100% useful info for what im currently working on. After hours of reading, watching, testing, changing, testing, rebuilding, testing charging, reading, watching, etc, etc... this video is probably the most clear and to-the-point bit of material ive come across on the subject of PID tuning. Thank you for the time and effort you put into these videos, you cant imagine how much they are appreciated by peeps like me. Rock on!
Thanks Alex! I've seen dozens of PID tuning videos in the last two years, but yours is the first where all the oscillations are demonstrated which is mighty useful, not having to guess what is what.
Nailed it man! I've read this stuff from a million different places, and although this video is not exhaustive, it is as close to a clear and accurate synopsis as I could hope for.
This is probably the best video I've seen explaining PID's. I had an understanding of how they worked but couldn't really explain it to people. I'm just going to give this to people now :)
Wow bullseye ! in 9 min you just gave the answers I needed ! Just built by 1st quad and been knee high in the PID mud for several days now. You just made it so much clearer, it toke me 60 min to get it to fly and be controllable ! not perfect but at least its not flipping uncontrollable ;-) Great job man !
That was by far the most comprehensive description of PID gains, finally understand why so many of my 'tuning' events resulted in quad-madness. Cant wait to test out the new knowledge.
Can you say more about your recommendations for P gain? I'm surprised that you have the racer's P gain lower than the casual flyer's P gain. I thought that an acro flyer would want P gain to be higher since they would want the copter to track the sticks more aggressively. Is that not right?
Joshua Bardwell Not at all. At high speeds and high powers, you want less p-Gain as this will keep the aircraft from oscillating. Racers will maneuver their aircraft harder and run faster. This will cause a high gain to over-correct. Thus, lower gains are better for racers.
I haven't tuned my 250 since I have not found any good explanation to what look for when tuning I and D, but now I finally have. Thanks for this upload!
Alex Greve why do we not use the 3/4Ghz bands for FPV? How come it jumps from 2.4 to 5.8? I can't find anyone who can answer but I bet you could? Great video and great addition to the 'How to be successful in FPV' series.
Thanks Alex! I have one question: is the relationship between I gain and P gain a universal constant? Because you said "start by setting I gain to 25% of P gain". I have my doubts if that applies to all flight controller firmware, but you could be right.
Wow, that is one helpful tutorial. Although PID's seem simple to understand, I think unless you have flown a really dialled in copter then how would you know how it was supposed to fly? Nice One IBCrazy I am going to think about this while I'm flying and then try some fine tuning.
5h1f7y - I agree that unless you experience a well-tuned multi, you really don't know how far you can go. in my case, I have only flown my own, so I started from rock bottom and am continually amazed at how much better I can get them to fly.
Thank you so much for addressing this subject Alex!! There is a total vacuum of Good information out there. Did I understand that you increase your gain until you see oscillation, then set your gain from 50 to 70% of that? Most people just say to back it off a little, which never seemed right to me. Thanks again!!
valiant360420 - Yes you back it off between to 50-70% of where you see oscillation. The reason for this is that there is a range of values that work for a multicopter and putting it right on the edge of oscillation will cause the vehicle to oscillate at higher throttle settings and harder maneuvers. Higher power systems require backing off further than lower power to weight systems.
+Alex Greve Your reasoning is very interesting and initially counter intuitive to me. Wouldn't lowering the p gain by a greater amount for someone racing rather than a beginner or photographer make the craft more sluggish for the racer who most likely wants the most agility? Especially now with the advent of TPA adjustments it seems you can get the best of both worlds?
+tmenet - This is how you get a base tune, not your perfect tune. Every vehicle will be different and thus this should be taken as a general guideline. I get more in detail in part 16 which explains how to refine your tuning. That said, the last thing a racer needs is an oscillation and thus running lower gains keeps your aircraft from oscillating. Remember that there is a range of values that work and there is no one perfect value.
awesome video.. I just have some questions as to how you start tuning your PIDS. I have a eachine racer 250 using librepilot. I have been flying in LOS right now and am trying to start tuning.. first question is.. do I start out with all my PIDS at 0 and enter in a number for P roll, and up it until it oscillates and then back down a little? and then do the same for P pitch? basically I'm asking if you have to adjust the roll and pitch for P gain I gain and D gain separately or do you adjust both roll and pitch at the same time for each P I and D? or do I start with the PIDS at the basic setting on librepilot and go up from there?... and one last question.. do I start tuning in attitude mode or do I have to be in rate or acro mode?.. thanks steve
No, don't start with all your PID's on 0, if you are inexperienced start with the defaults and go from there. I also worry about where to put my PID's when I'm starting, to focus on one at a time, but certainly on my FC (ardupilot) anyway I wouldn't try it with 0..
I have a QAV250 with a Taranis that I have in MultiWii (rewrite) and it flies fantastic in Angled mode, thanks to your video, and my struggling for 3 days, but when I switch it to Horizon mode it's very sluggish and slow, I have to give it tons of pitch or roll just to get it moving. How can I get it flying in Horizon mode as it does in Angled mode without messing with the tune in Angled?
+Aaron G - Your settings in angled mode (P-gain is the likely culprit) is too high. Horizon holds the value at 0 degrees where angled mode allows the resident angle to be manipulated by control. Thus, if your gain is too high on horizon, you cannot overcome the values with control input.
+Alex Greve Thanks for the reply Alex, I'm sure you're a very busy guy. But if I change the settings for it to fly well i Horizon won't it fly terrible in Angled? If I lower the P settings more in Angled it starts wobbling like a small boat in high seas. Sorry to keep bothering you but I tried everything lol. Thanks again.
+Aaron G - There is a P setting for horizon mode only. I assume you are using cleanflight, correct? In the menu, there is a horizon gain. Adjust that. The PID settings for acro are different than horizon mode.
+Alex Greve You are the man, that helped so much, I got it flying just how I wanted. Thank you for all the help, you got yourself a new sub. I will also give you a shoutout whenever I do the QAV250 review.
Alex what props are you using, they must be amazing as they use the correct pitch like my expensive carbon fiber system for larger copter? Rolling off correctly to give same lift at tip as at base? Thanks again STILL working on this 20th time I've watched but always wonder about your props...
+Anthony Cenabite Farr - I really like APC props, but HQ makes good props as well. Gemfan is ok for low power and so are the DAL props. Just be mindful that higher powers require better props.
Ty I was truly wondering what props I'm looking at in your vid? Been looking for props with this (correct) shape.? As I'm building a tri and I'd like to try them.. Thanks for all your help
great vid. I am using a Naze32 with clean flight with a 250mm mini H. I cannot seem to get my Yaw P gain to create oscillations even with the make value. any suggestions?
Eric Lindholm - Quads are a different animal than tricopters when it comes to yaw. In general, the oscillations will not happen unless the P gain is exceptionally high. Thus, there is a lot of room for error. if it seems pretty solid and is no oscillating, leave it alone and move on to I-gain. The P and I for yaw will be much higher than pitch and roll.
another thought...when tuning P sometimes the oscillations don't show until you jerk the controls. Then it will oscillate and die out. Other times it starts oscillating the minute its off the ground without any control input...your thoughts on this ?
TeamBlindMouse - Usually this jerk response oscillation is a combination of both P and I gain. This is why I recommend dropping the P-gain down significantly below the oscillation point (ie 60% or so). This usually mitigates those kinds of issues. Some might find they need to go further... especially those with a hexa or octocopter. Tuning is not a one size fits all. The purpose of this video is to explain how it works and how to get started in tuning. Everyone will have their own method... and that's the great thing about it ;)
***** In general, this helps. It is best to tune in rate mode rather than auto-level. It is harder to get the proper stability when the vehicle wants to sway around on you. This is especially true if your CG isn't correct. While not necessary, I feel it makes it easier to set your P-gain.
Do you know if this applies to the Haze a well as the CC3D? Have you used both? Same for the 3 main firmwares, CleanFlight, BaseFlight, and OpenPilot? For some reason I find it much easier to tune my CC3D in OpenPilot than CleanFlight. I can get my mini quad flyable pretty easily in OP, then try CF and just break props because defaults make my quad freak out on takeoff. I can't seem to get it in the air at all. I've done this twice now. I just want to get it to hover in CF! Then I could use auto tune. ;-)
Shawn Sato-Veillon - Yes it is the same with the Naze32 board. Indeed Open pilot is easier to use. The auto tune in clean flight is not very good. You are better off tuning yourself. However, you really need to read the Wiki on Clean Flight so you know what each setting really does.
Thanks Alex. Yep, seems so. It is interesting how I've read so many people say CleanFlight is easier to use. I think either it is because they already used to it, or the fact that is isn't as comprehensive. I'll take your advice and read the entire wiki - I have read some of it, but apparently not enough. It would be great if it had some sort of wizard, asking questions about the setup, and then giving some default starting point. Oh well. Thanks for the helpful videos!
Alex , what are your thoughts on tuning P with I&D set at zero, then tuning I with D still at zero(and P at whatever value it tuned to) and then finally tuning D ?
TeamBlindMouse - In auto-level mode this makes sense, but you will have two competing P-gain settings: one for auto level, and the other for the machine itself. However, in acro mode (rate mode) the vehicle can get out of control very easily without some I-gain in the equation since you removed the positional reference. Having a little bot of I-gain to stabilize it seems to help and also keeps you from over-tuning the aircraft.
