EEG Design - time for something totally different

So I've wanted to build an EEG since I was in college and have numerous times spec'd out parts etc but never really got going. I've followed the OpenEEG project for ages, but technology wise it has remained rather stagnant. However, I just recently came across the OpenBCI project which pointed me at the ADS1299 chip and that has reinvigorated my interest in building a simple EEG. It's a long shot, but my ultimate plan is to create a system like this:

One thing I wanted to change from the OpenBCI design was to use an STM32 processor, since that is what I'm more familiar with. My goal is to avoid a lot of the safety issues regarding powering from a computer by running the data via my phone. This also gives me free perks like wifi, screen, bluetooth, etc. I should be able to pretty easily use the Android GCS as a basis, especially since that already has a lot of the core code for handling data via USB (HID and Serial) as well as WiFi (great for running simulations).

It was pretty straightforward to base the design off Sparky which seemed like a pretty good starting point - both in terms of electrical and mechanical properties. I had to extend the board a bit to fit the ADS1299 on there without going to four layers (I'm cheap and would rather a bigger board for the prototype). This is the end result:

The final board is 50 x 35 mm. I can easily shave 5 mm off the right side, but figured I'd leave it this size in case the next revision has more channels and needs that space.  Currently it is single sided so lots of room for expansion. And it's nice because I can reuse a lot of the code from Sparky and just write a new PiOS driver that speaks to the ADS1299 chip. For laughs, I left the accel/gyro/mag chip in case I want to have some kind of head tracking. Either that or try and use it strapped to my arm like a Thalmic labs wrist band.

Here is the schematic. Sorry the layout is really ugly - I should redo the TPS60241 component so the power comes in on the left and goes out on the right.

Layout

The ground plane is split to keep the analog electronics and digital electronics as separate as possible. The top plane on the analog side is AVDD and the bottom is AVSS. This is where a four layer design would be a bit better but for now I'll save some money.

Top:

Bottom:

Features

• 8 Channels of single ended input, 1 driven bias electrode and one common reference electrode
• Small (36mm x 50mm)
• Utilizes ADS1299 chip
• Optional reference and bias electrodes
• Powered by USB. Small step up inverter generates a clean 5V after the input diode (probably not necessary if I never plan to use battery input). This can supply up to 25mA of current and the analog supply only requires 10.
• Single sided power supply design (so patient will be at 2.5V relative to the board)
• Solder pad allows using any electrode (routed out via SRB2) to serve as the negative input for the others. Alternatively use an additional reference electrode such as an ear clip.

Notes and quirks of the ADS1299

There were a number of things that were not obvious to me from a first pass through the datasheet. This is mostly a reference for me when I forget them after I came to a conclusion.

biasin - At first I assumed this input was something to do with computing the actual bias value. However, it is actually for situations where you want to reduce the number of electrodes going to the patient. In this case, you actually connect the biasin signal to the positive pin of a channel (i.e. electrode) which creates the path for the driven bias current. This will (obviously) introduce a small bias into the signal recorded on that channel, but this should be small and slow compared to the signals of interest.

srb1/2 -  Again, like with biasin, at first I assumed that both of these were meant to be treated as inputs. Either to provide a reference for the positive or negative inputs. However, (and again like biasin) this is to allow reducing the electrodes to the patient. In this configuration, you select an electrode (positive input) that will serve as the reference for others and route it out of SRB2. Then there is an external bridge from SRB2 to SRB1 which passes this signal back in and allows it to be the negative channel for all the other electrodes.

Input configuration

The ADS1299 has a pretty powerful analog multiplexer - but what is a tad counterintuitive or non-obvious is that channels can be mapped to inputs or outputs implicitly. The input mixer also gives lots of options. I'm going with what I think is the standard configuration and what seems most referenced in the EVM notes. This means the CHxSET register will be 000 for the mux "main" with SRB1 high. The positive values will all be routed out to the electrodes and the negative values will all go to SRB1.

Anticipated bias and reference

The full cable set will generally be used, at least at first. This means a reference electrode (e.g. earlobe) will come in to SRB1 and be connected to the negative input of all the channels. All of those channels will be averaged internally to compute the biasinv signal. The internal bias generation (2.5V) will be used for the positive input. The output (biasout) will then go to an electrode (via a protection resistor).

Thoughts

Inputs - OpenBCI used the negative inputs at SRB2 for their channels. I can't really understand why they did this (although it looks like it should work) but it won't be in "normal electrode" mode. 

I'm pretty tempted to switch a bipolar power supply like that in the EVM. The advantage of this is the patient is driven to zero volts relative to the digital electronics. This shouldn't ever be an issue provided everything is running from batteries but still is a potential safety improvement. It adds another chip though.

Isolation - one thing OpenBCI did quite nicely was solid isolation. This is important, especially if you want to make something commercial. I'm just goofing around and will try and only use it via phone or worst case when laptop is not powered by mains. To be honest, I just don't feel like putting all the components down for it.

Ground - Related to the the isolation issue, I hope I routed the ground appropriately. The ground plane is split - digital ground over the digital electronics and analog under the ADS1299. The digital grounds from the ADS go over to the digital ground plane and then are star grounded to the analog plane near the analog regulator. The AVSS and AVSS1 are both just connected locally to the ground plane rather than separately routed back to the star ground. The AVDD plane is similar.

Also one limitation of using my phone to collect the signals (provided processing power isn't the limit) is that I cannot simultaneously connect to a USB radio and control a quadcopter. However - if I get to the point that is the limit I'll happy redesign and use something like Freedom with more build in processing in the embedded board, or combine an RFM22b onto it.

Anyone have any comments on the schematic? I don't think I've got any glaring errors but you never know...

References

• http://www.ti.com/lit/ug/slau443/slau443.pdf
• http://www.ti.com/lit/an/sbaa188/sbaa188.pdf - good DRL notes
• http://www.ti.com/lit/gpn/ads1299
• http://www.openbci.com/technology-update/

Android Update

I pushed a pretty major update of the Android application to the Google Play Store today that I'm quite pleased with.  Felix Da Cat suggested using a Navigation Drawer to get around the app (he actually suggested it on the OP forums, which puzzles me since they don't have an android app). A year back I tried getting rid of the home page by using a spinner bar at the top and really wasn't happy with the results, but this worked much better.

So you swipe from the left hand side or touch the title bar and you can directly jump to the other pages. The back button should work fairly intuitively to the previous view. This also led to me converting more pages into fragments and really working on how they store state. The most useful of these changes was in the map, so that when you rotate the tablet or phone it will preserve your current zoom level and location. Also the zoom level is cached between runs so no more starting at an awkward way zoomed out distance.

Tablet mode got some aesthetic love. The armed/flight mode notifications were added to the PFD which cleared up some space. Also it didn't seem necessary to show the alarms (cropped) at the bottom of the page when you can just touch the alarm summary button. Getting rid of that junk made it easier to handle rotations in tablet mode too - which behaves reasonably for phones (at least on my N5) and tablets.

The browser was touched up. Editing and viewing both take place in the side view (which behaves well with rotation also) and the back stack will take you back to the previous UAVO you were viewing (but skips editing). This system seems to work well, although it is a bit tight in portrait mode on a phone.

Finally the most useful change (at least to me) is the map now shows a history of where the UAV was. This shows up as a small white trail. Currently rotating or changing screens will flush this which is unfortunate. Later I need to make this aspect of the logging happen at the level of the telemetry service so it can persist between views. But that is for another update. Unfortunately I don't have a screenshot from when I was outside so feel free to send me one :).

Easy logging with OpenLog

So one of my friends has wanted something light and easy that he can use to log data, and also Aqualuna on IRC has been wanting something similar. After seeing some of the logs that Ardupilot produces that my friend was getting with his Revo port I was really encouraged to just come up with something quick and easy that would just work.

And this was definitely easy. I think it took about 15 minutes from when I told Aqualuna I definitely wouldn't start working on this to having a working log file. I used an OpenLog and then connected it up to the flexiport on Sparky. Then I configured it to run at 57600 baud and first showed that it would work and collect data using the mavlink output. 

However, this has a limitation that there is no timestamping data (or handy parsing code) so I decided to adopt the uavorelay module to this purpose. Luckily from the Overosync module I'd added a native timestamped packet to uavtalk so I just modified it to send that type of message and periodically send AttitudeActual, Accels and PositionActual. Then I grabbed the file off the SDCard and ran it through the matlab LogConvert.m (from "make matlab"). Easy peasy, plot of the attitude of the board on my desk:

One of the nice things about this design is it is super flexible. For example, you could just have this attached to the line going to your telemetry system and have it redundantly log all the telemetry data in case you drop out (or don't always have it running) or you could even write a custom data format that is highly efficient to collect as much data as possible at a fixed rate. You can find the branch I'm using here. Hopefully at some point I'll tie it into our logging metadata so you can easily select which objects update at what rate. For the time being I'm just hardcoding it to a fixed rate for preselected objects.

For now I'll probably just make another module (don't want to break UAVORelay) that just allows streaming a few of the major data objects at a fixed rate.

If anyone wants to play with this, the code I used it located here.

Return to home and landing

So at the last Houston meetup at that Haymerchant I was talking with Oso Grande and he described the Naza failsafe sequence. Essentially hover in PH mode for a few seconds, then fly home, hover over home and finally land. It seemed like a good idea, but our current Vtol Path Follower didn't support any multi-step sequences like that. A day or two of thinking about it and a pretty clean implementation crystalized in my head.

Essentially for any "goal" the path follower might have, which range from simple ones like "hover here indefinitely" to the one described above, there would be an FSM that is switched in, which allows arbitrary levels of complexity. The basic description can be seen on GitHub.

I went ahead and implemented it this week and surprisingly it worked in simulation right away, so I got ambitious and went ahead and tested it at our Houston meetup this weekend. The results were pretty good. Also it was flown on Freedom, which has the benefit of providing high quality logs for offline analysis. I had one flight where it wasn't performing terribly well because I forgot to calibrate the mags (no fly away but some serious toiletbowling). However you can decide for yourself on the other flights:

And here is an image of the position data as it returned:

The location control was pretty good and the path home nice and straight. The altitude control wasn't great. It was meant to hold a minimum altitude of 15 m above home, which it tried but gradually dropped. However the altitude drop came from two issues:
1. the accels were biased so the estimate of vertical velocity was a bit off so when it was still it though it was already going up 0.5 m/s
2. the max vertical velocity was set at 1 m/s so it spend most of the time with this request pegged and not making much progress

First the good points. You can see there is less than 1-2 meters of error in the PH segments. However, you can also see how the altitude slowly droops down in the upper right panel, and the velocity is basically thinking it is going up (down at -1) 1 m/s squared so it thinks everything is peachy. After recalibrating the z-accel, this is the result:

Luckily we added a button the other week to facilitate this.

I also had the opportunity to see Naza PH and RTH, and I have to say I was impressed. It holds well and is nice and stable. It definitely has me questioning whether our dual loop controller for PH is the right design, as opposed to a directly PID control on position. We used to have an implementation of the later, so I'll probably try and resuscitate it and play this week. Also our altitude controller in general needs some love.

So a suggested change is that, like Naza, it also ascends in place when it is going to, before returning home. Also making the time it hovers in place programmable before RTH is important. Then add a simple geo fence and we should be able to get some users testing this. If we can make it robust it might actually be an option for failsafe in not too long!

TauLabs gets MultiWii Horizon mode

Some people were asking about Tau Labs getting support for a mode like MutiWii horizon. Credit goes to them for describing it and testing, as well as the MultiWii guys for the idea. In this mode as you move the stick it smoothly transitions from an attitude mode controller (in the middle region) to a rate controller (in the outer region). This means you can have the benefits of a pretty tightly locked self leveling mode with an aggressive rate mode.

Anyway, I whipped it up this morning and managed to get some tests in this evening. It's not identical to MultiWii but it's pretty similar. It felt really fun to fly and was neat having the best of both worlds. It was also fairly natural to fly.



This tricopter wasn't really tuned up so the rates are a bit slow and my loops had to be quite high. The leveling performance was good as always and it felt good in attitude mode regions of the stick range. The rate mode also felt good aside from not being fast enough.

I also tested it on a plane and had no issues there. Nice stabilization in the center and was able to do rolls and flips fine (although again it did not have a fast enough rate so they were fairly slow rolls). Hopefully you guys enjoy the few bloopers at the end. It's been a while since I did any flips.

(Updated) Here it is on Aggressor with Sparky doing some pretty nice tight flips

Horizon Mode - Flipping from James Cotton on Vimeo.

As an aside - we recently added an exponential to Tau Labs rate mode. This allows you to set a really high maximum rate (e.g. 500 or 600 deg/s if you set your gyro max rate high enough) while keeping good control in the middle region. Using this also works in horizon mode, so you can keep all your settings that you tuned independently in attitude and rate mode and it will just switch between them.

If you want to test it here is a version for OSX and here is a version for windows.  BTW there was a problem with the GCS on OSX 10.9 but that is fixed now. (Warning, the firmware in this doesn't flip properly and will do a split-S maneuver automatically. Use this updated firmware)

• Sparky
• CopterControl
• Quanton

If you want to test it here is a version for OSX and here is a version for windows.

Controlling a quad with a wave of your hand

So I just recently purchased a Leap Controller.  For those of you not familiar with it, this is an amazing device that can track your fingers and hands extremely precisely.  So of course, the first thing I wanted to do was use it to control a quadcopter!

The API is actually fairly straightforward to interface to C++ and you can access it here.  I wrote a plugin for Tau Labs GCS which would access the hand position and the roll-pitch-yaw value of the palm of hand. Right now I'm not doing anything with the position (other than checking a hand is there) and the RPY value is packaged into a LeapControl UAVO and relayed via the telemetry link to the FlightController.

On the flight controller side, I extended the ManualControl module to support another flight mode position (LeapControl). When the flight mode switch is in that position, then the LeapControl UAVO is used to set the desired attitude (yaw is used to set the turning rate, not absolute heading) and the throttle from the transmitter is used.  If a hand is missing it turns off the motors (a nice safety feature you'll see in the video).

The code changes can be found here although check the commit history since it requires some hardcoded paths to the Leap libraries.

The flight platform was my IconicX, which is beautifully light, has good flight times and only uses 8 inch props.  All really good things for indoor testing of a brand new and probably terribly advised control scheme.  It used bluetooth for telemetry to get the LeapControl information from GCS (more on that later).

The flight controller is the Sparky board that I designed for Tau Labs:

But of course, you really just want to see the video, so without further ado:

Controlling quad with a wave of the hand from James Cotton on Vimeo.

Conclusions

It's definitely flyable. Especially after a few minutes practice (I'd recommend a larger space).  A few things I need to change or would like to address in the future

• Increase the yaw sensitivity.  Right now the highest value is only a few degrees per second. However, it's a bit tricky because the Leap seems to zero yaw at the something random (related to what it first saw) and I'm finding "zero" seems a bit arbitrary.  Probably a dead band and some exponential on the GCS side would go a long way.
• Lag. This is the biggest blocker I think.  This quad is well tuned and reacts _really_ fast and so I can tell the limitations from hand controller. Of course the Leap itself has some latency, but I think something is going on in terms of the bluetooth serial port backing up data.  When I set the update rate too high there is a very clear backlog that occurs. Using a PipX might help since we have a bit more control about the lower level buffers.  Probably the serial code needs a goo review and for things like this we need a low latency protocol for sending the same object. I did something similar for the Tau Labs android app I wrote which made a huge difference for telemetry control (also bluetooth).
• Use hand position - I feel like tracking the position could allow a stronger communication of "oh shit" when getting near something.  Possibly if the hand moves rapidly then it applies a short lived roll signal to the quad to pull it away from the object but avoid the pilot induced oscillations from the latency.  And of course the Z position can also be used for ...
• Integrate this with the sonar based altitude hold (after some improvement) I described last week so that the hand height directly indicated the altitude.  That could create a really cool control scheme.
• Using overo on Freedom for optical flow then the X and Y position can actually indicate a rate to move at.  That would be really awesome.
• Try this outdoors.  I'm not sure how well the Leap will work there.

But it was fun.  I definitely won't give up my transmitter this week though.  If you are interested in this or our other developments, make sure to drop by Tau Labs and say hi.

Tau Labs Sonar altitude hold

So following up on Stac and scenkov's really nice work I added support for SMD-IO-UART sonar module which I had laying around and connected it to Sparky. This one is a bit different than the HCSR04 which encodes the range in the duration of a positive digital pulse. Instead, it had a negative pulse with a duration of 150µs when it finds an obstacle and the latency from the trigger to the pulse indicates the distance.  For more details check this out.

I mounted the module on the bottom of my silver hornet, which is a bit too big for the landing gear but works for now:

With the module working, I hacked up some code to pass that data into the altitude hold EKF.  I played around with the tuning a bit and can't get to the point where I'm super happy with it.  The next step will be trying a different module (e.g. I have an XL-MaxSonar sitting on my desk right now which people on IRC say good things about).  However it does work.

Sonar altitude hold from James Cotton on Vimeo.

One thing that is a problem with the current implementation is that if you go too high and the sonar goes out of range, then the EKF no longer has anything to correct the altitude. This means it goes into a straight prediction mode based on integrating the accels but depending on the direction of the bias this can make it think it is going down and correct by going up.

I suppose an alternative might be to keep correcting with the previous altitude (or max for that sonar) which should in any case be higher than what you engaged it at and cause it to come back down. A better approach would be to fill in that missing information with the baro.

If anyone wants to play with it, the code is currently at my Tau Labs github fork.