To M8N or not

[User Name]

Long story: (to skip, jump below to short story)

So I'm very much an "if it ain't not broke, don't fix it" type, and while I've never had a single issue with Solo's GPS, I find myself contemplating replacing it with an M8N. Here's why: When I first got the Solo, I wouldn't take off unless I had at least 8-9 satellites. I can fly just fine in stabilize/manual, but given the cost of the Solo, I wanted the added safety of a solid GPS lock, and I was happy to wait as long as it took to acquire one.

But before long, there was inevitably times when no amount of waiting was going to result in more than 5 or 6 satellites, and I cautiously got comfortable taking off with that few, even though I viewed this as exactly the type of letting down your guard that'll end up biting you in the rear sooner or later. Again, I'm fine flying in manual, I'm not afraid of doing so, but a good part of the attraction to a "smart" drone is the "smarts", which are all pretty much dependent on a solid GPS lock. So getting used to flying without a dependable GPS lock is somewhat defeating the whole point of having spent so much on a "smart" drone.

So, I'm planning to build another camera quad and ordered an M8N GPS from Drotek for it, and got the idea to test it against some others - the stock Solo 7N and a M8N from RTFQ. I hooked them each to a PX4 flight controller, placed them all in the exact same location, and tested them all within the span of an hour, so the conditions were pretty much identical.

Here's the results:



Solo's 7N never saw more than 4 satellites and the lowest hdop recorded was 1.98. Looking at the map in Mission Planner, the drift wasn't too bad, it reliably stayed within 20 feet over the course of about 15 minutes. Most worrying though was that if anything got anywhere near the GPS, it'd drop all satellites immediately. I couldn't even slightly preposition it, holding it only from the sides, not covering the ceramic antenna, without it loosing a fix entirely.

Next up was a Mini M8N from RTFQ. It quickly got 8 to 9 satellites in the exact same spot the 7N couldn't get more than 4, and before long it was bouncing between 13 to 15 with an HDOP of .71. Looking at the map, it did drift more than the Solo's, about 25 feet over 15 minutes. I could touch/preposition the GPS and it might drop a satellite or two, but still had plenty in reserve. Never dropped a lock.

Last was the M8N from Drotek. Like the RTFQ, it was quickly up to 8 to 9 satellites, saw 16 at max but spent most of the time with 13 to 15 and a had an HDOP of .69 at it's lowest, but averaged around .75. Like the RTFQ, I could touch/move the GPS without loosing more than 1-2 satellites and never lost a lock. Over the 15 minutes, it drifted less than 15 feet.

I've read that the number of satellites and HDOP aren't the only determiners of position data quality, and this would seem to bear that out. The RTFQ M8N had way more satellites and way lower HDOP, but drifted more than the Solo's 7N. It's lock was more resilient, but drift is drift. The Drotek though seems to offer both a far more resilient lock and lower drift. I know people are putting the RTFQ GPS into Solos, but I'm just not comfortable trusting $1500+ of quad and camera to a bargain basement GPS. But the Droteks have a much better reputation for quality, and I'm seriously considering swapping it for the Solo's 7N.

Short Story:

I tested the Solo's stock GPS against M8Ns from RTFQ and Drotek and found the RTFQ unit to be better in some ways, worse in others, but the Drotek seemed to be better all around. Why would I not want to replace my Solo's GPS with this? Are there any other testing criteria that I could use to determine if the Drotek would ultimately be a worst performer? Just saying "It's not all about satellite numbers and HDOP" is as dubious as saying it is - if it's not, then what does determine the best performance and how can I test for that?

Wrapping it up, as I said in the beginning, I've never had a problem with Solo's GPS, other than the low average number of satellites it picks up, it's always worked fine. But this year, I'm planning to film some rock climbers, which means I'll be flying in close to a solid rock face, which means a good portion of the sly will be entirely blocked. Given the low number of satellites Solo's GPS typically sees, blocking so much of the sky is a pretty good way to end up with a dropped GPS fix, hence my interest in the M8N to get a more resilient signal.

 

[Jubalr]

Let me start by saying I have not had a problem with my Solo GPS. I got my Solo when the first shipment landed and so I have an early unit. I generally have always got sat lock in the first minute, sometimes 2. Sat count is usually 10-12 while flying. But after all of this talk and seeing user rave about the M8n I installed 1 about 2 weeks ago. Yesterday I swapped it back to my stock unit. While the sat count was usually 18 or more, I did NOT get any faster locks as far as Solo telling me it was ready to fly. Sometimes it would show 12 sats before it would tell me ready. As has been explained long ago, Solo is not just looking for # of sats or HDOP. There are timing issues that Solo looks for as well. It was also not any better at holding position and actually holds better on the stock GPS. So after trying the M8n I am now back to the stock Solo GPS and have been quite happy with it. I am probably the only one to switch back.
YMMV

 

[User Name]

Curious what brand/model M8N did you use?

I too get upwards of 10 to 12 while flying, but it's the not getting more than 6 or 7 at best on the ground and how easily the fix is lost that bothers me. When I land to change batteries, as soon as I reach for the Solo, the controller indicates that it's lost GPS. If it's that easy to loose on the ground, as I said, if I'm flying close to a cliff face, it's probably going to loose the fix there as well.

I might give the Drotek a try, perfectly willing to go back like you did if there's no improvement.

 

[Jubalr]

Curious what brand/model M8N did you use?

I too get upwards of 10 to 12 while flying, but it's the not getting more than 6 or 7 at best on the ground and how easily the fix is lost that bothers me. When I land to change batteries, as soon as I reach for the Solo, the controller indicates that it's lost GPS. If it's that easy to loose on the ground, as I said, if I'm flying close to a cliff face, it's probably going to loose the fix there as well.

I might give the Drotek a try, perfectly willing to go back like you did if there's no improvement.

It is this one.. PM Sent..

 

[Ketch]

Let me start by saying I have not had a problem with my Solo GPS. I got my Solo when the first shipment landed and so I have an early unit. I generally have always got sat lock in the first minute, sometimes 2. Sat count is usually 10-12 while flying. But after all of this talk and seeing user rave about the M8n I installed 1 about 2 weeks ago. Yesterday I swapped it back to my stock unit. While the sat count was usually 18 or more, I did NOT get any faster locks as far as Solo telling me it was ready to fly. Sometimes it would show 12 sats before it would tell me ready. As has been explained long ago, Solo is not just looking for # of sats or HDOP. There are timing issues that Solo looks for as well. It was also not any better at holding position and actually holds better on the stock GPS. So after trying the M8n I am now back to the stock Solo GPS and have been quite happy with it. I am probably the only one to switch back.
YMMV

I have a thought. As a professional land surveyor I use RTK GPS with a base station and rover with a average measurement accuracy of 1/2" in a mile. This system does not depend on only the number of satellites, but the geometry of the satellites (strong or weak triangles). I wonder if solo has that feature built into it. I have had good GPS hover stabilization with my solo with fewer satellites, much better than my DGI. I have had good locks with the RTK with only seven, and no lock with 13 because of the geometry. Just a thought.

 

[Ketch]

I have a thought. As a professional land surveyor I use RTK GPS with a base station and rover with a average measurement accuracy of 1/2" in a mile. This system does not depend on only the number of satellites, but the geometry of the satellites (strong or weak triangles). I wonder if solo has that feature built into it. I have had good GPS hover stabilization with my solo with fewer satellites, much better than my DGI. I have had good locks with the RTK with only seven, and no lock with 13 because of the geometry. Just a thought.

What I am trying to say is that perhaps Solo has better "locking" software (geometrical computation ability) than others.

 

[User Name]

What I am trying to say is that perhaps Solo has better "locking" software (geometrical computation ability) than others.

It may very well, but as you say, a lot depends on the geometry of the satellites it's receiving. To that end, the more satellites it can pick up, the better the position fix, and this is the strength of the M8N. It would be handy to know exactly what it is about the 7N that makes it better, or in what way the M8N is supposedly worse. Is it something inherent to all M8Ns, or just "cheap" ones. As I said, I wouldn't trust my Solo to a cheap $30 M8N from RTFQ, but my interest is in whether or not a quality part will perform as good/better than the stock.

In other words, all else being equal, I don't see how more satellites could be anything but better. What's in question is specifically what's supposedly keeping any M8N from being equal to the Solo's 7N.

 

[Maddog]

1) If your going to be flying close to any obstruction, like a cliff, I would only fly in MANUAL. If you're using any GPS-assisted mode, a glitch, loss of signal, etc., could cause Solo to fly into the obstruction before you could react and take over.

2) in conducting your tests, are you saying you had Solo hovering in FLY mode (no pilot input) for 15 minutes and it wandered around about 20' from its starting point? (I hope I'm confused).

I have the RTFQ M8N installed. I have had Solo hovering for almost that long and have never seen it wander more than 2' from its' starting hover position.

 

[User Name]

2) in conducting your tests, are you saying you had Solo hovering in FLY mode (no pilot input) for 15 minutes and it wandered around about 20' from its starting point? (I hope I'm confused).

Oh, no, my tests were just the various GPSs attached to a PX4 flight controller and monitored through Mission Planner. It was just to see, with a 3D position fix, how much the determined GPS location drifted over time. Now Solo uses more than this to maintain a position, but it's a good test of the pure GPS performance. I also subsequently tested an XL Drotek M8N (it's on a 50mm x 50mm board with a larger patch antenna) and I think the drift over 15 minutes never exceeded 8 feet.

I have the RTFQ M8N installed. I have had Solo hovering for almost that long and have never seen it wander more than 2' from its' starting hover position.

How does that compare, if you recall, to the stock Solo GPS? And what motivated you to switch, if I might ask. For me, I'd just like the added assurance that a GPS position fix isn't 1 or 2 satellites away from being lost, and any M8N I've used on other quads regularly gets upwards of 2x the number of satellites that the Solo gets.

Of course, all this being said, I've never had a single GPS glitch with my stock Solo, which is probably why I'm so on the fence when it comes to changing it.

 

[Ian [P13]]

I wouldn't rush into buying an M8N, especially avoid the cheap and nasty ones.

There's some really high quality GPS units coming out for Pixhawk 2, I'm expecting we'll see these at NAB, and Philip Rowse who's behind the designs hasn't ruled out making one for the Solo, or he might just be saying that to stop me hassling him.

 

[pyrate]

agreed, waiting for a nod from the right people. Until then this one is to inconsistent

 

[User Name]

I wouldn't rush into buying an M8N, especially avoid the cheap and nasty ones.

There's some really high quality GPS units coming out for Pixhawk 2, I'm expecting we'll see these at NAB, and Philip Rowse who's behind the designs hasn't ruled out making one for the Solo, or he might just be saying that to stop me hassling him.

Yea, I bought the Drotek units for other multirotors, so no loss if I decide not to replace the Solo's. I'm waiting for April's firmware update, which CA said will include some improvements to GPS lock times.

I've been following the Pixhawk 2 information... definitely interested to see what it offers. I occasionally consider trying out other flight controllers, but always come back to APM/PX4 as the best choice for the kinds of 'copters I'm interested in building... in other words more than just racing drones.

 

[Ian [P13]]

Chris Anderson I've noticed on Twitter in the past gets a bit over excited about GPS firmware tweaks, there's only so much that can be done in the firmware for the GPS itself and I'm pretty sure there's not much more that can be done now.

However, I've seen it posted somewhere that the EKF2 arms faster, so whilst not truly being a GPS fix; we may well find the Solo is ready to fly faster than it was from a future update.

 

[pyrate]

Not sure how I feel about that.
I never minded waiting the brief time it takes mine to arm. and when mine takes a long time to arm I prefer to move locations or wait.
There is a reason Solo takes longer to arm, and why you do not hear the fly away issues with solo that you do some of those speedy arming ones

 

[Ian [P13]]

Not sure how I feel about that.
I never minded waiting the brief time it takes mine to arm. and when mine takes a long time to arm I prefer to move locations or wait.
There is a reason Solo takes longer to arm, and why you do not hear the fly away issues with solo that you do some of those speedy arming ones

I agree with you on that, however from reading about the new EKF I'm reassured.

Basically the new EKF is for want of a better word smarter, so it won't be cutting any corners if a lock is faster, it's just the logic is more precise.

Although the Pixhawk hardware has redundancy for example there's more than one IMU, the current EKF in the event of say an IMU failure won't use one of the working IMUs in its place where the EKF 2 is far more likely to be able to do so. So overall you'll be getting a much safer and robust Solo from the new code. If a magnetometer fails in the current EKF Solo will switch to fly manual, in EKF 2, it'll use one of the remaining working magnetometers instead.

The improvements in summary are:

• It can run a separate EKF2 for each IMU making recovery from an IMU fault much more likely
• It can switch between magnetometers if there is fault
• It can estimate gyro scale factors which can improve accuracy during high rate manoeuvres
• It can simultaneously estimate both gyro offsets and orientation on startup whilst moving and doesn’t rely on the DCM algorithm for its initial orientation. This makes it ideal for flying from moving platforms when the gyro has not been calibrated.
• It can handle larger gyro bias changes in flight
• It is able to recover faster from bad sensor data
• It provides a slightly smoother output.
• It is sightly more accurate
• It uses slightly less computing power
• It starts using GPS when checks pass rather than waiting for the vehicle motors to arm.

You can read the details here: EKF2 Navigation System — Dev documentation

The GPS prearm check is every bit as robust, in fact it looks moreso:

0: The receivers reported number of satellites must be >= 6

1: The receivers reported HDoP must be <=2.5

2: The receivers reported speed accuracy must be less than1.0 metres/sec (if available)

3: The receivers reported horizontal position accuracy must be less than 5.0 metres (if available)

4: The EKF2 magnetometer or compass innovation consistency checks must be passing. If these checks are failing, then the yaw estimate is unreliable

5: The rate of drift in the receivers reported horizontal position must be less than 0.3 metres/sec

6: The receivers reported vertical speed after filtering must be less than 0.3 metres/sec

7: The receivers reported horizontal speed after filtering must be less than 0.3 metres/sec.

Note: An unbroken pass on all selected checks for 10 seconds is required for the EKF2 to set its origin and start using GPS.

 

[cameraguy]

I spoke with the people who repair Solo's in San Diego. They also sell mods for the Solo. They indicated a new GPS was coming from them, I'm guessing after NAB.

This would act like the Phantom and give you access to both the US and Russian GPS system. No firm price, but they seem to sell most things around $100.

To me, that sounded like it was worth the wait.

Hope this helps

 

[pyrate]

I agree with you on that, however from reading about the new EKF I'm reassured.

Basically the new EKF is for want of a better word smarter, so it won't be cutting any corners if a lock is faster, it's just the logic is more precise.

Although the Pixhawk hardware has redundancy for example there's more than one IMU, the current EKF in the event of say an IMU failure won't use one of the working IMUs in its place where the EKF 2 is far more likely to be able to do so. So overall you'll be getting a much safer and robust Solo from the new code. If a magnetometer fails in the current EKF Solo will switch to fly manual, in EKF 2, it'll use one of the remaining working magnetometers instead.

The improvements in summary are:

• It can run a separate EKF2 for each IMU making recovery from an IMU fault much more likely
• It can switch between magnetometers if there is fault
• It can estimate gyro scale factors which can improve accuracy during high rate manoeuvres
• It can simultaneously estimate both gyro offsets and orientation on startup whilst moving and doesn’t rely on the DCM algorithm for its initial orientation. This makes it ideal for flying from moving platforms when the gyro has not been calibrated.
• It can handle larger gyro bias changes in flight
• It is able to recover faster from bad sensor data
• It provides a slightly smoother output.
• It is sightly more accurate
• It uses slightly less computing power
• It starts using GPS when checks pass rather than waiting for the vehicle motors to arm.

You can read the details here: EKF2 Navigation System — Dev documentation

The GPS prearm check is every bit as robust, in fact it looks moreso:

0: The receivers reported number of satellites must be >= 6

1: The receivers reported HDoP must be <=2.5

2: The receivers reported speed accuracy must be less than1.0 metres/sec (if available)

3: The receivers reported horizontal position accuracy must be less than 5.0 metres (if available)

4: The EKF2 magnetometer or compass innovation consistency checks must be passing. If these checks are failing, then the yaw estimate is unreliable

5: The rate of drift in the receivers reported horizontal position must be less than 0.3 metres/sec

6: The receivers reported vertical speed after filtering must be less than 0.3 metres/sec

7: The receivers reported horizontal speed after filtering must be less than 0.3 metres/sec.

Note: An unbroken pass on all selected checks for 10 seconds is required for the EKF2 to set its origin and start using GPS.

Wow now I see what all the fuss is about on getting to EKF2, totally agree

 

[Pedals2Paddles]

Would doing the M8N mod cause 3DR to deny unrelated warranty claims? Things like motor pod failures, IMU failures, and other issues that have nothing at all to do with the GPS.

 

[cameraguy]

I would guess yes. Rebuilding the Solo voided my warranty, and it didn't matter who did it, me or the official recommended repair place

 

[pyrate]

Would doing the M8N mod cause 3DR to deny unrelated warranty claims? Things like motor pod failures, IMU failures, and other issues that have nothing at all to do with the GPS.

Historically 3DR has not denied warranty on for example a motor if you modded your camera cable as an example.
Only if your mod directly impacted the failure, or you were flying in a way that would cause it