Having bought some of these recently and experimented with them I would like to share information for the benefit of others, this is relevant if you are building your own controllers or trying to understand why they do strange things!

These units are widely faked including the label so performance varies. The units I purchased most recently had digital innards, it would appear the genuine analogue SG90 is unlikely to be still available, all comments apply to this digital version. Unfortunately they are indistinguishable externally but the interior can be viewed through the rear case, if it has a single SO8 chip & SOT23  device with no large capacitors it is the digital version, this picture is the top view of the pcb inside-

I fear there are several versions of this digital pcb but the final key is the chip being marked AA51880 data here

This servo draws a peak current of just under 1 amp when moved and about 35mA idle.
I only use this servo over 90 degrees of movement, pulse width from 1 to 2mS, I have found some do not operate reliably at the extremes of 180 degree movement using pulse widths of 500uS & 2.5mS.

The idle non-moving current can become very large under some circumstances, if there is a fixed load like a spring trying to move the shaft to some other position than that commanded or if the servo is unable to reach the commanded position due to some obstruction. Either of these situations can rapidly heat the internals of the servo and maybe your power supply.

A way around this that I employ is to switch of the servo command impulses several seconds after the movement.

Noise immunity is fairly poor and can lead to instability particularly with long connecting leads, especially if they are of small cross section such as servo extension leads. I combat this by using large gauge (0.8mmsq) bus wires for 0V & +5V that feed each servo at it's connector, the peak current (and therefore voltage drop) in the 0V bus is also reduced by using a large (1000uF) capacitor at each servo and a small (6R8) series resistor to +5V.

 These servo's are very tolerant of frame rate, I tried all the way from 10mS to 250mS and they worked happily (it does not affect the servo speed as rumoured for the analogue ones) but the large frame rates do have an interesting effect near the end of the commanded stroke, they first overshoot slightly and on the next pulse reach final position, it maybe they always do this but at higher frame rates it's unnoticeable.

I have so far seen no evidence of power up twitches or any kind of jitter. Certainly if you have an oscilloscope check the control micro for software induced jitter, this if present will simply be acted upon by the servo. I also found any momentary supply glitches below 4.2V would upset it violently (sometimes a full stroke) so proper decoupling/filtering and large gauge wire helps to eradicate this source. If the problem persists it may be crosstalk from other servo channels or traction wires in the same cable bundle, a 10nf capacitor from signal to 0V at the servo may help or in extreme cases of crosstalk reducing the drive of the processor by adding a 100R resistor at it's output (before the long wire).
 
To be continued

Interesting findings Roger.

I use servos on one section of my layout but admit, I've never been inside one…………………..

Your comments about small section servo control cables and noise limitation might well be valid on mine.  I am quite surprised by how much noise the things actually make when operating - I'd been led to believe they were virtually silent as opposed to both Tortoise motors and the clunky solenoids.  My longest run from the controller is around 1.5 metres - I use (I think) 22 SWG Futaba type connecting cable.  It could of course, be the cheapo servos - SG90 purchased from Hobby King some time ago.  (I've just noticed how much the price has gone up since I bought mine !)

Using the Megapoints controllers, I haven't noticed any kind of twitching - except on power-up and whilst electrically centering them in programme mode when I'm guessing power is supplied continuously.  I think in "normal" mode, once they've completed their programmed run, power is cut.

I thought Roger was talking about electronic noise 'Chattering' interference Peter, interupting the pulses (Signal?) From the Digital board, along the thin cables, to the servo :hmm. I could be ( and very often AM) wrong  :It's a no no

Trying to follow along  :thumbs

Cheers

Matt

You  may well be right Matt  - I remember I was wrong once in the past …..

I was thinking electrical noise Matt but Peter reminds me the cheapo SG90's may be a problem on some baseboards that are rather to good a sound board. I remember a large O gauge club layout where the sound of many motor driven points (I forget the make) changing was excruciating and very distracting.

Edit, Peter I get no twitching on power up at all, even without any extra decoupling. I know nothing of your controller but I hazard a guess it is supplying the 5V for the servo's. Does your power up twitch still happen with just the native servo length lead ? and does it still happen if you remove the orange wire (signal) from the connector and leave it floating ? also does it happen if only one servo is connected to the controller ?

My servo controller is the Megapoints version Roger and yes, I get servo twitch for a few seconds every time I power up.

Reading Dave Fenton's notes (Mr Megapoints himself), I see there is a different servo he recommends if "twitch" on power up is to be avoided so presumably, the problem (if it is a problem) lies with the servos I use.  To be honest, that initial twitch isn't a problem for me so I'm not about to change them all ………………

When I bought them, from Hobby King in the States, they only cost me around £1 each - UK duty almost doubled the price !!

The twitching stops after probably 2 seconds, if that, then they're still.  My "noise" is just the common, electronically uneducated, audible noise which might be baseboard related although it doesn't seem to matter if the baseboards are chipboard or Sundeala.  My preferred surface is ply but, where the servos are installed, the boards (old)  just happen to be  either Sundeala or chipboard …………….. 

[user=6]Petermac[/user] wrote:

Reading Dave Fenton's notes (Mr Megapoints himself), I see there is a different servo he recommends if "twitch" on power up is to be avoided so presumably, the problem (if it is a problem) lies with the servos I use.

TBH Peter AFAIK there are many variations (clones) of the SG90 so I may just be lucky and maybe Dave recommends a servo that has less or even no clones so they are all guaranteed to operate the same. I guess usually a suppliers stock are the same variety so perhaps the process is buy one, test it and if ok buy the rest quickly hoping for the same variation.

Something else I notice is the odd servo from a batch seems stiffer then the rest (cannot be manually rotated with reasonable force on the horn when unpowered). These seem lower in pitch when moving and perhaps slower. Took one apart but cannot find anything wrong in the gearbox except the stiffness, it's almost as if one of the gears is to large or the shaft mounting off it's proper axis. Hard to see much amongst all the lubricating gunge! If one eventually fails I shall carry out a post mortem.

I know I'm a total philistine in these matters but I prefer the satisfying surge of current from my CDU and the immediate "thunk" of the point(s). Perhaps this is why I model 1930s GWR and WC&PR Steam and am allergic to Tortoises.

That aside, It is excellent work here and your technical expertise has me in awe but with no aspiration to follow down that path. I've enough on my plate already with things I can get my mind around.

Hi Colin, thanks for the appreciation, I guess we all have different talents  :) I remember well the H&M's that I was still using up till last year, guaranteed to rip any stiff mechanism to shreds!  As I had sold everything in the expectation I was giving up railway modelling I had to literally start from scratch anyway so bought a servo to experiment with, so far the results seem good with hopefully the bonus of being able to operate semaphores in the future too.

Regards
Roger

I too like that "thunk" as the solenoids try to wreck the points - it lets you know they've fired whereas the gentle "zzzzz" of slo-mo motors or sevos always leaves me guessing.

However, I know what a pain it is to replace the point in the middle of a complex junction when all are joined with insulated joiners so now mainly use either Tortoise or servos…………………..

I am waiting to find out what a servo fire smells like!! Will it be as satisfying as a solenoid smell, the same smoke I wonder, is the lubricating gunk in the gearbox perhaps flammable to add authenticity, only time will tell  :mrgreen:

Will it be a Havoline smell or a Castrol smell …………..  :hmm