I've killed another maple board. Help me understand why.

LeafLabs Garden » Support » Maple Hardware Support

I've killed another maple board. Help me understand why.

(12 posts) (5 voices)

Started 2 years ago by amatic
Latest reply from amatic

amatic
Member

Hi,

I've posted a few weeks ago about my dead maple http://forums.leaflabs.com/topic.php?id=74153
Since I liked the board, I purchased an olimexino because I thought it's more resilient, but I've managed to brick that one too.

Here's the setup:

I have 5 small motors, stall about 450 mA, controlled by SN754410 motor drivers. Driver logic power supply is 5V, (coming from an old computer PSU), common ground with maple. Drivers register maple's 3.3 logic level without problems. I use one pwm pin and two logic pins per motor. Motors power lines are isolated using 1N4001 diodes. There is a 470uF capacitor between driver shield's motor power line and ground. Motor shield's ground and maple are connected trough gnd pin next to Vin.

Everything works fine when motors are on 5V, but I needed more power. So I connected 12V. It actually worked for two days, but then the maple stopped being responsive. I still get the "power on" led. It blinks when I reset the board. After about a minute, the chip starts to get hot.

I've read that maybe the grounds should be connected further away. Why is that, and how do I connect them, while keeping usb connection with the computer? Also, do I need to somehow protect logic outputs from maple? Is there a possibility that SN754410 fried logic ports?

I only have a basic multimeter for diagnostics.

Thanks.

Posted 2 years ago #
hihohoho
Member

Have you measure the stall current at 12V? It might be more than 450ma. Are they 12V motors? Is the 12V actually 12V and not 13-14V as some batteries can be? 450ma for a 12V motor seems low. My small 3V motors have a stall current of several amps, but that tends to lessen if it's driving a gearbox as they gearbox means it will never really stall without shredding gears.

Otherwise it might be a good idea to post a picture of your circuit, with labels if it's messy and more experienced guys on here might be able to spot something wrong.

Posted 2 years ago #
gbulmer
Moderator

amatic - as well as some pictures, would you also draw the circuit? Even a picture of a drawing on paper might help.

"I have 5 small motors, stall about 450 mA,..."
I assume you mean the total stall current of the "5 small motors" is 450mA, and not the stall current of each "small motor"?

Are you confident the 'back-EMF' diodes are connected correctly? Are any of them connected to the logic +5V?

Have you tested the motor drive 'stand-alone'?

Disconnect the olimeino completely, and wire the SN754410 inputs to ground or +5V so that the motors run. Use a resistor to pull each connection to ground or +5V, and check that the voltage on each input (pulled via the resistor) is very very close to ground or +5V with your multimeter (use around 1K for the resistors). You'll need to connect the inputs with bits of wire so that you can easily connect and disconnect the inputs. Have a look for spikes when you start or stop the motors or change direction. You might even try wiring LEDs across the inputs just to see if there are any rapid spikes (which your multimeter might miss).

(I have a very vague recollection of having a problem once with the L293D, which is a relative of the SN754410, but I can't remember anything other than the suspicion that something weird was happening if the logic-level power got disconnected, while the motor power was connected)

Does the olimexino still get hot after a minute when it is completely disconnected from everything except USB?

(Full disclosure: I am not a member of LeafLabs staff.)

Posted 2 years ago #
amatic
Member

Ok. Thank you guys for taking the time.
Here's a picture of the driver board: http://imgur.com/oa3hRB4
Here's a hand drawing of the schematics: http://imgur.com/9gnFfBi

I hope my circuit drawing skills aren't as bad as my phone's camera quality. The schematics show just one SN754410, but I left out the two for simplicity, since everything is the same for them. The condensators on 5V and 12V rails don't repeat for each motor driver.

Yes, the olimexino gets hot even if everything but usb is disconected.
The motors I'm using are "micro gear motors", like this one: https://www.sparkfun.com/products/8910

I did not measure the stall current at 12V. I guess it's more than 450 per motor. I'll check the stall current and input spikes and report back.

As you can see if you look at the first picture, I'm using an Arduino Mega at the moment. It worked fine at 12V, but for now I'll just keep everything at 5V.

Posted 2 years ago #
gbulmer
Moderator

amatic - Would you add the connections from the olimexino to the SN754410, and anything else, on your schematic?

A possible explanation of why the olimexino has been damaged, may be 5V has got connected to it somewhere. That may have been enough to damage the STM32F, yet leave the Arduino unscathed.

Maybe put resistors in series, between a Maple/Olimexino pin and the SN754410 (a few hundred Ohms), to limit current into the STM32F pins in the event of a short-circuit.

Motors generate quite a lot of electrical noise.
The effect can be reduced by putting a 10nF to 100nF capacitor across the motor's terminals.

(Full disclosure: I am not a member of LeafLabs staff.)

Posted 2 years ago #
AdriHD
Member

This is a great example why buffers are handy (and cheaper to blow up)... Keeping a definitive barrier (proper isolation) between your board and the circuit is important.

A general bi-directional one I've used a lot of in the past is: http://www.fairchildsemi.com/ds/74/74AC245.pdf

You can get tubes of 10 of them for about £5

For your use it's probably overkill having something with 8 ports, but there are others with 2 and 4 too. Also to add, there's a few buffers with adjustable curet limiting on them too, one of those might be handy as a final resort.

But I can guarantee that utilising buffers will be the end of your maple murdering career haha

Posted 2 years ago #
gbulmer
Moderator

AdriHD - Thank you very much for posting.

I think it is always an interesting balance between using a microcontroller appropriately and protecting it.

I added quite a significant amount of protection to the I/O pins on a STM32F board I designed several years ago, and most people then said I shouldn't have bothered :-)

I don't think it is simple. Adding extra electronics in the hope of making a microcontroller's I/O pins "fool-proof", can give a false sense of security. For example, if that buffer were connected to 5V, instead of 3.3V it could destroy eight non-5V tolerant pins in one fell swoop. Also, if there is a problem with noise triggering it, then its ability to drive 50mA into a MCU pin might be significantly worse than nothing.

A few hundred Ohms of resistance in series with a pin can be very effective, and a zener diode to stop voltages getting too far outside spec.

Anyway, IMHO a helpful addition to discussions.

(Full disclosure: I am not a member of LeafLabs staff.)

Posted 2 years ago #
spas
Member

Amatic,
You can try opto isolators to completely isolate the controller from the higher voltages. Then there is absolutely no chance of any 5v or back emf from motor circuit finding its way back and destroying your beloved micro. Here is one example http://uk.rs-online.com/web/p/optocouplers/4965428/
There are plenty to choose from.

NOTE. Optp-isolators provide electrical isolation between 2 systems. That said, you will NOT need to connect the grounds of each system together.

Another alternative is to use a level shifter. There are plenty of circuit diagrams on google, 2 transistors, a few resistors and Bobs your uncle.

I hope this helps.

Posted 2 years ago #
amatic
Member

Thank you for the suggestions. The Arduino is holding at 12V, so I guess it must be the SN754410 chip leaking 5V somewhere. I couldn't find where, though.

Optoisolators seem like a very safe solution. Are they ok with fast PWM? However, I have 15 inputs to the motor driver shield, 3 per motor, and I don't like the idea of having to isolate every single input.

Could diodes protect from power surges? I could put diodes from Arduino/Maple to motor driver inputs. I'm afraid if there is voltage drop from 3.3V it won't trigger logic on the 5V SN754410.

Or are there any cheap 3.3V motor drivers? Can't seem to locate any that can supply 12V and about 500 mA+ per channel. Maybe I could go with 9V.

Posted 1 year ago #
gbulmer
Moderator

amatic - "Optoisolators seem like a very safe solution. Are they ok with fast PWM?"

How fast is the PWM? Assuming you are using 16bit timers, the 72MHz clock, and libmaples default 16bit count range,
then the PWM frequency would be 72,000,000/65536 = 1.1KHz

That should go through an optocoupler/optoisolater fine.

How small a change to the PWM frequency will you need?

For example, this one (at about $0.05):
http://uk.farnell.com/everlight/el1112-ta-vg/optocoupler-tr-o-p-sop-5/dp/1852690
says the worst case rise or fall-time is 18uS, which is less than 2% of PWM frequency.

"Could diodes protect from power surges?"
Yes, in similar configurations to the "fly-back diodes" on motors, except these will connect to ground and 3.3V

" I could put diodes from Arduino/Maple to motor driver inputs. I'm afraid if there is voltage drop from 3.3V it won't trigger logic on the 5V SN754410."
Diodes don't go in series with the signal.
Instead the configuration used 'clamps' the signal voltage so it can't go more negative than ground, or more positive than 3.3V (accepting there is a forward voltage drop, which the chip can cope with).

Have a look at this article on diode voltage clamps

"Or are there any cheap 3.3V motor drivers? Can't seem to locate any that can supply 12V and about 500 mA+ per channel. Maybe I could go with 9V."
If you can use surface mount parts, Toshiba TB6612FNG?

Toshiba and Allegro make quite a lot of interesting motor drive devices, and there are others.

Edit:
A couple of things I should have written about protecting the signal lines

1. Put resistors in the signal lines, especially If you do use diodes to protect lines. Use a value around 220 ohms. It should not make any significant difference to signal voltage, but will limit current so smaller diodes can be used safely.

2. Maybe use Schottky diodes because the forward voltage drop is about half a standard diode.

3. If you can use surface mount devices, and space is at a premium, there are packages which contain both voltage clamp diodes, e.g. Farnell Digikey [Warning, these were just examples, I haven't homed-in on a good specification]

4. A 3.6V Zener diode between the signal and ground, combined with a resistor in the signal path to limit current, would block higher voltages.

It would help a lot if the cause of the failure could be found. Without tools to track the problem down, I'd be very defensive.
I might put protection on the power supply to the board, as well as the signals. Optoisolaters/optocouplers with logic-level outputs are relatively costly, but may be the only solution if the cause can not be found.

Posted 1 year ago #
amatic
Member

Sounds great. Thanks gbulmer!

Posted 1 year ago #
amatic
Member

My project is slowly going forward. Thanks again for the suggestions gbulmer. I ended up using the TB6612FNG motor drivers, found them on pololu on a breakout board. I also used a Teensy 3.1 board instead of the maple.

I made a robot arm that implements a cascade of control systems to position the end point of the robot arm instead of using inverse kinematics. It is based on perceptual control theory. Works nice.
Here is a bit more detailed description on the Teensy forums if anyone is interested:

http://forum.pjrc.com/threads/25659-PCT-robot-arm-alternative-to-inverse-kinematics

Posted 1 year ago #

RSS feed for this topic

Reply

You must log in to post.