This has to be one of my favorite channels. Few channels take the time to learn, research, and dive this deep into multirotor control. Can't get enough!
The new NBD Hummingbird Pro FC uses the BMI160, FWIW. It needed new filtering from my other tunes, but is performing great and seems to handle knocks better in early testing.
As written some comments below - a good electric filtering is the key for proper operation, in my experience the ICM20602 seemed to be more sensitive to electrical noise compared to the MPU6xxx series. I mostly use a 470µF/10V @ the 5V rail (from which the 3.3 is cascaded in most cases). On Invensence linecard there is the ICM-20690 introduced as a replacement for the discontinued MP6xxx ("6-Axis DMP-Enabled Drone/VR/IoT solution, Targeted For All Current MPU-6000/6050 Users"). The main difference between ICM20602 and 20690 is that the 20690 have a higher range of sensitivity scale factors.
I have two similar quads, one with each gyro and they seem to the same to me. caring which gyro is like having a preferred color of internal wiring when purchasing a new juicer.
A few years back I jumped on the 32k band wagon and build my next quad with a Matek F405-AIO With the ICM20602 gyro, it never flew well, and had the twitch of death into a muddy ditch. I rebuilt the same quad and put the Matek F405-CTR with MPU6000 gyro. I ditched 32K and stayed on 8k, no more issues, built another two quads with the same gyro never had any more issues. I also build with 4 x 470µf 35v Rubycon low ESR capacitors for each individual ESCs and 1 more on the XT60, no more noise, no more random VTX failures, no need for any video filters and very smooth (well good enough for me) flying quad with very low noise.
Thanks for the video Mark. I had two iFlight F7 flight controllers (30x30 and 20x20) with dual ICM gyros (first gen of SucceX TwinG). They were a complete nightmare - could not get rid of yaw twitching. This was the reason to study all of your tuning videos :). One of the gyros gave a little better results opposed to fusion, but still unusable. I have soldered also 2 additional ceramic capacitors (100nF and 1uF) to each of the gyro power supply pins. Did not help either, so I gave up. Finally I have replaced these flight controllers with MPU6000 based flight controllers. Problem solved and lesson learned: I will never buy iFlight or ICM based FCs. But I wonder what will happen when MPU6000 stock runs out.
In the dual stack experiment, the comparison was between a MPU mounted directly to PCB and a ICM soft mounted in a casing on a silicon pad, right? - And there are still noise issues. Then it is not just some noise, it is also cost of the FC.
I have stuck with the clracing f7 dual v2.1. Out of the 5 flight controller's i have tried, i have found it works out the best for me. It is the only one i have had with the icm 20602. I dont know if it makes a difference having two of them on the board but i replaced my other boards with these so i exclusively use them for my kwads now.
The Kakute F7 has two versions now. V1 with icm and v1.5 with mpu. I have both and ICM drifts a lot more. Build identical just FC switch, baros covered with foam runing inav pos hold
get video Mark! Hey, I was wondering if some FCs are designed with little local capacitors for the gyro and whether that might account for some of the different experiences people have had. I know for instance that the Matek f405 (with the icm) has a brother PDB that has, I think, an on board LC filter/cap.
@@uavtech yes, of course the way you did it here is ideal. I was just thinking about people that have stated poor experience with one or the other and whether they might have been comparing apples to oranges, unknowingly, with a cap/lc filter in one case but not the other
@@uavtech frankly the on board electrical filtering is key and often under done.. ive seen and had plenty of mpu fcs that should be awesome and aren't....
@BWFPV I don't know if all, but at least some have this. For example the Sp Racing H7 Extreme has lists it in the description: "Simultaneous dual-gyro support already produces low-noise jitter free gyro output, but SP Racing raises the bar again as now the two ICM20602 gyros on the H7 EXTREME are supplied by their own regulator and are supplemented by an additional high-capacity, yet tiny, filter capacitor. Only this gives the gyros the noise-free power supply they need to work at their best. Each gyro has it's own dedicated SPI bus which allows for parallel data transfers." - source: shop.seriouslypro.com/sp-racing-h7-extreme
I have a matek f7 mini with both a MPU-6000 and icm20xxx. Don't remember which icm exactly, but tune it on the mpu6000 gyro, and then switched to the icm gyro, and it was significantly worse. I can recommend the board though. Has held up great for me.
Here is an issue you dont often hear talked about with the mpu6000: it generates heat all on it's own. Enough heat in fact to cause a noticable amount of drift during a flight from a cold boot. Pair this with flash chips, vregs, osd, stm which are all also producing heat and you can overheat a mpu gyro very easily. If you heat one faster than a certain threshold rate it will blow out each axis progressively with a massive drift event all at once. Heaven forbid you also put all this under a canopy with poor airflow. The mpu series really needs a heatsink. In contrast the icm gyro produces little to no heat on its own and is therefore significantly less susceptible to drift. All gyros drift with temp obviously and the icm is not immune from other things warming it up - I'm just saying the mpu gets a big fail from me since its creating the heat which causes it to drift. I have stopped using it in my own designs.
That is a good point - also explains why I've had better luck with gyro performance when I have nothing above it - I assumed that was EMI, but now realize that airflow for cooling might be as big of a contributor
I replaced a kakute f7 v1 using iCM because it started drifting pitch and roll with Kakute F7 v 1.5 using mpu no drift. No test was done on gyro specifically so can't be conclusive but same board otherwise one with mpu holds best atm.
@@anti-popfpv4638 that's a good point... the old kakute version had the icm20689. Looking at the icm20689 smd package, it has that same giant thermal pad on the bottom of the gyro for heat transfer just like the mpu series. I should be more specific in my statements about icm series and clarify that I have only specifically run my tests watching internal gyro temperature and evaluating drift on the icm20602. While not concluclusive and only my speculation, I suspect that the adding the thermal transfer die pad to the bottom of the 20689 gyro was done for a reason and this is a very different construction than the 20602. With the kakute placing the 20689 on the tiny soft mounted daughter board... it was really limited in having significant copper & pcb material to sink heat away from that gyro and not surprisingly it drifted from temp increase (my speculation of course). Its entirely possible this particular gyro can't withstand it's own heat generation too. Thanks for helping me realize I should not make blanket statements about the icm series when I have only tested the icm20602 in detail. Regarding drift on the mpu... I would agree that when the only heat source it is exposed to is itself... the resulting drift is manageable and often not noticed. The kakute v1.5 construction certainly does isolate it from other heat sources with the soft mount.
@@_NFE I now have both Kakute f7 v1 and 1.5. i was experiencing yaw drift with v1 pretty bad to point it was no longer fun to fly so thought I'd try v1.5. i am having the same yaw drift just not as bad. I assure you setup was fresh and it wasn't drifting until quad had a few abrupt impacts. Just my speculation. I don't have anyone outside from myself to check every single one of my settings. You do great work, i was just saying!
Hi, I have the Omnibus F7v2 since a while now in a 5" quad. when I first installed it I couldn't get it to run, had all kind of glitches and what not. I even bought a ne gyro from Airbot, because there was something with shielding to electrical noise going on with this board in the early days and they've made new gyro cases for that. But that didn't either help. Only since I have selected the MPU6000 gyro on the board, it is running smooth. Btw. it is throughout hardmounted to the frame.
Does the icm20xxx gyros require a cleaner electric layout? Asking if it is more important to have a really good 3.3v voltage regulator, to give a clean reading and signal.
love the info you give for this hobby thank you. i am so curious about the noise,and few other things on the talon fusion f7 from heli nation. It has the dual gyros mounted diffrently. Would be interesting video on dual gyros,etc.
Great testing and analysis, dude! 😃 It's pretty difficult, hum? After seeing all the difference between the noise accordingly to the high of the board... Yeah... I don't really know what to suggest to you. 😕 But there are frames with mounting holes for 2 or even 3 stacks... Like the FlyWoo Mr.Croc frame, it has 3 stack places... Who knows, maybe it could be easier to test? Anyway, stay safe there with your family! 🖖😊
So I am pretty sure I have an off the shelf quad with a BMI055, which is a pretty awesome sensor. The same underlying sensor is used in the BNO055 which adds Bosch sensor fusion, and the BNO070/080/085 which adds Hillcrest sensor fusion (those are basically just revisions of the same thing that required hardware changes) - the Hillcrest is awesome for VR folks like me, and 1K fusion rate is sufficient. (Source: worked for the BNO070 launch customer and helped tune it) But it's interesting to see evaluation of imus in other applications. Any idea why nobody uses ST? I never see them in VR either, but they seem to have a pretty prolific mems sensor division. All the hobby/maker vendors use them more frequently, I assume in part due to price. But the LSM6DSOX (I think?) is a suitably pricy and low noise unit - maybe it's too new for adoption? Interestingly, even assembly processes can affect the performance of these mems imus, so if you're seeing issues with just one board, it could be their processes. These little chips look innocent but they're pretty darn demanding in their handling needs if you want to use them for high performance needs, whether flight control or VR.
Probably because they shows the most promise in how to deal with the variation in the signal you receive in 32k mode and you need an ICM gyro for 32k mode. See -- facebook.com/UAVTech1/posts/2190748527852936 for details. So when 32k mode hype died, the benefit of sensor fusion dropped off. It is a good backup gyro in 8k mode and of course, doesn't hurt in 8k mode too. I covered this in my iFlight TwinG review: th-cam.com/video/aozT7BxP9ZU/w-d-xo.html
Hi Mark, @UAV Tech Matek Systems FC F722 SE Foxeer F722 Dual Flight Controller STM32F722RGT6 MPU6000 und ICM20602 Dalrc F722 Dual Stm32F722Rgt6 Flight Controller Mpu6000 and Icm20602 iFlight Revo Bee32 F7 F7 AIO Omnibus V2 Pro Flugsteuerung - F7 (both gyros in soft-silicon mount) AIKON F7 FC - OSD, DUAL GYRO (MPU 6000 / ICM 20602 SWITCHABLE), AIKON F7 MINI FC 20X20 - OSD, DUAL GYRO (MPU 6000 / ICM 20602 SWITCHABLE) So the lower the FC is mounted the better. Would it make sense to mount the FC under the ESC? I was not aware that nylon extensions influence vibration in a negative way. How would you recommend rebuilding stacks on a noisy quad? Steel screws with grommets? Nylon screws with grommets? O-Ring washers?
Holybro recently changed from ICM to MPU in their FC lineup (Holybro Kakute F7 v1.x -> v1.5). Would be interesting to hear what motivated this change for them. I have had the v1.x and now have the v1.5 and (subjectively!) the noise levels are much better
Crazy that you mentioned this. I have 2 similar 5" Quads. One Quad has the 1st generation Kakute F7 Mini with the ICM Gyro and the 1st Generation Tekko32 F3 45A Mini. The 2nd Quad has the New 2nd Generation Kakute F7 mini V2 with the MPU Gyro & B+ pad and the 2nd generation Tekko32 F3 45A Mini that runs at 96kHz. I'm tuning them at the same time with all the same BF 4.21 Settings, RPM Filtering, PID TUNING, ETC.. The 1st Quad with the ICM Gyro flies worse and is harder to tune. It's much harder to get the Quad with the ICM gyro to be critically damped on Inverted Yaw Spin Tick Tocks. The 1st Gen Tekko32 F3 45A Mini also runs much hotter than the new version, scary hot. After flying and tuning the new Kakute F7 mini V2 with the MPU6000 gyro, I'm getting the itch to buy another and replace the 1st gen stack on the other Quad all together.
DalRC F722 Dual has both of them. I have two of these FC’s and love them. Ran the ICM for a bit when BF supported 32k but went back to the MPU6000. You should give this FC a try.
Great thank u for the information im doing a build but for some reason the screw im using they bend a little because i force it so i order more screws. Then i start it doing it with the nilon stanoffs but now with ur tip im not going to. Im just going to wait for the screws thank u.
I think that the last resoults you showed could perfectly be because of that built in gyro soft mount of that particular FC. I think that simply hard mounted gyro is much better.
So I wonder if I can find mpu6000 gyros in my old cel phones and steal them for the bad gyro in my flight controller. They're probably using old stock or getting the old gyro chips cheaper. Out of all the flight controllers I've had I have yet to notice any manufacturer giving a crap about quality or reliability.
I have had only one board with the ICM CHIP and it was bad and I replaced it with another board with a MPU on it . On that board it was hard mounted and the fact that you see a lot of ICM chips being soft mounted says a lot to me . Sticking with the MPU6000 .
Hey quick question - I have noticed that some fc's have been coming stock with Yaw D-values enabled. I was under the impression that the Betaflight dev's had dropped Yaw-D from the code and that is why it was set to zero. Did this recently change? Thanks for your help.
It is there, but not advised to use. Yaw is way too slow and D-term adds noise to Yaw, which effects Roll and Pitch movements. So if not upside, but there is a down side, why use it? But as always BF gives folks options.
Was your testing with BF4.2?... if not then, I am curious on your next testing if the BF 4.2 will manage the ICM20602 gyro data better than the MPU6000 due to it's increased sensitivity - what do you think?
Tests were done looking at Gyro_Scaled (raw) noise. However, if noise is well defined and not broad is nature, yes, the improvements in 4.2 notch filter targeting (RPM and FFT) will crush it.
@@uavtech example: th-cam.com/video/sNDO3RqqHh0/w-d-xo.html . He does this in every esc noise testing/review. You should do the same thing, but instead of outputting directly to the motors, you should output to the pid controller the same set of inputs.
HGLRC removed the F722 dual from there product list. I have one, but used only mpu6000, which is the second gyro for choice. But these board had no(!) memory for blackbox recording and so I changed to the new HGLRC Zeus F722 DJI, which had only the mpu6000 and has memory on board. I fly only analog, but to have it DJI compatible is a nice option
12:39 talon 20x20 mpu6000 vs talon 20x20 fusion... no contest the mpu version has loads more electrical noise when swapped in and out of the same machine, manifesting its self in the pitch axis... IMO its not an issue of gyro type but an issue of how the recommended hardware spec has been implemented
I was troubleshooting pitch axis noise on two 3" quads recently, It was the damn mmcx axxi antenna hanging off the back. they flop around too much and cause lots of pitch noise. printed some mounts two keep them still, and no more pitch noise.
I use Kakute F7 ver. 1.1 with ICM gyro. It flies pretty good without any problem. There is a new version of this board ver. 1.5 where MPU gyro is used. Should I switch to the new version? Is it worth the change? Thanks mate for perfect comparison!
I know this is an old video but I figured I'd put my experience. I unfortunately just got a Lumenier Lux Ultimate F7. Out of the box the MPU6000 was bad. On flips and rolls the gyro would just go nuts for some reason. I switched the the ICM20602 and my problems went away. It is more twitchy for sure but atleast it worked for me.
I'm a bit disappointed you didn't mention about overflow bug in ICM gyros. Such an error is disqualifying in my opinion. Yes, I know this is patched in the BF software, but I am surprised that the manufacturer has not produced a revised version so far. Unless I'm wrong. ;)
i had assumed the icm was more sensitive just because its faster. even when you sample it slowly, the internals are faster. maybe the mass of the actual vibrating mems sensor is lighter, or just the measurement circuit is faster, im not sure. Also, i thought the modern filters in betaflight basically solved the noise problem. i mostly use the icm chips and i never had any issues. just one persons experience of course.
They are both sampling at 8k. You can set ICM to sample at 32k, but then it is a total mess. I use ICM all the time as well. Just seems like there is a valid reason the old MPU is still used. I have a better test coming soon! MPU and ICM on the same FC and a code tweak to log both at the same time.
Yup, and both hard-mounted on the board. It's a great FC ... if only the STM32F722 wasn't getting low on flash memory now. Seems that STM32F745 is the way to go.
So ICM isolated is cleaner but unisolated MPU is. Always a drawback somewheres. Drone company that creates good clean area for mounting would bees great.
I hope the 6000 wont be extinct soon... Some other company will make them... I hope... or a new one comes to play!? In my short experience I keep coming back to the 6000... need to try the F1 Ai H7 houdini stuff ... I believe frames really help on issues that we all relate to gyro management, at least that has been my experience... some frames just resonate more than others. My two cents im no expert!
1. most of the FC's using ICM20xxx have a flawed electrical design ! (this is a known fact for about 2 years, follow the link for technical details : www.rcgroups.com/forums/showthread.php?2934256-OMNIBUS-F4-Pro-Corner-from-AIRBOT/page35#post38947338 ) 2. the only board I am aware of that has implemented a sound design is the CLRACINGF7v1 (though it has other problems with the true PIT mode) 3. ICM20xxx are the better gyros - lower drift - higher precision ... etc. It only requires better designed FC's !
This has to be one of my favorite channels. Few channels take the time to learn, research, and dive this deep into multirotor control. Can't get enough!
The new NBD Hummingbird Pro FC uses the BMI160, FWIW. It needed new filtering from my other tunes, but is performing great and seems to handle knocks better in early testing.
what about sensor fusion and icm 20689 gyros?
th-cam.com/video/aozT7BxP9ZU/w-d-xo.html
As written some comments below - a good electric filtering is the key for proper operation, in my experience the ICM20602 seemed to be more sensitive to electrical noise compared to the MPU6xxx series. I mostly use a 470µF/10V @ the 5V rail (from which the 3.3 is cascaded in most cases). On Invensence linecard there is the ICM-20690 introduced as a replacement for the discontinued MP6xxx ("6-Axis DMP-Enabled Drone/VR/IoT solution, Targeted For All Current MPU-6000/6050 Users").
The main difference between ICM20602 and 20690 is that the 20690 have a higher range of sensitivity scale factors.
I have two similar quads, one with each gyro and they seem to the same to me. caring which gyro is like having a preferred color of internal wiring when purchasing a new juicer.
They both seem to work fine, in general, to me as well.
Loved it. Love to see bmi088 review.
I have the Hglrc dual gyro f7 board, works well for me so far, I'm going to try the mpu after watching this
Great info. Puts a lot of the points over the i.
A few years back I jumped on the 32k band wagon and build my next quad with a Matek F405-AIO
With the ICM20602 gyro, it never flew well, and had the twitch of death into a muddy ditch.
I rebuilt the same quad and put the Matek F405-CTR with MPU6000 gyro. I ditched 32K and stayed on 8k, no more issues, built another two quads with the same gyro never had any more issues.
I also build with 4 x 470µf 35v Rubycon low ESR capacitors for each individual ESCs and 1 more on the XT60, no more noise, no more random VTX failures, no need for any video filters and very smooth (well good enough for me) flying quad with very low noise.
Thanks for the video Mark.
I had two iFlight F7 flight controllers (30x30 and 20x20) with dual ICM gyros (first gen of SucceX TwinG). They were a complete nightmare - could not get rid of yaw twitching. This was the reason to study all of your tuning videos :). One of the gyros gave a little better results opposed to fusion, but still unusable. I have soldered also 2 additional ceramic capacitors (100nF and 1uF) to each of the gyro power supply pins. Did not help either, so I gave up. Finally I have replaced these flight controllers with MPU6000 based flight controllers. Problem solved and lesson learned: I will never buy iFlight or ICM based FCs. But I wonder what will happen when MPU6000 stock runs out.
Good stuff! Thanks
In the dual stack experiment, the comparison was between a MPU mounted directly to PCB and a ICM soft mounted in a casing on a silicon pad, right? - And there are still noise issues. Then it is not just some noise, it is also cost of the FC.
I have stuck with the clracing f7 dual v2.1. Out of the 5 flight controller's i have tried, i have found it works out the best for me. It is the only one i have had with the icm 20602. I dont know if it makes a difference having two of them on the board but i replaced my other boards with these so i exclusively use them for my kwads now.
The Kakute F7 has two versions now. V1 with icm and v1.5 with mpu. I have both and ICM drifts a lot more. Build identical just FC switch, baros covered with foam runing inav pos hold
get video Mark! Hey, I was wondering if some FCs are designed with little local capacitors for the gyro and whether that might account for some of the different experiences people have had. I know for instance that the Matek f405 (with the icm) has a brother PDB that has, I think, an on board LC filter/cap.
Right, but if you have two gyros on the 3.3v power rail (MPU and ICM) would they not see the same filtering?
@@uavtech yes, of course the way you did it here is ideal. I was just thinking about people that have stated poor experience with one or the other and whether they might have been comparing apples to oranges, unknowingly, with a cap/lc filter in one case but not the other
I TOTALLY agree!! Electrical filtering is KEEEEYYY!!!
@@uavtech frankly the on board electrical filtering is key and often under done.. ive seen and had plenty of mpu fcs that should be awesome and aren't....
@BWFPV I don't know if all, but at least some have this. For example the Sp Racing H7 Extreme has lists it in the description: "Simultaneous dual-gyro support already produces low-noise jitter free gyro output, but SP Racing raises the bar again as now the two ICM20602 gyros on the H7 EXTREME are supplied by their own regulator and are supplemented by an additional high-capacity, yet tiny, filter capacitor. Only this gives the gyros the noise-free power supply they need to work at their best. Each gyro has it's own dedicated SPI bus which allows for parallel data transfers." - source: shop.seriouslypro.com/sp-racing-h7-extreme
For the cleanest test with least variation you don't really have many options, it's - Matek Systems FC F722 SE
I have a matek f7 mini with both a MPU-6000 and icm20xxx. Don't remember which icm exactly, but tune it on the mpu6000 gyro, and then switched to the icm gyro, and it was significantly worse.
I can recommend the board though. Has held up great for me.
Here is an issue you dont often hear talked about with the mpu6000: it generates heat all on it's own. Enough heat in fact to cause a noticable amount of drift during a flight from a cold boot. Pair this with flash chips, vregs, osd, stm which are all also producing heat and you can overheat a mpu gyro very easily. If you heat one faster than a certain threshold rate it will blow out each axis progressively with a massive drift event all at once. Heaven forbid you also put all this under a canopy with poor airflow. The mpu series really needs a heatsink. In contrast the icm gyro produces little to no heat on its own and is therefore significantly less susceptible to drift. All gyros drift with temp obviously and the icm is not immune from other things warming it up - I'm just saying the mpu gets a big fail from me since its creating the heat which causes it to drift. I have stopped using it in my own designs.
That is a good point - also explains why I've had better luck with gyro performance when I have nothing above it - I assumed that was EMI, but now realize that airflow for cooling might be as big of a contributor
I replaced a kakute f7 v1 using iCM because it started drifting pitch and roll with Kakute F7 v 1.5 using mpu no drift. No test was done on gyro specifically so can't be conclusive but same board otherwise one with mpu holds best atm.
@@anti-popfpv4638 that's a good point... the old kakute version had the icm20689. Looking at the icm20689 smd package, it has that same giant thermal pad on the bottom of the gyro for heat transfer just like the mpu series. I should be more specific in my statements about icm series and clarify that I have only specifically run my tests watching internal gyro temperature and evaluating drift on the icm20602. While not concluclusive and only my speculation, I suspect that the adding the thermal transfer die pad to the bottom of the 20689 gyro was done for a reason and this is a very different construction than the 20602. With the kakute placing the 20689 on the tiny soft mounted daughter board... it was really limited in having significant copper & pcb material to sink heat away from that gyro and not surprisingly it drifted from temp increase (my speculation of course). Its entirely possible this particular gyro can't withstand it's own heat generation too. Thanks for helping me realize I should not make blanket statements about the icm series when I have only tested the icm20602 in detail. Regarding drift on the mpu... I would agree that when the only heat source it is exposed to is itself... the resulting drift is manageable and often not noticed. The kakute v1.5 construction certainly does isolate it from other heat sources with the soft mount.
I have always thought my quads running CL Racing F7 with ICM20602 were my best performing. This confirms I'm not crazy.
@@_NFE I now have both Kakute f7 v1 and 1.5. i was experiencing yaw drift with v1 pretty bad to point it was no longer fun to fly so thought I'd try v1.5. i am having the same yaw drift just not as bad. I assure you setup was fresh and it wasn't drifting until quad had a few abrupt impacts. Just my speculation. I don't have anyone outside from myself to check every single one of my settings. You do great work, i was just saying!
Hi, I have the Omnibus F7v2 since a while now in a 5" quad. when I first installed it I couldn't get it to run, had all kind of glitches and what not. I even bought a ne gyro from Airbot, because there was something with shielding to electrical noise going on with this board in the early days and they've made new gyro cases for that. But that didn't either help. Only since I have selected the MPU6000 gyro on the board, it is running smooth. Btw. it is throughout hardmounted to the frame.
Foxeer F7 dual had that, but it's discontinued
Does the icm20xxx gyros require a cleaner electric layout? Asking if it is more important to have a really good 3.3v voltage regulator, to give a clean reading and signal.
Maybe. It is important to have clean 3.3v power of course.
JHEMCU F7BT DJI flight controller with bluetooth has both gyros and retail around $30.
I’ve seen some sketchy stuff bout jhemcu in general. If u want a gud f7 get the hglrc zeus
love the info you give for this hobby thank you. i am so curious about the noise,and few other things on the talon fusion f7 from heli nation. It has the dual gyros mounted diffrently. Would be interesting video on dual gyros,etc.
th-cam.com/video/aozT7BxP9ZU/w-d-xo.html
Also, th-cam.com/video/mjRWuA7ICZ8/w-d-xo.html
Great testing and analysis, dude! 😃
It's pretty difficult, hum? After seeing all the difference between the noise accordingly to the high of the board... Yeah... I don't really know what to suggest to you. 😕
But there are frames with mounting holes for 2 or even 3 stacks... Like the FlyWoo Mr.Croc frame, it has 3 stack places... Who knows, maybe it could be easier to test?
Anyway, stay safe there with your family! 🖖😊
Matek F722-SE has both chips, thanks for the info!
so why don't you mount the two flight controllers horizontally on a stretched X frame and exclude the extra vibration due to height.
Someone else did the test. Part II woll be taking reading from both the IMC20602 and MPU6000 FROM THE SAME FC board!!
could a cap in the 3.3v rail help filtering electrical noise to the gyro?
It should. Hard to know if there is an issue there to solve until you add the cap.
It depends on the 5V DCDC converter,
if LC resonance is around 30kHz it won't help !
How would I black box log a board with no SD slot or on board storage? I know tethered is a thing but I can't fly tethered
store.myairbot.com/openlager.html
th-cam.com/video/eO2hIRroFSw/w-d-xo.html
So I am pretty sure I have an off the shelf quad with a BMI055, which is a pretty awesome sensor. The same underlying sensor is used in the BNO055 which adds Bosch sensor fusion, and the BNO070/080/085 which adds Hillcrest sensor fusion (those are basically just revisions of the same thing that required hardware changes) - the Hillcrest is awesome for VR folks like me, and 1K fusion rate is sufficient. (Source: worked for the BNO070 launch customer and helped tune it) But it's interesting to see evaluation of imus in other applications.
Any idea why nobody uses ST? I never see them in VR either, but they seem to have a pretty prolific mems sensor division. All the hobby/maker vendors use them more frequently, I assume in part due to price. But the LSM6DSOX (I think?) is a suitably pricy and low noise unit - maybe it's too new for adoption?
Interestingly, even assembly processes can affect the performance of these mems imus, so if you're seeing issues with just one board, it could be their processes. These little chips look innocent but they're pretty darn demanding in their handling needs if you want to use them for high performance needs, whether flight control or VR.
Unknown on the ST line. Maybe just because BF doean't have the driver support for them?? (yet)
Why are there no fusion boards with the mpu6000?
Probably because they shows the most promise in how to deal with the variation in the signal you receive in 32k mode and you need an ICM gyro for 32k mode. See -- facebook.com/UAVTech1/posts/2190748527852936 for details.
So when 32k mode hype died, the benefit of sensor fusion dropped off. It is a good backup gyro in 8k mode and of course, doesn't hurt in 8k mode too. I covered this in my iFlight TwinG review: th-cam.com/video/aozT7BxP9ZU/w-d-xo.html
Hi Mark, @UAV Tech
Matek Systems FC F722 SE
Foxeer F722 Dual Flight Controller STM32F722RGT6 MPU6000 und ICM20602
Dalrc F722 Dual Stm32F722Rgt6 Flight Controller Mpu6000 and Icm20602
iFlight Revo Bee32 F7
F7 AIO Omnibus V2 Pro Flugsteuerung - F7 (both gyros in soft-silicon mount)
AIKON F7 FC - OSD, DUAL GYRO (MPU 6000 / ICM 20602 SWITCHABLE),
AIKON F7 MINI FC 20X20 - OSD, DUAL GYRO (MPU 6000 / ICM 20602 SWITCHABLE)
So the lower the FC is mounted the better. Would it make sense to mount the FC under the ESC?
I was not aware that nylon extensions influence vibration in a negative way. How would you recommend rebuilding stacks on a noisy quad? Steel screws with grommets? Nylon screws with grommets? O-Ring washers?
THANKS!!
Steel studs with rubber grommets on the FC.
Is it possible that the icm gyro is better, but that the companies are damaging it by overheating or some other thing during installation on the FC?
Maybe. But ...
Holybro recently changed from ICM to MPU in their FC lineup (Holybro Kakute F7 v1.x -> v1.5). Would be interesting to hear what motivated this change for them. I have had the v1.x and now have the v1.5 and (subjectively!) the noise levels are much better
Crazy that you mentioned this. I have 2 similar 5" Quads. One Quad has the 1st generation Kakute F7 Mini with the ICM Gyro and the 1st Generation Tekko32 F3 45A Mini. The 2nd Quad has the New 2nd Generation Kakute F7 mini V2 with the MPU Gyro & B+ pad and the 2nd generation Tekko32 F3 45A Mini that runs at 96kHz. I'm tuning them at the same time with all the same BF 4.21 Settings, RPM Filtering, PID TUNING, ETC.. The 1st Quad with the ICM Gyro flies worse and is harder to tune. It's much harder to get the Quad with the ICM gyro to be critically damped on Inverted Yaw Spin Tick Tocks. The 1st Gen Tekko32 F3 45A Mini also runs much hotter than the new version, scary hot. After flying and tuning the new Kakute F7 mini V2 with the MPU6000 gyro, I'm getting the itch to buy another and replace the 1st gen stack on the other Quad all together.
@@GeorgeCee oh so im not the only one seeing crazy temps on the first mini.
DalRC F722 Dual has both of them. I have two of these FC’s and love them. Ran the ICM for a bit when BF supported 32k but went back to the MPU6000. You should give this FC a try.
I have in the cart. Thanks for the tip.
Great thank u for the information im doing a build but for some reason the screw im using they bend a little because i force it so i order more screws. Then i start it doing it with the nilon stanoffs but now with ur tip im not going to. Im just going to wait for the screws thank u.
Good move!!
I think that the last resoults you showed could perfectly be because of that built in gyro soft mount of that particular FC. I think that simply hard mounted gyro is much better.
Yeah, that is why i say Part I, because i need to do a Part - II that has the same mounting of the gyro on the board for both.
So I wonder if I can find mpu6000 gyros in my old cel phones and steal them for the bad gyro in my flight controller. They're probably using old stock or getting the old gyro chips cheaper. Out of all the flight controllers I've had I have yet to notice any manufacturer giving a crap about quality or reliability.
I have had only one board with the ICM CHIP and it was bad and I replaced it with another board with a MPU on it . On that board it was hard mounted and the fact that you see a lot of ICM chips being soft mounted says a lot to me . Sticking with the MPU6000 .
Hey quick question - I have noticed that some fc's have been coming stock with Yaw D-values enabled. I was under the impression that the Betaflight dev's had dropped Yaw-D from the code and that is why it was set to zero. Did this recently change? Thanks for your help.
It is there, but not advised to use. Yaw is way too slow and D-term adds noise to Yaw, which effects Roll and Pitch movements. So if not upside, but there is a down side, why use it?
But as always BF gives folks options.
Was your testing with BF4.2?... if not then, I am curious on your next testing if the BF 4.2 will manage the ICM20602 gyro data better than the MPU6000 due to it's increased sensitivity - what do you think?
Tests were done looking at Gyro_Scaled (raw) noise.
However, if noise is well defined and not broad is nature, yes, the improvements in 4.2 notch filter targeting (RPM and FFT) will crush it.
Maybe you can get better readings using something similar to Drone Mesh's simulated aggressive flight manoeuvres.
Link?
@@uavtech example: th-cam.com/video/sNDO3RqqHh0/w-d-xo.html . He does this in every esc noise testing/review. You should do the same thing, but instead of outputting directly to the motors, you should output to the pid controller the same set of inputs.
HGLRC removed the F722 dual from there product list. I have one, but used only mpu6000, which is the second gyro for choice. But these board had no(!) memory for blackbox recording and so I changed to the new HGLRC Zeus F722 DJI, which had only the mpu6000 and has memory on board. I fly only analog, but to have it DJI compatible is a nice option
Can’t betaflight devs give you a custom version to acknowledge both gyro data from a dual gyro FC?that would be great for testing
Already done. My buddy Ken gave me a tip on what code to change. Pretty easy.
UAV Tech that is great , will you test this 2 gyro in his way ?
@@Simofly yup.
12:39 talon 20x20 mpu6000 vs talon 20x20 fusion... no contest the mpu version has loads more electrical noise when swapped in and out of the same machine, manifesting its self in the pitch axis... IMO its not an issue of gyro type but an issue of how the recommended hardware spec has been implemented
I was troubleshooting pitch axis noise on two 3" quads recently, It was the damn mmcx axxi antenna hanging off the back. they flop around too much and cause lots of pitch noise. printed some mounts two keep them still, and no more pitch noise.
Exactly!!
I use Kakute F7 ver. 1.1 with ICM gyro. It flies pretty good without any problem. There is a new version of this board ver. 1.5 where MPU gyro is used. Should I switch to the new version? Is it worth the change? Thanks mate for perfect comparison!
I would NOT change just for MPU if ICN is working fine.
@@uavtech Thanks my friend!
I know this is an old video but I figured I'd put my experience. I unfortunately just got a Lumenier Lux Ultimate F7. Out of the box the MPU6000 was bad. On flips and rolls the gyro would just go nuts for some reason. I switched the the ICM20602 and my problems went away. It is more twitchy for sure but atleast it worked for me.
I'm a bit disappointed you didn't mention about overflow bug in ICM gyros. Such an error is disqualifying in my opinion. Yes, I know this is patched in the BF software, but I am surprised that the manufacturer has not produced a revised version so far. Unless I'm wrong. ;)
Good point. But it seems rare. Forgot about it, it is so rare.
i had assumed the icm was more sensitive just because its faster. even when you sample it slowly, the internals are faster. maybe the mass of the actual vibrating mems sensor is lighter, or just the measurement circuit is faster, im not sure. Also, i thought the modern filters in betaflight basically solved the noise problem. i mostly use the icm chips and i never had any issues. just one persons experience of course.
They are both sampling at 8k. You can set ICM to sample at 32k, but then it is a total mess.
I use ICM all the time as well. Just seems like there is a valid reason the old MPU is still used.
I have a better test coming soon!
MPU and ICM on the same FC and a code tweak to log both at the same time.
The matek f722-se has both gyros on it....
Yup, and both hard-mounted on the board. It's a great FC ... if only the STM32F722 wasn't getting low on flash memory now. Seems that STM32F745 is the way to go.
So ICM isolated is cleaner but unisolated MPU is. Always a drawback somewheres.
Drone company that creates good clean area for mounting would bees great.
Matek f722-se
I hope the 6000 wont be extinct soon... Some other company will make them... I hope... or a new one comes to play!?
In my short experience I keep coming back to the 6000... need to try the F1 Ai H7 houdini stuff ...
I believe frames really help on issues that we all relate to gyro management, at least that has been my experience... some frames just resonate more than others.
My two cents im no expert!
MPU6000 all day.
1. most of the FC's using ICM20xxx have a flawed electrical design !
(this is a known fact for about 2 years, follow the link for technical details :
www.rcgroups.com/forums/showthread.php?2934256-OMNIBUS-F4-Pro-Corner-from-AIRBOT/page35#post38947338 )
2. the only board I am aware of that has implemented a sound design is the CLRACINGF7v1
(though it has other problems with the true PIT mode)
3. ICM20xxx are the better gyros
- lower drift
- higher precision
... etc.
It only requires better designed FC's !
I like 20601 better than 20602
What board had that?
Only helio that I know of. But we did some testing a while back and preferred the 20601 to the 20602
They snag any data? Or just a feeling?
👍👀🇭🇷