Best way to power a 19 servo robot?

I am making a 19 dof humanoid robot. I'm going to use 18 mg996r servos and 1 mg90s servo. For the mg996r servo, the operating voltage is 4.8 to 7.2V and the stall current at 6V is 2.5A. For the mg90s the operating voltage is 4.8 to 6V and the stall current is 900ma at 6V (might be less, this based on what I found online). I'm going to use an esp32 (it's arduino compatible) and 2 pca9685's. Btw I don't have any of the parts yet. What is the best battery pack configuration for this type of build? I'm okay with splitting the servos more evenly between the servo drivers (9 + 10 vs 16 + 3), and multiple battery packs if it will help give a cleaner way to power everything. Links would be helpful.

So that is 18 times 2.5 A (mental arithmetic - 36 plus 9 = 45 Amps) plus another 900 mA.

Sounds like a large RC airplane 2S pack with a minimum 50 Amp rating and a little switchmode regulator to tone it down to 6 V for the MG90.

Do you have any specific recommendations? I'm having trouble finding some. Also would I be able to bring it down to 5v instead of 6? How would that affect the mg996R's?

I know you recommended LiPo but this is my first project and I'm not super comfortable using them. Would something like this work? 6V 3300mAh SC NiMH (5x1) RC Battery Pack Vapex | Component-Shop

Like having 9 servos (about 23A max) on one driver board and 10 on the other (24A max), then each board having their own battery pack so 2 packs in total. The pack I've linked says the discharge current is 30A. Does that mean 30A is the maximum current for the battery pack? Like would it be able to support the amps required for the servos in the configuration I'm suggesting?

Yes, any more will cause damage to the batteries.

So would two of those packs work if I'm using less than 30A on each servo driver? Like the scenario I described above. And is it enough headroom if 24A would be the maximum current drawn on 1 servo driver?

I too recommend lipo batteries. Just make sure any buck regulator is up to the job you ask of it, testing is essential as specs can be aspirational.

And be sure to test at the low point of lipo voltage sag. It may be necessary to use a 3s pack, 2s might go quite near 6 volts under load at the end of charge.

Use real batteries and real loads for tests. At least real batteries even if just resistive test loads.

Buy good regulators and you get closer to specified performance. Again, trust but verify.

You can probably cut you budget figures down by promise in software and design to not run every servo on board at its stall current requirement at once.


Tricky! :roll_eyes:

You are not in full control of the servo as you would be with a direct motor.

At least you can avoid some trouble by not telling all servos to go full speed at once.

And how is one in less control of a servo than a regular moto? If you stop telling the servo to do or try doing something, it stops… doing or trying. This could be informed by current measurement input(s) to the control system.

There is the possibility of software error or failure, so some extra precautions might be considered.

Planning to have current available for the maximum draw of every component is fine if you have the money, space and weight carrying ability. And some devices will obvsly need that.


2S LiFePO4 gives 6.4V, which seems a reasonable match. LiFePO4 is more forgiving than LiPo to abuse (not that abuse is ever a good idea with lithium). LiFePO4 has much longer life too. Big prismatic LiFePO4 cells are available covering a wide range of max currents and capacities.

Thanks for pointing this out.

I have only used LiFe, which I assume is the abbreviations for the chemistry you mention, in limited deployment scenarios and had the impression that when it comes time to deliver massive current it was unsuitable.

But it may also have been an issue of weight and volume versus stored watts.

Something to revisit for me.

Anything more forgiving than LiPo is to be valued! I’ve only had minor brushes with disaster; for amusement search youtube for the experiences of some who were less fortunate or more stupid.


Depends on the cell, many types are available, some are high discharge, others are not (true of any battery chemistry really).

we call this energy density and power density (or specific power):

THX for the link and further comments.

If my google-fu is what I hope, it looks like “lipo” are chosen for flight batteries more commonly as an issue of weight.

Of course now there are dozens of “lipo” battery choices for flying, but use of LiFe for ground station equipment and transmitters is common.

I put lipo in quotes as it isn’t usually specified beyond that, everyone just knows what it means…

Which I haven’t yet found, will keep looking, but what is the exact full name of the chemistry in the common hobby 1 to 6 cell flight batteries, 500mAH to 5000, bare naked packs we play with?


Depends on the model, but lithium polymer usually use cobalt oxide style lithium cells, the ones most likely to ignite if abused, but also the most energy-dense. They have the problem that lithium has a higher affinity for oxygen than cobalt, allowing a type of thermite reaction between Li and CoO2 to be triggered easily if the cell overheats. These cells are 3.7V nominally per cell.

The exact formulation will be proprietry, note, manufacturers don't easily give away their commercial secrets! For instance carbon nanotubes are often employed to increase the current handling (power density).

THX for the extra information. I suspect most of the lipos are coming out of the same factory, or at least whatever the raw materials or intermediate products that go into the manufacturing.

There are LHV cells, 3.85 volts nominal. I avoid them just because I don't want to have to wonder which I have, and to reduce charging mistakes. They don't really give that much more for the trouble, at least not to someone as conservative as I. Am.

There are (or were, was it a fad? haven't paid attention my batteries have lasted so long) "graphene" based cells, I assume/wonder if these are the carbon tubes you speak of. Again, from my perspective it just looks like paying more for a battery on some promise of something I don't need.

I've topped out by choice at <= 3s batts, <= 1500 mAH. So I don't spend too much, fly a bit less dangerously and sleep better than some I know with truly scary batts. Still plenty of all kindsa excitement.

I am amazed at what lipos can do, both the good stuff and the bad. :expressionless:


Not at all, there are numerous manufacturers its a growth industry and everyone's trying to win/maintain their share of the growing market. Even 5 seconds spent searching turns up:

Again, thank you. I should indeed have goggled, just now I spent more than 5 seconds and am fascinated by the whole manufacturing process.

I should not have extrapolated wildly; many of the batteries I use are obvsly mere rebrands, and most of them look extremely similar inside and out once fancy labels are removed. Or similar. Or alike enough.

But this is prolly a result of evolution - something that works, and is easier or cheaper to manufacture, starts becoming what all cells look like. I'll google that. :wink:

Also, my experience is very limited to batts that are of a certain size range and aimed squarely for a narrow use case, so it should not surprise me that they are all very alike.

Of course cells with the same chemistry are cranking away in a vast number of deployment scenarios and again, it shold be no surprise that many manufacturers are providing cells in all manner of physical forms made with differing (and proprietary, as you pointed out) details.

Now I go fly. :expressionless: