FY326 Q7 "red board"

Durete wrote: I compiled latest code for FY326 from Hexfet and uploaded to my personal Dropbox.
Devo 7e build:
www.dropbox.com/s/zeawt67tv9j72go/deviat...EST%20BUILD.zip?dl=0

As you know, I don't have this quadcopter so can't test. I'm only trying to help/learn as much as possible, because the 7e build require an "special" environment config to compile properly

You are a good guy, Durete! I'm sorry I cannot help in this case... I have a"green board" FY326 Q7!

I am working on some blog posts that would document some of what we did for the Hontai protocol. Pictures, diagrams, screenshots, videos... it seems there is alot of interest in *deviating* these days!

Thanks for testing. Please comment on how it behaves compared to the stock tx. For example, one difference in this implementation is that there's no dead-zone at center stick - no jump from center. Is that noticeable?

I've added the (suspected) headless and RTH controls on channels 9 and 10 respectively in accordance with the emerging sort-of-standard . That meant the expert mode had to move, and as a compromise I made it a protocol option which defaults to on. I doubt any reduced resolution will be missed if max angle is limited to small values, and didn't want to use up a channel for the control. But the option still gives control over that bit if you want it.

I've uploaded a test build for the 7e under Downloads|Test Builds|FY326 in the left column. The aileron channel is reversed in this build.

Could you also please test binding with other fixed ID values? as many as patience allows...

Hi

First, a big thanks to all the people involved and also to Greenfly for his neat program which was big help with the test ini's
I tried in a quick test a few ID and it binds with all. Last one I tried was 999999. It also binds with " Fixed id=None".
Compared to the stock tx, right stick for me personally feels seemingly a little softer, smoother, docile? I never noticed with stock tx a dead zone around right stick center. Perhaps it had such a zone I don't know but I sure liked the reaction from stock right stick as sporty:)

Rudder is reversed, aileron ok. Trims for right stick are working. quad on the ground and trimming al the way for example to the left, quad takes of in that direction. Same for other directions.
Flip channel 6 works. Tried ini's with throttle trim enabled and disabled but could at first sight not find a difference except for Devo complaint at next start-up, throttle not at -100
The motors do abruptly cutting out when I suddenly need to close the throttle a tad during some sporty flying. About that, could there be a difference with the first build?
I will test more tx id's in a few hour when I'm done with my day job.

Thanks for all the testing Nothing changed with the throttle code in the latest build. Can you explain the symptom in more detail? Just moving the throttle a small amount made the motors stop?

Changing the trims on deviation works through the deviation mixers so not necessary to test. I'll make a new build with the rudder reversed and implement driving the trim channels (maybe as an option just for testing). The current build has all the trim data going to the quad set to zero.

The expert mode is no longer controlled by a channel, so you can't assign it to a switch anymore. You can change it through the protocol option (click on the protocol name on the model config page).

When you say the cyclic seems docile, does it seem like the quad is not in expert mode? For example, does max pitch tilt the quad forward at the same angle as the stock tx in expert mode? Can you tell a difference in how the quad flies when the expert mode protocol option is changed?

The FY326 motors have a serious issue at his design. They start/stop very abruptly at some amount of throttle.
I don't have the FY326, but I have some sets of replacement motors for the FY326 tested at Syma X5 and NiHui U807 and both of them suffers this issue when are near the low end of throttle channel.
FY326 users complaints about this issue, and I guess they are 2 reasons for this issue.
1. Bad design at stock motors.
2. Probably, the 10% jump at low end of throttle channel we saw at SPI captures from stock TX.

Maybe with Deviation we can solve the second reason, but not the first one .

Thanks for all the testing Nothing changed with the throttle code in the latest build. Can you explain the symptom in more detail? Just moving the throttle a small amount made the motors stop?
Say you are at 80 % throttle and suddenly you need to close the throttle almost fully for what ever reason; a little above lowest stick end, the motors just cuts out. They have no minimal spin down or spin up behavior, It's almost like a switch, that's the nature of these particular motors I guess, but on the other hand they are fairly powerfull.
Changing the trims on deviation works through the deviation mixers so not necessary to test. I'll make a new build with the rudder reversed and implement driving the trim channels (maybe as an option just for testing). The current build has all the trim data going to the quad set to zero.
OK
The expert mode is no longer controlled by a channel, so you can't assign it to a switch anymore. You can change it through the protocol option (click on the protocol name on the model config page).
Yes I know it, for this build I switch no channel anymore, I have the protocol option "on" and I only use the Fmode switch for / with the channels Ail/Elev/Rud/ with "scalar=85" for fmode0, and switching to Fmode1 sets the max. is that not ok?
When you say the cyclic seems docile, does it seem like the quad is not in expert mode? For example, does max pitch tilt the quad forward at the same angle as the stock tx in expert mode? Can you tell a difference in how the quad flies when the expert mode protocol option is changed?
No problem, the way it flies, the quad is in expert mode, and the pitch looks the same after a very short moment, but the stock tx seems to start it faster instead of seemingly little more gradually with the Devo but after a short moment the pitch looks the same.
My day job is almost done and can test again this evening regarding protocol option on or off.

Is it a problem if I ask at rcgroups FY-326 thread if somebody else is also willing to do testing so we have more / different input?

I have a Huiying X6058 that has horrible throttle response with the stock transmitter. If you back the throttle off an amount that usually feels normal, the motors may stop or appear to, until you give it more throttle. The U807 is almost as bad, but both were cured on the Devo 7e. I tinkered with the throttle curve to soften it a bit below 50%, and that seems to help, or it may be that I just got used to it. Anyway, the throttle implementation improved immensely. I never had the impression that there was anything wrong other than the stock transmitter.

I have my own thoughts regarding the motor / throttle response which may or not be correct.

I do not believe the problem is caused by motor design. The motors do not start until a value of 0x1E is reached on byte 5 of TX payload. The sudden jump imposed by the stock transmitter is only to a value of 0x14. I would say (with my limited knowledge of these things) that the problem is therefore at the receiver end.

Not the end of the world though, because there are at least two good hardware modifications that cure the problem. I have been flying my FY326 for a few months with 7mm motors, and had forgotten all about 'low throttle dropouts', I just don't experience them any more. The quad does not take off until further up the throttle travel, and this keeps you away from the 'dead zone'. The other solution is adding weight of some kind to the quad.

Anyway, just my thoughts given in the hope that I'm incorrect!

My Devo 7e has arrived, BTW, but there's a learning curve and modifications involved, so I'll be a little late to the party.

Yes, you are right
I forgot a third reason, how the MCU handle the motor output.
This motors are driven by a Mosfet with a PWM signal from the MCU. If the internal MCU firmware doesn't implement a "soft start", no matter what motor you use.
BTW, I'm totally sure the motor start is a lot more abrupt than other motors set I used before. I have this behaviour fully tested.

Adding to my previous post, I tested the protocol option "expert mode" switching between on and off, it works like low and high rate like on stock tx but I will leave the option "on", no need for me to switch it off because if I get in trouble, it takes time to reach the protocol option if the wind takes the quad;)

Thanks for the testing and info.

Trying to visualize the "abrupt" and "switch"-like throttle behavior. There's always some point where the props change from stopped to rotating. Is it that they change immediately from stopped to some fairly fast speed, such as almost taking-off?

I've updated the test build to reverse the rudder channel and drive the trim values to the receiver. Code is pushed to the repo. There's a temporary protocol option Dyntrim to control this driven trims - default is on. Be careful because I may have gotten some of the directions wrong. That should manifest as reduced control range on the axis (roll, pitch, yaw, throttle). Please test on each axis with Dyntrim on and off and let me know which ones have increased control range, and which ones decreased. Based on the previous posts it's likely the throttle will be the same on both settings.

Feel free to recruit testers from elsewhere. Let me know if builds for other transmitters are needed. Most of my time on RCG is spent in the classifieds

Installed the new build, and I saw indeed in the protocol options"Dyntrim=On". But afterwards, after switching Devo off, and checking my ini the "Dyntrim=on" is not saved to my ini. Only the Expert mode got saved / save automaticly.

BTW, No reversed channels anymore, they are ok Had 3 flights no problem.

After those 3 flights, for some unknown reason ( had inbetween few times Devo off and on) it saved the Dyntrim parameter to my ini. I suggest others to do first tests INSIDE and not outside! Because the next flight...
Edit; Now I fully understand what Hexfet was telling doing it one by one;) hmm it shooted straight up in few seconds 20 meter high, only thing I could do was switch off Devo lol. Luckily the FY-326 shuts off instantly. phew it was really a rocket! I like that

- Throttle; After Devo bind, I always needed throttle stick up and down to arm the Quad. The behavior now is, after bind, I intend to arm and moving throttle stick a little upward, the motors start spinning and when trying to close throttle / stop position, they keep spinning at full speed.

- Aileron, moving stick, half to the left = ok quad flies left, but full stick to the left during left flight, quad shoots to the right. Stick to half and full right = ok. (Meaning it does no weird things, but seems slower then in the other direction?)
- Elevator, pulling stick half backwards = ok. Stick full backwards quad changes to forward motion. Pushing stick half forward is ok. Full forward is ok. (No weird things but slower?)
- Rudder, moving stick half to the left = ok. Full stick left, quad changes left turns to right turns. Moving rudder stick half to the right is ok. Full right is ok.(meaning no weird things but seems little slower then the other direction?)
Testing was a little difficult because of fairly strong wind.

The above behaviors I have it also with Dyntrim off which is weird and have it happen with scalar=125 When I change it to 100 no weird things happens with Dyntrim on. Maybe I missed how to use Dyntrim. Getting confused lol.
10th edit I sure hope Dyntrim can work ok because in the certain directions it is really responsive as long as I stayed away from stick travel end.

Thanks for the testing and good descriptions. Can picture the crazy behavior

I haven't seen any issues in the emulator with the model file being updated. What you saw with dyntrim off was due to the scale of 125. I forgot to limit the values sent to the quad. Please don't use anything greater than 100 scale till the next build. With these small quads anything over 100 only decreases the stick range anyway.

Don't know what's going on with the throttle trim. On the other channels it sounds like some internal values are being overflowed, which is consistent with the behavior you saw with scale at 125. Is it possible to make a capture on the stock tx when both stick and trim are full in the same direction, and also opposite? For example, set the roll trim to full right, then move the aileron stick full right, then full left. I'm interested if the tx limits the channel value based on the trim setting. If the stock tx also acts strange at full stick + full trim, then please reduce the trim to try to find the max trim value that still works as expected.

I just did a quick experiment.

1/ Set full LH roll trim, move RH stick fully left and capture. Byte 2 = 0x00, byte 9 = 0xEC.

2/ Set full RH roll trim, move RH stick fully left and capture. Byte 2 = 0x00, byte 9 = 0x14.

Does this tell you what you wanted to know? Do you need proper captures to be posted for this?

@ Hexfet, the captures I think S.Giles will do it if he reads your post because I have no logic analyzer.
About the scalar, i already had them back to 100 to save me from unwanted surprises. With the throttle mishap, it really took of like a rocket while I thought I was gonna arm the copter lol.

Yes, thanks. Looks like the stock tx just sends the full stick and full trim values. This is similar to what happened with the Hontai aircraft, so I made a new build with the same solution - cut the trim values in half. That may or may not make a difference. I also removed the driven throttle trim completely.

Please try the test build I just uploaded. If the controls are still wonky with the dyntrim enabled then I'll just set them to zero.

Is it this one; deviation-devo7e-v4.0.1-2b733e6 with the same name as the previous test build. And in testing, with dyntrim on, I need to keep the scalar at 100 ? Just asking to be sure

Yes, that's the right file.
Yes, please keep the scalar at 100 for testing. Though I've changed the code so values over 100 are handled correctly.

Flies ok with Dyntrim on and ok with Dyntrim off (both instances was with scalar 100).

Good to hear. When the right stick is at full deflection is the angle of the quad noticeably greater with dyntrim on? is the yaw rate faster at full rudder?