Thread Tools
May 06, 2016, 11:57 AM
Registered User
Thread OP
New Product

BrainFPV RE1: Next gen racing flight controller with OSD, Lap timing, LEDs & more


My company, BrainFPV, just announced two new products: The BrainFPV RE1 (Racing Edition 1) flight controller and its companion product the Mini Power Board (mPB V2). They stack seamlessly together without any wiring between the two, so you can build your own cutting edge racing drone.



Some of the highlights of RE1 include: Built-in graphical OSD, with advanced features like side-by-side 3D support and a menu, faster processor (STM32 F4, 180MHz), lap timing support (Trackmate, I-Lap, configurable using OSD menu), programmable LED support (up to 1024 LEDs, also OSD configurable), inverters for serial receiver and telemetry, stacking connector for future expansion boards, and more. Check out the product page for full technical details. [link]

RE1 runs dRonin, which is fully open source (GPL licensed) and does everything, including rendering the OSD. dRonin includes an advanced auto-tune feature that allows you to easily tune your PID gains for super stable flight.

We are super excited about RE1 and pre-orders will start soon. For a behind the scenes peek of the development of RE1 check out our blog post.

RE1 is an evolution of Brain, and we worked hard to make it both better and more affordable. The MRSP is $79 and it is assembled in the United States using a 6 layer PCB for optimal signal integrity. Sign up for our newsletter to get updates on the pre-order information.

Anyways, I hope you like it and feel free to ask questions .

Martin

Updates
5/16/2017: mPB V2 is now available. It brings 6S compatibility, lower stacking height (5mm instead of 8mm) and better voltage regulators
11/17/2017: Use the Logo Customizer to make custom splash screen and pilot logos for your RE1!

Reviews Etc
Last edited by HeliShredder; Nov 17, 2017 at 10:29 AM.
Sign up now
to remove ads between posts
May 06, 2016, 01:01 PM
Registered User
This is a really cool concept, made even cooler with the behind-the-scenes look into its development.

Glad to see that you're contributing directly to the dRonin code base.

You seem willing to share... now the next step is to corner the open source hardware market :P



I'll be the first to ask -- what are the LEDs sticking off the sides for? Infared for transponders or something?
May 06, 2016, 01:11 PM
Registered User
Thread OP
Quote:
Originally Posted by strg
This is a really cool concept, made even cooler with the behind-the-scenes look into its development.

Glad to see that you're contributing directly to the dRonin code base.

You seem willing to share... now the next step is to corner the open source hardware market :P

I'll be the first to ask -- what are the LEDs sticking off the sides for? Infared for transponders or something?
Yes, the LEDs on the side are for infrared based lap timing. They are included unsoldered and be attached by the user. They can also be mounted at a different location on the quad using wires (useful when IR receivers are on top of the gate).
May 07, 2016, 06:12 AM
Registered User
Hi
Very interested. Does the OSD work with
BlackBird 2 camera?

Did you test that?

Andy
May 07, 2016, 06:14 AM
Registered User
Thread OP
Quote:
Originally Posted by Morphin
Hi
Very interested. Does the OSD work with
BlackBird 2 camera?

Did you test that?

Andy
We only tested it with the NerdCam Mk.2, but the BlackBird 2 supports side-by-side 3D, so I don't see a reason why it wouldn't work.
May 07, 2016, 07:16 AM
Registered User
Maxzor's Avatar
While I loved the concept of produced in the US for the first Brain board, this F4 board still lacks important features!

There is no SD slot when Blackbox programs are so useful. The flash memory is something, but long flight times will need higher space.
Why put an IMU produced by a household electric company, with maximum gyro sampling at 6.4 kHz, when Invensense already features 8kHz data ouput rate? Raceflight features 8kHz gyro sampling/FC loop/multishot ESC.

Looking forward to hear about this board.
May 07, 2016, 07:30 AM
Registered User
Thread OP
Quote:
Originally Posted by Maxzor
While I loved the concept of produced in the US for the first Brain board, this F4 board still lacks important features!

There is no SD slot when Blackbox programs are so useful. The flash memory is something, but long flight times will need higher space.
Why put an IMU produced by a household electric company, with maximum gyro sampling at 6.4 kHz, when Invensense already features 8kHz data ouput rate? Raceflight features 8kHz gyro sampling/FC loop/multishot ESC.

Looking forward to hear about this board.
Thanks .

Regarding the sampling rate, 1.6kHz is actually plenty fast. The signal of interest on a typical racing multicopter is below 100Hz, so in theory you don't have to sample much faster than 200Hz. However, increasing the sampling rates helps with reducing latency, but there are diminishing returns, i.e., going from 500Hz to 1kHz helps quite a bit, 1kHz to 1.6kHz helps a bit, anything higher becomes questionable in terms of latency benefits. One reason why people run 2kHz/4kHz/8kHz on betaflight etc. is that the Invensense sensors require you to to disable internal anti aliasing filters when going above 1kHz, so you want to run at 4kHz/8kHz to make sure you don't get aliasing. The BMI160 on the other hand can have filters enabled (523Hz bandwidth) when running at 1.6kHz and going higher would only increase computational requirements with very little benefit.

RE1 has 128Mbit (16MB) flash built-in, which should be enough to log one flight fast and longer flights at reduced rates. However, SD card logging is something we will likely adding as an accessory in the future (this is why we added an extension port). Note that all current flight controllers that have an SD card slot have very limited logging rates due to memory limitations. The dRonin project is working on a new logger, openlager, that uses an STM32F411 processor and will actually support high rate logging. It will be a great combination with RE1 for people who like to log a lot.
May 09, 2016, 05:53 PM
Registered User
Hi Martin,

Sorry, a little off subject but the support pages for the previous iteration of the Brainfpv appear to be down. I've sold one of mine but the buyer can't access the set up information? Will they be back up again soon?

Thanks,

Peter.
May 10, 2016, 06:01 AM
Registered User
Thread OP
Quote:
Originally Posted by Peterev
Hi Martin,

Sorry, a little off subject but the support pages for the previous iteration of the Brainfpv appear to be down. I've sold one of mine but the buyer can't access the set up information? Will they be back up again soon?

Thanks,

Peter.
Sorry for that; you can find the support page for Brain here until the content has been migrated to the new site.
May 10, 2016, 09:45 AM
250FPV n00b
I like! I look forward to hearing from some people who've flown with it in various rigs to see what their feedback is, though that may be a while.

Do you have a projected release date for the first batch yet?
May 10, 2016, 09:53 AM
Registered User
Calling Bosch an household electric company is ignorance at its finest. They produce sensors for automotive and aerospace industries for ages.

--edit: a word.
Last edited by Glowtape; May 10, 2016 at 10:00 AM.
May 10, 2016, 11:09 AM
Registered User
Quote:
Originally Posted by HeliShredder
Regarding the sampling rate, 1.6kHz is actually plenty fast. The signal of interest on a typical racing multicopter is below 100Hz, so in theory you don't have to sample much faster than 200Hz.
I think that this is not correct.

The signal of interest on a racing multicopter is below 100 Hz. That's correct. In fact, the signal of interest is pretty much below 20 Hz. But that doesn't mean that there is no advantage to sampling faster. For example, there is a BIG peak in signal energy around 200-300 Hz, which is the prop frequency. If you are not sampling at at least twice THAT rate, you will get aliasing. So let's say you are sampling at 200 Hz as you suggest. If the prop frequency is at 220 Hz, you will get a big aliasing artifact at 20 Hz. Oops.
May 10, 2016, 11:11 AM
Registered User
In this video, I talk about what I think the minimum sampling rate for avoiding aliasing on a mini quad should be. Skip to about 7:00 if you already know what aliasing is and how that works.

Betaflight 2 kHz (and above) gyro sync explained (21 min 21 sec)
May 10, 2016, 01:32 PM
Registered User
Thread OP
In your video, you state "that everything above 25Hz or so up 150Hz is just noise and vibrations that the PID loop still needs to be aware of". This is simply incorrect. Usually, there is a digital low-pass filter in the flight controller with a cutoff frequency of about 50Hz to partially remove said noise. This is especially important for the D term, as the derivative operation will amplify high frequency (i.e. noise) components.

When I said that 200Hz sampling would be enough, I assumed that there would be an anti-aliasing filter present, so aliasing wouldn't be a major concern. Without a filter, you definitely want to sample faster, as you said. In the Invensense parts, the filer is disabled above 1kHz, this is why you would want to run it at 2kHz or higher.

In terms of latency, there is no real benefit of going much higher than 1.6kHz, as the overall latency of the control loop is about 10ms (time from command to motors until the gyroscopes detect a motion). Going from 2kHz to 4kHz makes a difference of maybe 250us, or 2.5% of the overall latency, which won't be noticeable in terms of flight performance differences.
May 10, 2016, 01:44 PM
Registered User
I'm just splitting hairs here, but hey, splitting hairs is fun so

Quote:
Originally Posted by HeliShredder
In your video, you state "that everything above 25Hz or so up 150Hz is just noise and vibrations that the PID loop still needs to be aware of". This is simply incorrect. Usually, there is a digital low-pass filter in the flight controller with a cutoff frequency of about 50Hz to partially remove said noise. This is especially important for the D term, as the derivative operation will amplify high frequency (i.e. noise) components.
Even with the low pass filter, the noise is still present and relevant. Assuming one was using a 1st-order lowpass with 6 dB / octave, and a cutoff of 50 Hz, you would be 3 dB down at 50 Hz, 9 dB down at 100 Hz, and 12 dB down at 150 Hz, adn 15 dB down at 300 Hz. I submit that the noise is still relevant to flight performance even at 15 dB reduction.

What I meant by "still needs to be aware of" is that, if you're not sampling that data, you're going to experience aliasing with potentially negative effects. If we had a perfect filter with a 60 dB reduction above the cutoff and zero phase delay, and if aliasing was not a factor, then we could probably ignore everything above 50 Hz entirely. Since aliasing is a factor, and since our filter is not perfect, my argument is that we need to raise the Nyquist Limit above at least about 300-500 Hz for best flight performance.

Quote:
When I said that 200Hz sampling would be enough, I assumed that there would be an anti-aliasing filter present, so aliasing wouldn't be a major concern.
I'm just not sure how effective the aliasing filter actually is. It's surely better than nothing, but given the choice between sampling with Nyquist = 200 Hz and an aliasing filter, and Nyquist = 500 Hz with no aliasing filter, isn't the latter better?

Quote:
Without a filter, you definitely want to sample faster, as you said. In the Invensense parts, the filer is disabled above 1kHz, this is why you would want to run it at 2kHz or higher.
Agree.

Quote:
In terms of latency, there is no real benefit of going much higher than 1.6kHz, as the overall latency of the control loop is about 10ms (time from command to motors until the gyroscopes detect a motion). Going from 2kHz to 4kHz makes a difference of maybe 250us, or 2.5% of the overall latency, which won't be noticeable in terms of flight performance differences.
I agree fully there. The advantage of going to 1 kHz and up sampling is not related to latency. Primarily, the advantage of going to at least 1 kHz is that you increase the Nyquist Limit above the prop frequency.


Quick Reply
Message:

Thread Tools