Search
Forum
Wiki
Images
Files
Member List
Help
04-16-2023, 03:15 AM
(This post was last modified: 04-16-2023, 09:10 PM by Dyne.
Edit Reason: More details, more accurate calculations
)
With the external cosmetics largely complete, I decided to try and sort out some mechanical and electrical issues in LD-F1's head.
First, however, I made my life a lot easier by wiring up a cable to let me power the droid from my benchtop power supply (ring terminals on one end, XT60 female on the other). This way, I don't have to fool with installing a battery in the droid just for testing stuff.
Then I removed the shorter white colored side of the head so I could see what's going on in there. Below is the view into the head.
![[Image: 20230415_171109-smaller.jpg]](https://www.niflheim.net/droidbuilding/images/LD-F1/20230415_171109-smaller.jpg)
On the right side, you will see a 9g servo on the back wall and a vertical white tube near the edge. The servo is new; it's intended to raise and lower the white tube (which is the stalk for the droid's antenna). Getting this working requires running a new signal wire into the head from the R/C receiver in the body, as well as designing a mechanism to translate the servo's rotation into lift, which is a bit awkward due to the suboptimal mounting position. I'm thinking of a disc shaped servo horn, with a little arm extending back toward the rear wall of the head. This arm sits underneath a horizontal arm that'd be attached to the base of the antenna, pushing it up or letting it down as the disc rotated.
The most obvious issue that I see with the rest of the stuff in the photo is that the ESP32 holder has broken free from the back wall and is hanging at an angle. It's not visible in the pic, but one or more jumpers/pins are also bent at weird angles and potentially loose. This would explain the issues I saw at Dragon Con, where the ESP32 rebooted whenever the head would look downward or tilt (presumably the board was losing power). However, when I took the last photo in my previous post, I noticed that the front LEDs were not powered on at all. This means either the microcontroller has died or something is entirely disconnected. This will be a bit awkward to fix without removing the head shell entirely, so I haven't gotten to doing much yet.
As mentioned back in post #54, when I reattached the servo horn that had come loose at Dragon Con (the one that's furthest from the camera) back in February, I did so without really re-centering the servo. I knew I'd eventually need to fix that (if I didn't, I'd have to keep the transmitter stick trimmed in order to level the head, which reduces the movement range). While looking at this, I also found that the tilt servo linkages were set to different lengths, so I fixed that.
While the head is open, I can easily access the servo on the port side of the droid's head. Unfortunately, the one that I really needed to fix was the starboard servo. I can tell you that it's a big pain to work on that one, whether the head is open or not. At least that slotted "vent" on the starboard side of the head allows me to see inside and sometimes stick tools in. But the only way to make the starboard servo easy to work on is to remove the exterior shell of the head entirely. I could do that, but as I recall, it was even more of a pain to get the head attached properly than it is to work on the starboard servo. (I may yet have to do it anyway to sort out the ESP32 issues, but I'd rather not.)
While moving the head around to test my fixes on the servo horns, I noticed that the bracket holding each tilt servo flexes a bit more than I'd prefer. I may print up and glue a brace between them, which I was considering doing anyway, or else I'll revise the parts for strength.
In the image below, you may notice that some of the wires that stretch across the head rotation gears are a little wavy. That's because they were getting caught in the teeth, which can't be good. I have printed some C-shaped cable clips, and I am gluing these to the interior head in order to secure the wiring away from the mechanism a bit better.
![[Image: 20230415_171153-smaller.jpg]](https://www.niflheim.net/droidbuilding/images/LD-F1/20230415_171153-smaller.jpg)
Stepping away from the head, I mentioned back in post #36 (and again in post #54) that I suspected I should have pulleys with higher tooth counts in the belt drive in place of the current 10T pulleys. This was because I knew this size made the print a bit thin and weak near the flange end. And as it happened, one of them did indeed shear off at that exact spot during Dragon Con.
This would make the belts a bit less likely to slip (more teeth to engage the belt and a tighter fit), and possibly less likely to break given a larger pully means the belt wouldn't be making as sharp a bend, but I also know doing this would change the gear ratio. I haven't committed to a change yet, but I did pop into Fusion and model a 12T version of my motor pulley, just in case I wanted to use it at some point.
I eventually want to convert all of my LIPOs to LiFePO4, which are substantially safer. Without building my own pack, the closest thing I've found to the current battery in size and capacity is a 3S LiFePo4. That'd lower my nominal voltage from 11.1 to 9.9, thereby decreasing the droid's raw speed quite a bit. But if I ignore the battery change, here's the impact a gearing change might have on the droids performance:
I originally calculated LD-F1's drive back in post #9, and arrived at a raw value of roughly 50 mph at nominal battery voltage, assuming that he had no gear ratio at all (therefore he'd have little torque beyond what the motor produces, and the motors would stall instantly under the load).
Since I wanted enough torque to actually move the droid, and a top speed closer to the range of 2 to 4 mph, I needed a gear ratio near the range of (50/2) = 25 to 1 (slower, more torque) and (50/4) = 12.5 (faster, less torque). Because the primary belt goes from a 10T to 30T pulley, and the secondary belt goes from a 10T to a 42T pulley, the droid is currently geared to a ratio of (30/10) * (42/10) = 3 * 4.2 = 12.6 to 1, so he'd be pretty close to the fast, lower torque end (max speed of 3.97 mph)
Except ... I was intending at the time to use a belt for the tracks rather than the current link + traction pad setup, so I ignored their thickness and got the raw 50 mph result by assuming a 4 inch diameter for the drive sprockets. In reality, the current modeled radius from the axis of the drive sprocket to the center of a traction pad is 54.665 mm, which gives a final diameter of 109.33 mm or 4.304331 inches. Thus, LD-F1's raw ground speed with no speed reduction is actually higher than ~50 mph. It's more like 54 mph. That makes the desirable gear ratio between 27 and 13.5 to 1. He is already below that, and with the current gear ratio his current max speed is more like 4.29 mph.
Replacing one of the 10T pulleys on each side with 12T would make the ratio (30/12) * (42/10) = 10.5 to 1, which would give LD-F1 a top speed around (54/10.5) = 5.14 mph. Replacing both 10T pulleys on each side would make it (30/12) * (42/12) = 8.75 to 1 ratio, with a top speed of (54/8.75) = 6.17 mph.
In truth, LD-F1 could probably stand to be a little bit faster than he currently is, at least if I can keep max speed from breaking/throwing his belts. But those increased speeds do come at the cost of decreased torque. I'm not sure how much loss of torque I could get away with for LD-F1, but I do know carpet can sometimes give him slight problems with turning as it is, as will some uphill climbing. Right now both are more due to belt slippage than stalling, but lowering the torque too far could easily exchange a minor problem for a major one, so I don't think I care to go as low as the 8.75 to 1 ratio.
I'd probably only want to replace a single pulley on each side to get the 10.5 to 1 ratio. I'd have to decided whether to replace the 10T motor pulley or the 10T part of the compound intermediate pulley. It should probably be whichever is more prone to shearing, which suggests the compound pulley.
Of course, going up to 12 teeth isn't the only option here. I could replace both 10T pulleys on each side with 11T versions, and I'd get similar numbers to what I get from a single 12T pulley (ratio of 10.42 to 1 and top speed of 5.18 mph). I'm just unsure whether 11T would give enough additional strength in the parts and grip for the belts to bother with.
First, however, I made my life a lot easier by wiring up a cable to let me power the droid from my benchtop power supply (ring terminals on one end, XT60 female on the other). This way, I don't have to fool with installing a battery in the droid just for testing stuff.
Then I removed the shorter white colored side of the head so I could see what's going on in there. Below is the view into the head.
On the right side, you will see a 9g servo on the back wall and a vertical white tube near the edge. The servo is new; it's intended to raise and lower the white tube (which is the stalk for the droid's antenna). Getting this working requires running a new signal wire into the head from the R/C receiver in the body, as well as designing a mechanism to translate the servo's rotation into lift, which is a bit awkward due to the suboptimal mounting position. I'm thinking of a disc shaped servo horn, with a little arm extending back toward the rear wall of the head. This arm sits underneath a horizontal arm that'd be attached to the base of the antenna, pushing it up or letting it down as the disc rotated.
The most obvious issue that I see with the rest of the stuff in the photo is that the ESP32 holder has broken free from the back wall and is hanging at an angle. It's not visible in the pic, but one or more jumpers/pins are also bent at weird angles and potentially loose. This would explain the issues I saw at Dragon Con, where the ESP32 rebooted whenever the head would look downward or tilt (presumably the board was losing power). However, when I took the last photo in my previous post, I noticed that the front LEDs were not powered on at all. This means either the microcontroller has died or something is entirely disconnected. This will be a bit awkward to fix without removing the head shell entirely, so I haven't gotten to doing much yet.
As mentioned back in post #54, when I reattached the servo horn that had come loose at Dragon Con (the one that's furthest from the camera) back in February, I did so without really re-centering the servo. I knew I'd eventually need to fix that (if I didn't, I'd have to keep the transmitter stick trimmed in order to level the head, which reduces the movement range). While looking at this, I also found that the tilt servo linkages were set to different lengths, so I fixed that.
While the head is open, I can easily access the servo on the port side of the droid's head. Unfortunately, the one that I really needed to fix was the starboard servo. I can tell you that it's a big pain to work on that one, whether the head is open or not. At least that slotted "vent" on the starboard side of the head allows me to see inside and sometimes stick tools in. But the only way to make the starboard servo easy to work on is to remove the exterior shell of the head entirely. I could do that, but as I recall, it was even more of a pain to get the head attached properly than it is to work on the starboard servo. (I may yet have to do it anyway to sort out the ESP32 issues, but I'd rather not.)
While moving the head around to test my fixes on the servo horns, I noticed that the bracket holding each tilt servo flexes a bit more than I'd prefer. I may print up and glue a brace between them, which I was considering doing anyway, or else I'll revise the parts for strength.
In the image below, you may notice that some of the wires that stretch across the head rotation gears are a little wavy. That's because they were getting caught in the teeth, which can't be good. I have printed some C-shaped cable clips, and I am gluing these to the interior head in order to secure the wiring away from the mechanism a bit better.
Stepping away from the head, I mentioned back in post #36 (and again in post #54) that I suspected I should have pulleys with higher tooth counts in the belt drive in place of the current 10T pulleys. This was because I knew this size made the print a bit thin and weak near the flange end. And as it happened, one of them did indeed shear off at that exact spot during Dragon Con.
This would make the belts a bit less likely to slip (more teeth to engage the belt and a tighter fit), and possibly less likely to break given a larger pully means the belt wouldn't be making as sharp a bend, but I also know doing this would change the gear ratio. I haven't committed to a change yet, but I did pop into Fusion and model a 12T version of my motor pulley, just in case I wanted to use it at some point.
I eventually want to convert all of my LIPOs to LiFePO4, which are substantially safer. Without building my own pack, the closest thing I've found to the current battery in size and capacity is a 3S LiFePo4. That'd lower my nominal voltage from 11.1 to 9.9, thereby decreasing the droid's raw speed quite a bit. But if I ignore the battery change, here's the impact a gearing change might have on the droids performance:
I originally calculated LD-F1's drive back in post #9, and arrived at a raw value of roughly 50 mph at nominal battery voltage, assuming that he had no gear ratio at all (therefore he'd have little torque beyond what the motor produces, and the motors would stall instantly under the load).
Since I wanted enough torque to actually move the droid, and a top speed closer to the range of 2 to 4 mph, I needed a gear ratio near the range of (50/2) = 25 to 1 (slower, more torque) and (50/4) = 12.5 (faster, less torque). Because the primary belt goes from a 10T to 30T pulley, and the secondary belt goes from a 10T to a 42T pulley, the droid is currently geared to a ratio of (30/10) * (42/10) = 3 * 4.2 = 12.6 to 1, so he'd be pretty close to the fast, lower torque end (max speed of 3.97 mph)
Except ... I was intending at the time to use a belt for the tracks rather than the current link + traction pad setup, so I ignored their thickness and got the raw 50 mph result by assuming a 4 inch diameter for the drive sprockets. In reality, the current modeled radius from the axis of the drive sprocket to the center of a traction pad is 54.665 mm, which gives a final diameter of 109.33 mm or 4.304331 inches. Thus, LD-F1's raw ground speed with no speed reduction is actually higher than ~50 mph. It's more like 54 mph. That makes the desirable gear ratio between 27 and 13.5 to 1. He is already below that, and with the current gear ratio his current max speed is more like 4.29 mph.
Replacing one of the 10T pulleys on each side with 12T would make the ratio (30/12) * (42/10) = 10.5 to 1, which would give LD-F1 a top speed around (54/10.5) = 5.14 mph. Replacing both 10T pulleys on each side would make it (30/12) * (42/12) = 8.75 to 1 ratio, with a top speed of (54/8.75) = 6.17 mph.
In truth, LD-F1 could probably stand to be a little bit faster than he currently is, at least if I can keep max speed from breaking/throwing his belts. But those increased speeds do come at the cost of decreased torque. I'm not sure how much loss of torque I could get away with for LD-F1, but I do know carpet can sometimes give him slight problems with turning as it is, as will some uphill climbing. Right now both are more due to belt slippage than stalling, but lowering the torque too far could easily exchange a minor problem for a major one, so I don't think I care to go as low as the 8.75 to 1 ratio.
I'd probably only want to replace a single pulley on each side to get the 10.5 to 1 ratio. I'd have to decided whether to replace the 10T motor pulley or the 10T part of the compound intermediate pulley. It should probably be whichever is more prone to shearing, which suggests the compound pulley.
Of course, going up to 12 teeth isn't the only option here. I could replace both 10T pulleys on each side with 11T versions, and I'd get similar numbers to what I get from a single 12T pulley (ratio of 10.42 to 1 and top speed of 5.18 mph). I'm just unsure whether 11T would give enough additional strength in the parts and grip for the belts to bother with.