Practical Quad Dynamics Sim - Page 2 - RC Groups
Thread Tools
This thread is privately moderated by AILERON8, who may elect to delete unwanted replies.
Jan 06, 2018, 12:45 AM
Registered User

PQD - v1.5


Here's the latest & greatest - PQD v1.5

I'll be adding projects(motor/prop combos) to the dB over the weekend with calibrated Cd's. I'm hoping to compile an executable this weekend as well.
Last edited by AILERON8; Jan 06, 2018 at 01:00 AM.
Sign up now
to remove ads between posts
Jan 06, 2018, 01:52 PM
Registered User
For now it looks like this will need to be run as an Access dB, but at least the MS Access Runtime is free:

https://www.microsoft.com/en-US/down....aspx?id=50040
Jan 06, 2018, 02:58 PM
Registered User
Here's the latest version: https://www.dropbox.com/s/mw5yemfwzr...1.5.accdb?dl=0
Jan 06, 2018, 03:43 PM
Registered User
I was just thinking of next additions, but am curious what folks thoughts are on the concepts below?
  • Throttle linearity: If a secondary calculation compared the max dist, accel, velocity, and jerk at idle, 25%, 50%, 75%, and 100% throttle I believe the latter 5 data points might give a good indication for throttle response when comparing motors, props, and frames.
  • Dynamic Efficiency: Using the same technique as throttle linearity I'm thinking motor/prop/frame efficiency may be estimated with relative accuracy by examining g/W differentiated with respect to throttle %.
  • KV & Prop Size Adjuster:I talked about this one before. If a motor only has data for a particular KV or simply doesn't exist yet I think it would be handy to have a way to estimate the quad dynamics of a different KV motor or prop size using a known motor/prop/frame configuration as a baseline. It's not a simple extrapolation of course, these are nonlinear relationships, but there are ways to minimize error for a "good enough" estimation.
  • [B]These are on the tip of my tongue, but there are others. Any requests? Thank you
Jan 06, 2018, 03:57 PM
Forward ever, backward never!
Brainstorm's Avatar
Quote:
Originally Posted by AILERON8
For now it looks like this will need to be run as an Access dB, but at least the MS Access Runtime is free:

https://www.microsoft.com/en-US/down....aspx?id=50040
Cool that MS Access Runtime is free! Of course, you still need Winblows to run it. Bummer!

I haven't run a Windows box in many years (that Win XT tower is buried somewhere in garage), and I refuse to install a Windows environment on my Mac. However, it might be time for me to pick up a cheap Win10 laptop. That would be barely more expensive than a separate license. And a dedicated Windows computer would obviously come in handy for the few Win-only programs that would still be useful to run, like this Access DB!
Jan 06, 2018, 04:52 PM
Registered User
Quote:
Originally Posted by Brainstorm
Cool that MS Access Runtime is free! Of course, you still need Winblows to run it. Bummer!

I haven't run a Windows box in many years (that Win XT tower is buried somewhere in garage), and I refuse to install a Windows environment on my Mac. However, it might be time for me to pick up a cheap Win10 laptop. That would be barely more expensive than a separate license. And a dedicated Windows computer would obviously come in handy for the few Win-only programs that would still be useful to run, like this Access DB!
Thanks! Yeah, I know, ms Access is terrible... lol At some point here I think I'll build this in C, Java or Python, but I'm slower in those languages and right now I'm in a rush to get a prototype built or just get my thoughts down. I know how to build an exe with this pretty easily actually, but it's not really legal, so .. oh well.

Here's where my dynamic throttle response charts idea is going for a FinX23-2600/Cyclone/tubular frame. The efficiency chart version is coming right up. I'm thinking a nonlinear regression curve could be used to evaluate each one of these charts when comparing motors/props/frames perhaps?
Jan 06, 2018, 04:55 PM
Registered User
RCG isn't letting me update my posts or attach anything right now.. error 505.. so more to come...v
Jan 06, 2018, 07:32 PM
Registered User
Here's a sneak preview of what the new "Advanced" features will include in the next version of the tool for an FinX23-2600KV motor with Cyclones on a tubular quad. After these additions I'll begin work on the KV & Prop-Size Adjuster.

Dynamic Throttle Efficiency

Throttle Curves

Dynamic Throttle vs Time
Jan 09, 2018, 05:28 AM
Registered User
Hey - thanks for making this available! Graphs look very cool.

Is the source code for this somehow viewable? I've tried to look at the access files but they seem to be binary (haven't installed access yet since I'm on a mac).

I'd like to look at the source to see how you've done it. Like does the prop efficiency get scaled up/down depending on the rpm the motor is running at (due to both variable prop and motor efficiency)? How is voltage sag of the lipo taken into account etc.

My hope is that something like this can revolutionize the way we measure components and build quads. Because a lot of whats being done doesn't seem to make a whole lot of sense - just to give a few examples:

- why measure each prop with each motor? Can't we measure thrust over rpm and moment of inertia for each prop once and then measure the properties of the motors separately?
- why measure each motor at different cell counts? Can't we measure torque/rpm over voltage, separately measure switching losses due to pwm wherever they occur and be done?
- a lot of this depends massively on the properties of lipos. Like how long can they supply how much current. Yet these aren't measured to that level of detail anywhere.

I'd hope that the core characteristics can be measured for all components in a quick and straight forward way. Which then allows to mix and match virtually and optimize to any criteria you want.

But yeah some day sitting down and trying different props out on your quad without ever unscrewing one. Very cool!
Jan 09, 2018, 06:55 AM
Registered User
It sounds like you get where I’m going with this. The worst part is how many sims and video games are out there with strikingly accurate quad physics, but not enough customizability for cause & effect analysis. There are plenty of available formulas to calculate one aspect of flight at a time, but I hadn’t seen something comprehensive. I’m hoping with more usage or if someone builds this tool or a facsimile in an online app the most important variables are identified, leaving the remaining as optional with dynamic confidence intervals letting the user know the impact to accuracy.

I’m on my cell now, but I’ll paste the vba source in a few hours.
Last edited by AILERON8; Jan 09, 2018 at 07:12 AM.
Jan 09, 2018, 07:59 PM
Registered User
I thought I would take a stab at each of the questions you wrote:

Quote:
Originally Posted by joelucid
Why measure each prop with each motor? Can't we measure thrust over rpm and moment of inertia for each prop once and then measure the properties of the motors separately?
The short answer is prop loading & efficiency. I created a term for the calculator called KVE that's been really useful so far. Rather than relying on the motor voltage constant Kv the parameter KVE accounts for prop loading at maximum rpm on average, but not due to longitudinal aerodynamic drag. The latter rpm is a function of a whole slew of variables including throttle position. After examining a few hundred motors using KVE I'm seeing some clear trends that align well with folks reports on propeller efficiency, speed, handling, etc. It improves the accuracy of the model considerably, but I'll admit that it hides a ton of complexity that could and should be explored further. Ultimate though, if the objective is to project how fast or efficient a fully built quad will fly, or how it handles on the throttle, then I do think using motors with props installed decreases error by reducing complexity. If the objective is to make a better propeller or optimize or compare a motor against another then measuring the motor & prop separately would be the optimal method though. Particularly when dynamic motor testing is commonplace.

Quote:
Originally Posted by joelucid
Why measure each motor at different cell counts? Can't we measure torque/rpm over voltage, separately measure switching losses due to pwm wherever they occur and be done?
I agree. Cell count is mostly irrelevant for dynamic testing other than to speculate what the maximum velocity or acceleration would be. I think testers are using different packs today based-on folks requests for various applications, e.g., they're aiming to please a broad audience who are using motors for a diverse set of applications.

Quote:
Originally Posted by joelucid
A lot of this depends massively on the properties of lipos. Like how long can they supply how much current. Yet these aren't measured to that level of detail anywhere.
True, and as far as I can tell LiPo's are one of the bigger sources of dynamic variation too. In the model I'm recommending folks use average minimum voltage while maximizing current draw, which translates to maximum RPM on average since the three variables drive one another in that respect, but also because I figured people would like to project for max velocity and acceleration which is usually going to sag the battery. Power is a huge variable though, but as long as ohms law is considered and prop loading taken into account its affects can be measured. It all comes down to initial conditions.. if you know you'll be running saggy LiPo's, lower the voltage while holding the current & rpm's steady. If you're planning to use a larger mAh LiPo than normal then add a bit to the voltage while holding the current & rpm's steady. If you're planning to keep it easy on the throttle for a long distance low rpm quad then it might be a good idea to use 75% or even 50% voltage/current/rpm figures since that's how you know you'll be using the quad.
Last edited by AILERON8; Jan 09, 2018 at 08:08 PM.
Jan 10, 2018, 02:23 AM
Brisbane, Australia
Quote:
Originally Posted by AILERON8
So bad news and good news... The method I was going to use to project quad dynamics using motors with untested KV, prop size, or Current draw by establishing K is too variable without including the internal resistance of motors in the equation. Since motor mfg's aren't publishing resistance and Static motor testers aren't either I abandoned the concept (for now) in favor of a technique that doesn't require resistance. The new method is differentiated with respect to Throttle and unfortunately I haven't been seeing static motor testers like MQTB or EngineerX publishing the smaller variety motors as often. So for the really small motors with unknown throttle curves the model won't be able to project quad dynamics without knowing Thrust for a known KV for the time being, but I'm not giving up. That said, the model will be up for the task for motor/prop combos that have been static tested in a few days.
If it's any help, you probably know this already but thought I'd mention it, it's not too difficult to measure internal resistance using a current-limited power supply set to say 1A, and a multimeter. I used to do that on motors I rewound but didn't use the data for anything so stopped doing it.

For people without a power supply, another possibility is to make a CCS with just 2 components and a heatsink, and power it with a 1S or 2S lipo. I could do a doc for it and place it here, parts cost would be maybe $5. It would require one lead of the motor to be unsoldered from the esc for the test. Maybe there's some cheap gadget already out there that will do the same function, I'll do a search if you're interested. That way, if you already have a motor you are interested in using, you can measure the Ri yourself. Just an idea.
Jan 10, 2018, 09:36 AM
Registered User
Quote:
Originally Posted by Ian444
If it's any help, you probably know this already but thought I'd mention it, it's not too difficult to measure internal resistance using a current-limited power supply set to say 1A, and a multimeter. I used to do that on motors I rewound but didn't use the data for anything so stopped doing it.

For people without a power supply, another possibility is to make a CCS with just 2 components and a heatsink, and power it with a 1S or 2S lipo. I could do a doc for it and place it here, parts cost would be maybe $5. It would require one lead of the motor to be unsoldered from the esc for the test. Maybe there's some cheap gadget already out there that will do the same function, I'll do a search if you're interested. That way, if you already have a motor you are interested in using, you can measure the Ri yourself. Just an idea.
Thanks, that would be a nice addition. Yeah measuring the internal resistance is really the only way to accurately project how a quad will perform with different Kv motors when thrust and rpm are unknown. Considering how simple it is to measure I wonder why manufacturers and testers aren't publishing it... I wonder if mfg's are worried it might expose the variability in quality..

You're right, I think I'm going to make the updates assuming folks can get ahold of Ri then. It sounds like dynamic motor testers will be publishing Ri as well, but if you don't mind making/posting a doc for that I would appreciate it? Who knows how long until dynamic motor tests are commonplace. Thanks again
Last edited by AILERON8; Jan 10, 2018 at 09:42 AM.
Jan 11, 2018, 04:21 AM
Registered User
Quote:
Originally Posted by AILERON8
I thought I would take a stab at each of the questions you wrote:

The short answer is prop loading & efficiency. I created a term for the calculator called KVE that's been really useful so far. Rather than relying on the motor voltage constant Kv the parameter KVE accounts for prop loading at maximum rpm on average, but not due to longitudinal aerodynamic drag. The latter rpm is a function of a whole slew of variables including throttle position. After examining a few hundred motors using KVE I'm seeing some clear trends that align well with folks reports on propeller efficiency, speed, handling, etc. It improves the accuracy of the model considerably, but I'll admit that it hides a ton of complexity that could and should be explored further. Ultimate though, if the objective is to project how fast or efficient a fully built quad will fly, or how it handles on the throttle, then I do think using motors with props installed decreases error by reducing complexity. If the objective is to make a better propeller or optimize or compare a motor against another then measuring the motor & prop separately would be the optimal method though. Particularly when dynamic motor testing is commonplace.
Right. But prop loading and prop efficiency is the same for a prop at one rpm across all motors. So how can additional insight be gained by testing one prop on several motors? A prop at one rpm will have one static thrust, one loading torque and resulting from this one efficiency. Plus one rpm independent moment of inertia. It simply doesn't matter which motor it's run on.

Then you can focus on the motor side on getting clean torque and rpm measurements. Plus measure its moment of inertia which might have a significant effect on low rpm acceleration where prop load torque is still low.

And by doing this you eliminate the combinatorial explosion that comes from measuring everything with everything. Which might give the time to think about how to measure dynamic thrust of props, an all important measure which is currently completely ignored.


Quick Reply
Message:
Thread Tools

Similar Threads
Category Thread Thread Starter Forum Replies Last Post
Discussion What exactly do you practice on the sim? FlightofPhx Simulators 4 Apr 14, 2017 07:07 PM
Discussion Practice Sims? FPVBadger Beginner Multirotor Drones 4 Oct 06, 2016 04:31 PM
Discussion sim to practice racing quads? razorseal Simulators 4 Dec 14, 2015 05:05 PM
Practice tip for flight sims! Keitha Parkflyers 1 Apr 17, 2002 03:26 PM