Power Supply

[Juan González Vázquez]

I want to buy a power supply for inmoov. The battery of 6V12Ah has few Amp or not? I want a compact power supply, I need 6V but how many Amp do I need? Because I read that each servo spend 3A approximately.

Inmoov has a 25 servos, would I need 75 A for Inmoov works fully?

Thanks.

[Richard Ryerson]

Servos use 3amp when they are outputting fully... not all servos will be running at once or at maximum... I would guess you'll need at least a 20A power supply minimum... Remember not all power supplies output what they say they output... Why use a power supply? Batteries are way cheaper and can output much more current... Be prepared to pay several hundred dollars for this kind of power supply....

[Keith McGerald]

This is what Gael uses for his InMoov. http://www.hobbypartz.com/26p-efuel-20a.html

--
You received this message because you are subscribed to the Google Groups "InMoov" group.
To unsubscribe from this group and stop receiving emails from it, send an email to inmoov+un...@googlegroups.com.
To post to this group, send email to inm...@googlegroups.com.
Visit this group at http://groups.google.com/group/inmoov.
For more options, visit https://groups.google.com/d/optout.

--

-Keith

[Mark Abrams]

Thanks for your rapid responses. So I gather that a battery with approx 3Ah functions properly because in fact only one servo at a time is drawing power (due to the programming model which blocks on servo.write(int))? Apologies if I'm asking novice questions I don't know much about electronics - this area is new to me.

[Hans Foerster]

No, I think you may of not understand what was said. Their are many instances that could trigger more then 5 servos moving at the same time.

Maybe not all servos are pulling 3amps because of the distance travelled, or whether it is 2 or 3 of the large servos and 2 or 3 of the small servos moving InMoov.

In theory when moving the hand to grasp an object and bring it up to the head you could experience 13 servos moving to achieve this task. Maybe not at full power or full movement but moving all servos at one time or other to achieve the stated outcome.

I guess if this sequence is programmed sequentially you could move one servo at a time but motion would seem choppy or not natural.

I would probably start with some 6 volt batteries with higher amp readings. You can provably set more then one battery on InMoov and start all servos using a zone type configuration.

Left arm
Right arm
Head, neck, cameras

Hope this helps.

Hans

[Keith McGerald]

Also just because the battery is rated at 3Ah doesn't mean it can only deliver 3A at one time.  The rating is more about total capacity of electricity it's storing. The max current available is usually rated differently.  Your 3Ah battery can deliver 3A for 1 hour, or 1A for 3 hours, or maybe 6A for 30 min.  That is an oversimplified explanation of an "ideal" battery though.  The links below provide a bit more info on the topic.

http://en.wikipedia.org/wiki/Ampere-hour
http://www.batterystuff.com/kb/articles/battery-articles/what-does-20-amp-hr-rate-mean-.html
http://www.allaboutcircuits.com/vol_1/chpt_11/3.html

--

-Keith

[Juan González Vázquez]

Thank you very much everyone.

In a principle I think I'll Buy batteries 6V12Ah

Thank you.

[ahmed....@imerir.com]

Hello,

I'm also interested into getting a power supply that would deliver enough V/Amps to power the inmoov(head, torso, arms, hands), are there any models you would recommend so i can get an idea about the prices ?

Thanks in advance.

[Kevin Watters]

I was looking for similar.  I ended up finding that LED power supplies for home led lighting strips actually have a lot of amps rated at 5v. 

I recently started using this 

http://www.amazon.com/Singpad-switching-display-Suppply-Transformer/dp/B00N751734

40A @ 5V.  There's a voltage adjustment, mine is actually outputting about 5.9V .  So far it's been great.  Even if all of the servos are moving at the same time, it has enough power to keep everything moving.

[gael langevin]

Hi all,

The one I use is the eFuel power supply 6V20A. You can buy it from various suppliers.

Make sure all your wires can hold the Amp drawn by the servos other wise you might get jittering servos.

--

[Dwayne Williams]

Gael as nice as that power supply is I have been waiting for 4 months for it to become available. I cant seem to find it stocked anywhere.

[Mark Abrams]

Interesting info Keith. I read the articles and found some others relating to ac adapters not batteries since it is my intent to use a corded solution for the time being. I have a bunch of old adapters laying around but nothing greater than 2A and my understanding is that drawing more than the rated current can overheat the adapter and It's not a good idea. What I also found though is that 3A max is different than 3Ah. http://en.wikipedia.org/wiki/Ampere vs the Ampere-hour page.

After some more research, I did some preliminary tests of the current draw and found that the whole system never exceeds 2.5A with 4 servos engaged at once. Once the ligaments are in, I expect I'll be able to do similar tests with all 6 servos as well as some servo stall current tests by gripping something. 

Thanks all for leading me in the right direction.

-- 

Mark

[gael langevin]

Hi Dwayne,

There seemed to be an update on the product which is difficult to figure what is the correct lowest output. Sites vary between 6V and 8V. It is not very clear...

If you trust Aliexpress there is this link:
http://fr.aliexpress.com/item/Power-adapter-eFUEL-20A-Switching-Power-Supply/575876057.html

[Petr Kruliš]

Hi, I am really new to inmoov so i dont know how many amps it drives. Truth is, the servos dont need those 3 amps all the time (they are not moving constantly and all of them at the same time). The 6v batteries can deliver many amps and are cheap. But I can imagine it can get really annoying sometimes.

The problem with 6V power supplies is there is not much of them, but you can easily find the 5V ones. A good power supply should have potentiometer for fine tuning which allows you to set maybe 5.2V - 5.4V. You actually dont need 6V to power the servos, they should keep running with 5V as well, it's just the torque what will be smaller a bit (check your servo datasheet).

What you are looking for is swichtching AC-DC 5V power supply which is able to deliver desired amperage. Those power supplies look like this one for example: http://www.ebay.com/itm/MW-High-Quality-5V-50A-360W-DC-Regulated-Switching-Power-Supply-CNC-/140726113423. I found this one in a few seconds, there are lots of those so keep looking. In general, those power supplies are great in matter of power/price/stability and the good ones are 100% tested for their max power so it should deliver as many as it is stated. But it is not a good idea to get the cheapest one (as always though).

But remember: The use of low voltage (5V, 6V, etc) to deliver a lot of power (and tens of amps is really a lot) leads to power loss. You should keep the wires from power supply as short as possible. Just google some more informations about this issue before buying anything.

If anything, you can use lets say 20A power supply for testing and development. If you are working on one hand for example, you dont need all the power to make the robot dance and it should simplifiy your life :)   

[Petr Kruliš]

So I found a power supply which should be suitable. It is Meanwell SE-600-5 (5V 100A - no chinese crap - this will give you those 100 amps). You can regulate the output voltage a bit by the potentiometer but I would expect it to give 5.5V max (according to specs also). You can get it for about 100 USD on ebay (new one).  

see:

http://www.ebay.com/itm/5V-500-watt-100A-power-supply-SE-600-5-/161543026743?pt=LH_DefaultDomain_0&hash=item259cb6f837

http://www.ebay.com/itm/1pc-Switching-Power-Supply-SE-600-5-5V-100A-500W-AC85-264V-247x127x63-Mean-Well-/130891366233?pt=LH_DefaultDomain_0&hash=item1e79bbbf59

or go with 60A:

http://www.ebay.com/itm/Mean-Well-MW-60A-5V-350W-AC-DC-Switching-Power-Supply-/121350275750?pt=LH_DefaultDomain_0&hash=item1c410a42a6

On Monday, December 8, 2014 12:23:30 AM UTC+1, Juan González Vázquez wrote:

[gael langevin]

Hi Petr Kruliš,

Thanks for all this info.

I have been, a few times, having issues with my power supply going in error mode, and that seemed since I have added length in the wire between the robot and the power supply. So your suggestions definitely explain the reason. I guess I will reduce the length to it's minimum.

Guys, you should use thick wires because the draw of the servos is important at stall and this can happen when the robot does very complex gestures which lets some servos under stress while others are doing other tasks.

5v power supply will do the job, although I really would recommend finding  6V because of speed. The gear boxes in the shoulders reduce the considerably the servo initial  speed and I noticed a serious difference between the two voltages.

--

[Petr Kruliš]

Right. 5v power supply will affect servos torque (approx. 20 % less of it). 6V power supply is better, but it's not a standard voltage so those are much more expensive. 

A relatively large capacitor will help to reduce the power loss and voltage drop. It should be as close to servos as possible. 

Since I started my inmoov project two days ago I cant come with definite and tested solution. Will do in future though :)

[Juan González Vázquez]

I bought this power supply , when I receive it I will tell you how it works .

http://www.aliexpress.com/item/Free-Shipping-6v-power-supply-60A-360W-LED-Power-Supply-CE-Approval/1533419426.html

[Petr Kruliš]

Looks promising. 60A should be enough (and even two of those would be for a good price :)). I'm curious about the quality, please let us know later :)

[Prof. B]

If you are buying a battery be sure that it is a deep cycle battery,  That is one that can stand multiple discharge and recharge cycles.   I've used  6 volt  golf cart batteries with good luck.   

Dave

[Prof. B]

I've got a 12volt one that I have had for a while, seems to work OK as a bench supply.  

Dave

[Petr Kruliš]

Hi,

if you want to calculate the voltage drop (this is what causing the servos to stop working properly), there are lot of tools for that.

See http://www.calculator.net/voltage-drop-calculator.html?material=copper&wiresize=13.17&voltage=6&phase=dc&noofconductor=1&distance=2&distanceunit=meters&amperes=70&x=35&y=8

You can get an idea of what is suitable solution before purchasing expensive power supplies etc. It is perfectly possible to achieve 50% and higher voltage drop (so 6v power supply would be 3v in the end). You need to use proper and short cables (larger diameter = lower resistance) and.

AWG wire size refers to American wire gauge standard, just google it for more info. It is all about the diameter of the cable.

[Kevin Watters]

Hello Petr, 

That's a great link to the voltage drop calculator.  It inspired me to write up a bit of the engineering and math behind this issue.  I posted a blog entry that talks about a way to think about sizing the wires and power supplies. In the blog post,  I simplify things a lot by talking about a DC motor under a constant torque, consuming constant power as would be the case when the HS-805BB's start "buzzing" 

http://myrobotlab.org/content/reflections-resistance-power-supply-lines-servo-motor

Hope it helps share some knowledge!  Happy roboting!

-Kevin

[Petr Kruliš]

You did a great job Kevin! That's exactly what had been missing here.

I want to mention one more very important fact. It's highly recomended to add a capacitor to each servo. And there is a good reason for that. Servos are not consuming current constantly and smoothly, they are taking small bits of much current. This effect takes place mostly when servo is starting to move as it needs a lot of current to accelerate. Let's say you want all the inmoov's fingers to move all at the same time. Then you've got 10 servos consuming approx. 20 amps together. And they need this power really instantly. Firstly this makes some undesirable noise in the circuit. Secondly power supplies are not good when it comes to those instant current demands, they can deliver the desired power, but it takes some time (a fraction of a second but that is still not fast enough).

That is why you yhould use decoupling capacitors with servos. They will help a lot to decouple the servos and cover those instant current demands. A decoupling capacitor shoul be placed as closed to servo as it is possible, so probably a place where the servo connector ends. Recomended value is 470uF or greater. It will make a difference.

I currently work o a small project where I try to modify normal hobby servo to be more usable with robotics. There would be some hardware changes to the servo itself and probably a library. I will also cover all of those things stated above more sophisticatedly. It should be capable of speed control and some easing (acceleration/decceleration) so the robot will move more naturely. Af course the best to use the proper robotic servo but those 5x times more expensive (but are also completely another league - worth the money) and that's a lot when you need tens of servos. But I dont want to promise smotehing now, I'm not sure about results yet.

[gael langevin]

Great work Kevin!
@Petr Kruliš, if you are working on improving the servo boards, you may have already found the Open Servo Project board. I bought two of those boards a few months ago but I haven't yet worked with them. My intention was to use them on the hip rotation servos. These boards have pretty good features.

[Petr Kruliš]

Yeah I recently bumped into teh Open Servo Project. But since it costs about $10 (and the servo itself is another 10), it may be a better option to buy the robotgeek servo for 20 (http://www.trossenrobotics.com/rg-180-servo). There is not a lot information about this servo but i want to give it a try and see what it is capable of. But Open Servo may be still better when you need more torgue, different size package etc.

As of my servo project, I'm trying to discover ways how to modify standard servo with no added costs (I mean higher than a few cents or so) with the use of components which everyone should already have at home. I'm not saying result will be something like open servo or so, just want to make little adjustments so it was easy to modify existing servos to give slightly better results. Maybe its not possible, we'll see :)

[Kevin Watters]

Hi Petr,

  Thanks for the feedback!  Adding a cap across the supplies of the servos is a good idea.  I'll try to incorporate that into the next build I do.  

I've been trying to make hobby servos much better with minor hacking.  I have some notes about modifying a hobby servo and reading the internal pot value to get feedback.

http://myrobotlab.org/content/closing-loop-hobby-servos-mrl

PWM as a control protocol is really not sufficient for real applications.  I aim to add proper feedback via tapping the internal potentiometer of the servo and using the arduino to read the value.  I'd love to get to a point where the servo could self-calibrate.  

  -Kevin

[Petr Kruliš]

Kevin,

reading the internal pot value is a great and easy thing to do. I tried it with my servo (towerpro MG946R). But as I want to make the servo speed controllable, it is not enough. For robotics it is really important to have a speed controllable servo to implement acceleration/decceleration (to prevent those nasty fast and jerky movements) or just make it slower.

My aproach was somehow similiar to those continuous servo hacks people do. So I unconnected the potentiometer from board completely (but left it at place) and connect it to arduino. It is a good thing to power the potentiometer from arduino 5V source because the servo is powering it with only approx. 2V or so. I could use the AREF pin as you did but different servos may use different voltage so I did it this way. I soldered a fixed voltage divider (two resistors) to the servo voard instead of the potentiometer. In this configuration I can control the speed of the servo easily with arduino (same way as with continuos servos) and I am still able to get readings from potentiometer. I left the servo stops at place, because I still want the servo to move only 180 degrees (otherwise it will damage the potentiometer by overturning it).

This should give me accurate readings (so position control) and speed control (so accel/deccel ability) without swapping out the controller completely etc. The main problem is the servo itself. It is metal gear cheap chinese servo and it moves really badly at low speeds. You will always get the vibrations and stuff from these hobby servos, but mine is really a piece of crap :D It cant even keep the same speed, the gears are horrible and sometimes the servo speeds up and down or stops completely etc.

I will buy more servos and try them. I have seen videos with servos moving really slow and still smoothly (well jerky but with constant speed without horrible sounds comming from it). I'm closing this project for now and will be starting again with new servos (in a month or so).

[RobotGuy6]

There is a way easier way to adjust servo speed than that.... Arduino has a servo library (I think it is ServoEx.h library) that can adjust speed on the fly it's an added parameter in the servo move command... I don't use arduinos much now since discovering EZ Robot.... full servo control is through software and it is a snap to use....

Enter code here...

[Petr Kruliš]

@RobotGuy6

Well, all those libraries (there are about ten of them I guess) are controlling the servo speed by moving it bit by bit. So the library tells the servo to move by one degree and it waits some time until servo arrives to move it another one degree and so on. The problem is you have no feedback from servo so noone knows when the servo arrived it's location. It's all about guessing. And even if you calibrate all the stuff to fit your servo, the conditions may change (more load on servo etc). So it's a good way in some situation and bad way in another.

Even with the feedback from servo's potentiometer it is still much better to remove the potentiometer from servo controller. When you need the servo to travel at lower speed you need to move only a little bit. When the servo controller thinks it is moving only a few degrees, it will not try to accelerate tu full speed and will travel  at lower speed. But moving by those bits can be problematic and jerky even with precise timing and feedback input. Better solution is to disconnect the potentiometer so the servo controller will think the servo is not moving. In that case, you can send one command to servo and will move at constant speed proportional to distance you told it to travel (shorter distance = lower speed). 

I tried to use a library only solution but it was good only for constant speeds and only in some situations. It is not good for easing in and out the servo movement. But if you just want to move servo at some constant speed, it is perfectly good and easy solution.

[Kevin Watters]

Interesting approach.. at that point. might as well just switch to a dc motor and implement the control on the arduino...  

In MyRobotLab, there is a separate thread that can be used to sweep the motor at lower speeds. as you mention in another post here, basically we have a separate thread that moves the servo a few degrees and pause.  The end result is surprising, but it does allow for variable speed control for the servo.  I've used this technique with an xbox video came controller (and a ps3 contoller) ..  moving the joystick a little bit and the servo moves slowly, move the joystick a lot, the servo moves faster.  Agreed, it's not perfect, but it does provide reasonable rate/speed control for a standard hobby servo...

On Sunday, January 11, 2015 at 10:34:03 AM UTC-5, Petr Kruliš wrote:

[Petr Kruliš]

I had been thinkink about a dc motor already, but in the end you will end up in a servo because you need the gears, the enclosure, the stops etc. And you would need to build a custom controller for a dc motor (mosfet and stuff), which would add costs and time. My approach is still intended to make use of your existing servo without any significant costs. If you need a perfect solution, you still need to go for a robotic servo or at least the Open Servo Board.

Open feedback loop libraries may serve well even for acceleration but you still cant implement deccelration. You could calibrate the servo and set the right delays but when you have whole humanoid, you will also experience different loads on servo in different situation and the timing approach will not be suitable.

I'm talking about my approach which implements servo easing. For constant speed, you should be good to go (with some decent calibration in real situation) with a open loop feedback and a step-delay-step library. Just to be clear.

[Petr Kruliš]

So...

As for the power supply: The best way is imho to use high voltage power supply (24V or higher) at the same place where is the 6V battery now. The power supply will be powered directly from mains. With this higher voltage there shouldnt be any problem with power delivery (almost no voltage drop/power loss) to the servos. Af course servos need 6v or 7V/7.4V/9V/12V - it depends, so you need to convert the hight voltage in something usable. Problem is servos use stranges non-standard voltages because they are intended to be used in battery power systems. But there are also stepdown regulators made for this purpose. They are called BEC/UBEC/SBEC (intended to power quadcopters electronic systems for example). They are cheap, powerful and good enough. Hobbyking has even 20A SBEC for approx 12 bucks (http://www.hobbyking.com/hobbyking/store/__40274__Hobbyking_YEP_20A_HV_2_12S_SBEC_w_Selectable_Voltage_Output.html). And 20A is more than enough to power all 5 fingers. Look just for SBEC/UBEC because those are swtiching regulators, not linear (switching regulators are far more effecient). This way, you can get wired inmoov without burned cables or slow servos because of voltage drop or similiar problems. I will definetly go this way with the inmoov. I just dont want to use batteries.

As for my servo upgrade research: Just stay tuned. I ordered tunrigy robotic servo for 20 bucks. I'm not sure about the quality or even features, so I need to test it. But if this is at least something like a real robotic servo, the price is awesome. It was deliver just a few days ago (it took more than a month to arrive from china). It was enough time to built the whole printer :)) I will test during weekend I hope and let you know (there are no reviews or anything about it yet). After that, I will continue with the hobby servo upgrades I was talking about.

[Di Wang]

Hi Petr, thanks for all the info. How did your test go with SBEC?

[Di Wang]

Hi Juan, I am interested in buying this power supply, how does it work for you?

[Di Wang]

I am at a point to hook up 6 servos for the right hand and wrist to my Arduino Mega, I am planning to get the power supply mentioned by Juan: http://www.aliexpress.com/item/Free-Shipping-6v-power-supply-60A-360W-LED-Power-Supply-CE-Approval/1533419426.html. Now, my question is how to connect this power supply to all the servos. I am in the US, so I don't have access to the Nervo board. I was thinking about connect all the servos through a breadboard, but it probably won't be able to handle that much current. So should I get a prototype PCB instead? And should I get some wire like 10 AWG to connect the power supply to the PCB in order to handle max 44amp current? Any help is highly appreciated!

[gael langevin]

Hi Di Wang,

We ship Nervo Boards to the US all the time... There is a new batch arriving in the coming week if postage traffic between France and China is okay.

That being said you can use a breadboard to test for the hand, it will handle the current.
And if not there is tons of servo shields for Arduino out on the Net.

Gael Langevin

Creator of InMoov

InMoov Robot

@inmoov

2015-09-15 21:17 GMT+02:00 Di Wang <robo...@gmail.com>:

I am at a point to hook up 6 servos for the right hand and wrist to my Arduino Mega, I am planning to get the power supply mentioned by Juan: http://www.aliexpress.com/item/Free-Shipping-6v-power-supply-60A-360W-LED-Power-Supply-CE-Approval/1533419426.html. Now, my question is how to connect this power supply to all the servos. I am in the US, so I don't have access to the Nervo board. I was thinking about connect all the servos through a breadboard, but it probably won't be able to handle that much current. So should I get a prototype PCB instead? And should I get some wire like 10 AWG to connect the power supply to the PCB in order to handle max 44amp current? Any help is highly appreciated!

--

[Di Wang]

Thanks Gael!  I would like to buy a Nervo Board to just support your great work:) Could you let me know as soon as they are available? Also, have you found a power supply or still using batteries?

[gael langevin]

Hello,

I use both, power supply or battery it depands of what I do. When InMoov needs to be moving around during events I will use the battery. But for everyday work, the power supply is handy.
Nervo Board is available on backorder, that means, if you place your order now, when the boards arrive, we will ship it automatically to you.

But feel free to evaluate what is best for your project, so don't buy it if you won't need it.☺

Gael Langevin

Creator of InMoov

InMoov Robot

@inmoov

2015-09-16 2:11 GMT+02:00 Di Wang <robo...@gmail.com>:

Thanks Gael!  I would like to buy a Nervo Board to just support your great work:) Could you let me know as soon as they are available? Also, have you found a power supply or still using batteries?

--

[Di Wang]

Thanks! I just placed an order for 2 Nervo boards:) Are you still using your efuel power supply? I read your blog that the efuel power supply won't provide enough current sometimes, have you found a better option?

[Hairygael]

Great!

Sometimes the efuel shuts down when too much power is drawn. It is also due to the dimension of the cable that connects the efuel to the Nervo board. Too much length and too thick cable will cause drop off. But during a day of demo it can happen 4 to 6 times.

Envoyé de mon iPhone

[juerg maier]

Hi

I use a Mean Well SP-240-7.5, 32 Amps at 7.4V, about 80 EUR from Reichelt, Germany. It has been able to burn any servo I used so far so be careful! You need ot limit voltage (easy turnweel) to the lowest servo specs you have connected in your robot - and make sure ALL your devices have a COMMON ground - I remember some ghostly movements before remembering that...

The power supply has a fan and if it runs for more than 5 seconds better switch of the power - otherwise it might start to smell..

[gael langevin]

@juerg, Nice power supply, hey but do you give  7.4v to your 805BB? They are built to get only 6V.

Gael Langevin

Creator of InMoov

InMoov Robot

@inmoov

[gael langevin]

Hello Di Wang,
Did you receive your power supply:
http://fr.aliexpress.com/item/Free-Shipping-6v-power-supply-60A-360W-LED-Power-Supply-CE-Approval/1533419426.html
Did you make some tests and does it handle the Amps required?

[Dean]

Yes, I did. It works fine with 6 servos for one hand and wrist, I haven't tested more than that since I only have one forearm finished right now

--
You received this message because you are subscribed to a topic in the Google Groups "InMoov" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/inmoov/3jwBqoN-H6g/unsubscribe.
To unsubscribe from this group and all its topics, send an email to inmoov+un...@googlegroups.com.