Servo requires "tapping"


 
I too had the shakes when my damper was controlling around the 0-10% open position. I turned on the 60hz filter and problem now gone. I have very smooth damper and control action now.
 
I spent a few hours testing various configurations of booster circuits and wire lengths and have come to the conclusion that no booster makes a difference here.

Booster Circuits
  • None - direct from Microcontroller
  • Voltage Booster - HeaterMeter v4.2/4.3 circuit with a 1k pullup to 5V on servo signal
  • Voltage Booster + Inline Diode - Same as voltage booster but with a diode after the boost
  • Current Booster - common collector amplifier / emitter follower / Ralph's linked circuit

Servo Connections
  • Directly to HeaterMeter (<100mm)
  • End of ethernet cable (~2m)
  • Long run simulation. This uses the 2m ethernet cable but also adds a 10 ohm resistor to the servo signal after the booster to simulate a longer run (~75m).

Of all the connections, the long run was by far the most problematic, with the servo going nuts all the time so most of my testing was done with this setup. The Current Boost and None setup did equally well. Both Voltage and Voltage + Diode did poorly, but only mildly better than the other circuits but I'd say none of them were acceptable in terms of performance. The Current Boost puts out a 2.7V signal with a 0.24mA current draw during the pulse followed by a -0.20mA dip as the pulse is discharged. The None was similar with a 3.3V signal, 0.24mA current, and a discharge that maxed out my current meter (-1.30mA). Voltage Boost put out a 5V signal but used almost twice the current 0.4mA and the same maxed discharge. All three had a high inrush current which maxed my meter >1.30mA.

I then spliced in a splitter to the servo end of the output and installed a socket for placing a capacitor on the power lines at the servo end. First capacitor, 10V 470uF, all circuits had zero problems. 10V 220uF - zero problems, 10V 100uF - zero problems, 25V 47uF - zero problems, 25V 22uF - zero problems. You can see a trend here. Then I tried 50V 0.1uF which had zero problems with the Current and None boosts, but showed rare glitches with the voltage boost circuit. Placing any of these capacitors on the HeaterMeter side of the 5V line had no beneficial effect. Without the capacitor, the 5V line measured at the servo varied from 4.14V to 6.18V.

Conclusion: No boost circuit works perfectly, but the Voltage Boost which is integrated on the HeaterMeter 4.2/4.3 boards is the clear loser in all testing. The best solution is to place a capacitor on the servo's 5V/GND pins as close to the servo as possible. Based on these observations, future hardware designs will certainly not use the Voltage Boost circuit. The Current Boost circuit as it did not show any beneficial effect above the None and given that the peak current draw and pulse current draw are the same with both of these, the Current Boost will probably not be used either.
 
I didn't go about it in a scientific manner like you did, but my cable is about 50ft, my servo spazzed like crazy. I added the current booster and the servo works perfectly. I put the current booster in and out of circuit many times and this was completely reproducible. With the voltage booster my servo works fine as well, though some Tower Pro models do jitter a bit while other servo's work fine. I leave my HM and RD3 connected and powered 24/7, been over a year maybe two, so I know the servo is not overheating, I also know without some sort of servo booster my servo would be DOA on this 50ft cable.
 
So Bryan, the results using a capacitor across the supply leads on the servo is very interesting and I will probably add a 25v 100uf cap from my parts bag to the blower/damper I am using. Did you try turning the filter on and see if you could get the same results as you would by adding the cap. Curious to see if anyone else had the same results that I did.
Glad your safe and that Irma left you with hopefully minimal damage. Hope the others in your state get back to enjoying their lives as they were prior to the storm real soon. Were thinking about you out west.
 
Yeah Ralph am not at all doubting the effectiveness if your circuit for your build-- it clearly works for you. I am just unable to reproduce the results quite as well. It seems odd considering the current needed is so low (1/4 milliamp) that a current boost can help. It must have something to do with the impedance of the circuit responding to the turn on, or the sharpness of the pulse edges. I couldn't see any skew of the signal in any configuration, although I was measuring closer to the HeaterMeter and not at the servo end for the pulses. Before I disassemble the test rig I am going to do that now that I think about it because the pulses might look wonky on the far end of the cable.

Gary, I did not try the 60Hz filter. The filter just changes how the analog inputs are read, trying to match them with the period of the AC frequency so it seems like a strange coincidence that it made a difference in your case.
 
I think you using a pure resistive load to simulate 50ft of wire is where the difference lies.
I also forgot to mention that the degree of malfunction of the servo went up with the size of the blower being used, and the effectiveness of the current booster did equally well to correct servo malfunction when using larger blowers.
 
Decided to try this out with a wonky servo I had. I put a 47uF cap across the servo power supply (fit nicely in the LAN connector under the cap) and it took every bit of "wonkiness" out, and it runs very smooth.

I'll be building all my dampers with this mod, now.
 
Will,
If you can, will you post a picture of how you attached the capacitor to the lan connector.

Pretty straight forward ...

WUnTASy.jpg
 
thanks for pic Will. My next question is does the pinch terminals on the cat connector hold both the cap leads and servo wires tight. concerned that the wire cross sections might differ enough to cause a flakey connection. Do the other wires just come out the side or do you still have enough space in the center to run the wires.
 
It fits with the wires. I leave the servo and fan wires on pins 4 and 3 a bit long and solder the cap leads to them on the outside of the plug.
 
It fits with the wires. I leave the servo and fan wires on pins 4 and 3 a bit long and solder the cap leads to them on the outside of the plug.

Yahbut do you maintain the twist to within a half inch of the punchdowns? :D (that's sorta important for high speed transmission.......)
 
but when you changed the servo control to "move and rest" mode I had a problem with the servo where I would find the damper open a bit when it should be closed.

Yeah, I had to revert to old firmware because I the move and rest mode is not compatible with my damper. I really wish that was an option in the GUI because servos are cheap--ruining meat due to a servo that didn't move is where the money is. And on top of that, my damper is several years old now I've never had a servo fail with the firmware that keeps power to it all the time.
 
You'll be happy to know that the current snapshot firmware leaves the servo on all the time (unless you switch the HeaterMeter mode to off). Just update the AVR from the online repository.
 
You'll be happy to know that the current snapshot firmware leaves the servo on all the time (unless you switch the HeaterMeter mode to off). Just update the AVR from the online repository.

Very cool! I'm going to build my other one (I have a full kit that I haven't assembled yet) and put the new firmware on it and take 'er for a spin! Thanks, Bryan. That is good news!
 
Thanks for all the research you guys are doing. I think I'll need to get a cap and try it out.

Bryan, I'll try not to bug you if I can help it.
 
Update:
I have a new servo and capacitor. The servo will work fine without the cap, but the minute I install the cap, neither servo will move. It appears that the Tower Pro SG90 servo works fine with no cap, but the Tower Pro MG90S does not work at all.

It is possible that I have a bad cap or that something was wrong with the install.

Thanks for all of your help. In the future if I have a servo problem, I'll just replace it with another SG90.

Edit: I think I just found my problem. I ordered a 470uF cap instead of a 47uF cap! Doh!
 
The size should be fine, generally more is better when it comes to power smoothing, although too much can create resonance instability with the power supply technically. Are you putting the capacitor across the 5V/GND lines or the SERVO/GND lines? It should be 5V/GND. I think I've tried a capacitor that large before and it worked but it has been a long time since I tested it and I believe it was with an SG90.

Edit: Also make sure the capacitor polarity is correct, although if it were backward I think it would just reboot over and over again.
 

 

Back
Top