Replies to This Discussion

Permalink Reply by Kyle on November 14, 2010 at 2:35pm

I had started to layout a circuit board to do just this but stopped when I realized it would have to be 4 layers and didn't want to front the cost for a super low volume board.

Think there is a big demand for this?

• ▶ Reply

Permalink Reply by Ivan on November 14, 2010 at 4:06pm

If it would be controlled through i2c instead of PWM, and cost around $50 for a kit, it would be demanded in thousands.

• ▶ Reply

Permalink Reply by Dean Lamborn on November 15, 2010 at 11:20am

It sounds good until you realize that if you blow one, you replace four.

Space becomes a problem. Instead of 4 ESC's that you can tuck wherever there is space, you have something 4 times larger.

There is no reduction in weight from the power section. You still have to get rid of the heat.

There would be a small weight reduction from the control side. I can't see how it offsets the other issues.

• ▶ Reply

Permalink Reply by Ivan on November 15, 2010 at 2:38pm

It sounds bad until you realise how much instability are we tolerating buy using PWM sub-par refresh rates. Keep in mind that using I2C we can:
- refresh desired motor speed several times faster
- read actual motor rotations, without supposing at which speed the motor is spinning

Last but not least, take a look at surgeries that turn ESC's from PWM to I2C, it is not a pretty sight, and it turns $10 ESC's into $30-$40 ESC's. Ridicolus. I don't think the weight and heat are real issues here, but I could be wrong.

• ▶ Reply

Permalink Reply by Kyle on November 15, 2010 at 7:08pm

Let me mention a few more things I was planning when I started my design:
1) You won't blow ESCs. Cheap ESCs with zero feedback and poor manufacturing "blow". My design has current sensing and temperature sensor per motor. Electronics don't spontaneously fail, instead they heat up from excess current and die. A more thought out design with i2c could sense a stressful condition, and an intelligent autopilot controller could throttle back the power to delivered to all motors allowing the user to figure out what happened and fix their problem. Or if a motor "sticks" and stops spinning as some have reported.

2) Heat, what causes heat? Cheap MOSFETs (and especially p-ch fets on the high side) with high rdson values. Instead, install better MOSFETs, this would be an enthusiast's board rather then a cheapo toy. What else causes heat? Slow and weak MOSFET drivers that slowly charge and discharge the gates of the MOSFETs to turn them on and off, resulting in crazy switching losses (--> more heat). Instead use an appropriate MOSFET driver. Now heat is not nearly as serious of an issue with proper component selection and proper design.

Furthermore, should you decide to build a heavy lift rig and want to run a lot of current through the drivers, I considered this with my prototype designs and placed all the FETs on the bottom of the board. This allows someone to heatsink the FETs to an aluminum baseplate and convect heat away on to a massive aluminum body (and possibly aluminum arms). You can use fancy carbon/fiberglass elsewhere, but use aluminum where it helps your cause.

All and all, I think heat is a non issue. And the less heat you generate, the more efficient the system and longer the flight time.

Furthermore, consider what Ivan said, I2C and potentially CAN (seems unreasonably expensive ~$2/motor for little to no payoff) would allow the auto pilot to command and query the motors. Tell motor 1 to spin at 2200 RPM, and it will rather then just to tell it to run at 33% (which means something different for each and every motor). The auto pilot can then query the motor drivers and see if they are reaching their desired RPM or not (in which case they are overloaded). Add in the response time to being able to change the duty cycle mid phase and you have tons of control the vast majority of the BLDC ESCs I've looked at don't have.

Oh yeah, and the real reason I like it is that now we have a reallllly clean quad copter with 2 battery wires going in to a power board, and wires coming out going straight to the motors. No ESCs and connectors flailing in the wind on the arms of a quad or tucked away somewhere slowly working their way out.

I'm itching to work on this again... buwahhaha I just convinced myself again that this is a good idea, anyone else interested? I'd be willing to design the PCBs, (share the schematics), print the boards and sell them as well as provide a full list of BOM that people can then order from Digikey/Mouser/Newark and assemble. And if there is enough demand I could investigate automated assembly. Software would be opensource (everything is better that way...) and it would run on STM32F100 microcontrollers. Form factor would be about 3.25x3.25" (designed to mount to a beagleboard)

Message me if you're interested, I'd like to get some feedback and see if I can get enough orders to make a first run and first attempt. :)

• ▶ Reply

Permalink Reply by tijuan aikens on November 15, 2010 at 8:35pm

I want first made for my project if no one places order pm me when ready

• ▶ Reply

Permalink Reply by Mauro Polgatti on June 14, 2012 at 10:11pm

Hi Kyle,

I've also had the same idea to design, build and market a PCB with integrated ESC, I agree especially the fact that it is better to rule the whole by I2C for obviously reason.

To be honest the PCB is just a part of our more bigger project to design, produce & realize a kind of "APM2 (clearly based on Arduino)" with an also integrated, ARM gnu/linux.

Our team consist of three people and we are trying to form a team of 6-7 minds, each with its own role, and bring our product to the market. 

P.s.= we have already the possibility of prototype printed circuit boards up to 16 layers available in 48 hours....

If you are interested in join us pls send me a pvt message

Mauro

• ▶ Reply

Permalink Reply by Brad Hughey on June 15, 2012 at 8:34am

Mauro, you realize this thread is over a year old?  But I feel compelled to offer an opinion on ESC design, considering I've blown up or fried more MOSFETs than practically everyone on DIYD combined.

Kyle is correct that it's not heat that usually kills MOSFETs - look at how many companies get away with shrink-wrapping the heat sinks on ESCs.  If you turn the MOSFETs on and off fast enough, and have an adequately fast flyback diode (or MOSFET used for that purpose), the heat is relatively minor, considering most have Rds' in the 5 milliohm range.  Want less heat?  Parallel a few more cheap ones.  That obviously lowers the apparent Rds as well as the Rth.  Heck, in some military high-current designs there are a dozen or more MOSFETs in parallel for this very reason.

What you need more than a big heat sink is a big line capacitance.  Exceeding the Vds avalanche voltage rating is what kills most MOSFETs, and the failure mode obliterates the gate channel control, latching the drain-to-source impedance to a minimum permanently (or at least until the silicon vaporizes).  Magically Obliterating, Smoke and Fire-Emitting Transistors they are! 

Controlling how much line inductance the MOSFET is subject to is nearly impossible as a hobbyist component part, so in practice there is no such thing as a mitigation capacitor that's too large.  Also, using the MOSFET itself to dissipate inductive field collapse surge energy is a good practice, with fast back-to-back zeners rated well below the avalanche voltage from the drain to gate (assuming low-side control with N-channel FETs).

Sorry, I don't wish to join you as I have "much larger fish to fry" right now, but I'd hate to see a group of dedicated enthusiasts set their workbench on fire.  :-)

• ▶ Reply

Permalink Reply by Mauro Polgatti on June 18, 2012 at 1:14am

Thank you for the answer and relative suggestion,

I Knewn the post was old, anyway I was hope to receive a reply from Kyle.

we are serious about building the board and do it with only the best components and quality design)... hopefully kyle will answer.

sorry for the question but now that you've put a flea in my ear, I must ask you about your "much larger fish to fry " ....

Do you have any sketch / drawing / component list or something like that about PCB ??

any help is really appreciated and at the end our project will be clearly public.....

Thank you Mauro

 

• ▶ Reply

Permalink Reply by Brad Hughey on June 18, 2012 at 8:31am

Here's a good primer on the use of MOSFETs in motor control.  The examples given are for brushed motors, but the application practices are the same (just X3).  I assume the gate drive issues are handled by the uP:

http://www.4qdtec.com/pwm-01.html

Which is to say, there are TTL-level gate MOSFETs now, which should allow you to forgo the traditional driver stage.  Here's one that isn't cheap, but is representative of the type:

IRLB4030

Here's my "big fish":

www.hecopter.com

• ▶ Reply

Permalink Reply by tijuan aikens on November 15, 2010 at 5:52pm

let me add a little more gas to some of the suggestions i have been reading
to solve the overheat issues can it be a liquid cooled?(or in some way work best at high attitude)
or pre programed to work on the motor its given?(maybe on a usb drive and kill the little turning knobs)! (and run the motor you may use now)
or to adjust the speed on a custom made motor?(for more speed build a multistage motor)
and to run at half the speed of a backup source with a low cost for turnaround ?

• ▶ Reply