I have three legs fully assembled now. I haven’t had nearly the time I would like to work on it. Midterms are over now, and spring break is only a day away though, so that should change soon. The build process has been slowed down considerably by my oversized servos. I tried several methods before settling on the method I finally decided to go with. That method has taken me a while, because I’ve had ot cut and drill spacers for every servo, as well as modifying the servo cases themselves. Hopefully I’ll have time to finish assembling it tomorrow.
The good news is that I got a new table, and have transferred all my tools and parts to it, and off of my computer desk. For the first time in about a month, I can see my desk again. It’s a nice feeling, and I think I’ll be more productive this way.
I also started putting together a website sometime last week or so, which I will be tracking my progress with. This should be a bit more convenient for me, as it will save me from having to type things twice, and it should also save Jim’s bandwidth, as I intend to post a lot of pictures, code, and anything else that is relevant. I’ll post abstracts here for those of you who don’t have the time (or desire) to read the full posts. I’ve still got a bit of work to do on it before it is ready, so the link won’t be added to my profile for another day or two, if not more.
The battery on my camera ran out while I was taking pictures of how to modify a Blue Bird servo to fit the LynxMotion brackets (details available on my website). Unfortunately, that is when I found out my charger no longer worked. A new one is on its way to me now, but will not be here for another week, so I don’t have any pictures of the completed robot yet.
My development diary can be found on my website, mike-winters.com.
I am still having power issues. I know what the problem is (servos draw so much current that the volts drop out), but I have not yet been able to solve it. It has been suggested that I look into using 18 large capacitors, one for each servo, to fix this problem. Anyone have any comments or ideas?
Well, the power snag has turned into a pretty serious issue. Nothing I have tried is getting all 18 servos to move rapidly in groups, as they will need to do for it to walk.
I am out of ideas at this point. I don’t know very much about electronics (very little, in fact), so I am totally at a loss as to what to do to fix this problem. For development and testing, I need to be able to power them off a wall supply. If my 250W computer PSU (discussed here) won’t cut it, what will?
More importantly, however, I need to ensure that whatever workaround or solution I go with now will also work when I start running this thing off batteries. Currently, the best batts I have been able to find are these. I plan to do as Andy has done, and step them down to 6v/5v. I figure I can fit at least 2, and possibly as many as 3 or 4 of these on my robot. The problem is that I am pretty much broke at this point, having just purchassed the parts for the hex, as well as the servos, so I won’t be able to afford these batteries and the associated charger, etc for at least a few months.
I really need help at this point. I am not a hardware guy, my focus is software. I need to get the hardware issues solved so I can start writing code. I’m out of my element here, and out of ideas. Any help you guys can provide will be greatly appreciated.
i am not sure if you found that thread. there is some discussion about power supplies as well. it might be of help…
However: i am just at the same problem as you: i fused my 5V PSU from an old computer PSU with (4amps, i think) and plan to use this for development. However: i am still waiting for my parts (and servos) to give it a try.
as for autonomous action, i have recently bought a 7.2V 3000mAH NiMH battery pack. hopefully, this is enough…
My problem is that I am past the point of thinking/hoping something will work. I have now hit the point where what I have DOES NOT work, and I am about out of ideas on how to fix it. As suggested by someone doing a similar project, I am going to try picking up some 1000uF capacitors today and see if they will be enough to prevent these brownouts from occuring.
Hey Mike, I’m not sure why you are still having trouble. I routinely control 18 HS-475 servos form a little bitty six cell “AA” NiMH pack. The pack is good for 1600mAh and I can run an EH3-R for over 20 minutes walking continuously. And this is without the separate 9vdc battery for the electronics. There is absolutely no reason you can’t control 18 servos from this battery packs big brother. Our 7.2vdc 2800mAh pack should control those servos with no trouble. If a PC powersupply is not capable of controlling these servos then you have another problem. I suspect wiring. What guage of wire are you using? It has to be 18 or 16 guage to carry the current. Also connections have to be solid. No wire twisted together, or little clip connectors. Think heavy duty here… I sense the frustration in your posts. Send images to me, or the forum showing your setup. We want to help…
I’m wishing I had just purchassed it all from you to begin with. You were right when you said those initial problems I had with the Blue Bird servos were just the tip of the iceberg.
I’m using 18ga wire, and the connectors are two pin molex style connectors I purchased from Radio Shack. The wires are crimped and soldered into them.
Unfortunately, my digital camera has a dead battery, and the charger stopped working (power issues are plaguing me, it seems ). I have a new charger on the way, but it won’t be here until the end of the week.
I’m going to try the capacitors later today, and will post my results.
Thanks,
Mike
Edit: Jim, I also don’t know why I’m still having problems. I figured the computer PSU would be plenty. I may look in to getting another supply, in the event that this one is faulty or simply inadequate (it was pretty cheap).
I just did another test. After carefully putting each servo to center one by one, I put it on the table (it was supported during the centering process), and the voltage dropped from ~5.3 to ~4.75v. Lifting one leg dropped it down to ~4.6v (one leg in the air, five on the ground). Trying to lift a second leg caused it to get twitchy again. I wasn’t metering it when it happened, so I don’t know what the drop was, but whatever it was, it was large enough to cause the servos to act up.
I’m off to try to find some capacitors now. We’ll see what I can come up with.
Mike, are you talking about continuously holding the voltage at 4.6 volts, or just it dipped down to that while the leg was moving? If the voltage is holding steady while under load then it is unlikely the capacitors will entirely solve your problem. They will help with spikes or surges of current by, in effect, buffering it out some but if you are sustaining the 4.6 volts under load the problem is probably not spikes. They might buy you some margin before you start seeing flaky behavior though, particularly if the voltage is dragged down and then you start moving a leg.
Question: Where did you take the voltage measurement? If not AT the pack then is it possible to measure the voltage AT the pack when loaded like this? I am asking to determine if the drop in voltage is from the packs internal resistance or the wiring between the pack and the servos. Also if you could actually measure the current coming from the pack when under this load it may be helpful to know as well. I am looking to see how hard you are drawing off the pack. Be aware this could easily be upwards of 10A with 18 servos under load so don’t pop the fuse in your meter, maybe start by supporting the robot and observe as you place it on its feet, and remember that the current meter will drop a small amount of voltage itself so flaky might happen sooner with it in place.
It was holding pretty steady. Small fluctuations, but only a few millivolts or so.
I measured the voltage directly off wires attached to the molex connector, and observed the same results. Generally, I am measuring off the screws on the SSC-32 terminal, because that is easier, and appears to be yielding the exact same numbers with everything I have tested so far.
I tested the current draw a few days ago under several conditions, and I think the number was about 3.5 amps for 6 servos. The meter did affect the performance, however, as the servos got twitchy at times when doing this, but when the meter was not connected in series to measure the current, the servos perfromed as they should.
I just got back from picking up a few caps, and I am about to wire them up now. Hopefully they help to solve the problem. If not, I don’t know what I am going to do.
You’re right again, Jim. The capacitors didn’t seem to help, although I only tried a few of them.
After that, I tried a 7.2v 2000mAh battery pack, and it seems to be more along the lines of what I’ll need. With this, the robot was able to stand on 6 legs, then pick 3 up and stand on the other 3, while keeping the 3 hovering legs in motion. The computer PSU could not handle standing on 3 legs with the other 3 in the air.
I did some slow moves (macro walking gait tests) and everything seemed to be working fine. Unfortunately, I think I may have cooked two of the servos. One happened earlier on, the second happened sometime during the move tests, but I am not sure when. I’m about to go to bed, and will have to figure out what the story is on those two servos tomorrow.
So, in addition to all the other problems associated with Blue Bird servos, it seems they are not as tolerant of over volting as HiTecs are. Hopefully tomorrow I will be able to figure out exactly what happened, and possibly even fix it (although I doubt it).
After ordering two Hitec 645MG servos to replace the cooked BMS servos, I have made another small update to my website, now with pictures of the assembled robot.
I’ll probably be focusing my time in the near future on the control software side of things. I have a lot of work to do on the sequencers and customizeable GUI before that is ready for release and beta testing.
I checked out your site. I stumbled onto the images of the blue bird servos. The image below I found interesting. mike-winters.com/Servo%20Photos/BMS%20660MG%20hexagon%20drilled%20close-up.jpg
I’m the reason the hex part was put into the mold. We tried to get a workable solution to the size problem, but they fumbled the servo horn part, and I could never get them to address the self destruct feature of the servos.
Yeah, the servo horn thing was a problem, until you pointed me to the Futaba horns, which (so far at least) have been working perfectly. Another nice thing about them is that they are black, and so look a bit nicer (but I’m much more concerned with functionality than appearance).
I have not yet encountered the self destruct “feature”, and hope not to, but it certainly wouldn’t surprise me if one day they all just explode and shower my cats with servo shrapnel. I may start switching over to Hitecs as these break or burn out, but hopefully that won’t happen.
With a bit of luck, I’ll be able to put the voltage regulator I just purchased from you onto this thing, and maybe post a video of it walking. I’m not sure how busy I’ll be with school, though.
I finally had enough time to make a small update to my site today. The last 6 weeks have been incredibly busy with school, and I only have about a week off before it starts all over again, but I’m hoping to make some significant progress on the RCS before I go back.
With a little luck, I’ll get far enough along that I can also start writing the IK code for my hex, but I’m not counting on that.
). I have a new charger on the way, but it won’t be here until the end of the week.
We tried to get a workable solution to the size problem, but they fumbled the servo horn part, and I could never get them to address the self destruct feature of the servos. 
