Has anyone had this issue? When I play back three recording of six servos the EZB will randomly disconnect from WIFI to my router. The EZB does not brown out or reboot. It simply disconnects. There are also other scripts running at the same time moving other servos and sound file playback on the robot.
There's my setup and what I did:
* I have three servos in each arm in my robot for a total of 6.
* I'm using EZ Robot's HDD servos.
* One servo moves a wrist, Second one opens and closes a claw and the third rotates the claw about 180 back and forth.
* I used the servo Pad to move them and used the servo Recorder to record them.
* I now have three recordings. One recording is of left arm moving the wrist and rotating the claw, second recording is the same but of the right arm, third is of both claws (right and left) opening and closing.
* The servos in the recordings are moving very fast.
* The recordings are playing back at the speed of 1.0 (normal speed)
Here's how I solved the issue. The EZB does not disconnect anymore:
* I reduced the playback speed to 0.6.
*I haven't tested raising the playback speed yet to see at what point I'll start getting disconnects again. It's a random problem and I don't want to spend hours finding that line. I'm ok with the speed of 0.6 for now but would like a more frantic look.
Why am I getting WIFI disconnects at full playback speed? Am I flooding the WIFI digital pipeline?
Upgrade to ARC Pro
Harnessing the power of ARC Pro, your robot can be more than just a simple automated machine.
I can't find anywhere in the caps datasheet that says anything about Aerogel. Here's the order page on Mouser of the ones I'm using.
In fact I'm having a hard time finding Aerogel Supercaps anywhere. If you don't mind, what brand are you using and where do you get them?
I was just lurking around today. Here are some pictures of my super caps that I’ve never had a problem. Sorry I don’t remember where I got them at but it might help with the name that’s on the caps. Name on caps say superstor .
This May not help or you already have this script running.
EDITED: I have my caps on ACD7
Datasheet here: http://static6.arrow.com/aropdfconversion/5595f83bc15c9d812700e2f0df7c9c90edf2c0c4/powerstorp_specs.pdf
They are available at Digikey and Arrow, among other places. I got mine from Solarbotics and Digikey.
Maybe it's the ESR value from the caps you have Dave? ESR (equivalent series resistance) will affect how fast a cap charges and discharges, the lower the ESR the better!
@Dave Maybe you should try using the same caps as Jeremy’s or the ones I’m using as a last ditch attempt.
Also I noticed in your video, I’m not sure if your sound amplifier have its own power source or getting it from the EZB and are you using the caps on the bottom EZB too?
last question please; on your B9 you are using EZB’s too, right. Is it the same Wi-Fi that you’re trying now on the B9 you’re working on?
I really hope you get this figured out as you have spent a lot of time and love of labor putting this B9 back together.
I just watched you video again. Have you tried pressing the reset button on that EZB, then reapply the client network for your WiFi network. I’ve had to do this once in the past for my EZB. I’m sure you also applied the EZB v41/2 firmware. Just a thought.
Of course still get the new supercaps that Jeremie posted.
@Jeremie, you mention ESR. I compaired the valus of both your's and mine from their datasheet and this is what I found:
The ESR of my current AVX supercaps are rated: 140 ESR Max @ 1000 Hz (m)
The ESR of the caps you point me to is : 1 ESR Max @ 1000 Hz (m)
That's a huge difference if I'm reading this correctly. I'll be ordering the ones you recommend and see if this works. I appreciate your help and time spent. I know you're a busy guy.
@Merne, Thanks for all the extra effort and advice. The pics really helped. I will follow your advice and get the caps you guys are using. T answer your questions:
No power for any device is being routed through the EZB. Everything, sound amps, lights, motors and servos are wired directly to the power source.
WIFI, Both the robots are using the same wifi network. In fact, the new EZB I installed in this second B9 (after I burnt up it's EZB) was pulled from my personal B9. It already had the sound modification done and proven to work. Saved me some time and effort.
I'm pretty sure it has the most recent firmware. I'll double check that. I'll have to relearn how to do that though. It's been years since I've done it. LOL.
I was asking a friend of mine who manufactures ventilator power supplies about this issue. Here area a couple of his suggestions I'll include for food for thought. I'd love to hear your thoughts on them:
I assume I only need to use one of these and not have to solder two of the 2.7v caps together?
EDIT: I think I found them. These are 2.5v, 1f Eaton supercapacitors but I don't see that they are Aerogel. What do you think about these?
I see these are the "B" series you are using. first link in this post that you pointed to is for the "P" series.
I would get the caps from your edited post, the Ultracapacitors 1F 2.5V EDLC B SERIES CYL, 2 of them.
They are fast charging and B series. I don’t think it matters what brand, but that would be a question for Jeremie. I have never had a problem with the caps I am using currently.
As for the Transient Voltage Suppressor I really don’t have any expertise in this area I believe JeremIm does. However I will look at the quotes out of curiosity.
Out of curiosity what if you took everything off the EZB like the amp And anything else connected to it. Even the super caps you have currently connected and try booting and see if you hear it connect to Wi-Fi every time.
PS, I am by far no expert in any of this. I am a very novice and been away from it a long time and it has changed since DJ form his new company has.
i’m slowly getting back into it myself and have to redo everything for my Inmoov currently. And I work at a very slow pace because I have a hard time focusing. LOL. I’m just giving suggestions that I hope work.
Edited: I forgot to say if it does connect to the Wi-Fi every time with everything disconnected, then I would first put the amplifier back on and see if it does the same thing. If it does it could be your amplifier.
DJ in the past has had me remove everything and put one thing at a time on to see which one is causing the issue that I was having at the time just food for thought.
I'm unsure what to think of your friend's advice. I don't see cross-talk noise ever being large enough to jump over and trigger the reset line. I have never seen this in all my career.
So capacitors are used for decoupling (shunting AC noise to ground) and for storage. Capacitors may look like a short at t=0 if they are filtering the first rising edge but that wouldn't affect anything because power is not on yet. Until the power supply is actually on and then short-circuited then bad things will happen but at t=0, who cares, it's as if there is not power applied. On switching power supplies, output stage capacitors are used for storage and to keep the output level steady. I am unsure about the ventilator world but in the robotics world, we are dealing with huge current demands and see brownouts happen very often. It's a common problem especially when using large motors. As mentioned, one solution is to increase the current capacity of your power supply but you could end up with an outrageously expensive and huge power supply to meet the demand. In robotics we are always worried about space, so another solution is to increase the output capacitor storage capacity (as you are doing). Yet another one, is to use battery power.
In my experience TVS diodes are not the answer. I have used them and when they see a transient voltage they sacrifice themselves and no longer work. I really wanted them to work because they sound awesome, but they failed me. In any case, TVS diodes are not the answer in this situation, they are not going to help with brownout conditions as this is a current starvation problem, not a transient voltage (high positive/negative voltage spike) issue.
I don't think that there's any reason to doubt that the supercap method works, because it does, both Merne and the Synthiam inMoov are proof that it does work. I'm not familiar with AVX supercaps and I'm not sure the reason why they wouldn't work for you, maybe Aerogel is the key.
@Merne, You have some real good points and tips there. Actually if I pull the supercaps off the EZB will boot and connect to the network properly. No matter what is connected to it. When I was first was troubleshooting this issue I actually had everything unplugged and fond no difference in behavior. Running bare and free the EZB booted and connected. As soon as I installed the AVX supercaps it would lock up. I'm going to order the caps I found in my edited text. They seem very similar but are not Aerogel. I'll also keep looking for the Aerogel Jeremie mentions.
@Jeremie, I really appreciate you sharing your knowledge, thoughts and insight. I'm learning a lot on this subject. You have helped me understand a lot of what is going. You also have helped me avoid some missteps. I'm looking forward to getting similar caps like you all have and see the difference. This has actually been a fun process so far.
@robo rad, Well, yes I actually have a lot of extra ezb stuff. However I've ran out of boards. LOL.
I have noticed additional behaviors with the caps I currently am using. If I leave them plugged into the EZB they will drain down to t=0 overnight. As I mentioned I have nothing pulling power through the EZB. Perhaps the EZB it's self is draining them.
#If the EZB is turned on with the caps attached in this state the EZB will lock up.
#If I pull the caps The ezb will boot and connect just fine.
#If after the ezb boots and connects and I plug in the supercaps in the state of t=0 it will disconnect and shut down. The top green LED will continue to flash but there bill be no lights down near the reset button.
@DJ has mentioned using a switch to connect and disconnect the caps after the ezb boots. I've simulated this. I've so far found that if the caps are or near t=0 the EZB will lock up when the switch is closed and the caps are added to the power rail.
I still need to test to see if the caps will shut down the ezb if they are charged when the switch is closed. Perhaps if the caps are isolated from the EZB with an open switch while it's turned off the caps will not drain down to t=0.
This morning I turned on the robot. The relay was still opened and caps disconnected. The EZB booted and connected to my network. ARC was started and connected to the EZB. Next I had ARC close the relay, connecting the caps to the rail. No issues at all! Everything stayed operational and connected.
I think as long as the Super caps don't discharge down to near T=0 then this should work. However I still have the new series B caps coming this Thursday from Mouser. I'll wire them in when they get here and see if there is any change in behavior.
#When the AXV super caps get down to near T=0 they are seen by the EZB as a dead short and shuts down part of the board, disconnecting it from my network.
#When the AXV caps are charged and disconnected from the EZB , the next day when I boot the EZB and connect I'll then close in the charged caps and everything remains stable.
#It seems that the EZB drains the caps when all external power is turned off. Perhaps the cap drain happens because the EZB is still trying to draw power to operate.
Questions still unanswered;
#Will the new Series B caps I have yet to use to replace the current AVX caps have this same behavior? New caps coming:
Supercapacitors / Ultracapacitors PowerStor / Eaton 1F 2.5V EDLC B SERIES CYL. These have a much lower ESR () (Equivalent Series Resistance) then the AXV caps.
#If the PowerStor / Eaton caps do seem to work for this application do I still need the relay to disconnect them from the EZB when external power is disconnected?
#If I still need to disconnect with the relay due to drainage then why is the EZB draining power and why are other builders having success with this modification and I am not?
This doesn't sound right Dave, dead shorts damage electronics. As I explained before, if the caps have a small amount of charge, 1.5V let's say, it will keep the chips have powered on, and likely the watchdog timers won't be able to reset the chips yet.
I wouldn't recommend this because you are placing a fully charged cap on a power rail, if it arcs and discharges you could see some damage. Remember you are dealing with a huge amount of capacitance 0.5F.
Yes, that is how the supercaps work and this is exactly what they are being used for. When there is a brownout condition due to current starvation of the power supply voltage the power rail dips because of Ohm's law (V=IR). If the current (I) goes down the voltage (V) goes down because of the linear relationship (the load (R) aka resistance, stays the same) and the supply browns out (too low of voltage to keep circuit powered). The job of the supercaps is to hold the 3.3V power rail up until the brownout passes. When you remove power the EZ-B the supercaps will try to keep the 3.3V up for as long as they can and slowly discharge as the power is used. This is what @DJ meant when he said they act like a battery. In the future, many devices will likely have supercap batteries, some toys already do.
No you don't need a relay, I wouldn't recommend it. As I mentioned, both Merne's inMoov and the Sythiam inMoov are both operating with the supercaps attached to 3.3V, nothing more, nothing less.
I am pretty confident that you'll see success with the new caps. If not, I would remove that other item you have connected to the ADC ports and test again.
The only way I've been able to get the AXV supercaps to work with the Super caps I currently have is with the relay. I'll get the new caps tomorrow and swap them out and pull the relay out of the circuit.
I read in the supercaps data sheet that these caps are very sensitive to soldering heat. They recommend soldering quickly to keep from damaging the caps. I'll be careful not to apply too much heat when soldering the legs together.
Your recommended removing the "other item" I have on the ADC header. I'll do that if needed but it's only an open switch. I don't even have a pull up resistor on it.
My the Robinsons be with you.
I'm not really sure what the secret is that made the difference. Weather it's the ESR, whatever the "B" series is pertaining to these new caps or I damaged the other set soldering them together. My "gut" tells me that Jeremie is correct that the caps ability to take a quick charge is key.
As far as an operational test I did get one lockup and wifi disconnect while checking things out after the install. I was running all the servos in the robot at the same time, moving them back and forth quickly and repeating this test quickly many times. I think I was depleting the caps and simply drained them of power faster then they could recharge. I slowed down the servos a little and doing the same testing this never happened again. All and all I'm pretty happy with this now.
Thanks again for all the help. I'll let you know if anything changes.
Recap and final working solution:
#I was having trouble with the EZB disconnecting from the WIFI network and needing to be rebooted before it would work again. This happened when servos moved in a very quick moving animation routine that I had recorded using the servo Recorder and servo Pad.
#Servos used are 6 EZ-Robot HD servos. They are all moving at the same time.
#This animation and others created by the servo Recorder using the servo Pad was being played back through an EZ script in ARC when the shutdown happens. There were three separate servo recordings being played back in that script involving the six EZ-Robot HD servos.
#I was advised to attach super caps to the 3.3 ADC power rail to shore up and stabilize any voltage sag caused by the servos drawing peek power when they moved.
#After installing and overcoming an issue with the Super caps not allowing the EZB to fully boot, the original issue of random EZB WIFI disconnection continued.
#Conclusion: The Super caps did not fix my issue in my case. After much testing to try to nail down when exactly this happened I have come to believe that my issue surrounds the way I recorded the servo movements. By me recording very fast and erratic servo movements using the servo Recorder using the servo Pad I believe I was overdriving the servos and the quick back and forth of the DC motors may have caused voltage spicks to feed back to the EZB controller thus causing some of it's components to shut down.
#Proof: I have other animations using the same HD servos created by the Servo Recorder using the servo Pad that have slower, more fluid servo movements. I have never experienced any shutdown issues when playing back these recordings through ARC using EZ scripts.
#Solution: I removed the Supercaps and I scraped the three recordings that were causing the shutdown issue. I then used the AutoPosition control in ARC to recreate the wanted animation of the six HD EZ-Robot servo movements. In the EZ Script that I was using I removed the commands to playback the three servo recordings and replaced them with the new AutoPosition action command. After testing this change many times I have had zero shutdowns or issues.
#Remaining questions: What really caused the shutdown? Was it the voltage spick of the DC servo motors backfeeding into the EZB? Was it simply to many Servos drawing peek power too often? Since the issue happens both with a large battery amp hour supply and a AC to DC power converter I'm going with the former. I've found a solution and am moving on.
Unless the EZB shutdown issue pops up again I'm going to say all is well now based on what I outlined above. I am a little disappointed in the results of recording servo movements with the Servo Recorder using the servo Pad in regards to causing my issues. However I don't blame ARC and it's controls. I blame the limitations of the devices I'm using and basic natural cause and effect.
I would love to hear any thoughts on this matter about any of this from you all. Thanks so much for the assistance in my efforts to resolve this. You guys rock!