New Info on 2009 control system, maybe

[Kevin Sevcik]

Just got out a training seminar at NI HQ a little while ago, and I thought I'd note the interesting bits I saw. I don't think I've seen this in any of the online documentation, though much of it seems to be in the docs on the install DVDs. Much of the detailed info seems to be in the C/C++ programming guide, so this might be more benefit to those not getting an early kit. Also, keep in mind that all this is without considering the Rules. Just because it's possible doesn't make it legal. Also, just because the GDC declares something legal doesn't make it possible. Rambling commences:

Ribbon cables work well as the digital sidecar cables. It's possible to use ribbon cables to move the analog breakouts off the top of the cRIO, though they're obviously unshielded and more noise prone.

There are currently no provisions to power the Driver Station off the robot, though it's being investigated. I think it's pretty certain there won't be any power over the ethernet tether cable, so any power solution will be a separate cable.

Some joysticks Just Don't Work with the DS. Joysticks that actually need custom drivers (not configuration software) to work just in Windows are highly unlikely to work on the DS.

DS digital inputs require active termination. You can't simply close a switch to ground or 5V and assume the internals will pull it the other way when the switch is open. You need your own pull-up/down or need to connect it to 5V or ground.

We were warned not to try powering the Power Distribution board from a bench power supply (in lieu of a battery). Given the switching power supplies on the PD board, it could cause entrainment and subsequent power supplies meltdowns.

They're rapidly researching the effects of running the practice field wireless on the 2.4GHz band to make practicing safer. They need to determine the effects on any FTC competitions.

E-Stops now latch in the robot controller. This means after E-Stopping, you MUST reset your robot for it to function again. This would be to increase safety so an E-Stop robot can only be dangerous again after the robot was power cycled, not after the field is cycled.

Random: Jaguar powered CIMs are quiet. To my "mosquito" ringer deaf ears, anyways.

All the cRIO modules are technically optional. Removing unused ones won't affect the operation of others. We were told the analog module in slot 1 would be required by rule, as it will be used to monitor battery voltage for field troubleshooting.

There currently isn't any method for displaying custom text on the DS LCD. Chances are, you're going to want a dashboard PC. Also, in non-competition mode, the dashboard pc can receive camera images and communicate directly with the cRIO for debugging. Neither will be the case in competition mode. The former for bandwidth concerns, the latter for safety.

The image processing in Labview looks pretty powerful for machine-vision tasks. Also, there's a Labview Vision Wizard program to help you develop your image processing algorithms on a PC, so you can parallel vision development with control code development. Also, you can probe images on wires while the program is running. Let me repeat that. You can probe the results of your image filtering and processing algorithms at any point you choose in the process. With a simple right click of a mouse. In near real time.

Finally, a list of the FPGA/WPILib supported special peripherals, as best I can recall them. And all of this is without costing any processing power:
four, 4x quadrature encoders, with pulse width measuring
eight, 1x or 2x quadrature encoders or counters, with pulse width measuring
two, analog signal accumulators, which means two analog gyros
two, I2C buses (with two sidecars)
two, SPI buses (with two sidecars) plus chip selects (on Dig I/Os)
serial communications through the cRIO serial port

And that's all the rambling I think I can manage tonight...

[synth3tk]

I'm not one of the team mechanics/programmers, but I thoroughly enjoyed this rambling*. Good read!

* It could also be attributed to my wanting to play with our new system soon....

[Alan Anderson]

Originally Posted by Kevin Sevcik

Ribbon cables work well as the digital sidecar cables. It's possible to use ribbon cables to move the analog breakouts off the top of the cRIO, though they're obviously unshielded and more noise prone.

Using ribbon cables would lower the maximum height of the cRIO. Replacing the provided digital cable with a ribbon cable would save some weight.

There are currently no provisions to power the Driver Station off the robot, though it's being investigated. I think it's pretty certain there won't be any power over the ethernet tether cable, so any power solution will be a separate cable.

Or you can roll your own simple, unprotected, non-robust pseudo-Power-Over-Ethernet scheme with a pair of breakout adapters at the ends of the ethernet tether. I am definitely planning to do this.

We were warned not to try powering the Power Distribution board from a bench power supply (in lieu of a battery). Given the switching power supplies on the PD board, it could cause entrainment and subsequent power supplies meltdowns.

Where did you get this warning? I don't remember hearing it in the main sessions, and Erik didn't mention anything about it during our somewhat lengthy discussion of the board.

They're rapidly researching the effects of running the practice field wireless on the 2.4GHz band to make practicing safer. They need to determine the effects on any FTC competitions.

The intent is to run the competition fields on the 5 GHz channels only.

Random: Jaguar powered CIMs are quiet. To my "mosquito" ringer deaf ears, anyways.

They sounded just like Victor-controlled CIMs with a good gearbox to me, but I didn't intentionally try to give them too little power to get turning.

All the cRIO modules are technically optional. Removing unused ones won't affect the operation of others. We were told the analog module in slot 1 would be required by rule, as it will be used to monitor battery voltage for field troubleshooting.

I think the digital module in slot 4 will be required too. Its associated sidecar provides the drive signal for the large amber signal light.

[Kevin Sevcik]

Alan,

Thanks for the corrections/extra info I missed. Also:

Originally Posted by Alan Anderson

Or you can roll your own simple, unprotected, non-robust pseudo-Power-Over-Ethernet scheme with a pair of breakout adapters at the ends of the ethernet tether. I am definitely planning to do this.

I was considering this as well, though the FRC legality of it is obviously unknown at this point. I hadn't brought it up because I hadn't looked up the ampacity of 24 AWG as would be in Cat5. Since it's 2.1A enclosed, and we have two pairs to work with, I think it'll be relatively sane. Provided you make certain to keep it the 12V away from the actual ethernet ports....

Originally Posted by Alan Anderson

Where did you get this warning? I don't remember hearing it in the main sessions, and Erik didn't mention anything about it during our somewhat lengthy discussion of the board.

At the risk of digressing into a personal conversation, I swear I heard Chris (Jennings?) mention this early in the meeting. I can recall a fair amount of the details. Someone brought up the point that the early kits hadn't shipped with any batteries, and any rookies wouldn't have any way to power up their new controller. It was asked if they could use a bench power supply in lieu of a battery, and Chris mentioned that they'd done their testing with batteries to avoid any possible interactions between the power supplies, plus it was more realistic. (Warning is perhaps a bit strong for this) It was then noted that they assumed any rookie teams actually motivated enough to grab an early kit would be capable of buying or bumming a battery.

Now, I'll grant you that Chris isn't the expert on the PD board that Eric Van Wyk is, so he could have been extremely overcautious. I personally think the actual likelihood of a good modern bench supply generating entrainment with the PD board is very low, but I felt I should recount what I'd heard. If someone more knowledgable would like to allay any doubts about powering the system with a bench supply, I'd certainly welcome it.

[Tom Bottiglieri]

Originally Posted by Kevin Sevcik

Jaguar powered CIMs are quiet. To my "mosquito" ringer deaf ears, anyways.

15 KHz doesn't hurt. Should make for some quiet low power drive, which has always been a huge annoyance.

[Abrakadabra]

I have a 10AMP 13.8VDC power supply (an older Radio Shack model). When I hook it up to the new PD board, I get nothing. This supply worked fine with the IFI controller and the old power distribution system (for driving only the low amp motors - e.g. Globes - of course).

Does the new PD have a minimum amperage it needs to see before it will even power up, possibly because of the needs of the additional power supplies on the new board?

[Cory]

100, 254, and 668 have been using ribbon cables successfully on our beta unit since nearly the start of the test. We also moved the bumper boards off the top of the cRIO-they seem far too liable to being broken if they remain there.

[Joe Hershberger]

Originally Posted by Kevin Sevcik

Finally, a list of the FPGA/WPILib supported special peripherals, as best I can recall them. And all of this is without costing any processing power:
four, 4x quadrature encoders, with pulse width measuring
eight, 1x or 2x quadrature encoders or counters, with pulse width measuring
two, analog signal accumulators, which means two analog gyros
two, I2C buses (with two sidecars)
two, SPI buses (with two sidecars) plus chip selects (on Dig I/Os)
serial communications through the cRIO serial port

One inaccuracy here. The SPI engine is routable to any DIO pins on either module... there is only one of them. The hardware supports streaming with 512 word buffers on both the send and receive side and buffer status interrupts (such as half full), but that is not implemented in the library yet. WPILib for LabVIEW supports single sample SPI transfers and WPILib for C++ has no support. This is a low priority system. Right now it seems there is little demand for it.

[EricVanWyk]

Originally Posted by Kevin Sevcik

Alan,
I personally think the actual likelihood of a good modern bench supply generating entrainment with the PD board is very low, but I felt I should recount what I'd heard.

Kevin -

You are correct, using a benchtop power supply is not recommended. However, it is safe. Some revisions that never saw the light of day would become unhappy when fed with a wimpy power supply. The revs that are shipping can survive with any input supply with a nominal voltage between -20 and 20V with any supply impedence, and can survive any load impedence that does not generate power. Don't expect it to work over this range: the range of ~3V to -20V is guaranteed not to work.

It is a "feel free to do it, it just might not work" situation. The reason for this is that the switching power supplies effectively have a "negative resistance". For a given power out, lower voltage in requires higher amperage in. This can interact with the power supply's internal protection, creating two stable points:
1) The good one, in which you are getting 12V and low current from the bench.
2) The bad one, in which you have hit the current limit of the supply, and have constant current at a low voltage. If the power out of this isn't beefy enough, it will latch in this state and won't work. Nothing will be damaged.

A kit battery doesn't really have a state 2, because it can push *way* more current that the power supplies could ever want or use.

Also, the momentary loads of starting a CIM or other big motor can easily brownout most bench top supplies. You will (briefly) take 10-100 amps from the supply. A supply that can handle that is impressive. I'm willing to bet that most supplies available to teams won't handle it. You will trip the brownout protection in the PD and DSC, which will disable the drive motors momentarily (in order to protect the cRIO and Wireless). When they are disabled, the gargantu-load of the motors will go away, which will allow the voltage to rise, which will enable the motors, which will repeat the cycle. Again, nothing is damaged, but it isn't working.

Neither of these situations occur with charged a kit battery. However, a battery in its death throes will exhibit the limit-cycle behavior described in the preceding paragraph. Don't worry, this happens after the battery has been drained beyond usability. In my testing, this occurred after the battery was unable to provide enough power to move a kitbot.

Long story short:
Don't use a bench top supply.
If you do use a bench top supply, don't complain that it doesn't work.

Originally Posted by Kevin Sevcik

Ribbon cables work well as the digital sidecar cables. It's possible to use ribbon cables to move the analog breakouts off the top of the cRIO, though they're obviously unshielded and more noise prone.

On the other hand, the cables that connect the sensor to the breakout aren't shielded either. Yes, adding a ribbon cable adds more opportunity for noise to get in. However, it might also reduce the length of the 3 pin connectors.

6 of one, half dozen of the other. For some teams, keeping it attached the cRIO makes more sense. For other teams, remoting it with a cable will make more sense. Some teams might even want to use a shielded cable. You (generic You meaning each team) won't know until you are designing the specifics of your robot. I won't pretend to know what a team needs better than they do, so I'm happy that these connectors provide more options.

However, you probably don't want to use a ribbon cable (unless you do).

Lastly, please review the rules before selecting an alternative cable. I am expecting that the connection to the Side Cars will be required to have all 37 pins connected straight thru. We'll find out soon!

[Mike Soukup]

Originally Posted by Kevin Sevcik

There currently isn't any method for displaying custom text on the DS LCD. Chances are, you're going to want a dashboard PC.

Ridiculous.

We used the 1 byte display on the IFI controller for quick pot calibration in the pits and at our practice field and for feedback about our autonomous programs. Now our drivers will have to hook up a laptop instead? Bad move FIRST.

[EricVanWyk]

Originally Posted by Mike Soukup

Ridiculous.

We used the 1 byte display on the IFI controller for quick pot calibration in the pits and at our practice field and for feedback about our autonomous programs. Now our drivers will have to hook up a laptop instead? Bad move FIRST.

I'm not satisfied either - I hope it changes soon.

However, I'd like to point out that for doing pot calibration, you will already have your programming laptop connected - it will be easier to probe the variable on the same computer that you will then use to modify your code.

[AdamHeard]

Originally Posted by EricVanWyk

I'm not satisfied either - I hope it changes soon.

However, I'd like to point out that for doing pot calibration, you will already have your programming laptop connected - it will be easier to probe the variable on the same computer that you will then use to modify your code.

He is probably referring to using it to calibrate the pot by mechanically moving it, or having a button reset a value stored in the eeprom. Many teams have done things like this in the past for the very reason of not wanting to download code or hook up a computer.

[EricVanWyk]

Originally Posted by AdamHeard

He is probably referring to using it to calibrate the pot by mechanically moving it, or having a button reset a value stored in the eeprom. Many teams have done things like this in the past for the very reason of not wanting to download code or hook up a computer.

Got it. Good point.

[Dave Flowerday]

Originally Posted by Mike Soukup

Ridiculous.

We used the 1 byte display on the IFI controller for quick pot calibration in the pits and at our practice field and for feedback about our autonomous programs. Now our drivers will have to hook up a laptop instead? Bad move FIRST.

Sounds like the driver's station is a real mess. Check out the thread that Joe Ross posted on the FIRST forums. When a joystick is unplugged, the controller keeps sending the old value! Clearly, they were more concerned with whiz-bang features on this system than basic safety. As a counter-example, IFI design IN HARDWARE a mechanism that if a joystick was unplugged, the output went to a safe value (so you know they were thinking about it from the outset). I grow increasingly concerned that safety seems to be an afterthought with the new system. We're dealing with such dangerous machines here - many people trust the control system to keep them safe. Joe mentioned a 2-second lag that sometimes occurs. Does this affect the robot disable signal as well?

Scary stuff. I once again suggest that everyone keep their bodies well clear of any robot powered by the new system whenever it is turned on. No more trusting the disable switch until the new system proves it is safe (which it clearly isn't right now).

To say I'm disappointed that they'd ship a system with these types of safety defects is an understatement.

[AdamHeard]

Originally Posted by Dave Flowerday

Sounds like the driver's station is a real mess. Check out the thread that Joe Ross posted on the FIRST forums. When a joystick is unplugged, the controller keeps sending the old value! Clearly, they were more concerned with whiz-bang features on this system than basic safety. As a counter-example, IFI design IN HARDWARE a mechanism that if a joystick was unplugged, the output went to a safe value (so you know they were thinking about it from the outset). I grow increasingly concerned that safety seems to be an afterthought with the new system. We're dealing with such dangerous machines here - many people trust the control system to keep them safe. Joe mentioned a 2-second lag that sometimes occurs. Does this affect the robot disable signal as well?

Scary stuff. I once again suggest that everyone keep their bodies well clear of any robot powered by the new system whenever it is turned on. No more trusting the disable switch until the new system proves it is safe (which it clearly isn't right now).

To say I'm disappointed that they'd ship a system with these types of safety defects is an understatement.

I agree wholeheartedly. I fear people will assume that because the safety and disable features of the IFI system were so robust and worked so well, that this new system must work the same way (I mean, all control systems must be the same, right? ).

I for one won't be trusting the robot nearly as much as before.