Servo Roundup Test #1 - Sound (unloaded)

UPDATE - Results posted here (DON’T PEEK IF YOU HAVEN’T VOTED YET!)

One of the most-asked about topics in OSR building is - What is the best servo?

It’s a very difficult question to answer as servos all have different characteristics in terms of noise, speed, smoothness, and power.

I have brought together several different servos that I have used over time that I feel can be helpful in drawing some comparisons and reference points. This is by no means an exhaustive servo test, but I have really focused on establishing a testing methodology that is scientific and repeatable.

The first test that I will present is focused on un-loaded (no weight on servo) SOUND. This seems to be the #1 reason people look for alternative servos. This was important for the original OSR2 and even more so for the SR6 which has TRIPLE the amount of servos of the original OSR2. I will follow this test up with a loaded sound test and then provide other tests that will display smoothness of motion, and speed.

I have some background in speaker design and measurement (not a professional), and know when it comes to sound - subjectivity can play a huge role. As a result, many consumers become fixated on empirical measurements such as decibels, watts, ohms, etc. as the nuance between “noise” can be difficult to discern. Most commonly, the usage of decibel readings for sound is very flawed. They don’t take the quality of sound or perceived loudness into account at all. For example - white noise at 60db may be considered pleasant, but a 2000hz saw wave at 60db would be annoying as fuck. Moreover, decibel readings don’t cover how transmissible sounds are - lower frequencies are more likely to travel through boundaries, whereas high frequencies can be more easily mitigated by covering them up with a blanket, closed door, etc.

It is for these reasons that I have decided to conduct my comparison wholly via controlled recordings and not provide any decibel readings. I looked into LUF measurement (becoming the new standard for loudness measurement), but figured that live recordings would be the easiest and most relatable way to provide my data.

I recorded 90 degree sweeps of movement for these servos at a fixed speed and voltage (6.25v via a 150watt Meanwell power supply). Recording was done via a Yeti microphone from 6" away. There is no processing, normalization, compression, etc. done on the audio clip (other than what Youtube does). Recordings for all 4 of the servos were done in a single take. I trimmed out the “dead air” between each of the servo swaps just to make the clip shorter.

As I mentioned, since sound is so subjective, I will not be identifying the servos for 5 days. I would LOVE if you would provide feedback, thoughts, and vote on the poll before the servo brand/model reveal. Thank you all for participating and I hope that this will be helpful.

Select the FIRST most pleasant/quiet servo from the test:

  • Servo #1
  • Servo #2
  • Servo #3
  • Servo #4
0 voters

Select the SECOND most pleasant/quiet servo from the test:

  • Servo #1
  • Servo #2
  • Servo #3
  • Servo #4
0 voters
11 Likes

Oh my that 2nd servo was horrible, would work great for showing a machine under stress in a movie like it trying to resist a weight come down on it. 4th servo to me sounded like something moving fast though noisier than 3 and 1, can’t quite place what it sounds like but maybe an alarm for just a small quiet room instead of doing the whole building.

3rd one for me wins since it is by far the least noisy when at anything but it’s peak speed from the sounds of it or when it’s in the middle of it’s sweep. While the 1st overall is consistent with it’s volume the sound it makes would mean that it could fade easily into the background if kept running forever but the 3rd one’s volume is less so overall that despite the changing sound it’s easier to deal with especially if you wear any headphones and maybe not as noticeable if in short use that someone would get annoyed enough to look for the source of the sound.

2 Likes

3 won by being the quietest and most pleasant. The rest all would be too loud for me personally. 1 sounded really cheap, and 4 sounded stronger to me, just my initial opinions.

1 Like

Keep in mind that this was taken from only 6" away :slight_smile: Not necessarily meant to be a representation of how loud the servos are individually, but more so in comparison against each other.

2 Likes

I have just gotten too used to the noise level of the handy I think. Have you thought of doing a twisting test on the servo gear to generate sound?
Also as a side note I’m currently and very slowly rebuilding my SR6 and splurged on some Flash Hobby servos. Do you have any experience with those (was it number #3 :smile:)?

whats a twisting test?

I meant just twisting them by hand and comparing the sound difference, but I don’t know how you would do that consistently or if it would provide any real data.

Hmmm. Not sure how that would be different than the controlled test I did. Other than being less accurate and consistent. Most servo manufacturers discourage doing that too to avoid damage to the gears

quite obvious i would say… and the votes show that too. amazing how much effort you are doing to get the best results… iam sold an will purchase one of the next drops if iam fast enough that is ^^

1 Like

I also really like 3. I am looking forward to your future testing. I have a Vorze A10 Piston but I have been incredibly interested in multi-axis content so I’ve been looking at the OSR2 and SR6. I think you’ll help influence my buying decisions!

1 Like

This would be more useful with performance specs and voltage used. If a servo is particularly weak, it doesn’t really help that it’s quiet.

I’d also say that while servo #3 was quieter, it was a more grating sound than servo #1.

I should have mentioned. These were all using the same voltage of 6.25v. I’ve amended the OP.

What sort of performance specs are you looking for? I don’t trust the KG rating from the manufacturers since there is really no verification for that.

I certainly don’t expect you to go measure the actual values, but the published figures from the manufacturer shouldn’t be entirely meaningless. Unless it’s a really cheap Chinese servo which exaggerates everything.

For example, the specs for my servos (Savox SB2292SG) are the following:

Torque:
22kg @6v
31kg @7.4v
45kg @8.4v

Speed (sec/60 degrees):
0.085 @6v
0.07 @7.4v
0.055 @8.4v

So at 7.4v, the servo is rated to have 31kg-cm of torque and take 70ms to move 60 degrees.

It is far from quiet, but it performs extremely well. So the question someone like me would have to ask is what compromise on performance is acceptable for a reduced noise level.

Gotcha - I’ll definitely articulate exactly which models were included in the test and they should all have available datasheets for folks to pore over. I still would encourage people to not get fixated on published specs - I don’t find that they really represent performance for our use cases.

1 Like

OKAY! Thank you to everyone who took the time to audition the comparison and vote! I really appreciate the participation.

Here is the big reveal (servo #'s match the video test #s) along with cost as of 3/31/2024 (Amazon cost if available)
#1 - FunOSR - AI-H200 - $41.78 + $29.16 shipping
#2 - Generic - DS3230 - $17.00
#3 - Flash Hobby - M45CHW - $41.59
#4 - Hitech - D954SW - $99.99

This was just the “sound” portion of my testing, so the comments below will be restricted to sound/noise, and I will discuss smoothness and speed in my other tests.

First Place:
The Flash Hobby M45CHW is one that I’ve been recommending for some time now. It is extremely quiet and (somewhat) affordable. I have tested some other Flash Hobby servos around the same price point, and some +$20, and they are all similar in noise. Some of you with good ears can hear a high pitch undertone - your ears are not tricking you. These are very quiet, but sound very strange. Thankfully, high pitch sound is very bad at transmitting through any sort of boundary, so headphones or earbuds totally mask the high pitch squeal, so it’s not an issue in real world use - just a minor annoyance for those of us who can still hear above 12khz.

Second Place:
FunOSR shipped me 3 samples of their AI-H200 for testing and really was the push for me to conduct this series of tests/comparisons. The community definitely owes them a thank you for helping kick this off. Unit to unit consistency was very good, and all 3 units performed identically to my ear. These are extremely quiet servos. The quality of motor noise was average, with some inconsistency of pitch depending on direction of movement. This could potentially be annoying to some, as the sound of the toy will vary more based on the use/movement. At a similar price to the Flash Hobby, I would be 50/50 in terms of recommending this servo. However, the shipping cost to the US, as well as the inability to easily return/replace servos makes it a tough recommendation to those located in the US like me. If FunOSR can get a local distributor to help eliminate shipping costs, and build a US based support network to handle warranty and RMAs, then they could have a strong contender on their hands for a highly recommended quiet servo.

Third Place:
The Hitech D954SW was one of the most highly recommended servos when I started diving deeply into the OSR2 world. Their reputation as a servo manufacturer is stellar and they have been in business for more than 50 years! It’s amazing how current quiet servos have surpassed this previous recommendation in just a few years. Even though they are not anywhere near as quiet as some of the recent quiet servos, I still love mine and use it in my personal machine. Some may have noticed (I see a few 2nd place votes for the Hitech) that the quality of noise is very good. It’s super consistent and sounds… precise. So while it may not be the quietest servo in the group, the quality of noise communicates a sense of high build quality overall. Unfortunately, it is priced at more than 2x the cost of the Flash Hobby and the FunOSR (without shipping), so making a recommendation of this servo based on noise is extremely hard to do. It could still be a good option for those that are annoyed by the high pitch sound of the Flash Hobby servos or want to purchase from a highly established company with a strong North America presence.

Fourth Place:
The DS3230 was included ONLY to provide a baseline for comparison. This is a variant of the standard generic China servos - they all sound pretty much the same. These aren’t “bad” servos for those on a budget and/or mostly watch PMV/HMV with loud music and headphones. While its easy to hate on how loud they sound - I do appreciate them for getting so many people into the hobby at a lower price point. It’s very hard to mentally make the jump to spend $300 on servos vs $45 in a DIY 3D printed toy without ever trying one. I still recommend these to folks on a budget who don’t have major concern with noise. IMHO, an OSR2+ with these cheap servos will outperforms the Handy in every way other than noise.

9 Likes

Do you happen to have tested any of the brushless Flash Hobby ones? They advertise them [M45BHW Brushless Servo] with an msrp of 40, so they could still be slightly cheaper than the competition and maybe even more ultra quiet :exploding_head:

Yes. They sound almost identical to the coreless ones. I tend to not prefer brushless since they are less smooth in terms of motion. Def not a bad choice although comparably priced brushless servos tend to be rated a bit less torque, or less resolution. Not sure if thats the case for these flash hobbies. Haven’t done a direct spec comparison since I’m not a fan of brushless.

1 Like

Thanks for doing this comparison!

1 Like

Update:

While conducting further testing, I encountered 2 issues with the AI-H200 from FunOSR.

  1. 2 of the 3 servos that I was delivered for testing/review developed shaking/vibration at rest. This is likely due to a bad encoder/potentiometer. Servos rely on their encoder to determine their position in their travel. If the encoder is giving back incorrect positional feedback, the servo will try to self-correct, sending itself into a feedback loop of shaking. This is not typical with new servos. I’ve seen it with servos after heavy use, but not ones that have only been bench tested. Anecdotally, another user has reported the same issue with a production unit that they purchased from FunOSR.
  2. The AI-H200’s sound signature has changed after periods of stress testing. Some users have theorized that their low sound signature is due to copious amounts of grease applied to the gears, effectively damping vibration and sound transmission. After a couple hours of use, I think that the grease may have migrated away from the gears, changing the sound signature of the AI-H200. It is not a loud servo at this point, but not as quiet as first measured.

Issue #1 is a big one - reliability is a huge focus point of mine. No one likes troubleshooting a sex toy. It’s the biggest boner killer of all time. Based on this, I’m not going to proceed with testing the AI-H200. I may put together another servo roundup in the future, but this is on hold for now.

1 Like

Hey, hoping this is okay to ask in terms of etiquette as I’m new here

I’m starting to source parts for my SR6 with very little knowledge on these kinda things (especially electronics), following the BOM spreadsheet and came across this post as sound is an important consideration for me. Does anyone know if using 6x of the 180 M45CHW and 1x 270 M35CHW would require swapping out any of the parts besides the power supply / work the same in the build as the typically recommended servos otherwise? Besides servos and the solderless power bus I’m looking to follow the typical instructions/bom pretty closely 1 to 1

Thanks <3