uBlox GPS touching Pixhawk: More test results

As the original thread on this topic got locked, here is a new thread with some additional information and attempted fixes.

To recap:

One of the connectors of the uBlox GPS module touches or rather presses against the I2C connector cable on top of the Pixhawk, also causing the Pixhawk not to be isolated from the frame anymore.
The result is that this connector squeezes the cables and damages them over time due to the vibration of the frame transmitted over the hull to the uBlox.
Have a look at this report which included this picture:

You can clearly see the damaged I2C connector cable.

After a closer inspection I found out the following:

  • The Pixhawk on the consumer version is position further to the front than on the developer version, thus moving it underneath the uBlox module. Have a look at this picture from the Developer Edition to compare.

So I started with a few ideas and tests to see how to fix this:

  • First I placed a small piece of rubber between the front of the Pixhawk (on top of the cables) and the GPS module. This is not ideal as there are still two problems with it:
  • The cables are still being crushed, just by rubber this time and not by metal or sturdy plastic.
  • As someone in the other thread pointed out: “Connecting " PIXHAWK with the UBLOX (even by foam), will affect vibration damping.” So this indeed doesn’t help with vibrations.

This is how vibrations looked like. Note: There is no difference in vibrations with the foam in between and without. That’s why I didn’t attach a picture without the foam.
The amplitude of the vibrations along all three axis looks pretty good. But there are two problems: z-axis vibrations have a lot of sudden spikes (which are supposedly a cause for barometer failures). And the x-axis vibrations seem to have a superimposed pattern, so it’s not really the white noise that it should be.
That leads to the assumption that the Pixhawk isn’t really freely dampened with this setup, but still connected to the frame.

After removing the rubber again, you could clearly see marks in the soft rubber of where the uBlox pressed into. Not very reassuring.

  • Next I investigated moving the Pixhawk to the back (just like in the Developer version). The problem here is that the Telemetry module is also mounted differently in the Consumer version than in the Developer version. It is therefore in the way of the Pixhawk and would need to be moved as well.
    That was a little bit to much of a hassle. So I had to find something else.

  • Next I figured to raise the hull by a few millimeter in order to remove the connection between Pixhawk and GPS module.
    In order to do this, I replaced the three screws holding the top hull with longer ones (M3x12mm) and placed rubber grommets on top of the screws between the hull and the top plate.
    The result looks like this:

You can clearly see the gap between the hull and the top plate. I placed electrical tape over this gap (not shown here) in order to prevent wind entering the hull. Also now the hull is only resting on three places, making it a bit wiggly. So you have to be a bit careful when carrying IRIS.

Also flying IRIS like this will give it a different sound. So don’t be scared.

The good news: Looking through this gap into the hull you can see that the GPS is not touching the Pixhawk anymore:

Thus the Pixhawk should not be directly connected to the frame anymore. As a result, this is how the vibrations look like:

The spikes on the z-axis vibrations are gone and also the x-axis vibrations don’t seem to be following a pattern anymore. Now the vibrations are really white noise, as they should be.
The drawback: The amplitude of the vibrations have increased.

This lets me believe that the vibration mount between Pixhawk and IRIS frame is to soft. I will therefore try to replace this foam with a gel mat (amazon.com/gp/product/B002U2GS2K/).
The other benefit of this gel mat is that it is thinner than the stock foam on the IRIS. With this I might be able to lower the top hull again.

@3DR: It would be great if you could chime in here and give some guidance on this topic. Am I moving down the right path with these workarounds?

Note: For everyone who isn’t comfortable taking his new RFT model apart in an extended arts and craft session, I recommend to wait until 3DR comes up with a recommended/permanent fix.

Just thinking, why not mount the gps on top of the iris?

Yes, you could. But then you have to make a hole into the upper shell for the cables and create some kind of mounting device for the GPS. That would mean quite a bit of work and change IRIS quite a bit as well.

As I have inferred in other posts, that raises the question if it’s worth it and also if that’s really the path you want to go down for with IRIS as a Ready-to-Fly platform.
For me personally the answer is No as I rather wanted to fly a Quad instead of perfection the hardware setup. I’m therefore trying to reduce my “fixing” attempts to a level that keeps the IRIS as close to it’s out of box character.

For the use case of extending a platform the 3DR Y6 (store.3drobotics.com/products/3dr-rtf-x8-2014/) anyways appear to be better choices as they give you more freedom and capabilities to extend. You don’t need to deal with cutting holes or gluing stuff to a plastic hull. You can just tie-wrap it to the frame. :slight_smile:

Keep us posted, and also try to snap some pics of the gel insert in place… Great ideas!

Please contact help@3DRobotics for a thinner piece of foam

Here is an update after installing the previously mentioned gel mat between the Pixhawk and the IRIS frame:

First I removed the foam between the Pixhawk flight controller and the top shell. The foam will most likely disintegrate doing so. At least mine did.
3DR uses 4 small blocks of foam in the 4 corners of the Pixhawk. I could have cut my gel mat also into 4 little squares and done the same in order to reduce weight. But instead I cut the gel mat to match the size of the Pixhawk for one big mat. In the end I had enough of the stuff and I wasn’t that concerned about weight.

While placing the Pixhawk on the gel mat and the gel mat on IRIS I made sure to move them as far to the back as I could before they would come to close of even touch the 3DR radio.

Here is how the result looks like:

My initial hope was that the gel mat was lower than the 3DR foam. Unfortunately it turned out that they were not. Look here to compare the gel mat with the 3DR foam:


They have about the same height.

Again looking into the raised hull at the front, I could see that the Pixhawk was still clear of the GPS module. I even reduced the raising by half to about 2mm. These picture are taken with the hull being raised by only 2mm:



Looking at the gap between Pixhawk and GPS I might even be able to remove the hull raising completely.
This also brings me to a question that came up frequently: Why are some people affected by this issue and others not?
It appears like it’s down to pure luck. As every fabrication process has slight deviation, so does assembling IRIS. And here it appears like mounting the Pixhawk within the IRIS a few mm too much to the front makes a big difference. So it’s pure luck how your IRIS’ Pixhawk was placed in the factory and thus whether it rubs against the GPS or not.

But to be honest: I rather safe a bit of safety room and play it safe. That’s why I will keep the hull raised for now. I might be looking into purchasing a gel mat that isn’t as thick as this one though.

Now what we have all been waiting for: How do the vibrations look like. Unfortunately I didn’t have the time to take IRIS out for a real flight. Instead I flew in my living room at about 10-20cm height (Kids, don’t try this at home! It’s tough to keep IRIS within a 2m x 2m space. And not having prop guards doesn’t make the thing more reassuring). But this way I was able to get enough test date to see how vibrations are now.

Here is the result:

I hope you are as stunned as I was, when looking at this. There are almost no vibrations along the x and y axis. And the z-axis vibrations are also great (These z-axis “waves” at the beginning are due to prop wash at that low height and me trying to get her to hoover).

I couldn’t believe the results, so I repeated the test flight three more times: All with the same great result.

The gel mat dampens vibrations so effectively that I will try to use it with the front camera mount next.

With that I’m hopeful that my next outdoor flights will see improved performance in AltHold and Loiter.

Here is another idea to the problem I’ve been thinking about:

It is possible to move the GPS module to the front by mounting it slightly tilted.
Like this:



There are a few challenges with this approach:

  • You have to be extremely careful tightening the screws and ensure that you don’t over-tighten them. Otherwise you could break the GPS board. So be careful!
  • The GPS module is now mounted in a tilted angle. I don’t know what (negative) effects this would have on the GPS reception and especially the magnetometer. For GPS you might now be able to pick up satellites behind you in a very orbit anymore. And for the magnetometer, I have no clue.
    Does anyone have an idea?
  • You are moving the GPS closer to a GoPro camera mounted on the front (if you have one there). This might increase interference from the GoPro onto the GPS.

Because of these challenges (especially the second bullet point) I didn’t try this yet.

Here is a workaround i found myself to solve this issue, simply rotate the GPS modul 90 degrees ending with the connectors pointing forward. this will give 2mm of additional space between the Pixhawk and the GPS modul. (if possible change also the screws by using some flat head screws) WARNING you will have to change the compass orientation to Yaw_90

Here the Compass screen were you will have to change the compass orientation. After that recalibrate the compass and the Accelometer

A lower gel mat looks like a good solution, 2mm lower would solve it for me. Is the gel mat or foam a special RC type? Or would something from a hardware store work?

Pomaroli and kanga, thanks for continuing to look into this. Pomaroli, is it absolutely necessary to calibrate the accelerometer after this change? Thank you.

If you rotate the GPS, yes its required to recalibrate or else it won’t know the correct orientation.

Sent from my HTC One using Tapatalk

Congrats guys ! Good ideas, great results. Concerning the thickness of the foam/gel pad : According to my AP experiences , it might be a good idea to cover a smaller area (split the gel pad/reducing foam pad size) along with reducing the thickness to achieve nearby vibration damping results . Using more flexible servo cables (ESC) , which don´t “fix” the PH, help the matter as well, as I explained at RCG (without a nice log doc as you did)

While pulling logs from my pixhawk i noticed today that one of the black leads on the I2C cable thats plugged into the pixhawk was worn through the insulation, I emailed 3DR and may stop by their San Diego facility(being local is nice) to see if i can pick up the thinner foam I’ve heard Craig posting people should email 3DR about, and maybe a new I2C cable. I confirmed that the pixhawk on my iris is mounted 75mm from the rear edge of the top plate just like the dev upgrade kit video(vimeo.com/90072591) shows on 3DR store page.

No, but it’s important to set the Compass orientation like shown to yaw_90, you can double check by turning the iris manually to yaw 0, the heading in the flight data screen should then point north.
Depending on how much the pixhawk was pushed down by the gps modul it will fly level or not. if it doesn’t fly level you have to recalibrate.

it’s a simple calculation, between the gps modul and the bottom plate you have 20mm of space, the pixhawk is 15mm high + 6mm foam gives 21mm and 1 mm of space you need at least for the cables. So we need at least 22mm of space. you can simple turn the gps to get the 22mm (if you use flat head screws for the gps mount you will have more space there were the cables are located on the pixhawk) or change the foam from 6mm to 4mm, the result will be the same. But i think to turn the gps will be the simpler solution until 3DR will show up with something better.

Just wanted to share my results. In every flight I made yesterday, 5-03-14, you can see large vibrations on the z axis]
]and deviations when comparing GPS: RelAlt with CTUN: BarAlt. I only made one flight today, but after taking the advice of pomaroli by rotating my compass 90 degrees, you can see absolutely no vibrations and the relalt and baralt track each other perfectly. Thank you guys for your continued input and what I think great help to many iris owners who may be frustrated by similar issues. You guys deserve medals!

yeah it’s a great improvement…i was having a lot of fun today, my Iris now flies like it should and from what i was expecting from 3DR… 3DR Team you need urgent to improve this.

I’d like to see 3DR’s thoughts on this, before I do it. But thanks for being a pioneer, Pomaroli!

As i wrote this is only a workaround, but for this it is working really well.
Today i was testing Loiter, AltHold, Stabilize and Auto mode with strong wind and what can i say it’s working almost worry free.

@Pomaroli: You are genius! Turning the GPS module is an awesome idea. Never even thought you could do that after seeing the arrow on the module.

I also turned the GPS module, changed the compass orientation to yaw_90.
With the hull in it’s original position (no raising) it appears that this gives you about 1 mm of head room between Pixhawk and uBlox. That might be enough for some of us for the cables.

Here is how it looks like peeking into IRIS from the back:
[attachment=2]Turned GPS.jpg[/attachment]

Also I went through the re-calibration of the accelerometer and the compass.

Unfortunately the test flight with Gel mat and turned GPS was a complete disaster:
The z-axis vibrations were the worst I have ever seen:


As a result the BarAlt vs. GPS->RelAlt tracking was also really bad:


No way to fly AltHold or even Loiter with that.

It appears that the Gel Mat can’t cope with z-axis vibrations that well, while the x- and y-axis vibrations can be compensated pretty well.

I didn’t have any tools and spare parts with me in the field to raise the hull again and see if it makes any difference. So I went home and did some more of an arts and craft session.

(More in another post, as I can’t post more than 4 pictures to this one).

After the gel mat disaster I followed the advice from gervais and cut the mat into 4 small square pieces for under each corner of the Pixhawk.
The result looks like this:


With this I headed out again to a park to do a quick test flight. The result was much more promising, although this time it was extremely gusty:

Vibrations along the z-axis were OK as you can see here:


And as a result the BarAlt vs. GPS->RelAlt tracking was also OK:


Just like Pomaroli I won’t dare to declare victory just yet. Even though I wish I could, after yet again another weekend where I spend more time fixing than flying IRIS.

It appears that Pamaroli’s idea gives us another 1-2 mm of head room, which seems to be enough for a lot of folks here. That is really great, especially as some folks here had the first successful flights with their IRIS as a result. Therefore big kudos to Pamaroli!
Unfortunately it appears that the Pixhawk and the GPS module might still be “connected” via the cables in between. Therefore not all vibrations disappear. Again, mileage might vary here for different folks.
(After having discovered that one of my prop nuts that came with IRIS is actually a different size than the others, I’m pretty sure that a reason why some folks are having bigger problems with some of the issues reported here than others are massive production differences).

I will ask 3DR to send me thinner foam pieces (along with new prop screws :wink: and will give this a try.
My gut feeling is that a combination of Pamaroli’s great idea of turning the GPS module, using thinner foam and moving the Pixhawk slightly to the back might end up being the best bet we have.

I really don’t want having to fly with a raised hull, although that is still a possibility. Maybe we can combine the raised hull with hotelzulumima’s “pregnant” IRIS, making her taller than she is wide. :slight_smile:

While I agree 100% with Pamaroli that we should get a tried and tested fix from 3DR within a reasonable time frame, I have the feeling that we are the ones who have to try and test this fix. So let’s keep working on it. Thanks again Pamaroli!