EEVblog #895 - BEC Pro Model Airplane Regulator Testing

[EEVblog]

Dave tests the Castle Creations CC BEC Pro battery eliminator used in model airplanes to see if it meets is rated continuous current performance claims.
http://www.castlecreations.com/products/ccbec.html
Also a bit of a how-to on testing regulators like this, 4 terminal load sensing error reduction is demonstrated.
Airflow and thermal camera measuremements are done.

Datasheets:
http://www.vishay.com/docs/88975/v20100s.pdf
http://cds.linear.com/docs/en/datasheet/3824fh.pdf

[piranha32]

A bit of airflow can indeed do wonders. I added tiny cooling fan to my Red Pitaya, and it dropped the temperature of the FPGA die by over 20C. Worked much better than I expected.
https://rroeng.blogspot.com/2014/03/keep-your-red-pitaya-cool.html

[Mr.B]

Well done Dave.
It would also be interesting to see how the heatsink performs with the airflow AND the original plastic case back on it.
I realize the fins are sticking out the top, but I would hazard a guess that the airflow would not be as good.

[FrankBuss]

Why do they call it a BEC? I know, it means "battery eliminator circuit", but this sounds silly to me, because it is just one application, like calling a general purpose switch "light on/off component", as if it can't be used to switch other things. Why not just "voltage regulator"?

BTW, is it true all model ESCs still use the old PWM servo signalling with analog 1 ms - 2 ms pulse widths? Someone should teach the manufacturers how modern digital buses like CAN work, would make wiring easier, too.

[Brumby]

It's just a name - and like any name, it's been coined to suit the target market.  My guess is that it was quite likely based on a term already in common usage.

Also trademarking 'battery eliminator circuit' would be impossible - but "BEC" would be easy.

[PeterL]

With a good synchronous regulator+design much higher efficiencies should be achievable.
See for example this:
http://cds.linear.com/docs/en/datasheet/3810fc.pdf
Linear claims around 95%, depending on the input voltage, and of course with a fixed output.

I wonder if this BEC uses a synchronous regulator, 2 mosfet's seem to suggest it.

The big diode is probably just for protecting against reversing the input.
 *edit
I just saw it uses the LTC3824. That's a bad design choice than if you ask me. To much power get's lost in the diode, because of the high Vf.

[TheSteve]

The "BEC" term originates with RC cars that used one battery pack to power the radio receiver/servos and a second battery pack to power the main drive motor. Adding a regulator "eliminated" one battery pack. I am pretty sure Futaba started using that term in the 80's as non tech people had no idea what a regulator was. Over the years the acronym has become common place in the RC industry.

[laneboysrc]

Why do they call it a BEC? I know, it means "battery eliminator circuit", but this sounds silly to me, because it is just one application, like calling a general purpose switch "light on/off component", as if it can't be used to switch other things.

In the early days of electric RC cars you had a main battery that ran the motor and a separate pack, usually 4 AA, to power the receiver and servo.


(main battery in the center of the chassis, receiver battery pack on the top of the image)

As technology progressed, RC receivers appeared on the market that included a voltage regulator (often a simple Zener and pass transistor), marketed as BEC, since it eliminated the need for the 4 AA.



Once electronic speed controllers became the norm, the BEC function moved into the speed controller. While a lot has changed, the term stuck for some reason.

cheers, Werner

[rollatorwieltje]

BTW, is it true all model ESCs still use the old PWM servo signalling with analog 1 ms - 2 ms pulse widths? Someone should teach the manufacturers how modern digital buses like CAN work, would make wiring easier, too.

They mostly still do, although modern speed controllers accept signals up to about 400Hz. They don't really care anymore about the low period, just the pulse between 1-2 ms.
There are some speed controllers with i2c interface, typically for multirotor use where update speed is a bit more important.

[Kilrah]

BTW, is it true all model ESCs still use the old PWM servo signalling with analog 1 ms - 2 ms pulse widths? Someone should teach the manufacturers how modern digital buses like CAN work, would make wiring easier, too.

For most common uses yes - There simply is no need for more, and completely moving away form a firmly established standard would cause more trouble than is worth. Most however do not require the usual 20ms period between pulses anymore and will thus accept faster refresh.

Some multicopter systems used I2C in the past, some closed commercial solutions like DJI stuff use CAN.

But for really high performance applications like racing multicopters any bus is too slow so these have recently introduced new standards based on the good old "pulse duration" principle again but adapted to the capabilities of today's fast uCs and their peripherals (Oneshot125, Oneshot42, Multishot) i.e. instead of having a 1000-2000us pulse at 50Hz refresh rate the pulse is (e.g. Multishot) only 5-25us long and allows for refresh rates up to 32Khz since the pulses can be repeated basically as fast as desired, no long dead time required. That's as fast as it gets and simple to handle given the large number of fast timer capture/compare modules in recent parts.
That field has also triggered a lot of innovation to improve speed regulation response speed of ESCs with active braking etc.

For linking the receiver to flight controllers SBUS is currently the main standard, i.e. a serial port sending packets of 16 10-bit proportional channels at 9ms refresh rate. There are servos and ESCs that read this natively too (that's what it was initially introduced for by Futaba) but it is only pretty rarely used for this purpose, the independent wiring of each servo/ESC to the receiver is still far more common in simple setups than the bus-style wiring. Sometimes both are mixed, e.g. SBUS from receiver to the wing in order to only have a single connector for all 4 or so servos that may be in the wing, then either SBUS servos or a decoder in there for standard servos, and the servos in the fusealge may be connected directly using the "old" method.

[daqq]

Setting up an automatic characterization setup would be a great tutorial - not set the values manually, but rather via a PC and connected test equips.

[FrankBuss]

For linking the receiver to flight controllers SBUS is currently the main standard, i.e. a serial port sending packets of 16 10-bit proportional channels at 9ms refresh rate. There are servos and ESCs that read this natively too (that's what it was initially introduced for by Futaba) but it is only pretty rarely used for this purpose, the independent wiring of each servo/ESC to the receiver is still far more common in simple setups than the bus-style wiring. Sometimes both are mixed, e.g. SBUS from receiver to the wing in order to only have a single connector for all 4 or so servos that may be in the wing, then either SBUS servos or a decoder in there for standard servos, and the servos in the fusealge may be connected directly using the "old" method.

Thanks, that's very interesting. Looks like the Futaba S-Bus is just a RS232 connection: https://developer.mbed.org/users/Digixx/notebook/futaba-s-bus-controlled-by-mbed/ At least it has a parity bit, but would be better to have a checksum. I guess in a plane with big motors there is a lot of noise. Two wrong bits and bad things could happen, especially with the digital channels.

I have a Phantom 3, it is all very closed source and proprietary (I wonder if they are using GPL software and so have to publish the rest their source code), but might be fun to play with the CAN bus.

[CJay]

It's just a name - and like any name, it's been coined to suit the target market.  My guess is that it was quite likely based on a term already in common usage.

Also trademarking 'battery eliminator circuit' would be impossible - but "BEC" would be easy.

Back when I bought my very first Tamiya RC car, the BEC was a feature that was important enough to be marked on the box, it had been around a few years then but prior the BEC features, the cars required a separate pack of four AA cells to run the receiver and servos so it's definitely a name that has been in common use for a few decades.

[Kilrah]

Thanks, that's very interesting. Looks like the Futaba S-Bus is just a RS232 connection: https://developer.mbed.org/users/Digixx/notebook/futaba-s-bus-controlled-by-mbed/ At least it has a parity bit, but would be better to have a checksum. I guess in a plane with big motors there is a lot of noise. Two wrong bits and bad things could happen, especially with the digital channels.

The probability of errors is very low (if you've got enough noise to disturb that you would also had disturbed the old PWM signal), and should one byte be corrupted it's going to be corrected 9ms later which is pretty much faster than anything mechanical behind can react anyway.
Big motors historically would cause interference on the RF side, not on the demodulated signal, but that's a thing that's been buried and forgotten since we have 2.4GHz digital RF.

Nowadays the main source of problems is actually power supply issues rather than signal problems, mostly because of the lack of awareness from people about how it works and how important it is and bad quality / lack of characterization of the components.
The thing is that when you build any standard setup there's pretty much always one of the elements that will have one of those "BEC circuits" (voltage regulators) built in so that things just work as soon as you plug everything together on the ground... but when you ask most people won't be able to tell you how or why it does, probably won't even know there's a voltage regulator somewhere, why it's there and what its limitations are so it could well not be up to the job once things start moving. It's only when nothing powers up because there is none that people start asking questions, and on the web they'll be quickly redirected by people who know "you need a BEC" but don't understand the parameters either to whatever crappy $3 Chinese regulator that may be noisy as hell and might or might not suit the setup's requirements (as someone already mentioned, what Dave describes as a "cheap voltage regulator" is actually among the most expensive ones you'll find on the market that don't have other functions, and it's comparatively very well characterized, usually on cheap units you only get no-load voltage and an overrated max current and just have to hope for the best for the rest).

DJI stuff is excellent for its plug and play aspects, awesome capabilities and seamless integration, but scores below zero for the DIY interest. Closed and proprietary, the whole system is extremely complex and made of so many components it's virtually impossible to figure out anything and each product is significally different technically so any findings you could do would be obsolete 6 months later.

[rx8pilot]

With a good synchronous regulator+design much higher efficiencies should be achievable.
See for example this:
http://cds.linear.com/docs/en/datasheet/3810fc.pdf
Linear claims around 95%, depending on the input voltage, and of course with a fixed output.

I wonder if this BEC uses a synchronous regulator, 2 mosfet's seem to suggest it.

The big diode is probably just for protecting against reversing the input.
 *edit
I just saw it uses the LTC3824. That's a bad design choice than if you ask me. To much power get's lost in the diode, because of the high Vf.

My synchronous designs are all >95%. I don't have any specific analysis of this particular model, but it is non-synchronous and seems to be using 2 FETs in parallel. Parallel FETs is not the best way to get more current since it causes other issues like lower efficiency from the additional gate capacitance. Most of what I design, can not have fans and only tiny heat sinks, so I have no choice but to fight for efficiency. ~75% to me means there is a poor layout and poor loop control. Even with the limitations of that controller - it should (or could) be substantially better.

[bigsky]

Dave, at the end of the video you said that you could obviously spend a lot more time characterising a converter module. You just seemed to be talking about steady-state conditions. But if you are ever in the mood, I would love to see something about the transient performance of a switching regulator like this. I'm sure it would be quite interesting to capture the output voltage on a scope at the point the load goes from 0 to 15A, and vice versa.

[rx8pilot]

Agreed, that would be fun to see as well as the gate drive.

Sent from my horrible mobile....

[FrankBuss]

The probability of errors is very low (if you've got enough noise to disturb that you would also had disturbed the old PWM signal), and should one byte be corrupted it's going to be corrected 9ms later which is pretty much faster than anything mechanical behind can react anyway.

But there are digital channels in the S-Bus protocol as well. One signal could be "turn off the motors", which triggers a flipflop, until another bit says "turn on the motors" and this needs to be there only for one cycle. I know, at least for the Phantom with the odd combination-stick-command (CSC) feature this is not a problem (if the CSC is evaluated at the receiver side), but could be a problem for similar things.

And you are right, looks like many hobbyists don't know much about even basic electronics. Maybe Dave should build a copter (recently I bought a used DIY racer with a 250 board for EUR 300, (cheap) FPV goggles with camera, sender receiver, batteries etc., and remote control (but of course not a Taranis) included) and do a fundamentals Friday about all the electrical things of it

[Kilrah]

But there are digital channels in the S-Bus protocol as well. One signal could be "turn off the motors", which triggers a flipflop, until another bit says "turn on the motors" and this needs to be there only for one cycle

Whoever did it that way would deserve a serious kick, you just don't do that. Just like for your CSC, what you do is check that bit (or channel, or set of channels) over a significant number of transmissions and only trigger your function if ALL of them give you the right command over a "long" period, around 1 second or >50 transmissions for CSC.
The standard RC link is not based on transmission integrity but on the principle that if a packet is lost or corrupted it's fine because its content would be obsolete by the time it was retransmitted, you're better off just forgetting it and sending/waiting for the next one. That means that the single packet where the bit was off may never be received... i.e. you can't ever rely on a single sample. Anything that takes data from those systems is aware of it and all is fine.

I doubt Dave would be able to just pick up a set of gear and make a video out of it - as much as modelers would need quite a bit of basic electronics training to understand how things work Dave would also need to learn the basics of R/C equipment to understand how things are and why, how everything interoperates (or doesn't) at a system level and with gear from 20 different manufacturers etc before he could do something really knowing what he's talking about... that's nearly a specialty in itself, it's really not always easy and seeing some of his videos he'd likely get pissed off and throw everything away pretty quickly 

[rollatorwieltje]

Dave did try to make a QuadCopter, a very long time ago:

[Towger]

The EEV canyon tracking quad copter must be the first started, but not finished project.

[rrinker]

 Been a while since I watched that video, but didn't the Quad belong to Dave's brother-in-law or something?

[FrankBuss]

The EEV canyon tracking quad copter must be the first started, but not finished project.

I guess most engineers have many unfinished non-client projects (including me, I'm a master in this art), because interests changed etc.

Good idea in the video how to find the short by measuring the resistance. Might work better with a better meter with kelvin connection. But I never had a shorted capacitor, only shorted tracks or pins on ICs, or burned ICs. Good to see that in the end it flew like 1 m, so not a complete fail

[Kilrah]

Hah I'll have to watch that...

I guess most engineers have many unfinished non-client projects (including me, I'm a master in this art)

Don't dare telling me about it. 

[Towger]

Don't dare telling me about it. 

[emoji57]