REV 2025 Beta

[jfabellera]

See the latest announcement here

Original Post

Apologies for the long wait! REV's 2025 beta is finally here. We look forward to hearing your feedback.

• REVLib 2025.0.0-beta-1
• Instructions for updating Hardware Client and firmware

For information on how to migrate to 2025 REVLib, please see our documentation

Warning

After updating your SPARK MAX or Flex to 25.0.0, you will be unable to communicate with them through the Hardware Client if using a Power Distribution Hub or Pneumatic Hub as the USB CAN bridge. A later update will be available for the PDH and PH that fixes this. For now, connect to the Hardware Client via a SPARK MAX or Flex.

[person4268]

Setting some parameters in sim (signals, followers) results in a crash with 2025.0.0-beta-1:

Error at com.revrobotics.spark.SparkBase.configure(SparkBase.java:276): Unhandled exception instantiating robot com.revrobotics.spark.SparkBase java.lang.IllegalStateException: Parameter type mismatch for parameter id
        at com.revrobotics.spark.SparkBase.configure(SparkBase.java:276)
        at frc.robot.Robot.<init>(Robot.java:39)
        at edu.wpi.first.wpilibj.RobotBase.runRobot(RobotBase.java:301)
        at edu.wpi.first.wpilibj.RobotBase.lambda$startRobot$0(RobotBase.java:418)
        at java.base/java.lang.Thread.run(Thread.java:840)

I sent some more details to support@revrobotics but I can send them here or whereever else is needed. Just figured I'd post the high-level cause somewhere so other people can see it.

[jfabellera]

Same issue as #28

This will be fixed in the next release which should be coming soon

[person4268]

Oh 🤦, that got posted after I finished debugging but before I wrote it up. Thanks you for fixing it.

[jfabellera]

Fix is out. See #27 (comment)

[jfabellera]

REVLib beta 2 and SPARK MAX pre-release 6 have been released

• REVLib 2025.0.0-beta-2
• Instructions for updating Hardware Client and firmware

Warning

After updating your SPARK MAX or Flex to 25.0.0, you will be unable to communicate with them through the Hardware Client if using a Power Distribution Hub or Pneumatic Hub as the USB CAN bridge. A later update will be available for the PDH and PH that fixes this. For now, connect to the Hardware Client via a SPARK MAX or Flex.

Issues fixed

• REV Hardware Client won't set any parameters #32
• Setting certain parameters in simulation causes crash #28

[pigrammer3]

Does SparkMax.configure (and SparkFlex.configure) wait to verify that the configuration is applied? That is, is it a blocking call or does it run in the background? Do we need to wait some time for the configuration to save after calling the method, or can we use the motor immediately?

[jfabellera]

Yes it does verify that setting the parameters was successful (and retries if necessary). It is a blocking call, so once that's done, you're able to use the configured controller if no errors or exceptions come up.

[viggy96]

How many retries are attempted, and that number configurable?

And is the total amount of time to wait for the operation to complete configurable?

[jfabellera]

How many retries are attempted, and that number configurable?

It retries a maximum of 5 times and can be configured with setCANMaxRetries() (I recognize that the docs will need to be updated for the new system but it's still inherently correct)

And is the total amount of time to wait for the operation to complete configurable?

What operation are you talking about? All of configure() or individual parameter set commands? There isn't one for configure() as that would be directly related to the number of parameters being set from the configuration. Individual blocking calls that require a response (like the parameter set calls that configure() make) have timeouts that can be configured with setCANTimeout(). See my other comment about this method: #27 (reply in thread)

[viggy96]

I guess its taken care of via setCANMaxRetries(), and the timeout on each one attempt determines the maximum amount of time configure() will take.

[person4268]

If I want to toggle a motor from brake to coast mode, would it be faster (and semantically correct) to create a new SparkFlexConfig with just idlemode set to the correct value?

[jfabellera]

Yes and you would pass kNoResetSafeParameters to preserve the rest of the configuration and preferably kNoPersistParameters since that operation takes a considerable amount of time especially for the SPARK MAX, and blocking for any amount of time during robot operation should generally be avoided.

[NoahStricker]

Is there no longer a non-blocking way to set idle mode?

[jfabellera]

Not at the moment. The effect of setting the CAN timeout to 0 (as documented here) is no longer present due to some infrastructure changes. The ability to do certain commands asynchronously will be added back before kickoff.

[viggy96]

Would it be too much to ask for some wrapper methods for the more mundane tasks, such as setting idle mode, and inverting the motor?

[BR88C]

Great to see these improvements, thank you to the team that worked on this! Just a few feature requests/nitpicks:

• It appears that you can no longer get the fault/warning bit field without shims, which is a regression from last season. Is there a chance this could be added back, possibly just as another field of the Fault/Warning classes? It was useful for logging as it could easily be packed into a WPILib struct.

• MAXMotion looks really promising, however I am weary of the lack of feedforward types other than kV. For some applications, the current interface is mostly sufficient (static kG for elevators, and kS for any mechanism, is easy enough to implement on the RIO and pass via arbFF) however mechanisms like arms could benefit from an on-controller cosine based kG, and kA may also be useful in some cases. Are there any plans to implement more on-controller feedforward gains, or will kV remain as the only available option for this season?

• Last season, absolute encoder velocity by default would return rotations per second, unlike the motor's internal encoder which would return RPM (this inconsistency was undocumented). It appears that the documentation for the absolute encoder's config has been updated to now specify that the default base unit is RPM, however I didn't see anything in the change log specifying this fix. I haven't gotten the chance to test this yet on a real robot, was the documentation change an oversight that occurred while migrating the configuration API, or was a fix actually implemented?

[viggy96]

I'd say it would also be kinda weird in the code to use the WPILib feed forward for the other values, but REV feed forward for kV.

[jfabellera]

Thank you for the feedback!

• I definitely see how that can be useful, and we can add a bit field for that. We've had some reports that the short being returned by getFaults() wasn't very usable for some, so that's why we moved to returning a struct of human readable faults.
• I can't make any promises about adding additional feed forward gains this season, but it has definitely been on the list.
• The original documented RPS was actually a mistake in documentation. The native units returned by firmware was RPM, and it is still RPM. No changes to firmware have been made for this, only documentation.

[viggy96]

What is the recommended way to do feed forward then for various mechanisms? Assuming MAXMotion is doing motion profiling, having the controller do feed forward calculations is kinda necessary, as the feed forward is based on the setpoint, and a motion profile usually works by continually changing the setpoint. So the controller should be able to calculate not only the correct setpoint to apply given the motion profile at that point in time, but also the correct feed forward for that setpoint.

[narmstro2020]

Throwing another voice in the mix.
Doing a mix of RIO feeforward math and controller feedback math is not really ideal. You can only add the arbFF every 20 ms from the RIO while the feedback calculations are running faster than that.

Question: Can the controllers support ks, kg, kv, and ka with a cosine component for arm?
It seems like kV exists as kFF (which is really confusing and should be renamed to kV like every other usage in FRC).

Also if I'm not mistaken kFF doesn't quite act the same way as WPILIB's kv and either needs to behave more like WPI or the usage (especially with SmartMotion) needs to be documented.

[narmstro2020]

And I don't want to beat a dead horse on this topic, but the intended use for set Reference is to set when you need to set. Sure you could do constant sets and most likely will, but if you have a separate RIO led feedforward then you have to bury it in a periodic loop anyways. Where as integrated it into setReference would keep the same set it and forget it nature (though of course at the millisecond level

[viggy96]

Is there an easy way to query the Spark for a config setting? I don't see any getters to that effect.

Also, I've previously had a "SparkHelpers" class that had an identical package to REV's SparkBase so that I could create certain getters that weren't available via normal API, as well as some other methods that weren't expose either. Some examples include, getting smart current limit values (current, chop cycles, RPM), enabling/disabling center aligned mode on the absolute encoder (this allows the sensor to report -0.5 to +0.5 instead of 0.0 to 1.0), and converting the Spark fault bits to a human readable string.

https://github.com/lasarobotics/PurpleLib/blob/master/src/main/java/com/revrobotics/SparkHelpers.java

Are all those methods now exposed via the normal API?

[BR88C]

Is there an easy way to query the Spark for a config setting? I don't see any getters to that effect.

• Spark MAX
• Spark Flex

[jfabellera]

I believe the only one from the list that is not in the API is support for the center aligned mode for absolute encoder. There were reasons why we didn't include it in the past, but I believe we should be able to add that now. Let me know if I am missing anything.

[viggy96]

What about the following method?

  /**
   * Enable mode which sets the output of the PID controllers to be voltage instead of duty cycle.
   *
   * <p>To disable, change or disable voltage compensation. Those settings will overwrite this one
   */
  public static REVLibError enablePIDVoltageOutput(CANSparkBase spark) {
    return REVLibError.fromInt(CANSparkMaxJNI.c_SparkMax_SetParameterUint32(spark.sparkMaxHandle, 74, 1));
  }

[NoahStricker]

Is the using the config accessor a blocking call?

[jfabellera]

Is the using the config accessor a blocking call?

Yes, as documented in the Javadoc

[viggy96]

Is there any methodology to tuning MAXMotionPosition/MAXMotionVelocity that teams should be aware of? Will a high level explainer of REV's implementation of this algorithm be released?

[jfabellera]

A docs page for MAXMotion will be added before kickoff and will include a tuning guide and best practices.

[viggy96]

How do I change the status frame period in the new API?

[jasondaming]

It’s part of the signals config. Some signals are always enabled (status frame 0) and most are disabled by default. Calling any getter will automatically enable the associated signal, so in most cases no manual configuration is required. The config object lets you force enable any signal and customize periods.

Shamelessly stolen from @jwbonner when he answered my question about the same thing.

[viggy96]

What are the default measurement period and average depth for the internal motor encoder for a NEO Vortex and NEO 1.1?

I believe the optimal values that we previously found were 32ms/8 for a Spark Flex, and 16ms/2 for Spark Max.

[jfabellera]

The defaults are documented in the EncoderConfig javadocs.

One thing to note is that the NEO Vortex uses the quadrature set of parameters and the NEO 1.1 uses the UVW set of parameters.

[viggy96]

Can these defaults be changed to reflect the more optimal values?

As it stands these settings have to be changed in order to achieve better velocity PID performance, especially on the NEO v1.1.

And less experienced teams won't know to do that.

[viggy96]

Those values I specified were found by former REV employee Will Toth by the way.

[viggy96]

Will the API provide detailed feedback on which portion of a config failed to apply if any?

[jfabellera]

Errors including parameter set failures should be logged to the driver station with the parameter that failed.

[viggy96]

Is there a toString() method for a Faults object?

[jfabellera]

Not at the moment.

[viggy96]

Is there a method to automatically reset the motor encoder or external encoder to a user specified value upon connected limit switch activation?

Right now we can technically do this by creating a Trigger, but it would be best if this functionality was in the Spark firmware itself and its internal control loop.

[jfabellera]

Not at the moment. We can add this to our backlog.

[viggy96]

Can SparkModel and getSparkModel() be made public?

[jfabellera]

We will add it to the list.

[viggy96]

Is there a method to limit supply current?

[jfabellera]

There is no input current limiting at the moment.

[Liam-Stow]

Is the spark simulation expected to work in C++ yet? I have tried the bare-bones test below and this does not change the position or velocity of the spark in sim gui at all. The equivalent code works in Java.

void Robot::TeleopInit() {
  _sparkFlex.Set(1);
}

void Robot::SimulationPeriodic() {
  double rpm = 100;
  double batteryVoltage = 12;
  double deltaTime = 0.02;
  _sparkFlexSim.iterate(rpm, batteryVoltage, deltaTime);
}

[jfabellera]

REVLib beta 3 and SPARK MAX and Flex pre-release 7 have been released

• REVLib 2025.0.0-beta-3
• Instructions for updating Hardware Client and firmware

Warning

After updating your SPARK MAX or Flex to 25.0.0, you will be unable to communicate with them through the Hardware Client if using a Power Distribution Hub or Pneumatic Hub as the USB CAN bridge. A later update will be available for the PDH and PH that fixes this. For now, connect to the Hardware Client via a SPARK MAX or Flex.

Issues fixed

• SparkMaxRelativeEncoderSim and SparkMaxAbsoluteEncoderSim do not default Position and Velocity Conversion Factor to 1 (as documented) #44
• SparkMaxRelativeEncoderSim iterate not properly setting the position #43
• Spark Max rapidly oscillates output when switching control types #41