor method of injection control for injection molding. Calculations are shown in Table II.
Mercury-Filled Mold or Nozzle Pressure Transducers
Mercury capillary, bonded-foil strain gauge pressure transducers have been available for many years and are used as both direct cavity and nozzle transducers. Compared to silicon on sapphire transducers, they are 100 to 500 times slower in response time and at least ten times slower than the maximum 5 milli-sec required to catch the Ppeak in the same place every shot. In a recent test on a 300 ton all electric machine, this transducer could not improve the shot weight distribution over the normally supplied standard linear positioning control system. Additionally, they possess less resolution due to hysteresis, a larger error band (1% vs 0.5%), poorer repeatability (.2 vs<.1%), have a lower fatigue life (1 million vs >10 million cycles) and poor abrasion resistance of the stainless diaphragm compared to a silicon on sapphire diaphragm transducer.
The calculations for mercury transducers were without error calculation for response time, which is very difficult to estimate. Even without this estimate, absolute error was +960 psi, or +4.0% at the nozzle, while repeatability error was +341 psi, or +1.4% at the nozzle. This explains why this type of nozzle or mold transducer does not improve the precision molding capability of injection machines. Morever, the slow response time of 100-200 milli-sec is reason by itself.
Quartz Piezoelectric Direct Mold Cavity Pressure Transducer
Quartz piezoelectric cavity pressure transducers are used in nozzles and directly in cavities. For this pressure transducer, the direct cavity unit was selected for study. Like Silicon on Sapphire, the quartz diaphragm is located in the transducer tip, near the plastic melt, but it is covered with a stainless steel cap, preventing direct melt contact with quartz.
A piezoelectric pressure transducer's signal output responds to dynamic change but decays with static pressure. A piezoresistive type transducer's signal output is always directly proportional to the pressure it measures, therefore piezoresistive type units measure both static and dynamic pressures, with response time being the essence for dynamic measurements. While response time for quartz is very fast, the combined error and repeatability of the quartz diaphragm type is much larger than the direct cavity Silicon on Sapphire piezoresistive type. This means that they are less accurate. They are also subject to abrasive wear by glass and other abrasive fillers in the plastic.
This quartz type of pressure transducer is very sensitive to changing temperature environments as seen in the specs of the manufacturer. For this study, the estimated temperature change on the quartz transducer tip was estimated at 50 deg F, instead of the typical full range of 100 to 450 deg F or 350 deg delta temp. of the plastic melt. While the stainless cap is a good thermal conductor, the amount of heat that can be transferred in a short time is limited. While possibly too conservative, these calculated errors were +425 psi, or +2.1% in the mold for absolute pressure variation and +308 psi, or +1.5% in the mold for repeatability. It is obvious that for longer cycle times and in high temperature molding, these errors will be larger, while for fast cycle molding they might be lower.
Load Cell Mold Pressure Transducers
Load cell mold pressure transducers are employed under ejector pins. These units measure pounds or kilograms of force and do not directly read pressure. To obtain pressure, the force must be divided by the ejector pin cross-sectional area in the mold. Users of ejector pin load cells report a fairly high level of breakage and damage, partially due to handling and partially due the mechanical dynamics involved in their location.
For this study, a mid-range, 0-550 lb load cell was selected to cover a pressure range of 0-22,000 psi, operating at 20,000 psi peak pressure (Ppeak/M). For this load cell, the sensitivity will be from the added friction of the ejector pin plus any hysteresis this contributes. A force of 5 lbs was added to the calculated value of error to account for the ever present friction of the ejector pin because they are truly additive. Thus, absolute error was calculated as +384 psi, or +1.9% on the plastic melt in the cavity, while the error for repeatability amounted to +258 psi or +1.3% in the mold. It must be remembered that ejector pin friction will vary greatly throughout a run due to clearances, pin bending and lubrication uniformity. Even so, unless friction was estimated too low, it appears that absolute error and repeatability are superior to hydraulic pressure transducer control, as well as mercury direct cavity and mercury nozzle pressure control transducers.
This explains why molders are using load cells in molds to advantage over standard hydraulic controls and getting better results. These results however, are not as good as can be obtained with silicon on sapphire nozzle transducers.
Sapphire Nozzle and Mold Pressure Transducer
Among the unique characteristics of this new pressure transducer is their ability to measure both melt pressure and temperature in one melt penetration, superior accuracy, lack of dead bands and hysteresis, as well as the fastest response times available for catching the peak pressure in the same place repeatedly. Their ability to monitor temperature while controlling pressure is ideal for ISO 9000 certified molders and for monitoring the condition of the plastic melt generated by the screw. The big advantage for silicon on sapphire transducers in injection molding for the utmost precision is their ability to zero out thermal errors on pressure readings in the machine. This is done by zeroing pressure on the melt after start-up by stopping the machine with the shut-off valve open for a short time.
For this study, the silicon on sapphire nozzle pressure transducer is operated with a +4oF nozzle melt temperature variation, in which absolute error calculated to +150 psi, or 0.5% of span at the nozzle, while repeatability error amounted to only +38.4 psi., or 0.13% of span at the nozzle. These are the lowest of any evaluated, with one magnitude of improvement over all the other units discussed. Because this repeatability error is significantly lower, being 12 to 15% of the other transducers studied, it indicates that significant improvements in process repeatability and precision will occur in any machine in which sapphire units are used to control injection pressures. Indeed, that has been the case for all the installations made to date. A secondary benefit is that machines run for weeks at a time without daily multiple shift adjustments to control the fill/flash ratio. This is a big advantage for unmanned factories working throughout the night without personnel. Example 2 shows the benefits of outfitting an old machine with a silicon on sapphire nozzle transducer.
Figure 4 shows the results of a large sample evaluation of shot weights to six significant figures, which was made by an independent laboratory for Van Dorn in Example 3. The comparison to the shown broader distribution hydraulic curve is a generic one, not based on actual data (which was not available), but it probably is really is much broader than shown. Also, the calculated repeatability for Example 3 is +30.2 psi, due to the temperature variation being within +2 F, instead of +4 F as used for the calculation in above. Figure 5 shows the correct manner of installing nozzle transducers.
Nozzle and Mold Melt Pressure Temperature
For measuring nozzle or direct cavity plastic melt temperatures, there are two transducers available; an infra-red unit of the Vanzetti type and the sapphire SRTD (silicon resistance temperature device). The major differences are that the SRTD is combined with pressure in a dual pressure/temperature model, while the infra-red is a stand alone temperature measuring unit. Additionally, there are differences in precision measurement capabilities of +1/2 deg F for the sapphire SRTD models versus +5 to 8 deg F for the infra-red. Response times are 10 milli-sec for the infra-red and 200 milli-sec for the SRTD units. Both response times are adequate compared to thermocouples which are 500 to >1,000 times slower. The infra-red unit costs about the same as a silicon on sapphire pressure plus temperature unit.
Summary
Table III summarizes the data obtained in this study. The calculated errors are surprisingly high compared to manufacturer's specifications. This shows that the idealized conditions in which transducers are tested and measured at the factory have no relation to real world molding machine operating conditions. Most of the variations seen are the result of thermal effects. The sole exception to the excessive variations calculated was the silicon on sapphire transducer which exhibited zero deviation from specified absolute error of 150 psi. Silicon on Sapphire also had the lowest repeatability error of only +38.4 psi, or 0.13% vs 0.10% spec, compared to +258 psi for the nearest other transducer and +292 to +341 psi for the others transducers evaluated. Thus, the use of a silicon on sapphire nozzle pressure transducer will provide 7 to 9 times improved accuracy over the others evaluated.
This superior performance of silicon on sapphire nozzle pressure transducers is attributed to the user's ability to zero out pressure temperature effects at the melt temperature, for which 4 F was used in the silicon sapphire calculation compared to the &2 F variation measured in the Van Dorn Example 3.
Table III Summary - Effect of Pressure Transducer Errors on Melt Pressure Variations in the Nozzle or Mold
Transducer Location Error Calculation at Nozzle
or Mold
| Absolute Error | Repeatability Error | |
| Hydraulic Hydraulic | 819 psi(3.4%vs1%) | 292 psi(1.5%vs.1%) |
| Mercury Nozzle | 960 psi(3.2%vs1%) | 341 psi(1.4%vs.2%) |
| Quartz Mold | 425 psi(1.4%vs1%) | 308 psi(1.0%vs1%) |
| Load Cell Mold | 422 psi(1.9%vs1%) | 261 psi(1.2%vs.1%) |
| Silicon on Sapphire Nozzle | 150 psi(0.5%vs.5%) | 38.4 psi(.13%vs.1%) |
Figure 6 shows a cost/performance plot of repeatability for all the transducers evaluated. This clearly shows that silicon on sapphire nozzle transducers are the most cost effective type that can be employed for precision molding injection pressure control. Not only are they most cost effective for performance, but also for durability due to their wear resistance to abrasive fillers.
Conclusions - Why Measuring Nozzle Melt Pressure and Temperature With Silicon on Sapphire Transducers is Important for Precision Molding
- They define the pressure and temperature condition of the plastic melt leaving the screw and being injected.
- Melt pressure and temperature are the only two dependent variables in the process.
- Hydraulic pressures are not indicative of nozzle pressures due to friction losses, which can amount to 10-30%.
- Barrel temperatures are not indicative of plastic melt temperatures because most of the heat energy is supplied adiabatically (frictionally) instead of from heaters. Differences as great as 70 deg F have been measured.
- Nozzle pressure is closer to the mold and can be controlled within 30 to 40 psi, whereas hydraulic errors are multiplied by the pressure intensifying piston to cause variations of 300 to 400psi at the nozzle.
- Shot weights and uniformity are directly controlled by Ppeak and Ppack, and not any other process parameter.
- The closer to this point that pressures are controlled, the better the precision of control.
- Linear LVDT transducers do not indicate pressures and are indirect methods of process control, unless used only for injection speed control.
- There is potential for Cpk 6 to 7, or better, shot weight control, and better on defects/rejects.
- Fits perfectly with ISO-9000 principals of process control and monitoring, fullfilling a need that no other type of transducer can provide.
- In the absence of inability to use direct sapphire cavity pressure transducer control, which may be best, sapphire nozzle control will be greatly improved over standard hydraulic control methods as well as all the other types of transducer control, as seen in the evaluation data.
- Hydraulic transducers would have to be made 10 to 30 times more accurate to equal the performance of a sapphire nozzle or direct cavity mold transducer.
- Machines Operated with Silicon on Sapphire Nozzle or Direct Cavity Melt Pressure Control will significantly out perform all other types of control for precision molding, providing zero rejects, reduced operator attention and substantial savings in operating costs!! Machine Savings of $30,000 per year have been reported with Silicon on Sapphire Nozzle Units.
Acknowledgements
The authors express their thanks to Mark Blevins, Van Dorn Demag, for garage door gear molding data.
Also to Robert Dray, Integrated Molding Corp., and Brian Ricci, US Valves, for bumper structure molding data.
Author's Comment
The authors sincerely hope that this paper will help provide a better insight into injection machine performance and control to owners and operators of injection machines as well as manufacturers of the machines and their controls.
Bibliography