Of late a few guys have made the accurate comment that info on V1.2 firmware is hiding in too many places. To aid this situation, I shut down the other thread. In this one I'm reposting and tweaking the information that was in the other one an transposing the set up notes for V1.2 FF mixing and head response. I hope if we do it this way, guys can use use the search engine and, quite simply, the title of the thread to find what they need and to ask questions.
First, here are the FF mixing and head response notes:HEAD RESPONSE
When the CGY750 was designed, its engineer worked from the ground up to have the system be as fast (low latency)as possible. Futaba made the system faster still (10-15% more) with the 1.1 firmware.
Some customers have had problems with non Futaba servos running too warm when paired with the 750. The problem has often been made worse or even completely induced by overvolting servos. Factory suggestions for power systems have been published here on RR and elsewhere and have been further clarified by myself and others. There was a period when it was suggested by customers that the device's operating frequency to the servos was excessive. This was proven to NOT be the case in interim beta firmware versions with adjustable servo driving frequencies similar to what we have in V1.2. Servo frame rate should still be set to mfg's specs, but reducing frame rate below the rating of a given servo typically does not appreciably reduce its operating temperature.
So, what's the deal with this head response stuff? Well, extensive testing has shown that there is one downside to having a system as fast as the 750. It turns out that the system can feed a data stream to some kinds of servos at such a high rate that the servos really can't even translate it into useful function. That portion of the data stream effective is seen as white noise by the servo, and it does little more than heat up the motor and increase current consumption. The head response parameter basically tunes the data stream to the rate the servos can effectively accept and translate to useful motion. For the fastest and latest of the Futaba HV servos, the head response works fine when set to 2 or even 1. Other servos in the Futaba product line like the BLS 252/253's or 451's will be better optimized for head response values of 2 or 3. You can easily assess the benefits of head response tweaking by observing the decrease in current consumption with values of 2 or 3 as compared the 1 (the only option for V1.1).
Some of you were possibly confused by the description of how to tell how high to set the head response value. It's a bit tough to give hard recommendations for what values to use for every servo and model out there, because there are so many variations. It gets even more complicated with some guys running 6V rated servos at 6V while some try to run 6V servo on straight 2S. Fortunately, as comlex as these different combinations can be, setting this value is not. If you are running Futaba servos @ or below rated voltage, then head response can be set to 2 or 3. You're done. Those guys who have had issues with other servo brands running too warm should first make sure you're driving them at or a bit below rated voltage. After doing that, I would set the head response to 3 and see how the temps and current consumption look. If it's good, you can try dropping back to head response of 2. It is COMPLETELY normal for cyclic servos to get a bit warm in a well performing model. They're working hard. Your rudder servo gets warm too. What we want to do is tweak the head response and use proper voltage to achieve servo temps that aren't too warm after a flight and with current consumption per flight in the expected range (typically </= 500 maH).
What happens if you run the head response up too high? NOTHING SCARY AT ALL. If you get the value too high, all you'll begin to see are inconsistencies in the way the model flies. For example, you might see rolls start to lose axial-ness or consistency from one roll to another because the system isn't feeding the data stream to the servos as fast as it should.
I need to make it clear that head response IS NOT dumbing down the system. It simply optimzes the control loop to the servos and the model. When it's set right, the model feels good like it always did, but you'll see what it's doing for you in the background when you go to recharge. It also has some benefit to stops, but I'll touch on that later on.AIL>ELE FF mixing (ELE Expert menu) and ELE>AIL FF mixing (AIL Expert menu)
FB model helicopters have certain common characteristics that are present in all brands due to inherent helicopter aerodynamics. For example, a model with a CW rotor direction with no background mixing will often try to lift its nose during the inverted portion of right rolls or tuck its nose in the inverted portion of left rolls. That same model when inside looping will tend to roll off to the left; it'll do the opposite with an outside loop.
When we see these issues in FB models, we can use a program mix in the transmitter to correct the deviation. That mix may be a simple linear affair or a more complex multipoint mix. The mix may differ in direction and magnitude for either side of neutral of the mix's master channel
Now, for the most part, most FBL control systems allow a model to to roll in a fairly axial fashion. The FBL system adds the same corrections in the background to suppress the same tendencies the pmixers do with a FB model. Just how perfect that roll is is highly dependent on model, servos, control ratios, system tuning, FBL system reaction time, and the alignment of the planets . I have absolutely seen the same FBL system roll better in one kind of model than the other. When I first started flying FBL, I worried about what I would do if my given set up didn't roll as perfectly as I would have liked. Having competed for the better part of 20 years, I'm used to being able to tweak a mixer to achieve a model capable of a perfectly axial roll. I worried about what I would do if my FBL system didn't roll perfectly, because you cannot apply transmitter based program mixers to FBL control system trimming. Fortunately, the 750 rolls quite well in most models.
This all said, let's say your model with 750 installed has a slight tendency to lift its nose in a right roll and/or tuck in a left roll. The AIL>ELE mix in the ELE Expert menu exists to fine tune any adverse tendency. To apply this function, first put you model into rate mode on the bench. It's easier to see how the mixer is working. Next, crank up the mixer percentage for the right or left AIL mix direction to 100%. Doing this allows you to see the mixer direction the most clearly. If the model is lifting its nose a bit in right rolls when inverted, the you'll want the mix direction (+ or - sign in front of the %) set so when you give RIGHT AIL command, you'll see the 750 add some UP ELE. For a nose tuck in the inverted portion of LEFT rolls, you'll want the mix direction set to give DOWN ELE with LEFT ail command. Some models need a little mix in only one direction. Some models need it in both. Some need no mix at all. I have one each of the latter two. Now, when you have your mix direction(s) set, go back and reduce the mix percentage(s) to 2% as a starting point AND return your gyros to AVCS. Increase in 1-2% increments until you see the effective improvement of the mix on the way your model rolls.ELE>AIL FF mixing
This mixer is conceptually easy to get your head around. In some cases with a CW rotation model, you may observe you model trying to roll off to the LEFT during inside flips and to the RIGHT during outside flips. In theory, you could observe the behavior during loops, but that has not been my experience; the CGY750 does them well with no added mixing. CCW rotation models, of course, would exhibit opposite roll off behavior if it's present at all.
If you happen to see this roll off when your model is flipping, configure the ELE>AIL FF mixing to provide the needed RIGHT ail with UP ele and LEFT ail with DOWN ele. Try about 2% for up and down elevator commands to start and adjust in 1% increments. It is very easy to observe the benefits of this little mix. If your model has not been flipping as cleanly as you think it could, ELE>AIL FF mixing will fix it handily.
It's about time for a repeat of my usual disclaimer. Do not assume you have to use a mix if your model is already flying well. My model flips cleanly with no added ELE>AIL FF mixing. The mixes are there for IF you need them, not because you HAVE to use them.
Some of you may be aware the the 750 has from the start had adjustable swash timing. In a way, this feature can mimic the effect of the AIL>ELE and ELE>AIL FF mixing. I would suggest you NOT use it for the purpose for two reasons. First, the use of the phasing gives you mix for both left and right AIL commands in the same amounts. This isn't good because it assumes that all models need mix for both left and right rolls AND that the mix value will be the same. This is rarely the case. Second, when you use swash phasing, you not only get AIL>ELE mixing, but also ELE>AIL mixing. The problem here is that the correct mxing for a CW or CCW rotation model for rolls is the exact opposite of what may be needed if the model happens to not flip or loop absolutely perfectly. In effect, the phasing can create one problem as it improves another. Thus you should use the swash phasing only for adaptation to multi-blade heads. Keep it set to zero for usual 2 bladed set ups.PIT>ELE FF mixing, PIT.DAMP, and PIT.ACCE UP and DWN Mixing
Any of you that have flown "NOBAR" are aware that the inherent tendencies of a non-electronically stabilized FBL model are to pitch up on the pitch axis with positive collective input and pitch down with negative collective input.
Most low latency FBL control systems do a reasonable job of supressing this inherent tendency, by you need to understand that since a FBL control system is reactive in nature, there are situations where the nose of the model will lift or dip a bit before the system has the chance to supress the deviation. To effectively address this tendency, V1.2 firmware now has PIT>ELE FF mixing.
The basic PIT>ELE FF mixing should first be configured and confirmed to give DOWN ELE with collective movement above half stick and UP ELE with collective movement below half stick. As a start, set each percentage to 5%. To tune the UP mix, climb the model upright and make sure the head loads just a little bit. Adjust the mix until the model climbs perfectly straight and the nose doesn't lift at all. Most models don't take much mix, so adjust in 1-2% increments.
Several methods exist to tune the PIT>ELE mix for collective pitch ranges below half stick. One way is to fly along at a modest pace and enter a fairly agressive descent. The nose will likely dip a little. Tune the mix percentage until this deviation ceases to exist. You can then average this value against that value obtained in an inverted climbout. Remember; UP ele keeps the nose down in an inverted climbout out, so the mix directions do jive whether the model is upright or inverted with stick positions below half collective.
NOW, what the model does with smooth collective inut may not be the same as it does with rapid pitch pumps in that 1/4-3/4 stick range. Once you have the PIT>ELE mix dialed, go back and do some pitch pumps. Watch the nose. If you see it still try to pitch up or down intially and then come into line or simply deviate as though the basic PIT>ELE is not enough, then dial in some PIT UP and/or DWN acceleration mix. To make sure you've got the mix directions going the right way, crank the mix percentage up to 100%. With the stick at half, at some postive collective pitch. You'll see the system add some extra ELE, and then it will take it back out after a second or so. The mix direction is set by the + or - in front of the mix number. Overall mix direction should be the same as that for the overall PIT>ELE FF mixing outlined above. The PIT.Damp parameter sets how long the system keeps the extra acceleration mix input in place. For now, keep it at 10n (default). I think you can get an idea of how to set this value by having the model on the bench and then cycling the collective at typical rate. The rate of the mix coming in and then tapering off should roughly match the rate you're moving the stick up and down. Again, you can turn the mix percentage wide open to see it better. Just make sure to turn it back down to about 5% (for both the UP and DOWN sides of the mix!) as a start if you feel like you need to use it to completely tidy up the ELE axis during collective pitch pumps.
Next up will be PIT>AIL FF mixing, a feature that will be especially useful to those who have been troubled by their model drifting during tic tocs or rainbows.PIT>AIL FF mixing, PIT.DAMP, and PIT.ACCE UP and DWN Mixing
In certain cases, some pilots have observed a roll off tendency of their model during certain maneuvers such as rainbows and tic tocs. We can use PIT>AIL FF mixing to help address some of these issues. Before you use this mixer array, PLEASE make sure that any adverse behavior is not being created by unintentional pilot input. Before writing this, I consulted with Nick Maxwell about certain adverse FBL behavior that can be created during maneuvers such as tic tocs if the timing of control inputs is not correct. He confirmed that apparent adverse FBL behavior can be duplicated in some cases by pilot error. I am NOT trying to insult anyone's piloting ability here. Just please, guys, if you try to mix out apparent adverse behavior, make sure first that you're not causing it unintentionally. If helpful, have a friend fly your model and confirm he feels the same thing going on.
The PIT>AIL FF mixing is configured conceptually the exact same way as the previously described PIT>ELE FF mixing and associated acceleration mixing. First, examine the model closely in flight and confirm it holds a line in long climb out both inverted and upright and during long rainbows. Adjust the direction and volume of the mix until the model stays straight on the roll axis in the described flight envelope(s). NOW, the acceleration mixing for PIT>AIL can be tricky. If you have the model climbing and/or rainbow'ing straight BUT it seems to roll off just a tad with quick collective input, then you can try SMALL amounts of acceleration mixing to clean this up. Don't use a lot or keep it in for long with the DMP value turned up too high or the model will look like it's doing a belly dance as you come in and out of the collective. REMEMBER; when ANY mixer is working correctly, you can't see it working beyond the model simply flying well._________________________________________________________________________
For the most part, those having issues with wobble or bounce are seeing it with stiff rotor head set ups and with the use of HV servos. These set ups react very quickly to FBL control system input, so high(er) D gain is contraindicated.
Try the following combination:
I gain on AIL and ELE set to no more than 40%.
D gain esp on AIL set to anywhere between 0 to as much as 50%.
Start with AIL and ELE total gain at ~65%. By tweaking I and D as noted, you should be able to get the total gain back up to where you had it.
Control OUT delay on ELE set to 8-10; 6 works pretty well on AIL. Remember to do both sides of stick travel. Of late myself and others are liking 0 (zero) for the control IN delays on AIL and ELE.
Head response set to 2 or 3.
All Futaba digital servos on the cyclics will run fine at 285 Hz. I think that most of you who are using non Futaba servo AND who have had servo heating issues will see the greatest benefit from increasing the head response to 2 or 3 AND by avoiding overvolting your servos. If for some unusual reason your servo persist in running too warm, you can decrease the servo driving frequency to less than 285 Hz. I cannot give specific frequency recommendation for non Futaba servos, but I have seen those data in charts posted elsewhere. Consult them and adjust the frequency as indicated.
Play around a little with headspeed. Different head dampening and blades are sometimes happier at one headspeed(s) versus another and especially at hover.
Other things that overall make the CGY fly better appear below. Your results may vary, but try and see what you think. These are EXPERT menu items, but they are useful to pilots of all skill levels.
1. AVCS.DMP set to 100% on AIL and ELE .
2. If you go into the charts in the back of the manual, you will see a value called angular base that refers the the maximum controlled cyclic rate for a given flight mode. You need to go out and fly a series of hovering rolls and elevator flips and let the system measure the roll and pitch rate. The measurement can be found by scolling over in the home screen. It resets with each fresh reboot or you can do it manually. See the book for a picture of that screen and the flow chart to reach it. Let's take for example flight mode 3 for AIL with a 600/700 class model. You have 300d/s of angular base. Take the value for you STATIONARY roll rate and divide that by 300. The resulting value x 100% is what you should set the ANGULAR RATE in the EXPERT menu to in order to maximize consistency of roll (or pitch) rate between stationary and FFF. Once this value is set, you can fine tune the consistency with RATE CONSTANT. If the model does not have enough stationary roll rate, increase the static cyclic pitch to 9 or 10 degrees, but maintain the swash AFR's in the 50-60% range.
3. Increase the AIL and ELE control feel to somewhere between 7 and 10, inclusive. If the model start to feel "heavy" on the stick, you went a little too far. Control feel is useful for eliminating any tendency towards abrupt pitch up with ele commands in FFF. The model grooves better with a bit more than the default setting of 5, esp. on the pitch (ele) axis.
4. In some cases, increasing the ELE Stop Delay improves ELE bounce. The higher the I gain, the more tendency for bounce. Run I gain no higher than needed for good straight line tracking and stability. Too much will induce or worsen ELE bounce with agrgressive cyclic input.
5. Some people feel like decreasing the control IN delays to little or nothing improves the linearity of the control feel and makes the model feel less "sticky" off center. I agree. You may not, but try it. It's not dangerous or spooky in result. In contrast, cranking up the control OUT delays too high can be disconcerting, so tread lightly there if you go higher than suggested earlier.
This is a work in progress.....
Team Futaba Team Kontronik USA