As many of you know - the ultimate goal of the OSR2+'s I build is to do everything I can to help ensure reliability and ease of maintenance. In my opinion, it doesn’t matter how well a unit performs if it doesn’t hold up long term. This has been the driving factor behind my component selection, print settings, assembly practices, and setup guides.
With this upcoming #9.0 launch, I will be introducing 2 changes to the g90ak OSR2+ that I’ll be reviewing below.
Arms
I’ll start with the more minor change to the arms of the OSR. The 2 screw points that affix the main and pitcher arms to the servo horns have classically relied on the plastic to support the screws. Many users really like to torque these screws down a LOT. They may think it helps improve performance, but it results in the screws crushing the plastic and “blowing out” the backside of the arms around the mounting point. Additionally, it is challenging to print the “shelf” for the screw head to sit against since having a perfectly flat shelf is impossible to print, as printers cannot print floating floors without support. This is a big reason why the original arms had “cone” shaped mounting holes to allow for easier printing. However, the cone shape is prone to crushing when screws are aggressively tightened.
I’ve mitigated these issues by using washers and locking washers, but I still have had a few reports of blown out arms here and there. I went back to the drawing board and have improved things by really looking at distributing the screw force across a greater area and relying less on the plastic to support the screw head. Rather than the cone shaped hole face, I’ve moved to a flat hole face. A big enabler for this is swapping to a dual nozzle printer, which allows me to print the supporting material out of a dissimilar plastic, allowing for clean support removal without contaminating the plastic of the actual part. With that out of the way, I’ve selected a much wider fender washer to help distribute the force. The new fender washer I’m using has more than 3.5x the surface area of my previous washer and pressure distribution is MUCH better. Yes, it doesn’t look as clean as the smaller mounting holes, but the reliability will be MUCH better going forward.
If any of my previous owners would like to upgrade to these new arms - I am happy to upgrade them for the cost of shipping + $5 to cover the cost of the hardware components. Plastic and labor costs are on me.
Check out how clean the screw mounting surface is - perfect home for the wider washer.
New H300 Servo
The second change is a more major one that requires even more explanation.
In short, I will be moving to FunOSR’s H300 servo in a “pilot/release candidate” fashion over the next few drops with the goal of a 100% cut over to the H300 if successful. I will call out that I have been testing the H300 for 4 months on a daily basis, so please don’t view the new servos as a blind test or beta. I’ll go into the results of my testing after my explanation below.
Servos are the most important component of any OSR. They’re responsible for ALL of the brute force muscle to move the toy. Additionally, servos are really amazing pieces of technology. Not only are they able to create fast moving torque out of an extremely small package, they are also positionally aware - this sets them apart from many other types of motors - they know where they are at all points of travel. This positional awareness is what makes them so well suited for sex toys, but is also one of their biggest weaknesses and modes of failure.
Of the thousands of servos that I’ve sent out into the wild, it is EXTREMELY rare for me to see any failures due to gearing or control electronics. By FAR the most common failure is due to bad potentiometers. This is what I’m trying to address with a servo model change.
What the hell is a potentiometer anyway? The potentiometer or “pot” is responsible for dictating the servo’s position. This is how the servo knows how to move up or down or anywhere in between. Potentiometers use a small metal wiper that travels over a track. Based on the arm’s reading, the servo knows where it is at any given point in time. It is important to note that this is a PHYSICAL contact point that is used. Pots are present in SO many devices, volume knobs, game controllers/joysticks, DJ equipment, and more. However, the physical contact of the wiper can wear out over time or get contaminated leading to issues.
In order to explain common behavior of a bad potentiometer, I’m going to use an analogy of a mail carrier trying to deliver mail to a series of 5 houses on a street simply numbered 1-5 (In reality, a servo has many more steps, but I’m keeping things simple). Imagine your OSR moving from bottom to top.
In a properly functioning servo/pot, each house is clearly addressed, allowing the mail carrier to easily move from one house to another, stopping properly along the way. The mail carrier is never confused and can smoothly move along without issue.
With a bad pot, one of the houses is mislabeled (house #4 has a #5 instead of its correct address). For most of the carrier’s route, everything is fine, but when they get to house #4, they now read they’re at house #5. They rush back one house, then realize they’re at house #3, they then attempt to move forward to house #4, run back to #3, and then repeat over and over again.
This exact behavior is why some servos will start jittering/vibrating at certain points of their travel. Additionally, they can also display random “jerking” at certain points. A bad pot can be very hard to detect for 2 reasons. Firstly, OSRs have all of their servos tied together mechanically at the receiver connection points. A single servo vibrating will transmit that vibration to the others, making it hard to tell if a specific servo is causing the issue. Secondly, as demonstrated, MOST of the pot track may be still good (all houses except for #4), making it hard to reproduce. For these reasons, my method of troubleshooting is to disconnect the receiver, allowing the servos to each move freely. I then increment the servos travel using MFP 1% at a time to determine where in the travel the servo may start producing vibration/jitter.
So, we’ve established that the analog potentiometer is a failure point. This is very similar to how many game control sticks get “drift” after use/time. The answer there is to move to hall effect sensors, eliminating the physical contact/wiper that potentiometers use. I’ve been searching for a servo equivalent for several years without much luck until now. Digital/magnetic encoders within servos are fairly rare. The servos that I’ve found that have this technology typically require a different (non-PWM) control signal and don’t come in the form factor that we need. I wrote them off as a dead end until I came across the H300 from FunOSR.
FunOSR’s H300 is advertised as featuring a magnetic encoder. This is the main reason I purchased a set and have been testing them. Simply put, the magnetic encoder could be a gamer changer for us. Since it has a magnetic encoder instead of a potentiometer, there is no longer a physical wiper to wear out. In theory, this should provide a much greater degree of reliability. I haven’t disassembled one of these servos to verify, but I wouldn’t be able to identify a magnetic encoder vs a traditional pot anyway, so I’m going on their word. I’m excited to be implementing the H300s into my OSR2+‘s as it will ultimately bring servos’ reliability up to the standards that I hold my design/plastics/fasteners to.
H300 Impressions
Across the past 4 months I’ve been testing the H300 servos in my personal OSR2+ and have been happy with them. I run my OSR2+ daily for 15-30 minutes with a mix of scripts, but mainly fast action high stroke distance ones. Here are my impressions of their performance.
Speed - Plenty fast, keeps up with the fastest scripts I have, even at 1.5x speed. Zero complaints here.
Noise - MUCH quieter than generic servos. Just a LITTLE louder than the Flash Hobby M45CHW servos, but the tradeoff for reliability is worth it. I’d be hard pressed to pick the louder one between the H300 and the M45CHW in a blind test.
Smoothness - A bit better than the Flash Hobby M45CHW in terms of smoothness, and much more smooth than generics (no comparison at all). Not as smooth as my favorite Hitech D954SW, but I’ve recently had 3 Hitechs end up with bad potentiometers, AND they cost $110 each, so really no comparison there.
Heat - Not any more noticably hotter than any other servo tested other than the Perfect Pass Monster servos which run very hot. No worries about heat issues with the H300 in my testing.
Notes - These are a bit more power hungry, I run them at 8v, whereas I reccommend using ~6.5v for the M45CHW and D954SW servos. Also, they don’t seem to come homed from the factory. Definitely a pain in the ass and an extra step to home them before installing the servo horn. I’ve reached out to FunOSR to see if this is intentional or if it’ll be fixed in the future.
As I mentioned above, the servos are really the heart of the OSR, so this change was not made lightly. This is why I’ve been testing these across such a long period of time before putting them into production units. I really want everyone’s OSRs to be as absolutely reliable as possible - no one likes troubleshooting in the middle of a session. This should be a good jump for the g90ak OSR2+'s long term reliability. Please let me know your thoughts and questions.
These changes will go live with Drop #9.0 on 9/1/2026. Thank you so much!
