YGE-120LV and CGY750 gov

[Steve Evans]

Having had some issues with the Align RCE-BL60G ESC on my 500ESP not being able to run at full power without causing instability under high load I've just fitted a YGE-120LV VBar ESC using the following adaptor cable I've just made to make the phase sensor output suitable for the CGY-750.



Under the skin it looks like this.


I was rather pleased that this all worked first time and a log of the RPM the first time I spooled it up on a 3S battery is shown below. A little noisy at full throttle, but pretty good.



I'll be reporting back soon with the results of gov testing.

Steve

[exfokkerflyer]

Very interesting! Keep us posted...

Tom M

[Andreas]

So a simple pullup resistor doesn't work again? Same issues as your phase sensor project Steve?

[Steve Evans]

A pull up wouldn't help as the YGE has plenty of high drive. The issue is that the CGY750 has a limited max input frequency and the fact that phase sensors put out 3 pulses per motor rev rather than the one of a magnet sensor exceeds this limit.

The pre scaler in the cable above is effectively the same as used with the Hyperion phase sensor but with a resistor network on the input to drop the 5V signal from the ESC down to 3V.

Steve

[Steve Evans]

Proper bench test time. First thing to note is that the YGE ESCs have six response speeds which use the same naming as the spool up speeds on the programming card; heli slow/mid/fast and plane slow/mid/fast in increasing speed. For VBar operation plane mid/fast is generally recommended which is in keeping with general VBar advice to set ESCs for swift response.

The CGY750 governor is clearly a different beast.

My test setup has now standardised somewhat and I use two set speeds of 2500 and 2720 RPM, switching between them to observe the governor response. It's the edges in the plots below which are most telling; rate, overshoot, ringing etc all being things to tune, so a 200rpm step is really useful.

I initially set the gov to that which has worked reasonably well with the Align RCE-BL60G ESC; quick gain 20%, up delay 6, down delay 16, startup delay 2.

I started with the fastest setting, "plane fast" and was disappointed to see the following.


Then "plane mid" which wasn't much better.


At his point I took a break for a think, and then decided to try "heli slow".


Very different, and although clearly hunting, no longer with the instability, so I tried "plane slow"


"heli fast"


"heli mid"


Of all of these "heli fast" suited my 500ESP with it's 500M motor best. This makes me wonder if I should have tried my Align ESC with a slower response as the instability seen here with "plane fast" is very similar to that seen with the Align RCE-BL60G.

I then explored gain. I'll post those graphs shortly.

Steve

[Steve Evans]

It's just been pointed out that I should have applied the formula "(kv * voltage * poles) / 20" to select a switching frequency. I had been aware of this, but my mental arithmetic had failed me resulting in me selecting 8kHz. With a 1600kV 6 pole motor on 6S this equation suggests 10.7kHz would be better. I have the choice of 10 or 12kHz though, so which would be better? Should I always round up?

Steve

[joe1l]

Originally Posted by Steve Evans

It's just been pointed out that I should have applied the formula "(kv * voltage * poles) / 20" to select a switching frequency. I had been aware of this, but my mental arithmetic had failed me resulting in me selecting 8kHz. With a 1600kV 6 pole motor on 6S this equation suggests 10.7kHz would be better. I have the choice of 10 or 12kHz though, so which would be better? Should I always round up?

Steve

Too high a pwm can cause your motor to run rough. this is how i first learned of the PWM calculation and why I started applying it to all my YGE esc's. I was using a motor which had a recommended pwm of 12. What i found was when the pwm was set to 12, i saw issues holding headspeeds at higher throttle positions (above 75%). I learned of the pwm calc and started seeing big improvements...in power, temp and efficiency. When the PWM was too high like in the one specific case i gained power and a smoother running motor...and when it was set too low, i saw motor temps go down and flight times go up by raising it to the appropriate value.

[Ranger71d]

Very nice formula. I wish I'd have known this earlier. I like using science vs. The Black Arts to find the correct settings for things.
Looking at my 450 Pro V2 set up, Scorpion 2221-6, Ice Lite 50, running 3s 2.2aH 40c packs. I've been running 8 pwm. Using the formula provided (4400kv * 11.1v * 6poles = 14,652.
My current settings have the esc and motor coming down barely warm with the packs warm when holding them. Im only sport flying so not a lot of stress on the system. But seems like I can improve efficiency and flight time with better settings. Windy the last couple days so I will check this out when the weather decides to cooperate.
Thanks guys.
Sent from my SAMSUNG-SGH-I727

[Steve Evans]

I like science too, but theory often doesn't measure up to practise.

I've been continuing my experimentation. I've settled on the following test procedure with the heli clamped down.

"low" pitch set to -6deg
"high" pitch set to -9deg

Test passes consist of spoolup followed by gov @2500rpm then gov @2720rpm, repeat, and then two 6deg pitch pumps and two 9deg pitch pumps.

This provides a repeatable basis for comparison with step changes to help set the up/down delays, and consistent loads to compare the effects of various settings. In this case, the setting I'm exploring is ESC PWM frequency and timing.

I started with 8kHz and 18deg timing.


On the 6deg pitch pumps HS dropped to 2596rpm. 9deg pitch pumps dropped to 2554rpm.

I then tried 10kHz and 18deg timing.


On the 6deg pitch pumps HS dropped to 2667rpm which was a significant improvement. 9deg pitch pumps dropped to 2579rpm which is better, but not a great deal.

I then tried 12kHz and 18deg timing.


On the 6deg pitch pumps HS dropped to 2661rpm which was much the same as at 10kHz. 9deg pitch pumps dropped to 2610rpm which is a distinct improvement compared to 10kHz.

I then tried 14kHz and 18deg timing.


On the 6deg pitch pumps HS dropped to 2650rpm, a little worse than with 12kHz. 9deg pitch pumps dropped to 2600rpm, again a little worse than 12kHz.

I then tried 12kHz with auto timing.


On the 6deg pitch pumps HS dropped to 2667rpm, matching fixed 18deg timing with 12kHz. 9deg pitch pumps dropped to 2617rpm, a little better than 12kHz fixed timing. There was also a little less overshoot on the rising edge of the HS after the dip. To be honest there wasn't much to choose between fixed and auto timing, but auto had the edge, so I settled for 12kHz/auto.

Steve

[Steve Evans]

I then decided to try Plane-Slow response again.


Quite a bit of instability is evident.


This is in contrast to the same (9 deg) pitch input using Heli-Fast.


So 12kHz PWM, auto timing, Heli-Mid startup and Heli-Fast response are my final ESC settings.

Steve

[Steve Evans]

I played around with up/down delays, but found that the original setting I'd settled on at 8kHz (after my first post in this thread, although I never posted the plots) was good. I'm using an UpDelay of 8, a DownDelay of 14 and a StartupDelay of 2.

Next was to experiment with feed forward through use of a throttle V curve.

I found that an 83-79-75-79-83 curve worked well for the 6deg pitch pumps, the HS dropping to 2713; i.e. not at all!



And 85-80-75-80-85 was better for the 9deg pitch pumps, but too much for 6deg. At 9deg the HS dropped to 2683rpm, but it should be noted that the battery was at 3.7V per cell at this point, and so this is worst case.


Based on my normal max pitch range of +/- 12deg I'll set the throttle curve to 87-82-75-82-87 for my flight tests.

I'll also apply cyclic to throttle mixing, which judging from my previous flight logs is at least as much load as collective.

In summary then:

Final settings:

ESC:
HeliMid/HeliFast
12kHz PWM
Auto timing

CGY750:
Gear.Rt 8.10t
ServoType DG:1520
Yaw.Comp CW/TOP
Wrk.Mode Governor
Response Quick
Gv.Gain 20%
THR.mode Tx.Curve
Rv.UpDly 8 Frm
Rv.DnDly 14 Frm
StartDly 2 Frm
On.Revo 60%
LLmt.Hov 25%
LLmtIdup 25%
Low.Revo 1000rpm

Tx:
Throttle 87-82-75-82-87

Steve

[Steve Evans]

It's worth noting the the throttle range of the YGE ESC has been calibrated such that 1100uS PWM input is zero throttle, and 1940us is max throttle.

The steady state value is 1820us (86%) at the end of a flight (3.7V cell voltage) and 1770us (80%) at the start of a flight (4.2V cell voltage). Based on the advice that a pinion should be selected such that the gov is set at 80-85%, it would appear that my choice of a 15T pinion is spot on.

Steve

[Ranger71d]

I just tested 4 flights with the PWM set to 12 and the timing set to 10*. They were: PWM 8 and Timing at 5* (setting listed on Scorpion's website for my motor).
Today is about 10* warmer than the previous logs I compared today's flights to.
Ran my Sky Lipo 40C 2200mAh packs for 5min 30 sec each, 4 flights. I noticed two things.
The temps in the CC log showed that my average temps were about the same 140-146*F range. But my max temps (seems to only peak at this temp for very short periods of time) went from 200* down to 170*.
Also noticed that I used between 100 and 150 less mAh during these four flights as compared to previous flights. End of flight voltage per cell was 3.78v as opposed to the previous average of 3.7v per cell. I gathered this information from my charger.
I'm going to try PWM set to 12 and timing back to 5* and test this setup out.
I'm not being as detailed as you are Steve when in comes to ensuring near identical flights. But my flying skills only allow for so much variety at this point. So there definitely could be differences in the temps and voltage/mAh usage do to this.
Thanks for your posts on this . It is definitely helping me out. Getting a better grasp on how to go about properly setting up the electronics.

[MrMel]

Steve, VERY nice wrap-up!

It will be interesting to see when you do some testing inflight as you add one more component and that's sudden unload with non-static "recovery", sometimes you unload the heli quick and its still some load on the heli, next time its negative load since gravity helps.
That will put another dimension on the testing and requirements from the governor.

Ranger71d, Do note, PWM and Timing differs from one brand ESC to another brand ESC, because one ESC likes one thing the best it's not given the next will do the same, that is one thing I learned in my testing.

[Ranger71d]

Thank you Mr. Mel. I'll keep in mind that the esc Steve is using in his tests are not the one Im using.
Im using his info as a starting point to compare different settings. Before I just used what someone else used for settings. Not really fond of that. As each heli will have differences.
From the work Steve and you have been doing I have been learning quite a bit. Im becoming more confident in testing things out vs. going with the flow.
Thanks for the hard work guys.

Sent from my SAMSUNG-SGH-I727

[Steve Evans]

Traditional April showers rather got in the way today, but I managed to get out for a short flight at lunchtime. Following a quick test flight after which the low voltage warning (a double beep) was sounding, I fixed the battery voltage monitoring in my logger. In the chart below I show the total voltage/6 to show the average cell voltage such that the scale works for all the traces.



The flight includes hover, a left flip, a back flip, left and right piros, three pitch pumps, a right flip, a circuit, another left flip, left flip to inverted, left flip out of inverted and then a rain induced landing.

Overall it can be seen that the head-speed held pretty well, except for the three pitch pumps. All significant manoeuvres as listed above resulted in a dip in battery voltage, although none should really have triggered the under-voltage warning from the ESC. I shall probably disable that as I judge my flight duration using the tx timer and I log all flights anyway.

The flips and piros look good in closeup. Note that I added +15% AIL->THR and ELE->THR mixing to provide some feed-forward. The flips were the smoothest I've ever experienced with no noticeable head speed drop at all.



A pitch pump in closeup shows a clear problem. Well predicted Mr Mel!!!!!!



As the head is unloaded at the top of the climb the head-speed rises and the ESC PWM signal backs off. The head-speed then drops, overshooting, and the governor then ramps the ESC PWM signal back up (to full throttle) to recover the head-speed.

I had hoped to get outside again this evening to try some more pitch pumps with different settings, but alas the weather was against me.

I'd appreciate thoughts as to what the cause of the errant pitch pump behaviour might be. I've already had input from one fellow HeliFreak member who's more knowledgable on control systems than me who's suggested running the CGY750 with a Moderate rather than Quick response and then potentially speeding up the ESC response from Heli-Fast to Plane-Slow. It is apparently better to have faster plant (i.e. that being controlled) and a slower controller than the other way round (as it would appear to be at present).

A closeup of the pitch event clearly shows the steps in the ESC PWM signal (red) (taken at 7ms intervals) and the resulting head-speed changes (green). These are rather noisy compared to the battery voltage (blue) and throttle/pitch inputs.



Slowing down the CGY750 to smooth out the ESC PWM signal may help matters.

An alternative I intend exploring first is to increase the DownDelay from it's current value of 14 to see if that will prevent the head-speed undershoot.

Unfortunately, as can be seen in the above, in response to a full pitch pump the ESC PWM hits its max limit. This represents saturation in the control loop, and it never a good thing. The hope is that by slowing down the CGY750 and speeding up the ESC the command to the ESC to speed up will be delivered sooner and thus max throttle will not actually be required. The hint that this may work is shown below where a 9 deg pitch pump caused saturation on the ESC PWM signal (blue) with Heli-Fast response.



But not with Plane-Slow even though it was noisy.



Steve

[Steve Evans]

A spot the difference puzzle...

Down delay of 16


18


20


I think we can conclude that this hasn't made much difference! It is unclear from the manual exactly what the delays do. Previous experimentation HERE has clearly demonstrated that the up/down delays affect tracking of the set point (i.e. the selected head-speed) when switching conditions, and the above has confirmed that there is no discernible effect on the response in maintaining a constant head-speed.

I also tried dropped the governed speed to 2650rpm to help prevent saturation of the ESC PWM output and it didn't help.

Steve

[joe1l]

Hi Steve, are you doing all of your testing with v throttle curve? Have you tried using a flat line now that you are using the YGE? Also is your throttle type still set to 1520? It should be set to analog to work properly with the YGE.

[Steve Evans]

Originally Posted by joe1l

Hi Steve, are you doing all of your testing with v throttle curve? Have you tried using a flat line now that you are using the YGE?

Hi Joe,

I introduced a V curve in post #11. I played around with PWM/timing etc with a flat throttle curve.

Steve

[Steve Evans]

More testing this evening. Lot's of runs, and not much success, but something particularly worthy of note is worth sharing.

It's been suggested that I try PlaneSlow rather than HeliFast response, and also to try Moderate rather than Quick CGY750 response. Neither has proven particularly good on the bench, however the difference between using Analog and DS1520 gov servo settings is noteworthy. This was with PlaneSlow, Quick, gain 10%.

I'd been using Analog as recommended and shown below. The tail was twitching quite a bit on load changes.


I then tried digital and the twitched disappeared and the heli sounded much better, with no fluttering noises, however the corresponding plot looks less stable.


Not sure what to make of this. I guess some flight experimentation is going to be called for.

Steve