2DOF Build Log Thread

And it's finished! Just have to do a final check of the footprints/connections, then send out an order.

Hopefully work will let me use some of their fancy high-current test equipment to test thermal/current carrying performance once I get the boards in .

Boards Arrived today! I decided to go for white because you never see PCBs in that colour. I still have to order the relays/barrier terminals, but hopefully I can get to testing soon.

It is true that I had never seen PCB of this color
Do not hesitate to put it under plexiglas so that it remains visible !!

Well all the pins seem to fit together, and all of the important signals seem to go where they need to go. Work has given me permission to use their test gear, so I'll post how hot the board gets under various current loads once I've got that stuff all set up. After that I'll order the relays/connectors and test the extra features that I added. If all goes well I'll release the board files somewhere on here for anyone to use if they want them.

I have a white UNO they do look nice in white

I did some thermal testing today:



I didn't use MMs or the relays, I just wanted to test the raw carrying capacity of the board (hence the wires jumping between the through hole pads). All of the testing was done in a room sitting at 22 degrees C, with very little ambient airflow. I tested from 10 to 50 amps of continuous DC current (until it reaches a stable temperature, usually about 20 minutes) through one channel, which is far more than a single MM is capable of outputting for any long period of time.

• 10 Amps: Peak temperature of 28.1 C
• 20 Amps: Peak temperature of 37.3 C
• 30 Amps: Peak temperature of 51.0 C
• 40 Amps: Peak temperature of 71.0 C
• 50 Amps: Peak temperature of 100 C

The main hotspot in each of the tests was the barrier terminal input pin. The terminals I ordered are only rated for 30 amps continuous, so this isn't a big surprise. The rest of the board was on average about 10-25 degrees cooler than this hotspot.

While some old graphics cards could reach 95-100C and continue working just fine (cough cough, Fermi, looking at you old GTX570), I wouldn't want the PCB to be running quite that hot during normal operation, however this really shouldn't happen unless you have your motors locked at stall current the whole time, and even then, the MMs should throttle if they go past 30A continuous, and your motor coils would likely also run into trouble (unless it's a monster and/or you run water cooling).

That being said, a sim is likely to draw a smaller average current, with intermittent big spikes, which is easier to thermally dissipate. Even further, once the whole board is set up, with fans and relays, peak temperatures likely won't reach nearly this high. I've also got images from a thermal camera of the current flowing which I'd like to post, but I need to get the OK from the boss first.

Also, I finally found a mistake with the board. The relay footprints are upside-down . They'll still work, they just have to sit on the bottom of the board unfortunately. I'll correct this at some point.

So I came across something interesting the other day. My school machine shop has some junk/spare parts that the students can sometimes take home. There are huge (1-5hp+) AC and DC motors, VFDs, power supplies, tons of bearings and mechanical parts, sensors, etc. I can't usually take any of this stuff because it's either too heavy, too expensive for them to give me, or I don't have a real use for it (I have enough interesting junk lying around as is).

However the shop supervisor gave me something interesting the other day that I think would be very useful to tinker with. It's not a full system, but he gave me the housing and internals of a small hydraulic dynamometer (dyno).


I'm not sure of the exact part number, the manufacturer no longer sells this exact model. The most similar one I could find on their website is good from 2 to 100hp, so sadly my motors are too weak to test, but perhaps this one can do lower power levels. I'll need to investigate more.

edit: it appears to be a DY series by Go Power
http://gopowersystems.com/wp-content/uploads/2016/01/DY-Series-Manual.pdf
I haven't found a minimum horsepower spec in the manual (although our motors are basically near zero power as far as the levels that these are meant to be used at).

Could be interesting to try and measure how powerful the common motors actually are, rather than the rated specifications.

So while sketching up the design for the folding base frame and aluminium top frame that I want to make, I decided to try and calculate what exactly the optimal angles for the motors and rod arms are, and I came across some interesting results that you guys might be able to get some useful information out of:

​

So it turns out that the optimal angle for the motors to be mounted at will actually vary from build to build. It depends on the distance and angle made between the platform pivot point (the U joint), and the point of the applied force (where the rose joints are mounted to the frame).

For maximum force transfer between the motor rods and the frame, the applied force should be perpendicular (90 degrees away from) the line drawn from the pivot point to the point of applied force.

If the force is not applied directly in-line with the optimal angle, as shown in the image above, the actual (useful) applied force to the frame can be found by calculating:

Actual Force = Applied Force * cos (Difference in Angle)

In the example above, the difference in angle is about 36 degrees, and cos(36) = 0.809, so about 81% of the applied force will go to actually pitching the frame forwards, the extra 19% is wasted applying tension between the frame and the U-joint.

The last factor that needs to be taken into account is the length of the line between the pivot and the point of applied force, as this will give us the final equivalent torque at the pivot point (or the force with which the frame will turn).

This is simply:

Torque = Actual Force * Distance Between Pivot and Applied Force

So even if a given angle is inefficient for force transfer, if the U joint and the point of applied force are far away, it gives the motors a lot more leverage to work with. This is mathematically why a shoulder mount sim can use faster motors with less torque compared to a platform one.

The more interesting/alarming thing that I found was this:

​

If you look at the roll axis on a typical shoulder mount rig, as shown in the above image, the difference in the optimal angle and the actual applied force is much larger than in the pitch example.

In this case, the difference in angle is about 65 degrees, cos(65) = 0.423, so only 43% of the force applied in the pitch direction is actually going towards moving the frame, the rest is wasted applying tension to the frame.

This was just something interesting that I found. The roll motion being inefficient is inevitable I think, just based on the geometry of how the sim has to be built. To get the optimal angle the motors would need to be much farther apart than is feasible, or the rose joints would need to be mounted much further down on the frame, which would mean poor efficiency and leverage for the pitch axis. Even with only 43% efficiency, the leverage gained by moving the mounts to the shoulder is enough to give good side to side movement, and we tend to want a more powerful pitch since it might need to move more weight (and who doesn't love those snappy gear changes). This could be something that someone that wants to build a flight oriented sim might want to consider, however.

On another note, I'm writing "Sim Calculator" application that would help people (me included) do these kinds of calculations when designing their builds, along with other stuff like CTC angles, movement speed, linear force, maximum weight etc. I'll upload a copy once I get a first build working.

TL;DR: Optimal motor/rod angle depends on U joint and Rose joint placement, most shoulder mount frames have poor efficiency in the roll axis.

Awesome Tim, Keep it up.

A shoulder mount makes only sense if you use linear actuators like SCN5/6, which are too long to mount under the seat.
Otherwise a compact design with the motors mounted under the seat is much more efficient, because the point of the applied force is almost at the same height as the pivot point.

This is true, a compact sim can have near 100% efficiency if designed properly, however the shorter lever angle means that high-reduction gearboxes need to be used to get the required torque to drive sim, and the higher you go in gearbox reduction, the more mechanical losses there are inside of the gearbox (worm gears aren't efficient to begin with).
Plus driving the sim from the shoulder allows for smaller vibrations to be more easily transferred to the user, they won't be damped through the frame like in a foot mount design.

I think each way of building has its own unique trade-offs and advantages, it just depends on what you want your rig to be best at.

When all is said and done, I think I'm going to go the DIY actuator route eventually, because the amount of free play in my CTC levers is kind of driving me nuts right now.

I agree, but only if the rods are mounted directly to the seat like in this case:

@Bernd Manger, true true, I have a set of bolts that clamp my seat back directly to the frame at the point of contact near the shoulders to maximize this effect. Probably not as good as bolting directly to the seat, but a proper rigid bucket seat costs a bit more than I was willing to pay .

As far as the second question goes, someone like @noorbeast would be better suited to answer, I don't have much experience with compact sim frames. But I believe it comes down to the fact that at some point you need to reduce your gearbox further, because your CTC lever would end up needing to be impossibly short.

The lever I have on my rig is about 5cm long, with an effective lever length of about 95cm from the rose joint to the platform pivot. If I were to mount my motors directly to the bottom of the frame, my lever length from the point of application would only be about 35cm, which would mean that for the same equivalent torque at the pivot, I'd need a CTC length of about 1.4cm, which is smaller than I can fabricate because the holes would overlap.

Hello Tim, it's clearer now:
The main reason to chose a higher reduction when building a compact sim is the speed getting too high, not the lack of torque.
My sim is a seatmover with wheelchair motors ratio 23:1, ctc 3cm, an effective lever length of about 30 cm and it's very very fast. But combined with a compact fullframe design this ratio would be too fast.

This problem i have reduced by moving the pivot point a bit out of balance, so there is always pressure on the gearbox, but going the DIY actuator route is definitely a better idea.

Sorry for "desturbing" your threat.

@Bernd Manger No need to apologise at all! I love these kinds of discussions

@all: SimCalc has been moved here https://www.xsimulator.net/community/threads/simcalc-by-tim-mcguire.11277/ and is sticky.