RC Servos and decoupling caps

BrendanP

Senior Member
I have 2 RC servos in a complex pcb design running off the same V REG as the picaxe.

I know the servos generate a lot of noise in their supply when in operation.

Will decoupling ceramic and electrolytic caps on the servo lines filter out this noise to an acceptable level? What cap levels are recommended, the ubiquitous 100n and 10n?
 

inglewoodpete

Senior Member
I'd put a 100n as close as possible, on the board, to each of the two servo headers. Also, one as close as possible to the supply pins of every other chip on the board.

Also, a single electro (47 or 100uF) close to the servo headers to act as a 'tank' supply for the servo motors. The 100n's should be closest to the headers.
 

Dippy

Moderator
Brendan, these little caps aren't magic 'cure alls' . You have to think about the reasons they are used.

Probably depends on Servo spec and your PCB design?
Have you determined that extra capacitors are needed? ('scope on power rails).

Have you tried a cap or two? Quicker than waiting for an answer.
I would be too impatient to wait. I'd just give it a go.
It won't explode.

Obv. you already have decoupling next to PICAXE.

You'll be using the 'star' method for power and ground PCB tracks I assume?

I would guess that if your PSU is slightly weak or your track layout not good then you'd need a lot more capacitance of the right sort. Maybe a 'reservoir' >100uF lo-Z electro as close to motor as possible in adition to ceramics.
Ceramics for tinkywinky and sparklies and electro for bongawonga transients and surges and dips.

If I were doing it I'd dig out the 'scope and my box of caps. I'd have a go rather than relying on Forum theorising.


Just remember, that your PCB physical design and where you put the caps is more important than just throwing any old caps at the problem.
And remember also, there are a dozen different specs of ceramics. They range from cheap rubbish all the way up to COG and N0P dielectrics.
You'll be needing X7R or better ideally.
They musn't be confused with poly-whatever caps which have different specs but are available in similar values.
 
Last edited by a moderator:

BeanieBots

Moderator
Everything that Dippy says.

But why try to make a bad design work by plastering it with caps.
Use a seperate regulator for PICAXE and the Servos.
As Dippy states, correct layout will sort out most issues. By having seperate supplies, even if the servo supply dies completely, your PICAXE won't even blink. Also, a seperate regulator will use less space and cost less than the large electrolytics required to hold a line high when your servos pull several amps each.

How many servos?
What is the peak current?
Can your regulator supply that level of current?
What size cap required to supply that level of current?

There are TWO issues to consider.
Noise - decoupling.
Collapse - electrolytics/regulator.
 
Last edited:

BrendanP

Senior Member
Thanks guys.

I hadn't even thought in terms of using electrolytics to prevent V drop during high I drain during servo operation. Thanks for raising that issue. I had thought only in terms of using a V reg. with plenty of margin. I am familiar with using a low ESR electrolytic in this role because a GSM module I did lot of work with required it to prevent V level slump and hence module auto shut down during burst data txing at max. power.

I didn't even know what what star distribution was for pcb layout. Thanks Dippy. I found this doc helpful. Also thank you for pointing out the fact there are different quality/grades of caps., I wasn't thinking of that. I will specify the high grade caps you suggest.

http://focus.ti.com/lit/an/szza009/szza009.pdf

I propose to use this line of capacitors for decoupling Dippy.
http://www.mouser.com/catalog/specsheets/x8r.pdf

I have decoupling on the picaxe supply pins and on the reset.

I am going to do as has been suggested and use a separate reg. for the servos and the mcu.

I am using 2 X Hitec HS-311 servos to operate a pan-tilt on which the pcb is mounted.

relevant Hitec HS-311 specs.
Operating Voltage: 4.8-6.0 Volts
Current Drain (4.8V): 7.4mA/idle, 160mA no load operating
Current Drain (6.0V): 7.7mA/idle, 180mA no load operating

At 5V the servos pull 630mA stalled according to my trusty bench top power supply.

The deal is I have a hardware engineer doing the pcb layout. He has come back to me with a partial schematic and a lot of questions. Some things he is very knowledgeable about, others not so good, I know more and my BA degree is in humanities. His specialty is DSP (digital signal processing) which I would of thought would be highly sensitive to just the sort of design issues we are trying to ameliorate here so I'll point out to him the issues.

The upside is the development costs for the pcb are low enough that I can afford a bit of hit and miss. I know that good design should obviate the need for "a have stab at it and see what happens approach" but I have the margin there to do just that if need be.
 
Last edited:

Dippy

Moderator
That's a good document Brendan.

A lot of design is a experience-related common sense.
For example, why have a reservoir capacitor?

Think of a long wire and a relay coil at the end , switched by a transistor.
That's just like a PCB track and your PICAXE switching a relay.

When you trigger the tranny (and switch the coil) the coil current has to come all the way along the wire / track.
3 things.
1. V=IR drop in related to the current and transient, means a voltage drop if your, e.g. , PICAXE taps off from that shared line.
2. Radiation from that line ... is it near a sensitive signal line?
3. When the relay coil is switched off a reactive transient appears on that line.

So, now stick a reservoir/tank capacitor just by the coil. Now what happens?
The initial transient surge comes from the capacitor now, so vastly reduced current surge in wire/track.
This reduces radiation.
And the switch-off reactive transient is mostly absorbed by that capacitor.
A low-Z electro is better due to better HF performance and lower ESR.
It acts like an RC filter to a certain degree.

Result: Less rubbish on the power line. Happy PICAXE and smiling user.


All the noise/transient stuff is a big subject which someone should write up as a Tutorial and save in the specially allocated Forum section for Tutorials. The AXEIKI.



A slight red herring (or is it Read hearring on this site?)
A PCB design document I have for RF circuit design reminded the user about the orientation of certain SMD caps when sitting on PCBs with ground planes. The oritientation of the capacitors plates wrt the board.
Of course, it's obvious after you've been told isn't it.
"Yes, well I knew all about that as I have worked in circuit design since 1643."
 
Last edited by a moderator:

BrendanP

Senior Member
A lot of good info there Dippy. I didn't appreciate the full benefits of using capacitors as a "reservoir". I'm very grateful.

Btw, why isn't a back back emf suppression diode used with RC servos given that the servo is just geared dc motor?
 
Last edited:

Dippy

Moderator
It's also handy if your PSU is a bit feeble.
Allows a relay to switch more cleanly if the power supply is a bit droopy (within reason).
(Within the time constraints of charging the capacitor of course.)
 

westaust55

Moderator
Btw, why isn't a back back emf suppression diode used with RC servos given that the servo is just geared dc motor?
Because there is a controller board between the supply and the bi-directional motor which should look after such "things" for RC servos
The internal controller looks after position, motor speed and motor direction
 
Last edited:

slimplynth

Senior Member
A slight red herring (or is it Read hearring on this site?)
Am i the only person who felt disappointed in themselves for trying to click that non-existent link, an eggsellent read hearring :)

Good stuff (reading) by the way, i just picked up Ferric Chloride and Copper Clad board to make my second PCB (Led Tester circa 1992 was first)
 

BeanieBots

Moderator
Btw, why isn't a back back emf suppression diode used with RC servos given that the servo is just geared dc motor?
Are you talking about a "Hobby Servo" as used by radio control models or are you talking about a "servo motor" as used within "servo systems"?

If the former, then it's what Westy says. (the diodes ARE fitted internally) and there's a LOT more to it than just a geared dc motor.
(how else does a pulse stream become a position demand?)

If it's the latter, then the same rules apply as to any other inductive load but many 'true' "servo motors" have caps built into each winding so that they don't appear as inductive.
 

westaust55

Moderator
Sorry for the confusion, I'm referring to radio controlled aircraft style servos like the Hitec units I gave the data for earlier on.

I am using them to activate a small pan/tilt similar to this design.
http://www.servocity.com/html/spt100_pan___tilt_system.html

Just to clarify, there are back EMF suppression diodes within the servo case?
Basically YES.

The servos that I have as purchased from Jaycar and modified for 180 degree operation internally have 4608 Silicon Limiter Diode packages to which each of the DC motor terminals are connected.

In effect there is a whole miniature processor inside each servo.
The servos I have include internally an ATMEL Atmega8L microcontroller internally which takes the signal from your PICAXE, monitors via a feedback pot the current position of the output shaft and from those two variables plus how far from the desired position sets the motor to operate in the correct direction and at a speed that reduces as the output shaft approaches the desired position.
 

BeanieBots

Moderator
Just to clarify, there are back EMF suppression diodes within the servo case?
The servo electronics will take care of the motor back emf for you.
There are no direct connections to the motor available to the end user.
All you can do by way of suppression is add decoupling (electrolytic + ceramic) to the servo power lines.
Due to the internal electronics, most hobby servos will have good internal motor suppression including a snubber across the motor to help prevent RFI as well.
Just concern yourself with your layout and decoupling and leave the motor control to the guys who designed the hobby servo.
 
Top