Possible actuator problem

Waggly tails
Mike_K
Posts: 351
Joined: 16 Feb 2018, 06:35

Re: Possible actuator problem

Post by Mike_K » 18 Nov 2019, 19:51

Where do I start?

Tobe is correct to say that most Tobe Rand actuators don't suffer from the problem. So if you aren't suffering from the "glitch/freeze" problem don't try to fix it! And how brief the "glitch/freeze" is, you'll probably not notice it when flying. I noticed it on the bench, but still went flying and didn't notice it in the air.

One of my two Tobe Rand actuators stopped misbehaving once I'd freed the linkages/hinges up, the problem definitely shows up more when the actuator is working against a load. The other one still occasionally "glitched/froze" every 5 to 10 minutes, but a lot less than before. But for the tests, I reverted to the stiffer linkages to see if the diodes/resistor were helping.

It was Phil's idea to try a TVS diode and I received mine today and they seem to fix the problem. I bought mine from Farnell, but RS has them too:

https://uk.farnell.com/stmicroelectroni ... EN-SKU-MDC
https://uk.rs-online.com/web/p/products ... lsrc=aw.ds


As well as the TVS diode there's a 200R 0.6W resistor across the motor terminals.

Before I got the TVS diodes, I used two zener 5V6 1.3W, back-to-back with the 120R across the pair as it was all I had. It/they work well too. You could use higher-rated zeners, but these ones survived over an hour of running without failing.

https://www.rapidonline.com/vishay-bzx8 ... de-47-3058

I also tried separate batteries for the actuator motor and Rx (linking the GND but splitting the Vcc and Vin) and it worked OK without any diodes/resistors. But again I'd still fit the diodes/resistor as well.

If you are programming your own ATtiny85 to drive the Pololu DRV8838, I've modified Phils program to separate the Enable and Phase pins, and turn the enable off to let the motor "coast" fractionally before changing direction. I shamelessly copied this from what Martin (ceptimus on this forum) has done with his pulse recoder/driver. It solves the "glitch/freeze" problem and the motor runs cooler too. And you can still fit the TVS or zener diodes for "belt and braces". Rudder and Elevator are still on P0 and P2 of the ATtiny85, the enable is P1 and the Phase on P3. But you can change the order to suit.

And finally, I have tested all the above on the bench for hours, but I haven't had a chance to test fly any of them yet. And the above worked for my actuators, but as we know, they are all slightly different, most work well without any modification, but we might find that the above doesn't solve the problem for all.

Mike
Attachments
digispark_gg_mjk_v2a.ino
(7.46 KiB) Downloaded 35 times

Mike_K
Posts: 351
Joined: 16 Feb 2018, 06:35

Re: Possible actuator problem

Post by Mike_K » 19 Nov 2019, 07:00

Phil_G wrote:
18 Nov 2019, 17:30

Incidentally, has anyone tried a simple decoupling capacitor across the receiver power rails (like the famous Spektrum "capacitor-on-a-plug"), say at least a few hundred μF ? (I think the Spekky ones are 4700μF)

Cheers
Phil
Hi Phil

I tried plugging in an old Spektrum capacitor, but it makes no difference. But the component that is causing the problem is the DRV8838, not the PIC/Arduino or the Rx. I tried connecting a 220uF Low ESR Electrolytic capacitors across the GND and Vcc/Vin on the DRV8838 and it worked well, but unless you can get one that is physically smaller, then it's not so practical.

In fact I'm still not certain what is causing the DRV8838 problems. Is it resetting, or is it one of its many self "protection" functions kicking in. If you read the TI data sheet, there is over current, over temperature, low voltage etc, etc. My storage scope is not quite quick enough to look at the transients, I really should buy my self a new one.

Mike

jmendoza
Posts: 123
Joined: 18 Feb 2018, 23:07

Re: Possible actuator problem

Post by jmendoza » 22 Dec 2019, 21:01

I now use the 220 ohm, and the chokes (reduce peak current draw) and a minimum of 1000mAh capacity 3S Lipos. The other thing was I put in a softer spring Tobe gave me as I had some of the very first GG Tobe actuators with the stiffer spring. After those mods, I have not had a glitch, but that could just be luck as they are random.

From my understanding, if the current limit feature of the DVR board gets activated, you get the glitch. The trick is to limit the peak current, but that is not always possible to do successfully due to the peak limit being slightly different on each DVR board. I confirmed this with my storage scope and found the PPM output never drops out from the PIC, it is the DVR that shuts off.

In the past, when I did get the glitch, it was random, and independent of the battery voltage, as it could occur in the first minute of flight, or later after a few flights on the same battery. One thing I did note however was I never got the glitch on landing approaches as the throttle was low or off.

When I have gotten the glitch it is very un-nerving. because thankfully I had enough altitude to recover. Once, I almost racked it in, it was too close for comfort.

Many people never said anything to me about the glitch, leading me to beleive maybe they have not used their GG Tobe in flight. So I'm amazed to read all the stories here confirming many have had the glitch, apparently it is not as rare as I once thought. What would be nice is to find a brushed motor driver with a higher current limit threshold to cure the problem on existing Re-codes people have already built/purchased.


One thing I have noticed is that some actuators are much more free mechanically than others, meaning if the top rudder plate is manually pushed to one side and then released, some actuators will return to neutral, while others stay there. This means when the DVR does shut down and glitches, the severity of how the plane reacts is due to if the spring pulls the plate back, or if the actuator stays to one side because of a frictional bind, as I pointed out above. This may be why some people do not notice the glitch, or say it is not very severe, while in my experience I have had some really close calls when the actuator stayed locked hard over. Both Rand, and Tobe actuators vary quite a bit as far as their mechanical friction and ability to quickly return to neutral on spring power alone. In my experience with Rands, it is the gear mesh that is the source of this friction, and with the Tobe, it is the gear drive that is the variable.

Post Reply