I was thinking about your comments on the size of a bot…
I get the impression there is a maximum sensible size beyond which motors/gearboxes need to be used.
Especially when you consider the speeds that some of the more powerful servo’s are capable of moving at.
I’ve been looking at geared stepper motors myself but as yet have very limited knowledge of whats needed to produce accurate, repeatable movements. The cost could escalate quite rapidly.
That is a relatively easy answer, basically a microstepping drive controller with torque compensation. Still a stepper will not get you out of the need for position feedback though since they slip if you over torque them. At the same time forcing them to slip doesn’t break, burn, or otherwise chew them up either. Cost… well not inexpensive but the whole concept of expensive is relative to what it is you are trying to build. For a 12" high SES based biped I’d say prohibitively expensive but for a biped the size of a 3-yo kid… not so much. 
I was just looking at the Farnell catalogue as you wrote that… For £145 you could build a nice geared stepper motor. The recommended power supply and electronics wouild require a rethink on space as you suggest. A small bot it would not be. Steppers don’t seem to be suited for battery operation either as the power consumption appears high.
Oh I don’t know… I hear folks are blowing 3 to 4A MOSFETs out of servos when they ask them to hold near rated torques for more than a few minutes at a time. With a nice stepper drive control you could probably tell it to reduce the phase currents when idle since their holding torques are generally way more than the moving ones. 8)
It would be good to see your calculations on the stepper motor. I see that available stepper motors are capable of less half the the power of these servos, and yet weigh in at ten times the weight.
Yes the stepper solution is "nice"and safe, but a limited capabilty !
I sometime worry that we fail to appreciate how leading edge these servo products are. In a couple of years these reliability issues may be resolved, but only if we suffer !, improve and provide feedback to manufacturers.
Brushless motors, intelligent controllers, active cooling will become standard, and “safe”, but the world will have moved on by then, so maybe still unlikely to represent state of the art performance.
This site has made clear both the massive opportunity and the limitations of these high performance “athlete class” servos.
Hitec make “kick butt” servos, sometime they get burned. Life on the edge ! Servos on steriods !
Factor in a realistic 20% maintenance budget for servos, and have load of fun.
The trick is in differentiating leading edge from bleeding edge and your budget really needs to drive where you want to be.
Personally if a $115 servo operating at 60-70% “rated” load bought the farm after 3 minutes of use, and the data sheet for the servo didn’t make it blatently clear that you needed to de-rate that torque spec for use beyond that 3 minute mark I would be… hmm, what’s a family appropriate word… a VERY angry ex-customer. 
As for steppers… what are you calling “available” steppers? What’s your selection criteria? Are we talking “real” stepper motors, or crap salvaged out of a printer and bought from a bargain bin?
Just off the top of my head go look at something like this linengineering.com//site/pro … inipak.htm
The smallest drive_motor combo which has a holding torque of 9.2 oz-in and, just using the data on their site which is only for 24V, is delivering over 6 oz-in at 15,000 steps per second. it’s a 1.8deg stepper so 200 steps per rev, put a 200:1 gearbox on it, that’s 37.5 rpm, still real fast by servo standards at about 6.7mS for 90 degrees, but you’re at 1200 oz-in of torque. Now the motor and drive are roughly 4 oz, figure another 2 oz for a gearbox, that’s 6 oz for 1200 oz-in for torque. The $115 wonder servo that needs a smoke after 3 minutes weighs 2.6 oz and delivers 333 oz-in at 1/30th the speed (0.15s for 60deg) … while it lasts. The stepper+gb probably cost 3x what the servo does, requires custom mechanicals, and the end product will cost a lot more and be bigger and heavier because of it. But this is sort of what I started off by saying to Paul… costs are relative to what you are trying to build. If my budget could take it I’d go for the biped the size of a 3-yo built from composites and titanium running off 6S 30A LiPo packs in a heart beat, wouldn’t you? 8)
I looked at that also but steppers can have some quite high rpm rates so you can afford to add a gearhead or gearbox. On a larger bot the speeds are apparently somewhat lower and the tradeoff would be quite a robust and powerful one.
I have been playing with R/C stuff for 20 odd years now and the servo is still basically the same, a niche market product that would require a mass market to force serious investment. The only real market for the hi-performance servo’s appears to be based in Japan but is (very) slowly appearing elsewhere.
Agreed, and when you watch the way the robo competitors treat their bots they are obviously aware of how frail the servos are. There are never any long extreme demonstrations. Usually very quick and to the point.
I’ve no experience of Hitec but the Kondo ones have taught me that lesson. £100 ($200) in 4 months so far…
Darn Eddie beat me to it again… 
Gotta learn to type faster…
Scary part is I’m doing it in between simulation runs… I worry about keeping the same train of thought in my posts between those pesky work interruptions.
BINGO. I still haven’t seen an answer as to whether these things could run under “normal” use at 7.4V for 15-20 minutes… is it because no one knows?
Is that maintenance cost per month, per year, per time you turn on the robot? 20% of a lot is a lot!
This is for Kondo KRS788 10.8V servo 10Kg.
I know its not Hitec but these are uprated and heres what mines survived…
Standing with full weight on 1 leg for 3+ minutes quite happily.
20 minutes (full battery life) of activity, some of it extreme but mostly just standard moves.
The recovery move from a fall requires the 2 hip servos to lift the whole upper body through 90deg. I would say thats almost 1Kg moving rapidly from Horizontal to Vertical. Probably ranks as the most strain I can apply.
And it survives…
The only 2 servos I have blown are the Head which has no load whatsoever and the forearm which at most is used doing pressups, Possibly about 1Kg max…
So the answer is not No or Yes, it very much depends on circumstances etc.
I dare so if I tried to do the extreme moves when the servos were hot they would blow
Just curious, is that a 9-cell NiMH? That’s quite a lot of voltage for a standard size servo.
Yep, 9 of AAA rated 800mAh NI MH…
the AA would give a longer life but the weight and volume are a problem
Guys take it easy. Ribotson is a very smart guy. He singlehandedly reverse engineered the Hitec digital servos, a feat not to be taken lightly. I am a big fan of his work! So there may be a stepper solution that will perform as well as or better than servos. I personally have my doubts, but whatever… But the majority of folks in hobby robotics are using servos due to lots of reasons.
**The fact is no one should design any mechanical device using the stall torque. **
Manufacturers have been publishing stall torque values forever knowing full well the motor can’t perform at this capacity at 100% duty cycle. I’m talking motors, not servos. Lots of places list “theoretical” torque values (motor torque x gear reduction) even though the gears can’t handle the torque. I have done my best to make sure what we publish is accurate!
When you are ready to build the cool robot of your dreams, you need to have realistic expectations, and um, well… hopefully have some fun in the process.
Apologies if my posts appeared to be flaming, thats not what I intended. I was attempting to sympathise as I have been watching comments both here and at Robosavvy about blown servo’s and can understand where he is coming from. Specs are specs and phrases like “theoretical limits” and “within acceptable limits” are too often used to cover up inadequacies in the design.
Normally thats enough to put me off a product and go looking elsewhere but as you and Ribotson have pointed out budget and other limitations make the super-servo the best option.
Again, apologies if I’ve appeared offensive, compared to many on here I’m a newbie and an amateur one at that and I wouldn’t force my opinion on any situation. Usually coz I’ve jumped the gun or mis-understood it myself and I’m miles off target…
I get it right sometimes… I think…
AMEN! I could not have said it any better. 
I’m going to be posting assembly pictures of one of my dream (or was it a nightmare?) robots soon.
8-Dale
Which robot?
I’m not sure where the take it easy part came from actually because I was mostly just talking numbers. I’ll admit I played on words a little bit to make some humerous innuendo but it wasn’t really intended to be overly disrespectful. I also believe there are several very smart people frequenting this board with a diverse set of background, life experience, and personal interests. In my opinion Mr. Ribbotson’s post didn’t reflect his technical skills applied to decrypting the digital servo protocol. I did see a request to see my "calculations on the stepper motor so I banged out something straight forward and to the point. Really though he made a marketing pitch… kick butt servos living on the edge learning and burning. Enthusiasm is great but we’re in a thread the topic of which is I’m tired of throwing away $100 servos when I didn’t do anything wrong, another contributor is concerned about the (sounds like) two grand he just sunk into servos for his new project, and yet another person is examining the question of at what design threshold do we accept that we have exceeded the practical design limits of even “athlete” servos and start to examine alternatives. Even you yourself Jim are in the process of designing a servo product to replace even exotic class “normal” servos because the application you want to develop is limited by those normal servos by either performance or outright cost. Why bother if there isn’t a need? Because there IS a need and your own learn and burn has identified it.
You just also happen to be in the excellent position to do something about it. 
No harm meant by my posts in this thread. I just tend towards taking more of an engineering approach to figuring things out then a try it and go from there one. That IS the world I live in but I am also trying to relax a bit so I might enjoy this stuff as more of a hobby than yet another extension of my job. I do still have that tracked ROV being built on my bench downstairs with the WAG gear motors on it… hoping it does more than sit there and smoke the first time I get to put power to it all. 
8)
Thanks for all the informative responses guys. Just to clarify. I’m aware just how much of a marvel these little servos are. There is alot of power packed into a tiny package.
I also realize that servos can’t be operated at stall continuously. And that they will never actually produce their stall torque for more than a few seconds before burning out. What I think, is that servo manufacturers should be displaying their specs the same way Lipo batteries do. With a continuous and a burst rating. A 300oz stall servo should probably be called a 200oz continuous or 300oz burst. Since that is the reality. To assume that people will only be using their servos for short bursts is not realistic. Hitec posts their specs in such a way for the purpose of putting the highest numbers they can out, then adding some fine print somewhere saying “posted torque at 7 volts for 2 seconds on the surface of mars”.
I was running my 5995 on an SES style arm, with 5 inch long segments and a 5 oz motor on the end of the hand. The servo was in the elbow joint, no restricted movement and very little load. I was running through some very basic movements in RIOS for about 4 to 5 minutes. Then poof, up in smoke. I’m using just 1 HSR-5995 on the base of the arm, it was barely warm. And at 6.0 Volts!! The last one was at 7.2 Volts and burned up about 30 seconds into some load tests where it was near stall, not stalled.
I’ve used and abused HS-645’s and 805’s where they are repeatedly being stalled for extended periods and never had one fail. So why is it that me, and others, keep smoking the 5995 when it is pushed as being so powerful and wonderful. Especially when the price reflects that.
I can’t help but feel like there is nothing special about this servo at all. They tossed titanium gears and a metal arm onto a servo and said “the super robot servo, 417oz of torque!!! Wow, come buy one” When really they could have just said “run your HS-5645 at 7.2 Volts and you get loads more torque…[size=75]for 10 to 15 seconds[/size].”
If the chips and motor can’t handle a moderate load current for more than a couple minutes or a stall current for a few seconds, then they are pretty useless. I mean thanks for the heatsink and the thermal shutoff on the HS-5990 but I think you can keep it. Especially if it’s the same servos with a shutdown and heatsink added. So I should buy some so that instead of smoking the servos, they just constantly shut down from overheating. That’s alot less frustrating.
I love Hitec servos. But come on. The HS-5995’'s are a joke. They labeled them something they are not to set the bar in the market for a powerful servo. When really all they had was a moderate servo that they recommended everyone overvolted. I have notice some vendors don’t even post the higher torque spec at 7+ volts. Some advertise them as a 333oz/in servo at 6.0V. Which is what they are. Almost.
And this is not an attack on you Jim. You always put the customer first and I love that. It’s not like you made the servos. You have to sell whatever they offer. We are all at the mercy of that. I just wish Hitec would realize that I would pay $200 for a servo that could do 333oz/in indefinitely and reliably, as apposed to a 417oz/in servo that only works for 30 seconds then commits suicide.
I had one blow up running at 6vdc with no load, just sitting there. Of course they replaced it. I just do not know what the real problem was. We may never know. The fact that they discontinued the servo speaks volumes. Curious, did you ever contact Hitec about the servos you blew? It’s one thing to complain, but if they are willing to make it right somehow, it may be much to do about nothing.
Um, a heatsink would effect how long it takes to go into shutdown eh? All of these “super servos” no matter who makes them MUST dissapate heat. So this falls under the reasonable expectations I was talking about earlier. Do you really think this isn’t a significant improvement over the 5995?
I appreciate that! Hey, nobody’s perfect, but I can say without a doubt, that my list of faults DO NOT include trying to cover up product faults to make a buck. 
I am only interested in providing the most accurate information I can so people can make intelligent purchasing decisions.
