OSR2 servo arms fail mid script

Finished building my OSR2 and running into a problem where during an intense section of the funscript… the servo arms just drop straight down like it’s giving up. Then spike back up and repeat until the scene calms down.

Slow strokes have no issues… except there’s considerable jitter (vibration?) during the down stroke.

Any help would be greatly appreciated.

i experienced a similar issue look at the power supply during those intense sections and see if the light goes out if so its the power supply

Thanks for the quick response. You nailed it right on the head.
I got a 5v 10Amp PSU from amazon… didn’t think I’d need more than 50watts!

I am using an ESP32 and sharing the ground leg onto the powerbus. ESP is powered by USB to my PC. I double checked all connections and even removed all the power connection + bus out of the OSR shell to remove any vibrations. Still slumps down during intense sections and the green light on my power supply brick turns off.

It’s labeled as a 5v 10amp PSU. Amazon.com is the one.

The servos I’m using are AGFRC 32KG (red/black) and out of curiosity I threw in some cheaper servos , 20kg from amazon, and the system is stable w/ the cheaper servos. Although much louder. Assuming the AGFRC 32kgs take a lot more power to do the same thing.

Yeah… power supply claimed ratings are always tough. I’ve run into a lot of ones that may be able to output power in a sustained manner, but can’t deal with “spikey” loads and will just go into protection mode. Ended up just using a Meanwell unit, and 3d printing an enclosure, and I swear that my OSR2 responds quicker and smoother with the same servos.

Thanks for the link. Ordered and will do the same.

1 Like

Make sure you crank the voltage adjustment (with a multimeter). It comes set to 5v, and can be adjusted up to around 5.8v.

I have come across some servos needing 7v to move, hopefully yours will be okay with 5.8v.

Definitely report back - I really have faith in Mean Well usually and have been headache free with mine since I trashed my questionable power bricks.

1 Like

Will do. It should be delivered tonight- ty amazon.
I saw the voltage adj potentiometer and planned on going as close to 6v as possible.

The servos I got are: https://www.amazon.com/dp/B07T23FNPT and it says it works at 4.8v… will report back :slight_smile:

I run my servos at 7.5V, and have measured their power consumption at the wall to be higher than 60W during intense action.

Even before doing that, however, I guessed based on specs that the two servos I’m using (Savox Sb2292Sg) would be too much for a single power adapter, based on what I was finding for the latter while hunting on Amazon.

So I modified the OSR2 design to use dual power inputs and power each servo from its own adapter. It’s possible to use a single power adapter, of course, but finding one isn’t exactly easy. Most adapters capable of that amount of power draw are designed for laptops, and put out much higher voltage.

The linked PSU is another option, but keep in mind that servos go slower with lower voltage. They all have a range they’ll operate within, and their max performance specs only apply to the top voltage.


OK, so results. Replaced the PSU with the suggested Meanwell LRS-100-5 (5v 18amp).
All was going well and surprisingly the motors ran smoother and quieter on this PSU vs the generic 5v brick I was using. I did turn the voltage up to 5.75 (highest it would go).

Tests with just the FL case mounted worked great and it no longer shutdown on intense scenes. Then I installed the actual FL sleeve inside the case… and during intense scene it just slumps down again and the Meanwell green light turns off.

I simply don’t believe these 2 servos are pulling 15amps… so returning the servos and trying a different set. I did test with 2x cheapo amazon no name 20kg servos and they work perfect even with the FL sleeve installed. So something’s up with these AGFRC A73CHLW V2.

Any suggestions that wont break the bank? The Savox look nice but are way out of my budget :frowning:

I’m happy to hear that I’m not the only one noticing a diff with a higher quality power supply than the generic bricks! I thought I was crazy.

I am NOT a servo expert, but I’ve built a buncha OSR2+s and I’ve tried a few diff servos, and I really only have 2 recommendations. The cheap red servos, or the Hitec D954SW (90 bucks). Everything I’ve tried in between has either had heat issues, required a TON of power, or was not smooth. The Hitec is more than fast enough for me, and very very smooth and quiet(er). Expensive though…

Are you totally unhappy with the cheapo servos? They could get you by until they burn out and maybe the Hitec’s could fit into your budget.

1 Like

My only issue with the cheap red ones are they are loud and other people in the house can hear them. But maybe the new PSU will quiet it down enough.

I’ll look into those HiTecs and start saving some money. Thank you!

1 Like

Update. I decided to throw more power at these AFGRC A73CHLW V2 servos and found this adjustable 480w PSU: https://www.amazon.com/dp/B0B18JVDGP

I set the volts at 6v and ran through the the intense sections again… this time… no issue. I slapped a Kill-a-Watt on the AC side to see how much these are pulling from the wall… and at peak… it’s only 25 watts. My only guess is these servos at peak load cause so much resistance that the cheap brick and the Meanwell PSUs see them as a short… and this 480w PSU just plows through that short. Only a guess.

Anyhow I’m able to run through entire scripts now and with this PSU I can adj volts between 0 ~ 12v which is pretty handy. While running through a script I can bump from 5 → 7v and really notice a difference. Things get faster the higher the volts… and also smoother and quieter… but maybe servo dependent.

Calling this case closed for now. Thanks for all the help everyone!

The servo peak power consumption is much higher than the wall power consumption measured with a kill-a-watt. I tested my SR6 clone with 6 powerHD 1501mg servo’s at 6v. The max power consumption observed at the display of my lab PSU is 40w (~1.1A per servo), but I need to set it to 75w (~2A per servo) if I don’t want it to sound the overcurrent alarm constantly. The theoretical max according to the datasheet is 2.5A per servo. Your servo’s list a max current draw of 3.3A at 6.4v.

The problem is that the display shows the power consumption averaged over 1s, but the current spikes that trip your PSU can be a couple ms.


That makes a lot of sense. Thank you!

Wonder if there’s the possibility to add a bulk storage cap between the PSU and servos to help with these spikes and potentially improve performance. I saw some folks on the tempest discord adding small ones, but I’m guessing we’d need to use something a bit bigger? Not sure if you have the EE experience to make any sort of recco though.

I was thinking the same thing… I know card audio guys add those giant caps to sustain the sudden power draw. I’d imagine there’s a right size cap for our voltage/current… but totally unsure on how to calculate that.

A quick google yields: If the Servo Misbehaves | Arduino Lesson 14. Servo Motors | Adafruit Learning System

I’ll start looking into this next. Would be great to go back to using the smaller Meanwell PSU.

Yeah… car audio caps are a bit of snake oil… those are more to make up for weak alternators or bad wiring - shouldn’t be necessary. (spent a past life deep in the car audio game).

But you do bring up a good point - batteries can be better at dealing with peak loads than switching power supplies, perhaps there is a battery based solution that could do the same thing…

We’re probably not discussing anything that hasn’t been vetted on the Tempest discord though.

Lithium cells like 18650 perhaps as the buffer?

The manufacturer or my servo control board (adafruit) recommends 100uf per servo as a starting point. I suspect the exact value depends on the stability of your PSU, wire thickness/length brownout voltage of the servo, but I lack the knowledge to make specific recommendations.

Perhaps someone should add a few capacitors to their setup and test this.

You can use lithium cells as a buffer, they do that for some high-powered motors drawing many hundred amps at startup. But you need specific servo’s accept some multiple of 3.7v, and if you aren’t careful and monitor the cell temperature you might start a fire…