Trajectory Scanning with the Newport MM4005, Newport XPS, Aerotech Ensemble, and Pro-Dex MAXv Motor Controllers

Contents

• Overview
• Implementation
• Safety
• Notation
• Defining a Trajectory
• Building a Trajectory
• Executing a Trajectory
• Reading Back a Trajectory
• Interaction with EPICS Motor Records
• Communication with the controller
• Hardware Notes
• Installation
• Loading the database
• Starting the SNL program
• Startup script
• MEDM Screens
• Autosave
• Example IDL Procedure
• SPEC Interface
• Restrictions

Overview

The Newport MM4005 and XPS, the Aerotech Ensemble, and the Pro-Dex MAXv motor controllers are capable of executing complex coordinated motions. Trajectories can be defined that move any or all of the axes through any set of complex motions. (Currently, the Ensemble trajectory support can only control a single axis.) The controller will coordinate these motions, keeping each axis very close to the theoretical position during the entire motion. The controller can output logic pulses during the execution of the trajectory, permitting external equipment to be synchronized to the motion. These capabilities are ideally suited to “on-the-fly” data collection.

At the APS one application of the MM4005 and XPS is to drive the large Newport diffractometer. This diffractometer has rather long settling times because of the mass of the moving elements. However, with this trajectory scanning software one can use SPEC, for example, to compute a set of diffractometer positions for a scan in HKL space, and then download these positions to the controller. The entire scan can be executed without stopping, collecting data in a multi-channel scaler or other triggered buffering detector. This can reduce data collection times dramatically relative to traditional step scanning.

This document describes an EPICS interface to the trajectory capabilities of the MM4005, XPS, Ensemble, and MAXv. This interface is completely general for the controllers, it is not specific to the Newport diffractometer. The interface lets any EPICS channel access client do the following:

• Define the total number of trajectory elements.
• Define the absolute or relative position of each axis for each point in the trajectory.
• Define the time for each element of the trajectory, or alternatively the total execution time with equal time per trajectory element.
• Define the total number of output synchronization pulses. The MAXv can only send output pulses at trajectory points.
• Define the trajectory elements where pulse outputs begin and end.
• Build and verify the trajectory, checking for errors.
• Define a total time scaling factor from .01 to 100 that will speed up or slow down the trajectory execution relative to its original definition (MM4005, MAXv, and Ensemble only).
• Execute the trajectory, checking for completion and errors. This can be done repeatedly without rebuilding if the only changes are in the start position or the execution time scale factor. The MAXv must load the trajectory before every execution.
• Read back the actual position of each axis when each synchronization pulse was output. The Ensemble and MAXv cannot do this. For these controllers, position data are usually recorded with a multichannel scaler to which motor pulses (MAXv) or PSO pulses (Ensemble) are sent. The Ensemble can record and read back position error with an internal scope.
• Read back the following error (actual-theoretical positions) when each synchronization pulse was output. The Ensemble and MAXv cannot do this.

Implementation

The EPICS implementation consists of the following:

• A database file, trajectoryScan.db. This database contains almost no "logic" with no links between records in the database. The records are simply variables that channel access clients and the State Notation Language (SNL) program use.
• SNL programs, MM4005_trajectoryScan.st, XPS_trajectoryScan.st, MAX_trajectoryScan.st, and EnsembleTrajectoryScan.st. These programs implement all of the logic for communicating with the controller and with channel access clients via the database.
• MEDM screens, trajectoryScan.adl, trajectoryScanDebug.adl, trajectoryPlot.adl. These screens are used to control the building, execution, readback, debugging and plotting of trajectory scans. MAXv and Ensemble currently use a different display: MAX_trajectoryScan.adl.

Safety

The MM4005 and XPS are used at the APS to control the large Newport diffractometer. This device is capable of moving large masses at high speeds. The trajectory scanning software does not use the EPICS motor record, and so does not obey any software limits defined in the motor record. It is very important that:

• The controller be programmed via the front panel (MM4005) or Web interface (XPS) to have safe software limits to prevent collisions whenever possible. The trajectory scanning software does obey the MM4005 internal soft limits. However, the MultipleAxesPVTExecution command on the XPS does not obey the soft limits! The XPS trajectory scan SNL code checks that the limits of the trajectory do not exceed the soft limits, and will not execute the trajectory if they do.
• Care be taken to avoid trajectory execution whenever personnel could be harmed. The following error thresholds for each axis on the controller should be set tight enough that the motor power will turn off when a significant resistance is met. An emergency stop button should be within reach whenever personnel are working on the diffractometer, and the switch should be in the Stop position whenever personnel are working for extended periods or in vulnerable positions on the diffractometer.

Notation

The database is loaded with $(P) (prefix) and $(R) (record base) macros. For example, $(P) might be 13IDC: for the name of the IOC, and $(R) might be Traj1: for the first trajectory in this IOC. The prefix and record bases are omitted from the Process Variable (PV) names in this document, but one should be aware that, for example, Nelements is really $(P)$(R)Nelements or in this case 13IDC:Traj1:Nelements.

There are 8 motors in the database, and thus 8 similar records for many functions. For example, the records defining the trajectory positions are M1Traj, M2Traj ... M8Traj. These are referred to in this document either as M1Traj ... M8Traj or as MnTraj.

Defining a Trajectory

The MM4005 and XPS always define a trajectory in terms of displacements (i.e. relative positions) of each motor for each element of the trajectory. Each trajectory element has an execution time associated with it, and hence there is a velocity defined for each motor (displacement/time) for each trajectory element. The MM4005 supports trajectories of up to 2000 elements, while the XPS supports an essentially unlimited number of elements.

The MAXv controller allows a maximum of 2500 trajectory elements. The number of elements supported by the Ensemble controller depends on the number of global double variables specified in the controller's configuration. The default configuration supports around 70 trajectory elements; the maximum is around 40,000 elements.

During execution of the trajectory the MM4005/XPS can output a user-definable number of logic pulses.  The trajectory elements where these output pulses begin and end can also be selected. On the MM4005 these pulses are evenly spaced in distance along the trajectory, which is a distance in up to 8-dimensional space. On the XPS the pulses are evenly spaced in time. The MAXv can output a pulse only at a trajectory point. Ensemble controllers can output pulses equally spaced in distance; some models (ML, HPe, HLe) can output pulses at each of an array of user specified positions.

These output pulses are typically used for the channel-advance of a multi-channel scaler, allowing detector pulses to be collected in hardware, with no software overhead at each point. At the time that each pulse is output the controller captures the theoretical and actual (encoder) position of each motor. These positions can be read back when the trajectory is complete. The MAXv and Ensemble controllers do not record positions at each output pulse.

It is important to remember that the number of data points in a scan is determined by the number of output pulses, not by the number of trajectory elements. For example, a constant velocity theta/2-theta scan over 10 degrees can be defined with a only 1 trajectory element, but could be used to collect 1000 data points at .01 degree steps.

The following table lists the EPICS Process Variables (PVs) that are used to define the trajectory. Some of these are explained in more detail below.

The EPICS interface permits the motor positions for each element of the trajectory to be defined in one of 3 ways. This flexibility can remove the burden of converting absolute positions to displacements from the EPICS channel access client and let the SNL program do the calculations. MoveMode can have the following 3 values:

• Relative. This mode maps directly to the way the MM4005 and XPS work internally. Each trajectory element is a displacement or distance for the motor to move during that element.
• Absolute. In this mode each trajectory element is an absolute motor position to which the controller will move at each point in the trajectory. In practice the SNL program computes an internal trajectory with Nelements-1 elements where InternalTraj[i] = MnTraj[i+1]-MnTraj[i]. The SNL program drives all the motors to the first point in the trajectory, waits for them to get there, and then executes the internal trajectory. The disadvantage of Absolute mode is that a new trajectory needs to be downloaded and built each time the absolute position of any motor in the trajectory is changed. 
• Hybrid. In this mode the trajectory is defined in absolute coordinates, as in Absolute mode. Again, an internal trajectory is computed from MnTraj[i+1]-MnTraj[i]. However, when the trajectory is executed the motors are not moved to the position of the first point of the trajectory. Thus, Hybrid mode permits a trajectory to be defined in absolute coordinates, but executed as motions relative to the current positions of the motors when the trajectory is executed. Thus the motors can be moved using the EPICS motor record, and the trajectory executed at a new absolute position without downloading new values to MnTraj or rebuilding the trajectory.

Note that when programming TimeTraj in Absolute or Hybrid mode TimeTraj[i] is the time to execute the move from position MnTraj[i] to MnTraj[i+1].

The MAXv and Ensemble currently do not support Hybrid mode. MAXv may, and Ensemble always does, send the motor to a starting point that is the acceleration distance from the start position, whether the trajectory mode is Absolute or Relative. The acceleration distance is AccelTime*velocity[0]/2, where AccelTime is the acceleration time, and velocity[0] is the average velocity over the first trajectory segment. Currently, the Ensemble support reads the acceleration time directly from the motor record.

The MM4005 requires the number of elements in a trajectory to be a multiple of 4. This means that Nelements should be a multiple of 4 in Relative mode and a multiple of 4 plus 1 in Absolute or Hybrid mode. However, this can be an inconvenience for EPICS channel access clients. The SNL program works around this restriction as follows. If Nelements is not a multiple of 4 for MoveMode=Relative or a multiple of 4 plus 1 for MoveMode=Absolute or Hybrid, then up to 3 padding elements are automatically added to the user defined trajectory to satisfy the requirement. The padding elements always have a time of 0.1 seconds. The displacements of the motors in these padding elements is computed to maintain the same velocity as the last element in the original trajectory. Thus the padding elements will cause the trajectory to execute for up to 0.3 seconds longer than requested, and the motors will move slightly farther than requested. However, there will be no velocity change, and hence no unexpected accelerations during the padding elements.

The MM4000 has a concept of accelerating onto the trajectory, so that the time per trajectory element will actually be longer than the specified times for the first few and last few trajectory elements, as the motors accelerate and deccelerate. This is typically normalized out during data processing by having a scaler channel that measures the time (or integrated incident intensity) during each trajectory element.

The XPS does not do this acceleration automatically, rather the acceleration must be explicitly defined in the trajectory file that is downloaded to the XPS. The XPS SNL program adds one padding element at the beginning of the file and one padding element at the end of the file. The first padding element is calculated to be the minimum time to accelerate all of the motors from 0 to the velocity of the first actual trajectory element. This time is calculated using an acceleration of 90% of the maximum allowed acceleration for each axis (to avoid roundoff problems). It also calculates how far each axis moves during this acceleration time. Similarly the final padding element is calculated to be the minimum time to deccelerate all of the motors from the velocity of the last actual trajectory element to 0. These padding elements cause the actual range of motion of each of the motors to be somewhat larger than the range in the trajectory itself. This means that trajectories cannot execute right up to the software or hardware limits for an axis, because there will not be room for the acceleration motion.

The MAXv and Ensemble controllers also do not automatically accelerate onto the trajectory. The SNL programs add an element to the beginning of the trajectory and one (MAXv) or two (Ensemble) elements to the end. (Ensemble always does this; MAXv does it if AddAccelDecel=Yes.) These elements cause the actual range of motion of each of the motors to be somewhat larger than the range in the trajectory itself. This means that trajectories cannot execute right up to the software or hardware limits for an axis, because there will not be room for the acceleration motion.

The number of trajectory elements, Nelements is limited as follows.

• The MM4005 allows a maximum of 2000 trajectory elements.
• The MAXv allows a maximum of 2500 trajectory elements.
• The Ensemble allows a maximum of 70 trajectory elements as delivered, but its configuration can be modified to support up to 40,000 elements.
• Channel access on EPICS 3.13 is limited to 2000 points, since it has a 16,000 byte limit, and all arrays are double precision, requiring 8 bytes per element. In EPICS 3.14, the number of points Channel Access can send is the environment variable EPICS_CA_MAX_ARRAY_BYTES divided by eight.
• There are 9 double precision arrays that are dimensioned MAX_ELEMENTS in the SNL program, and 9 waveform records (MnTraj and TimeTraj) that are dimensioned NELM in the database. If 2000 points are allowed then the total memory used in the IOC is 2000*9*8*2 = 288,000 bytes. This is a significant amount of memory for an IOC. MAX_ELEMENTS is presently defined to be 2000 in the SNL programs. This value can be changed (up to 2000 for the MM4005) and the SNL program recompiled. The value of NELM can be set to any value up to MAX_ELEMENTS when the database is loaded.

The number of output pulses, Npulses is limited as follows.

• The MM4005 allows a maximum of 2000 output pulses.
• The MAXv can only output pulses at trajectory points, so Npulses is ignored.
• The Ensemble controller can send output pulses equally spaced in position, and imposes no limit on the number of such pulses. Some Ensemble controllers (ML, HPe, and HLe) can also generate output pulses at user-specified positions. Aerotech documentation does not specify the maximum number of positions.
• Channel access R3.13 is limited to 2000 points, since it has a 16,000 byte limit, and all arrays are double precision, requiring 8 bytes per element. In EPICS 3.14, the number of points Channel Access can send is the environment variable EPICS_CA_MAX_ARRAY_BYTES divided by eight.
• There are 16 double precision arrays that are dimensioned. MAX_PULSES in the SNL program, and 16 waveform records (MnActual and MnError) that are dimensioned NPULSE in the database. If 2000 points are allowed then the total memory used in the IOC is 2000*16*8*2 = 512,000 bytes. This is a significant amount of memory for an IOC. MAX_PULSES is presently defined to be 2000 in the SNL program. This value can be changed (up to 2000 for the MM4005) or and the SNL program recompiled. The value of NPULSE must be set to exactly the value of MAX_PULSES when the database is loaded.

Building a Trajectory

After a trajectory has been defined by setting the values of the PVs described in the previous section it must be built before it can be executed. Building the trajectory consists of downloading it to the controller and checking it for errors such as excess velocities or accelerations.

The following table describes the EPICS PVs used for building a trajectory.

Channel access clients should do the following to build a trajectory:

• Set Build=1
• Repeatedly read Build, wait for it to go to 0=Done.
• When Build=Done check BuildStatus. If it is not 1=Success, then something went wrong.
• BuildMessage can be used to determine what the error was, although this will probably require a human rather than a program.

If the build fails then it is useful to look at the trajectoryScanDebug.adl MEDM screen to examine the maximum velocity and acceleration values. See if one or more motors is being commanded to move too fast.

Executing a Trajectory

After a trajectory has been successfully built it can be executed. The trajectory execution consists of the following steps:

• Remember the initial positions of all the motors.
• Move to the start position defined by MnTraj[0]. This is only done if MoveMode=Absolute. (MAXv moves to a start position if MoveMode=Absolute or if AddAccelDecel=Yes. Ensemble always moves to a start position, because it always adds a accelerate-onto-trajectory element.)
• Trigger the detector start PV.
• Execute the trajectory in either Real or Simulate mode, with the execution time scaled by TimeScale. Poll during execution and post channel access monitors for the current element being executed, for the current positions of the motors and for any errors.
• When the trajectory is complete:
  • Trigger the detector stop PV.
  • Command the EPICS motors to move to the final position in the trajectory. This does not actually move the motors, but syncs the EPICS motor VAL and RBV fields to the actual motor positions.

The following table describes the EPICS PVs used for executing a trajectory.

Channel access clients should do the following to execute a trajectory:

• Set Execute=1
• Repeatedly read Execute, wait for it to go to 0=Done.
• When Execute=Done check ExecStatus. If it is not 1=Success, then something went wrong.
• ExecMessage can be used to determine what the error was, although this will probably require a human rather than a program.

The execution can fail because the velocity or acceleration is too large, even if the build succeeded, if TimeScale is less than 1.0.

Reading Back a Trajectory

After a trajectory has been executed it is possible to read back from the MM4005 or XPS the theoretical and actual positions of the motors when each synchronization pulse was output. The EPICS interface presents this information as the actual positions and the following errors (actual position minus theoretical position) since these are usually of most interest to the user. Obviously the theoretical position can be computed from the actual position and the following error.

Reading back this information from the MM4005 (but not from the XPS) is rather slow, but in many cases this does not need to be done for each scan. Once it is established that the following errors are small enough it is possible to execute scans without reading back from the MM4005 each time. The readback time is determined by the speed of the communications interface to the MM4005. Each point returned from the MM4005 is about 200 characters. Using RS-232 at 19,200 baud this requires 0.1 seconds per point, where the number of points is equal to the number of output pulses. This is thus 30 seconds for a scan with 300 output pulses. The XPS uses FTP over Ethernet to read the "Gathering.dat" file containing the theoretical and actual motor positions. The time to copy this file is very short, even for a large number of points.

The MAXv controller cannot read back trajectory data. The Ensemble controller can record actual motor positions periodically during a trajectory. Currently, this is a compile-time option. (See #define USE_SCOPE 1 in EnsembleTrajectoryScan.st.)

The following table describes the EPICS PVs used for reading back a trajectory.

Channel access clients should do the following to read back a trajectory:

• Set Readback=1
• Repeatedly read Readback, wait for it to go to 0=Done.
• When Readback=Done check ReadStatus. If it is not 1=Success, then something went wrong.
• ReadMessage can be used to determine what the error was, although this will probably require a human rather than a program.

Note that the readback command reads the global trace buffer of the MM4005. It can be used to read back this trace buffer even if the previous operation was not a trajectory execution. This can be useful for debugging operations in general.

Interaction with EPICS Motor Records

The trajectory scanning does not use the EPICS motor records, but rather talks directly to the controller. In order to keep the motor records in sync with the actual motor positions the SNL program always sends the motors to the final position in the trajectory scan, i.e. the positions where the motors actually are, after a trajectory execution is complete. This should not result in any actual motor motion (except for small following errors), but serves to resynchronize the EPICS motor VAL and RBV fields with the actual motor positions. One should not move any motors on the involved controller with the EPICS motor records while a trajectory scan is in progress. For the MAXv, one must not move any motor on the involved controller.

When a motion is aborted with the Abort PV it will be necessary to home the motors and synchronize the motor records with the actual positions of the motors. (This is not necessary for the MAXv and Ensemble controllers.)

Communication with the controller

The communication with the MM4005 uses asyn. It can be used with either the RS-232 or GPIB interfaces. Debugging can be enabled with the asynSetTraceMask and asynSetTraceIOMask commands. Communication with the XPS also uses asyn over a TCP/IP socket connection. Debugging can be enabled and disabled in the same way.

The timeout for communication with the MM4005 is set to 30 seconds, because some commands can take a very long time to response. This was found to be necessary to allow for the long time it takes the MM4005 to respond to the TB command after a VC command is issued when verifying the trajectory. However, it would probably be better to determine empirically how long it takes the MM4005 to verify trajectories as a function of Nelements and Npulses and have the SNL program wait that long after sending the VC command and before sending the TB command. These measurements have not been done yet.

Hardware Notes

The synchronization output pulses from the MM4005 are provided on pin 12 of the DB-25 Auxilliary Connector. The synchronization output pulses from the XPS are provided on pin 12 of the GPIO2 connector. We have found it convenient to make DB-25 connectors with a short BNC or LEMO pigtail coming from these pins. The pulse outputs from both the MM4005 and XPS are open-collector circuits. The maximum rating in both cases is 30V and 40 mA.

When using the pulse output as the channel-advance input of the Struck 7201 or SIS 380x multi-scaler some modifications are required. The SIS 380x manual states that with the LEMO TTL input configuration the inputs are pulled up to +5V with 1K Ohm resistors in a resistor pack. For our module the factory configuration was actually a 4.7K Ohm resistor pack. These resistors did pull the open-collector output up to +5V. However, when driving the cable over a long distance (~80 feet) the rise time of the signal was quite slow, about 8 microseconds to go from 0V to the TTL threshold. This slow rise time caused the SIS 380x to double count the channel advance signal most of the time. By replacing the 4.7K Ohm resistor pack with a 200 Ohm pull-up resistor pack the rise time was reduced to about 2 microseconds, and the module does not double count. 200 Ohms is within the spec of the MM4005 and XPS, since it will result in a current of 5V/200 Ohm = 25 mA, which is less than the 40 mA maximum.

The synchronization output pulses from the MAXv are sent on one of the pins IO0 - IO15, selected with the PV OutBitNum, which must be in the range[0,15] inclusive. If the number is outside this range, no output pulses will be generated. The MAXv is able to start a trajectory on the rising edge of an input pulse, selected by the PV InBitNum, with the same range. If InBitNum is outside this range, the controller will start the trajectory as soon as it has been loaded. These signals are available from the 50-pin SCSI connector on the front panel of the MAXv.

The synchronization output pulses from the Ensemble are sent on the "Auxiliary Marker Output". Different controller models have different connectors and pin numbers for this signal. This signal is RS422. Some Ensemble controllers implement RS422 in such a way that the differential signals can be used as complementary TTL signals.

The Ensemble controller must be prepared with the AeroBasic program motor/motorApp/AerotechSrc/doCommand.ab. You must copy doCommand.ab to into the user-file directory of Aerotech's Motion Composer program, load it into Motion Composer, and compile it to produce the file doCommand.bcx. This file must be copied to the controller, so it can be executed immediately when EnsembleTrajectoryScan.st sends the command PROGRAM RUN 1 "doCommand.bcx". The controller also must be configured (using Aerotech's Configuration Manager) to increase the number of global integers and doubles. For N trajectory points, the number of global doubles (the GlobalDoubles parameter) must be at least (N+3)*3 (for one motor), and the number of global integers (the GlobalIntegers parameter) must be at least N+50. Also, the controller must use the same units as the motor record that talks to it.

Installation

The source files for trajectory scanning are in the synApps motor module , in the motorApp/ tree.

NewportSrc/MM4005_trajectoryScan.st
NewportSrc/XPS_trajectoryScan.st
OmsSrc/MAX_trajectoryScan.st
AerotechSrc/EnsembleTrajectoryScan.st
Db/trajectoryScan.db
op/adl/trajectoryScan.adl
op/adl/trajectoryScanDebug.adl
op/adl/trajectoryPlot.adl
op/adl/MAX_trajectory*.adl

Loading the database

The database file is called trajectoryScan.db.. This database is completely general for the MM4005 and XPS, it makes no assumptions about the motors defined on particular axis. Thus it can be used with the Newport diffractometer or any other set of up to 8 motors. It takes the following macro parameters:

Macro parameter	Description
$(P)	PV name prefix
$(R)	PV base record name
$(NAXES)	Number of axes to be used. 6 for our Newport diffractometers.
$(NELM)	Maximum trajectory elements
$(NPULSE)	Maximum number of output pulses

Starting the SNL program

There are different SNL programs for the different controllers. The MM4005 SNL program is defined as follows:

program MM4005_trajectoryScan("P=13IDC:,R=traj1,M1=M1,M2=M2,M3=M3,M4=M4,M5=M5,M6=M6,M7=M7,M8=M8,PORT=serial1")

The XPS SNL program is defined as follows:

program XPS_trajectoryScan("P=13BMC:,R=traj1,IPADDR=164.54.160.34,PORT=5001,"
                           "USERNAME=Administrator,PASSWORD=Administrator,"
                           "M1=m1,M2=m2,M3=m3,M4=m4,M5=m5,M6=m6,M7=m7,M8=m8,"
                           "GROUP=g1,"
                           "P1=p1,P2=p2,P3=p3,P4=p4,P5=p5,P6=p6,P7=p7,P8=p8")
                           

The MAXv SNL program is defined as follows:

program MAX_trajectoryScan("P=13IDC:,R=traj1,M1=m1,M2=m2,M3=m3,M4=m4,M5=m5,M6=m6,M7=m7,M8=m8,PORT=none")

The Ensemble SNL program is defined as follows:

program EnsembleTrajectoryScan("P=13IDC:,R=traj1,M1=m1,M2=m2,M3=m3,M4=m4,M5=m5,M6=m6,M7=m7,M8=m8,PORT=tcp1")

Startup script

XPS

# Database for trajectory scanning with the XPS
dbLoadRecords("$(MOTOR)/motorApp/Db/trajectoryScan.db", 
              "P=13BMC:,R=traj1,NAXES=6,NELM=2000,NPULSE=2000")

iocInit

dbpf("13BMC:traj1DebugLevel","1")
seq(&XPS_trajectoryScan, "P=13BMC:,R=traj1,M1=m33,M2=m34,M3=m35,M4=m36,M5=m37,M6=m38,
                        IPADDR=164.54.160.124,PORT=5001,GROUP=GROUP1,
                        P1=PHI,P2=KAPPA,P3=OMEGA,P4=PSI,P5=2THETA,P6=NU")

MM4005

# Database for trajectory scanning with the MM4005
dbLoadRecords("$(CARS)/CARSApp/Db/trajectoryScan.db", "P=13IDC:,R=traj1,NAXES=6,NELM=2000,NPULSE=2000")

iocInit

# Trajectory scanning with MM4005
seq(&MM4005_trajectoryScan, "P=13IDC:, R=traj1, M1=m25,M2=m26,M3=m27,M4=m28,M5=m29,M6=m30,M7=m31,M8=m32,PORT=serial13")

MAXv

# Database for trajectory scanning with the MAXv
dbLoadRecords("$(MOTOR)/motorApp/Db/trajectoryScan.db", 
              "P=13BMC:,R=traj1,NAXES=6,NELM=2000,NPULSE=2000")

iocInit

# MAXV trajectory scan
seq &MAX_trajectoryScan, "P=xxx:,R=traj1:,M1=m1,M2=m2,M3=m3,M4=m4,M5=m5,M6=m6,M7=m7,M8=m8,PORT=none"

Ensemble

The Ensemble trajectory support uses the same infrastructure that the motor record uses to talk to the controller. Here's are typical commands that set that up:

	dbLoadTemplate("AeroAsyn.substitutions")
	drvAsynIPPortConfigure("tcp1","164.54.51.77:8000", 0, 0, 0)
	EnsembleAsynSetup(1)
	EnsembleAsynConfig(0, "tcp1", 0, 1, 50, 1000)
	drvAsynMotorConfigure("AeroE1","motorEnsemble",0,1)

	file "$(MOTOR)/db/asyn_motor.db"
	{
	pattern
	{P,     N,  M,        DTYP,         PORT,   ADDR,  DESC,          EGU,  DIR,  VELO,  VBAS,  ACCL,  BDST,  BVEL,  BACC,  MRES,   PREC,  DHLM,  DLLM,           INIT}
	{xxxL:,  17,  "m$(N)",  "asynMotor",  AeroE1,    0,  "motor $(N)",  mm,   Pos,  .1,     0,    .2,    0,     1,     .2,    1.25E-5,   7,     0,      0,  "SCURVE 100"}
	}

# Database for trajectory scanning with the Ensemble
dbLoadRecords("$(MOTOR)/motorApp/Db/trajectoryScan.db", 
              "P=13BMC:,R=traj1,NAXES=6,NELM=2000,NPULSE=2000")

iocInit

# Ensemble trajectory scan
seq &EnsembleTrajectoryScan, "P=15iddLAX:,R=traj1:,M1=aero:c0:m1,M2=none,M3=none,M4=none,M5=none,M6=none,M7=none,M8=none,PORT=tcp1"

MEDM screens

The following show the MEDM screens with which the user can view and modify the trajectory scanning parameters.

trajectoryScan.adl is the main screen used to define, build, execute and read back trajectories. The only thing that cannot be done in MEDM is to edit the MnTraj and TimeTraj arrays, since MEDM does not provide a method to edit arrays.

trajectoryScan.adl is called with macro parameters P, R, TITLE, and M1 ... M8. P and R are the prefix and record base used when the database was loaded. M1 ... M8 are the names of the EPICS motors. These are used to label the plots in trajectoryPlot.adl using the .DESC fields of the motors. For example trajectoryScan.adl in this screen shot was called with P=13BMC:, R=traj1, TITLE=Trajectory Scans, M1=m33,M2=m34,M3=m35,M4=m36,M5=m37,M6=m38,M7=m25,M8=m26 .

trajectoryPlot.adl is used to plot the requested trajectory in position and time (MnTraj, TimeTraj), the readback positions (MnActual) and the following errors (MnError).

trajectoryScanDebug.adl is used to display detailed information, useful for debugging.

MEDM displays for the MAXv and Ensemble are very similar, but also include start-position PVs written by the MAXv sequencer program, and some controller specific PVs on the separate displays trajectoryScan_MAXv.adl, and trajectoryScan_Ensemble.adl. These displays can be called up from the "More" menu button on the MAX_trajectoryScan.adl display. The plotting displays for these controllers are different in that they plot as functions of time, rather than trajectory point.

Autosave

The user-modifiable PVs that control an instance of trajectoryScan.db can be autosaved by including a line like the following in the autosave-request file:

file trajectoryScan_settings.req P=$(P),R=traj1:

Example IDL Procedure

The following IDL function illustrates how an EPICS channel access client can define, build, execute and read back a trajectory.

function trajectory_scan, traj, positions, relative=relative, hybrid=hybrid, $
                          time=time, accel=accel, npulses=npulses, $
                          build=build, execute=execute, read=read, $
                          actual=actual, errors=errors

;+
; NAME:
;   trajectory_scan
;
; PURPOSE:
;     This IDL function loads and execute a complex trajectory
;     using the EPICS trajectory scan databases and SNL programs. The EPICS
;     support current exists for the Newport MM4005 and XPS motor controllers.
;
;
; CATEGORY:
;     EPICS trajectory scanning
;
; CALLING SEQUENCE:
;     Result = TRAJECTORY_SCAN(Trajectory, Positions)
;
; INPUTS:
;     TRAJ: The EPICS record name for this trajectory, for example '13BMC:traj1'.
;
;     POSITIONS: [NELEMENTS, NMOTORS]. The positions of the motors at each element in the trajectory.
;         By default the positions are absolute motor positions.  If the RELATIVE
;         keyword is present then the positions are deltas, i.e. the difference in position
;         from the previous point.
;
;
; KEYWORD PARAMETERS:
;     RELATIVE: Set this keyword if the positions are deltas, i.e. the difference in position
;               from the previous point.  Default mode=ABSOLUTE.
;     HYBRID:   Set this keyword if the trajectory should be executed in "Hybrid" mode, i.e.
;               the positions are absolute rather than deltas, but the trajectory should be
;               executed from the current motor positions without moving to the first point
;               in the postions array.  Default mode=ABSOLUTE.
;     TIME:     If this keyword is a scaler, then it specifies the total time to execute
;               the trajectory.  If it is an array then it specifies the time per element.
;               Default=1 second per element.
;     ACCEL:    The acceleration time for the trajectory.  Only for the MM4005. Default=1 second.
;     NPULSES:  The number of output pulses during the trajectory.  Default=NELEMENTS, the number
;               of points in the input Positions array.
;     BUILD:    Set this keyword to build and verify the trajectory.  This is the default.
;
;     EXECUTE:  Set this keyword to execute the trajectory.  This is the default.
;
;     READ:     Set this keyword to read back the trajectory into ACTUAL and ERROR. 
;               The default is to not read back.
;
;     NOTE: Any or all of these keywords can be set.  If none is set then the
;         function does not do anything.
;
; OUTPUTS:
;       Result:     This function returns a status indicating whether the
;                   selected operations were successful or not. 0=success,
;                   anything else is a failure.
;       ACTUAL:     An array of [Nactual, NMOTORS] containing the actual positions of
;                   each axis.
;       ERRORS:     An array of [Nactual, NMOTORS] containing the following errors of
;                   each axis.
;       NOTE: The ACTUAL and ERROR outputs are only returned if the READ keyword it set.
;
; SIDE EFFECTS:
;       This procedure can move the motors.  Be careful!
;
; EXAMPLE:
;       positions = [[1,2,3],[.1, .2, .3], [0,3,4]]
;       status = TRAJECTORY_SCAN('13IDC:traj1', positions, /read, actual, errors)
;       plot, actual[*,0]
;       oplot, errors[*,0]
;
; MODIFICATION HISTORY:
;       Written by:     Mark Rivers, December 15, 2006
;-

    t = caget(traj+'NumAxes', maxAxes)

    MoveMode = 'Absolute'
    if (keyword_set(relative)) then MoveMode = 'Relative'
    if (keyword_set(hybrid)) then MoveMode = 'Hybrid'
    t = caput(traj+'MoveMode', MoveMode)

    if (n_elements(build) eq 0) then build=1
    if (n_elements(execute) eq 0) then execute=1

    if (keyword_set(build)) then begin

        dims = size(positions, /dimensions)
        nelements = dims[0]
        naxes = dims[1]
        if (n_elements(npulses) eq 0) then npulses = nelements

        t = caput(traj+'Nelements', nelements)

        t = caput(traj+'Npulses', npulses)

        ; Default is 1 second per element
        if (n_elements(time) eq 0) then time = nelements * 1.0
        if (n_elements(time) eq 1) then begin
            t = caput(traj+'TimeMode', 'Total')
            t = caput(traj+'Time', time)
        endif else begin
            t = caput(traj+'TimeMode', 'Per element')
            t = caput(traj+'TimeTraj', time)
        endelse

        if (n_elements(accel) eq 0) then accel = 1.
        t = caput(traj+'Accel', accel)

        ; The first naxes motors will move.
        for i=0, maxAxes-1 do begin
            axis = traj + 'M' + strtrim(i+1,2)
            if (i lt naxes) then begin
                t = caput(axis+'Move', 1)
                pos = positions[*,i]
                t = caput(axis+'Traj', pos)
            endif else begin
                t = caput(axis+'Move', 0)
            endelse
         endfor

        ; Trajectory is now defined.  Build it.
        t = caput(traj+'Build', 1)
        ; Wait for the build to complete. Wait 0.1 second between polls.
        repeat begin
            wait, 0.1
            t = caget(traj+'Build', busy)
        endrep until (busy eq 0)

        ; Make sure the build was successful
        t = caget(traj+'BuildStatus', BuildStatus, /string)
        if (BuildStatus ne 'Success') then begin
            t = caget(traj+'BuildMessage', BuildMessage)
            print, 'Build failed, error = ', BuildMessage
            return, BuildStatus
        endif
    endif

    if (keyword_set(execute)) then begin
        t = caput(traj+'Execute', 1)
        ; Wait for the execute to complete. Wait 0.1 second between polls.
        repeat begin
            wait, 0.1
            t = caget(traj+'Execute', busy)
        endrep until (busy eq 0)

        ; Make sure the execution was successful
        t = caget(traj+'ExecStatus', ExecStatus, /string)
        if (ExecStatus ne 'Success') then begin
            t = caget(traj+'ExecMessage', ExecMessage)
            print, 'Execution failed, error = ', ExecMessage
            return, ExecStatus
        endif
    endif

    if (keyword_set(read)) then begin
        t = caput(traj+'Readback', 1)
        ; Wait for the readback to complete. Wait 0.1 second between polls.
        repeat begin
            wait, 0.1
            t = caget(traj+'Readback', busy)
        endrep until (busy eq 0)

        ; Make sure the readback was successful
        t = caget(traj+'ReadStatus', ReadStatus, /string)
        if (ReadStatus ne 'Success') then begin
            t = caget(traj+'ReadMessage', ReadMessage)
            print, 'Read failed, error = ', ReadMessage
            return, ReadStatus
        endif

        ; Read the actual and error arrays into IDL, return to
        ; caller
        t = caget(traj+'Nactual', nactual)
        actual = dblarr(nactual, maxAxes)
        errors = dblarr(nactual, maxAxes)
        for i=0, maxAxes-1 do begin
            axis = traj + 'M' + strtrim(i+1,2)
            t = caget(axis+'Actual', temp, max=nactual)
            actual[0,i] = temp
            t = caget(axis+'Error', temp, max=nactual)
            errors[0,i] = temp
        endfor
    endif

    return, 0
end

This is an IDL main program that defines a trajectory and executes it using trajectory_scan.pro above.

; This program builds a trajectory and executes it.

; The trajectory definition is hybid mode, meaning the positions are
; definined in absolute coordinates rather than displacements from on
; element to the next. However, the motors do not move to the absolute
; position of the first element before executing the trajectory.

; 101 elements in the trajectory. We use 4N+1 since we are defining the
; trajectory in Hybrid mode
 nelements = 101
 
; We will move the first 2 motors (Phi and Kappa)
 naxes=2
        
; Define array of positions
positions = dblarr(nelements, naxes)

; The Phi trajectory is a sin wave with two complete periods and an
; amplitude of +-8 degrees
positions[*,0] = 8.*sin(findgen(nelements)/(nelements-1.)*4.*!pi)
        
; The Kappa trajectory is a sin wave with one complete period and an
; amplitude of +-20 degrees
positions[*,1] = 20.*sin(findgen(nelements)/(nelements-1.)*2.*!pi)

trajectory = '13BMC:traj1'

; Total time for trajectory
time = 20.

status = trajectory_scan(trajectory, positions, /hybrid, /build, /execute, /read, $
                         time=time, npulses=300, actual=actual, errors=errors)

end

These are the IDL commands to execute the above program, and plot the results:

IDL> .run trajectory_test1
IDL> iplot, actual[*,0]   
IDL> iplot, actual[*,1], /overplot
IDL> iplot, errors[*,0]   
IDL> iplot, errors[*,1], /overplot

These are the resulting plots:

SPEC Interface

A set of SPEC macros allows SPEC to utilize trajectory scanning with EPICS and the MM4005 or XPS. It requires a multi-channel scaler or some other detector that can be triggered and has buffering.

The implementation is done at a low level, so that all of SPEC's standard scans can be done "on-the-fly" utilizing this trajectory scanning software. This was done by providing replacement macros for:

_ascan    # Used by all ascan and dscan macros
mesh
hklscan   # Used by hscan, kscan and lscan
_hklmesh
_hklline  # Used by hkcircle, hlcircle, klcircle, hkradial, hlradial and klradial
_scanabort
resume
_loop

traj_index        # Converts a SPEC motor index to an MM4005 or XPS motor index
traj_build        # Builds a trajectory
traj_exec         # Executes a trajectory
traj_read_counts  # Reads the data from the multi-channel scaler
traj_read_actual  # Reads back the actual MM4005 or XPS motor positions
traj_scans_on     # Enables trajectory scanning
traj_scans_off    # Disables trajectory scanning, uses step scanning

It is easy to switch back and forth between traditional step scanning and trajectory scanning. traj_scans_on turns on trajectory scanning for all subsequent scans. traj_scans_off reverts back to traditional step scanning.

It is possible to have the motor positions and HKL values that SPEC prints on the screen, plots and stores in the SPEC data file be based upon the theoretical motor positions during the scan. Alternatively SPEC can use values based upon the actual motor positions at each point in the scan. Using the actual motor positions is slower, because the values must be read from the MM4005 at the end of the trajectory execution. Set the SPEC global variable TRAC_USE_ACTUAL=0 to use the theoretical motor positions, and TRAJ_USE_ACTUAL=1 to use the actual motor positions. <

Restrictions

The following are restrictions and problems with the trajectory scanning.

• At short trajectory element times with velocity changes the XPS can exceed the maximum acceleration, even when a simple calculation (delta velocity/delta time) says that it should not. The fix will be to use the actual spline function that the XPS uses when calculating the velocities in the trajectory file. We have not run into this problem at trajectory times larger than 0.01 second.
• The MAXv controller does not actually have a trajectory capability, but instead has a variable velocity capability that can be programmed to produce the effect of a trajectory. This works well when there are many motor pulses between trajectory points. When there are few, the times at which the motors hit trajectory points become erratic.