Servo Mechanisms

[Mark Bresler]

Hi,
this is my first time using servos and am getting little bit confused. I am trying to interface a HiTec HS-422 servo to my IsoPod board. I am under the impression that midpoint of rotation is a 1.5 ms pulse every 3 ms, with a range of motion corresponding from 1 to 2 ms. Therefore I put entered the following line, aiming for 333 Hz :

7508 PWMB1 PWM-PERIOD

using the following I am able to change rotation:

HEX nnnn PWMB1 PWM-OUT

Where nnnn is a hex number, my results are below
Hex F000 max CW loud buzz
Hex D800 max CW soft buzz
Hex 2800 max CCW soft buzz
Hex 0000 max CCW loud buzz

The servo is also getting warmer than I think appropriate for no-load situation. What am I doing right and what am I doing wrong?
Any help would be appreciated.
Thanks,
Mark

[Mark Bresler]

Hi,
This is a repeat of something I posted Thursday but have not seen in the Forum.
I am am away from my office so I do not have as much information. I am trying to operate a small Hitec analog servo with the IsoPod. I am assuming that midpoint is a 1.5 millisecond pulse every 3 milliseconds, with full range varying from 1.1 milliseconds to 1.9 milliseconds. I used 7508 (decimal) as data for PWM-PERIOD, trying to get a 333 Hz frequency. I was able to move the servo by sending numbers from 2800h to E000h. The servo would always buzz, louder if I went outside these limits. and got warmer than I thought normal for a no load situation.

What should I be doing differently?
Thanks Mark

[RMDumse]

I am under the impression that midpoint of rotation is a 1.5 ms pulse every 3 ms, with a range of motion corresponding from 1 to 2 ms.

Here's your problem. RC Servos pulse widths are 1.5 mS nominal for center. However, since the purpose of RC Servo is to be able to multiplex several channels of pulses up a single radio channel, the repetition rate is NOT 3 mS. The repetition rate should be around 20 mS.

Without slowing down the processor, we can repeat in 13 mS using a full width delay. Therefore we suggest the following line, aiming for 67 Hz :

32767 PWMB1 PWM-PERIOD

Then using:

HEX nnnn PWMB1 PWM-OUT

Where nnnn is a hex number,
5000 is min (1mS)
7500 is mid position
10000 is max (2mS)

Note, by careful experimentation you can go a bit above and below 1mS and 2mS until you hit the stops, and there, a buzzing will begin. If it is not backed off, the gear train may tear up. But you can generally extend the range beyond 90 degrees, up to (usually) 170 degrees or in some servos more than 180 degrees.

[RMDumse]

I am am away from my office so I do not have as much information.

Sorry Mark, I too was away from the office as well, up working in Chicago at Harpo Studios on Oprah's Panavision Supertechno movie crane. It has IsoPod(TM)s in it.

We're running a bit shorter handed than we like to now in the office. So my response was a bit delayed.

Try the

32767 PWMB1 PWM-PERIOD
7500 PWMB1 PWM-OUT ( 1.5 mS * 5000 for mid point

and I'll bet you're much happier with your results.

[RMDumse]

Then using:

HEX nnnn PWMB1 PWM-OUT

Where nnnn is a hex number,
5000 is min (1mS)
7500 is mid position
10000 is max (2mS)


Oh. Appologies. I was following your example and missed tha you were showing hex values. All the values i used were decimal. so

DECIMAL nnnn PWMB1 PWM-OUT

Where nnnn is a decimal number,
5000 is min (1mS)
7500 is mid position
10000 is max (2mS)

( 1mS * 5000 = 5000
( 1.5mS * 5000 = 7500
( 2mS * 5000 = 10000
( Because the system clock runs at 5MHz or 5000 cycles per mS

[Mark Bresler]

Thank you,
I was wondering what happened, I though you might have had a computer blip, I usually get very prompt responses to my questions. I will try to implement your solution Monday when I return to my office. Are you going to be able to post more information about what you did at Harpo Studios?
Mark

[RMDumse]

Sure. Since you ask, I'm happy for a chance to brag.

Up Wed. eve., up Thurs. at 6AM to get to the set, worked till ~11Pm, back, up Fri. at 5:30 to make flight home. No time for anything but the job at hand.

Panavision had me work on their controller on their movie crane out at Harpo Studios on the Oprah set. This control system is called Back Pan Plus. What the 'Pod's are doing is reading the movements of the grips (guys who move the crane with the camera around) and coming up with compensating commands, that are inserted into the existing camera operator controls, so when the crane moves, my system generates back commands to keep the camera pointing at the same spot in space.

However, whenever the operator wants to move the camera to another point of interest, actor, or what not, he just uses the joystick to change what is the center of attention, then the crane tracks it instead.

The system greatly eases the camera operators job, allowing them to do the artistic framing and pointing of the camera, while the grips move the crane from position to position.

Panavision had been working on this problem for about five years when they brought me in. Up 'til then, they'd been working on essentially a motion control system, but this wasn't flexible. If an actor did anything "creative" or was off his time beat, etc., they'd have a bad shot, or in the case of the movies, take off the $400 canister of film and throw it in the trash and try again.

Using IsoPod(TM)s and IsoMax(TM), the system I came up with for them in less than a month's total work, uses calculus-based relative motion, so the actors can change what they're doing mid shot, and the camera man can still track them. Plus the grips and the camera man don't have to be in absolutely perfect sync to keep the motion fluid.

It's pretty impressive overall. In essence, the system is doing full inverse kinematics to figure out from the camera operator's commands just where he is looking, and then uses forward kinematics as the grips move the crane body and thereby the camera to generate back pan and tilt to keep the camera tracking the spot the operator had picked. All this is running on a 'Pod running the interactive IsoMax(TM) language.

I went out for them, and helped show the crane for the 2005 Cinegear show. I don't remember them coming by, but I was told Oprah's people saw it on display then, and felt they had to have it.

Oprah has had her crane about two years now, and I'm told she loves it. This lets them do some pretty creative shots never before done on live TV. If you are looking at an Oprah show, if you see a long sweeping shot looking like there's a camera just floating around over the crowd, but still staying pointed at the center actors, that's probably the crane.

We were there this past week chasing some phantom problems, where the tracking would cut out suddenly, we hope we've caught. There was a watchdog that was preventing too aggressive moves by the system, and we think after having it a while and getting used to what it could do, the operators were pushing the limits. So we opened out the limits on the watch dog so they could do more without making the watchdog "suspicious".

[Mark Bresler]

I just tried out your settings, they work like a charm. I am able to have 180 degrees of movement by varying the numbers from 2400 to 12600. No buzzing, no getting warm.

On the camera crane, is one person running the camera and some axes of movement, and a second operating other axes. Is it too much for one person to run?

Thanks again
Mark

[RMDumse]

On the camera crane, is one person running the camera and some axes of movement, and a second operating other axes. Is it too much for one person to run?

Yes, the good old days of the camera man on the end of the crane arm are gone. Today he sits in a little tent, and may not even see the actors (line of sight). He has the two original controls he had when he sat at the camera, a pan wheel and a tilt wheel. Just running those two knobs is very much a full time job. It's so hard to do, a good camera man make hundreds of dollars a day (typ. $800). Yet he is still in charge of the entire movement of the crane. So he has one or two "grips" who work for him and they do the boom, swing and telescoping of the crane.

So it really is a trick now days to run the camera. Seat of the pants feeling from being on the moving crane is all gone. So now the cameraman and the grips have to practice and practice to get their moves coordinated before they even bring the actors out. Then they bring the actor out. Usually they bring in the stand in who stands in while they practice. Then they bring in the real actor. Now if the actor does the same thing every time, they've got it. But of course, actors are known for their egos and instability, so when the actor "gets into his art" and the camera and crane crew get surprized, they throw $400 cans of film in the trash over and over.

So the system I did for them, takes lots of these problems away, and so movies and television programs can be short more efficiently with less waste of film and/or time.

Now the cameraman and his one or two grips can make good shots on the fly, with high efficiency, and few errors.

Oprah is one of the first live TV programs to see the possibilities. But then, that's probably why Oprah's camera crew won an Emmy last year... although I'm told they didn't specifically mention the crane...