"The Better Mousetrap" Servo Tester

[mikronauts]

I've made a new build log page my new servo tester at my site, as I thought others may be interested



You can find my build log here:

http://www.mikronauts.com/the-better-mo ... vo-tester/

I'd love to get your feedback! (and I'd be happy to answer your questions)

[mikronauts]

Servo Jitter Solved!

My MG-995 servos were jittering with my “The Better Mousetrap Servo Tester”, and today I decided to get the bottom of it before I started testing HexPi, my RoboPi + Raspberry Pi based hexapod robot.

When I hooked up a scope, and looked at the 1500us centering pulse and the servo power, here is what I found:



Time for some troubleshooting… more @ http://www.mikronauts.com/the-better-mo ... vo-tester/

[BMS Doug]

Another pitfall of the cheap servo, well worth finding now.

I'd probably consider adding the capacitor in each servo, you may be getting volt drop down the servo wiring (another cost saving exercise, under-rate the wiring) in which case each individual servo would perform better with its own capacitor. Or replace the wiring to each servo...

You get what you pay for, unfortunately.

[mikronauts]

Thanks Doug, I agree 100%

One of the reasons I am going nice and slow with this build is to find the pitfalls ahead of time... had I rushed ahead, connected all 18 servos, and tried to get some IK code running.... I think I'd have torn my hair out.

Having the "Kentucky Fried Servo" experience, and finding another dud (2 DOA out of 50) strongly suggested a slow approach - which melds nicely with how little time I have available for this project.

Right now, I am torn between adding 3300uF+100nF caps at each servo, or adding say 33000uF per group of eight servos, and an individual 100nF per servo. Plus I have to get some bigger electrolytic caps, the largest I have in stock is 1000uF... I feel another Digikey order coming on. I'll try different arrangements to see what works best. Blogging my experimental results hopefully will help others stuck with less than stellar servos.

This is slowing down my HexPi build, but I am certainly learning a lot about how to get cheap servos working!

I totally agree re/ getting what you pay for... but I could not get wifey approval for $12-$20 per nice servo (x20-x50 servos, I bought two hexapod chassis') into my toy budget so I am reminded of an old saying:

"When life hands you lemons, make lemonade"

or

"When you have cheap servos, see if you can use them anyway"

p.s.

Silver lining: I am finding my new servo tester quite useful!

Another pitfall of the cheap servo, well worth finding now.

I'd probably consider adding the capacitor in each servo, you may be getting volt drop down the servo wiring (another cost saving exercise, under-rate the wiring) in which case each individual servo would perform better with its own capacitor. Or replace the wiring to each servo...

You get what you pay for, unfortunately.

[BMS Doug]

Thanks Doug, I agree 100%

One of the reasons I am going nice and slow with this build is to find the pitfalls ahead of time... had I rushed ahead, connected all 18 servos, and tried to get some IK code running.... I think I'd have torn my hair out.

Having the "Kentucky Fried Servo" experience, and finding another dud (2 DOA out of 50) strongly suggested a slow approach - which melds nicely with how little time I have available for this project.

yep, I know how that feels, there's never enough time for everything.

Right now, I am torn between adding 3300uF+100nF caps at each servo, or adding say 33000uF per group of eight servos, and an individual 100nF per servo. Plus I have to get some bigger electrolytic caps, the largest I have in stock is 1000uF... I feel another Digikey order coming on. I'll try different arrangements to see what works best. Blogging my experimental results hopefully will help others stuck with less than stellar servos.

This is slowing down my HexPi build, but I am certainly learning a lot about how to get cheap servos working!

I guess the easiest route is to try the bigger capacitor on the group first and see if it works. Good luck.

I totally agree re/ getting what you pay for... but I could not get wifey approval for $12-$20 per nice servo (x20-x50 servos, I bought two hexapod chassis') into my toy budget so I am reminded of an old saying:

"When life hands you lemons, make lemonade"

or

"When you have cheap servos, see if you can use them anyway"

p.s.

Silver lining: I am finding my new servo tester quite useful!

Yep, I knew you'd say that, getting hold of the toys is the first step in using them.
My wife is fairly tolerant of my Dalek project, perhaps because technically the dalek is hers...

[mikronauts]

yep, I know how that feels, there's never enough time for everything.

Too true!

I guess the easiest route is to try the bigger capacitor on the group first and see if it works. Good luck.

Thank you. Yesterday I made a dinky diode based board for dropping the voltage, and today I added 3x 1000uF and 3x 100nF caps to filter the power.

Testing all servos on HexPi shows that is enough for about 12 of the servos to behave, six still jitter - I should have big caps arriving on Thursday, I'll update the thread with results using them.

Yep, I knew you'd say that, getting hold of the toys is the first step in using them.
My wife is fairly tolerant of my Dalek project, perhaps because technically the dalek is hers...

DALEK???

Do you have a link to your project? I'd love to see it.

"Exterminate..." MG-995 servos...

[BMS Doug]

I haven't got any photo's yet.

All of my work so far is documented here but it doesn't look like a Dalek yet as I haven't put it into the case.

I'll have to organise a proper project page, but I really don't have anything to show yet. (just a pair of wheels on a standard base, a raspberry pi, a L298N and an 12c 16 channel PWM output board). I'm still fumbling my way through programming everything up and won't put it into the case until I'm a bit further along (I need to get my battery working, I took a twin 3.6v ( so 7.2v) 4ah battery from an emergency light that was being thrown away and also took the charger circuit. I need to reconnect the two halves, add a plug and connect the charger (in such a fashion that it's safe from wandering children from poking their little fingers in)).

I get about 2hrs a night that I can use for working on stuff, or I can watch TV instead... so when I'm tired slobbing out takes priority and no work gets done.

I'm learning Python as I go, I may eventually re-do everything in C, which would mean learning that language too.

[mikronauts]

Thanks for the link Doug, I'll read that thread!

I picked up a selection of larger capacitors for my servo tester, I am bound and determined to get the cheap "less than stellar" quality servos I picked up for HexPi to work well enough for HexPi. My current approach is to filter the &@#$&^! out of the servo power - it does help a lot. I blogged a bit more about it on my site.

I want to solve the problem to the extent that *at most* HexPi will have a slight tremble when holding position.

I haven't got any photo's yet.

All of my work so far is documented here but it doesn't look like a Dalek yet as I haven't put it into the case.

I'll have to organise a proper project page, but I really don't have anything to show yet. (just a pair of wheels on a standard base, a raspberry pi, a L298N and an 12c 16 channel PWM output board). I'm still fumbling my way through programming everything up and won't put it into the case until I'm a bit further along (I need to get my battery working, I took a twin 3.6v ( so 7.2v) 4ah battery from an emergency light that was being thrown away and also took the charger circuit. I need to reconnect the two halves, add a plug and connect the charger (in such a fashion that it's safe from wandering children from poking their little fingers in)).

I get about 2hrs a night that I can use for working on stuff, or I can watch TV instead... so when I'm tired slobbing out takes priority and no work gets done.

I'm learning Python as I go, I may eventually re-do everything in C, which would mean learning that language too.

[mikronauts]

I had an interesting thought... would the GWS MT-1 servo tester still behave better (ie almost no jitter) if it was powered the same way that TBMST was?

The answer is NO! It is the the power supply!

As soon as I come up with a clean enough power source, and I can resume my HexPi build!

It should not take long, I have several different DC-DC converter modules, and many types of batteries in my lab.

More details at:

http://www.mikronauts.com/2014/06/09/tb ... t-results/