2020 Project Status?

[ToddG]

Peter, this looks like an incredibly cool project. I have a boat, and I'd love to build one of these as the commercial options are expensive, bloated, and gobble up energy.

What's the status of this project? Have you built and used one, and if so, what has been your experience using it? Do you want/need contributors to this project? For one thing, a blog, and/or some current documentation might be nice.

So this ticket is really all about asking for a quick refresh of the docs, project status, project needs, etc.

[boelle]

:-( would have been nice thou (if it was that easy - but then it would have been done many moons ago)

[peterantypas]

A couple of updates:

3D printed antenna cap fits well and will undergo accelerated aging at a local lab using 3 different sealing methods (2 types of heat shrink tubing plus self-amalgamating silicon tape).

Latest board is almost ready, with a couple of issues I would like to resolve before releasing it, so it will take another spin. Also, here is full TX power consumption and SWR:

[rszemeti]

Halfwave?

On Sat, 12 Sep 2020 at 01:18, Peter Antypas notifications@github.com wrote:

A couple of updates:

3D printed antenna cap fits well and will undergo accelerated aging at a local lab using 3 different sealing methods (2 types of heat shrink tubing plus self-amalgamating silicon tape).

[image: IMG_0824] https://user-images.githubusercontent.com/1565933/92982340-986d0a80-f452-11ea-9b83-7794ba0d2fea.jpg

Latest board is almost ready, with a couple of issues I would like to resolve before releasing it, so it will take another spin. Also, here is full TX power consumption and SWR:

[image: IMG_0830] https://user-images.githubusercontent.com/1565933/92982391-c7837c00-f452-11ea-9438-f26d31fb7a07.jpg

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/peterantypas/ais_transponder/issues/13#issuecomment-691362676, or unsubscribe https://github.com/notifications/unsubscribe-auth/ALFTBRCV77OZ6EJNFTJTG43SFK46DANCNFSM4MBGCNFQ .

[peterantypas]

1/2 wave flowerpot:

[rszemeti]

Good choice, anything better than -20dB is more than good enough for S11. I like 1/2 waves on boats, the minimal counterpoise requirements are very useful.

On Sat, 12 Sep 2020 at 02:09, Peter Antypas notifications@github.com wrote:

1/2 wave flowerpot:

[image: ANT1-19Turns-41+43cm-S11] https://user-images.githubusercontent.com/1565933/92983686-d9b4e880-f459-11ea-9628-e6990ad2eefe.png

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/peterantypas/ais_transponder/issues/13#issuecomment-691372435, or unsubscribe https://github.com/notifications/unsubscribe-auth/ALFTBRGTGNWLBKKB42VFX4LSFLC3RANCNFSM4MBGCNFQ .

[marcelrv]

A couple of updates:

3D printed antenna cap fits well and will undergo accelerated aging at a local lab using 3 different sealing methods (2 types of heat shrink tubing plus self-amalgamating silicon tape).

Good choice for small production run. Which material did you end up choosing for the printing?

[peterantypas]

I started with ASA, but given that this part will be sealed in UV protective material in its entirety, I can use ABS or even PLA next.

[SV-Oceant]

Hello all, Just found your postings yesterday looking for an inexpensive AIS transponder. I'm very interested in a mostly complete kit. Maybe something that has all the parts with the hard to solder components already complete.
I've had good success with cable glands being waterproof down to 130' saltwater on DIY video lights. I use them with electrical cable where the jacket is formed around the wires leaving no voids. I realize this option might be slightly big but I found this the other day looking for a waterproof housing for an RPi. http://openh.io/rubicon/

[peterantypas]

Update: Latest board works as expected with a couple of caveats:

1. The total output power at the fundamental is +32 dB, so not quite 2 Watts. I could raise it a bit, but:
2. There is a significant 2nd harmonic component that is best suppressed with a notch at the very output, suggesting that the RF switch itself is responsible. Even with the notch and duty cycle adjustments, the level of this harmonic would still violate FCC rules given that it falls within a restricted band (aviation).

That said, I'm not particularly worried about this at the moment. I will validate one more spin of the PCBA and should be ready to update github in the next couple of weeks. I will do this using "git move", so the entire legacy project will move under a "Legacy" subdirectory.

The new firmware will have a proper BSP architecture to support new board iterations with little effort. Also, the bootloader will go away and it will rely on ST's ROM bootloader for DFU over UART. I will include a Python script for this.

[rszemeti]

That sounds perfect. The bootloader did give me some issues, but I did manage to load code over it (I think I wrote some Perl to do this in the end, as there was no obvious loader script) but it made life harder than it needed to be, so I won't be sorry to see it go.

On Wed, 23 Sep 2020 at 17:58, Peter Antypas notifications@github.com wrote:

Update: Latest board works as expected with a couple of caveats:

1. The total output power at the fundamental is +32 dB, so not quite 2 Watts. I could raise it a bit, but:
2. There is a significant 2nd harmonic component that is best suppressed with a notch at the very output, suggesting that the RF switch itself is responsible. Even with the notch and duty cycle adjustments, the level of this harmonic would still violate FCC rules given that it falls within a restricted band (aviation).

That said, I'm not particularly worried about this at the moment. I will validate one more spin of the PCBA and should be ready to update github in the next couple of weeks. I will do this using "git move", so the entire legacy project will move under a "Legacy" subdirectory.

The new firmware will have a proper BSP architecture to support new board iterations with little effort. Also, the bootloader will go away and it will rely on ST's ROM bootloader for DFU over UART. I will include a Python script for this.

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/peterantypas/ais_transponder/issues/13#issuecomment-697695202, or unsubscribe https://github.com/notifications/unsubscribe-auth/ALFTBRHXVIJRDOD6MN7UT6DSHISKBANCNFSM4MBGCNFQ .

[peterantypas]

There was another issue with the current board: Spurious emissions due to improper ramp up and ramp down:

The solution required both a hardware and a firmware change, but things look much better now:

The updates will be in the next revision. Note that there is a 10dB pad at the input of the SA, so add 10dB to all measurements.

[marcelrv]

indeed significant improvement in the 2nd one. Now only very minimal spike left indeed

[peterantypas]

The last issue that makes the design non-compliant (for FCC anyway) is the harmonic signature:

All evidence suggests that the main culprit is the (rather inexpensive) Skyworks RF switch I used (13330-397LF).

I was able to get the second harmonic where it is by constructing a single pole notch filter at the output. I am designing a 3-pole notch for the next PCB, so hopefully it will get close to the level of the 3d. That said, I'm not going to worry too much about it.

In case you're wondering, there is an extra 10dB of attenuation at the input of the SA, so add 10dB to all measurements ;)

[peterantypas]

Here is why I'm not going to worry about the harmonics:

It turns out that the antenna is such a horrible radiator for the 2nd harmonic that it acts as a secondary notch filter, with an additional mismatch loss of 6dB (or thereabouts). I can see why this would be a concern for a device that connects to a user-supplied antenna, but this is not the case here.

So at 3 meters, assuming a normal free space path loss of 32dB, the radiated power of the 2nd harmonic should be -2.1 - 6 - 32 = -37.1dB, and that assumes 100% antenna efficiency. This is still 12dB above the FCC limit, but there is no way that this 27ms "click" in the aviation band will interfere with an aircraft flying hundreds or thousands of feet above your boat :D

[rszemeti]

Would not a simple lpf be easier and more forgiving to component tolerance than a notch? Swapping the multilayer ceramics in the output filter for AVX C0G types might help as well.

On Fri, 25 Sep 2020 at 22:58, Peter Antypas notifications@github.com wrote:

Here is why I'm not going to worry about the harmonics:

[image: Antenna-Broadband-S11] https://user-images.githubusercontent.com/1565933/94318868-4cd05b80-ff3e-11ea-8214-5344cdde5bb3.png

It turns out that the antenna is such a horrible radiator for the 2nd harmonic that it acts as a secondary notch filter, with an additional mismatch loss of 6dB (or thereabouts). I can see why this would be a concern for a device that connects to a user-supplied antenna, but this is not the case here.

So at 3 meters, assuming a normal free space path loss of 32dB, the radiated power of the 2nd harmonic should be -2.1 - 6 - 32 = -37.1dB, and that assumes 100% antenna efficiency. This is still 12dB above the FCC limit, but there is no way that this 27ms "click" in the aviation band will interfere with an aircraft flying hundreds or thousands of feet above your boat :D

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/peterantypas/ais_transponder/issues/13#issuecomment-699175622, or unsubscribe https://github.com/notifications/unsubscribe-auth/ALFTBRGOYDKO3LZXIZCEZT3SHUHBVANCNFSM4MBGCNFQ .

[peterantypas]

I need at least 30dB of total attenuation in the 2nd harmonic to be compliant, and it can only be effective past the RF switch. I have not been able to devise a low pass filter that will achieve this without at least 1.4dB loss at the fundamental as well, but definitely want to explore this option too.

I sent a bunch of boards to JLCPCB for fab yesterday. There are 5 different RF switches and 2 filter configurations in there that I want to try out, so hopefully one of them will work well enough. We may just end up with a design that only transmits at +31dBm, which is still OK, it will have a few miles of range.

[rszemeti]

The downside to a filter after the switch is that the loss also degrades the noise figure of the receiver. Maybe a discrete PIN diode switch would perform better than the skytronics switch?

On Fri, 25 Sep 2020 at 23:58, Peter Antypas notifications@github.com wrote:

I need at least 30dB of total attenuation in the 2nd harmonic to be compliant, and it can only be effective past the RF switch. I have not been able to devise a low pass filter that will achieve this without at least 1.4dB loss at the fundamental as well, but definitely want to explore this option as well.

I sent a bunch of boards to JLCPCB for fab yesterday. There are 5 different RF switches and 2 filter configurations in there that I want to try out, so hopefully one of them will work well enough. We may just end up with a design that only transmits at +31dBm, which is still OK, it will have a few miles of range.

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/peterantypas/ais_transponder/issues/13#issuecomment-699201313, or unsubscribe https://github.com/notifications/unsubscribe-auth/ALFTBRHSXFGTUKREF5HRRODSHUOA3ANCNFSM4MBGCNFQ .

[peterantypas]

Yeah, I really wanted to avoid anything between the switch and the antenna. But I got the result I wanted with a pretty sharp, high ripple Chebyshev LPF added to the PA output matching network. So I don't even think I'll need to evaluate other switches.

Remember to add 10dB to all these values since there's always an external pad on my SA:

[rszemeti]

That looks pretty good to me. 2nd harmonic 45db down should be enough, as you noted, S11 of the antenna will assist as well.

Slightly jealous that your SA display is much crisper than mine, 8563A?

On Mon, 28 Sep 2020 at 02:44, Peter Antypas notifications@github.com wrote:

Yeah, I really wanted to avoid anything between the switch and the antenna. But I got the result I wanted with a pretty sharp, high ripple Chebyshev LPF added to the PA output matching network. So I don't even think I'll need to evaluate other switches.

Remember to add 10dB to all these values since there's always an external pad on my SA:

[image: IMG_0850] https://user-images.githubusercontent.com/1565933/94382313-11c25980-00f1-11eb-89ba-0a8b0b03a108.jpg

[image: IMG_0851] https://user-images.githubusercontent.com/1565933/94382317-1424b380-00f1-11eb-9fb9-475d66ff75fc.jpg

[image: IMG_0852] https://user-images.githubusercontent.com/1565933/94382321-1981fe00-00f1-11eb-9d9b-6b2825c48c7a.jpg

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/peterantypas/ais_transponder/issues/13#issuecomment-699725378, or unsubscribe https://github.com/notifications/unsubscribe-auth/ALFTBRFMSEQNWLNEKXKJ2UDSH7S6ZANCNFSM4MBGCNFQ .

[peterantypas]

Update:

I got the ROM bootloader to work fine and adapted a Python script from another one of my projects. Unfortunately, the current board cannot enter DFU mode over USART1 because the GNSS is tied to USART2 and is continuously sending data. So even though firmware can enter the ROM bootloader via a serial command, the bootloader immediately shuts down every interface except USART2 because it thinks that's where the firmware update will come from. Oops!

I did another board spin to validate the solution, but I totally forgot about the Chinese Harvest festival, so that will take a couple of weeks to get here.

[peterantypas]

The branch "new_design" is already started. I will continue updating it over the weekend and finally push to master.

[rszemeti]

the 6.1 board looks good, The GPS block looks like it might be difficult to source though.

its way too late here for my brain to work well though ...

On Sat, 3 Oct 2020 at 03:56, Peter Antypas notifications@github.com wrote:

The branch "new_design" is already started. I will continue updating it over the weekend and finally push to master.

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/peterantypas/ais_transponder/issues/13#issuecomment-703036357, or unsubscribe https://github.com/notifications/unsubscribe-auth/ALFTBRGI564AZYJDKUFSAKTSI2HGTANCNFSM4MBGCNFQ .

[rszemeti]

Layout looks pretty good. I notice you have swapped to 0603 instead of 0805 after the PA, which should reduce parts inventory ...

the only thing I can see is a few earth vias in component pads that should be corrected for production I guess, as it makes a paste mess under the board eventually, but apart from that, nice work dude.

[image: image.png]

On Sat, 3 Oct 2020 at 05:09, Robin Szemeti robin@redpoint.org.uk wrote:

the 6.1 board looks good, The GPS block looks like it might be difficult to source though.

its way too late here for my brain to work well though ...

On Sat, 3 Oct 2020 at 03:56, Peter Antypas notifications@github.com wrote:

The branch "new_design" is already started. I will continue updating it over the weekend and finally push to master.

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/peterantypas/ais_transponder/issues/13#issuecomment-703036357, or unsubscribe https://github.com/notifications/unsubscribe-auth/ALFTBRGI564AZYJDKUFSAKTSI2HGTANCNFSM4MBGCNFQ .

[peterantypas]

That's definitely fixable, do you want to take a stab at it? I can give you commit privileges.

[rszemeti]

More than happy to do that if you like, or I can fork and you could pull

On Sat, 3 Oct 2020 at 16:25, Peter Antypas notifications@github.com wrote:

That's definitely fixable, do you want to take a stab at it? I can give you commit privileges.

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/peterantypas/ais_transponder/issues/13#issuecomment-703120360, or unsubscribe https://github.com/notifications/unsubscribe-auth/ALFTBRCJE73KTHEZ24FWLDLSI465JANCNFSM4MBGCNFQ .

[rszemeti]

Also ... can you check the layer stackup? It looks to me like you have L2 as the 3.3V layer and L3 as the RF groundplane ... I suspect you wanted those the other way around ...

On Sun, 4 Oct 2020 at 03:20, Robin Szemeti robin@redpoint.org.uk wrote:

More than happy to do that if you like, or I can fork and you could pull

On Sat, 3 Oct 2020 at 16:25, Peter Antypas notifications@github.com wrote:

That's definitely fixable, do you want to take a stab at it? I can give you commit privileges.

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/peterantypas/ais_transponder/issues/13#issuecomment-703120360, or unsubscribe https://github.com/notifications/unsubscribe-auth/ALFTBRCJE73KTHEZ24FWLDLSI465JANCNFSM4MBGCNFQ .

[peterantypas]

That works too.

On Sat, Oct 3, 2020 at 7:21 PM Robin Szemeti notifications@github.com wrote:

More than happy to do that if you like, or I can fork and you could pull

On Sat, 3 Oct 2020 at 16:25, Peter Antypas notifications@github.com wrote:

That's definitely fixable, do you want to take a stab at it? I can give you commit privileges.

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub < https://github.com/peterantypas/ais_transponder/issues/13#issuecomment-703120360 , or unsubscribe < https://github.com/notifications/unsubscribe-auth/ALFTBRCJE73KTHEZ24FWLDLSI465JANCNFSM4MBGCNFQ

.

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/peterantypas/ais_transponder/issues/13#issuecomment-703190030, or unsubscribe https://github.com/notifications/unsubscribe-auth/AAL6J3M3MZBQ5FM2CJWHER3SI7LY3ANCNFSM4MBGCNFQ .

[peterantypas]

RE: Layer stackup, that's intentional. I don't remember the exact reason but I had much better luck with this a few years ago than the alternative, so it kind of stuck. I think it had to do with traces being too thin to achieve 50 Ohms between L1 and L2 in a coplanar waveguide configuration. I could be wrong though ...

[peterantypas]

I remember what it was: I wanted my RF traces to be wide enough to encase 0603 pads without creating stubs. Could not possibly achieve that with 0.1mm height between L1 and L2 with any material. I would have needed 0201s instead. Only way to get the "hot" trace to be wide enough at 50 Ohms was using L1 and L3.

[rszemeti]

Could be, it just seems strange to see the 50R output trace with a 3V3 trace between it and the groundplane ... there are no buried vias though so it would be easy just to rename the layers and try the other way around when assigning layers at JLC. At the end of the day, its only VHF, so unlikely to make a whole heap of difference.

On Sun, 4 Oct 2020 at 04:26, Peter Antypas notifications@github.com wrote:

RE: Layer stackup, that's intentional. I don't remember the exact reason but I had much better luck with this a few years ago than the alternative, so it kind of stuck. I think it had to do with traces being too thin to achieve 50 Ohms between L1 and L2 in a coplanar waveguide configuration. I could be wrong though ...

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/peterantypas/ais_transponder/issues/13#issuecomment-703195891, or unsubscribe https://github.com/notifications/unsubscribe-auth/ALFTBRCQ57JOQDA35M3K5RTSI7TMVANCNFSM4MBGCNFQ .

[rszemeti]

Fair enough, as I said, its only VHF, the track lengths are less than 1% of a wavelength anyway, so whether they are 100R or 25R "width" probably is irrelevant. Would be significant at higher frequencies obviously, but at 150mhz, the tracks are too short to be an issue. .. I'm not sure with the 3V3 trace underneath it will quite work as expected, maybe assign a block of "keepout" on L2 under the RF tracks to prevent the fill from the L2 polygon?

On Sun, 4 Oct 2020 at 04:35, Peter Antypas notifications@github.com wrote:

I remember what it was: I wanted my RF traces to be wide enough to encase 0603 pads without creating stubs. Could not possibly achieve that with 0.1mm height between L1 and L2 with any material. I would have needed 0201s instead. Only way to get the "hot" trace to be wide enough at 50 Ohms was using L1 and L3.

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/peterantypas/ais_transponder/issues/13#issuecomment-703196599, or unsubscribe https://github.com/notifications/unsubscribe-auth/ALFTBRDNLOVKY77QOGAAKLLSI7URVANCNFSM4MBGCNFQ .

[rszemeti]

What about ... L2 before:

[image: image.png] L2 revised:

[image: image.png]

On Sun, 4 Oct 2020 at 04:40, Robin Szemeti robin@redpoint.org.uk wrote:

Fair enough, as I said, its only VHF, the track lengths are less than 1% of a wavelength anyway, so whether they are 100R or 25R "width" probably is irrelevant. Would be significant at higher frequencies obviously, but at 150mhz, the tracks are too short to be an issue. .. I'm not sure with the 3V3 trace underneath it will quite work as expected, maybe assign a block of "keepout" on L2 under the RF tracks to prevent the fill from the L2 polygon?

On Sun, 4 Oct 2020 at 04:35, Peter Antypas notifications@github.com wrote:

I remember what it was: I wanted my RF traces to be wide enough to encase 0603 pads without creating stubs. Could not possibly achieve that with 0.1mm height between L1 and L2 with any material. I would have needed 0201s instead. Only way to get the "hot" trace to be wide enough at 50 Ohms was using L1 and L3.

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/peterantypas/ais_transponder/issues/13#issuecomment-703196599, or unsubscribe https://github.com/notifications/unsubscribe-auth/ALFTBRDNLOVKY77QOGAAKLLSI7URVANCNFSM4MBGCNFQ .

[peterantypas]

That's already the case. There is no power plane anywhere near the RF traces:

BTW, I don't see any of the attachments in your messages.

[rszemeti]

OK, maybe I don't have the latest version from GIT or maybe my eagle interprets it differently ... but L2 extends right under the RF in my version .. not sure why github is removing pasted images, thats .. annoying.

On Sun, 4 Oct 2020 at 04:57, Peter Antypas notifications@github.com wrote:

That's already the case. There is no power plane anywhere near the RF traces:

[image: Screenshot from 2020-10-03 20-56-25] https://user-images.githubusercontent.com/1565933/95006583-ffe22a00-05ba-11eb-860c-8cfedc02ecb0.png

BTW, I don't see any of the attachments in your messages.

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/peterantypas/ais_transponder/issues/13#issuecomment-703198169, or unsubscribe https://github.com/notifications/unsubscribe-auth/ALFTBRFRDS5SYO2AJDQXYPLSI7XCJANCNFSM4MBGCNFQ .

[rszemeti]

For whatever reason my Eagle is not picking up that restrict polygon under the RF for L2, I suspect it is too old a version :( ...

On Sun, 4 Oct 2020 at 05:03, Robin Szemeti robin@redpoint.org.uk wrote:

OK, maybe I don't have the latest version from GIT or maybe my eagle interprets it differently ... but L2 extends right under the RF in my version .. not sure why github is removing pasted images, thats .. annoying.

On Sun, 4 Oct 2020 at 04:57, Peter Antypas notifications@github.com wrote:

That's already the case. There is no power plane anywhere near the RF traces:

[image: Screenshot from 2020-10-03 20-56-25] https://user-images.githubusercontent.com/1565933/95006583-ffe22a00-05ba-11eb-860c-8cfedc02ecb0.png

BTW, I don't see any of the attachments in your messages.

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/peterantypas/ais_transponder/issues/13#issuecomment-703198169, or unsubscribe https://github.com/notifications/unsubscribe-auth/ALFTBRFRDS5SYO2AJDQXYPLSI7XCJANCNFSM4MBGCNFQ .

[peterantypas]

I’ll add Gerber exports to git

Sent from my iPhone

On Oct 3, 2020, at 9:06 PM, Robin Szemeti notifications@github.com wrote:

 For whatever reason my Eagle is not picking up that restrict polygon under the RF for L2, I suspect it is too old a version :( ...

On Sun, 4 Oct 2020 at 05:03, Robin Szemeti robin@redpoint.org.uk wrote:

OK, maybe I don't have the latest version from GIT or maybe my eagle interprets it differently ... but L2 extends right under the RF in my version .. not sure why github is removing pasted images, thats .. annoying.

On Sun, 4 Oct 2020 at 04:57, Peter Antypas notifications@github.com wrote:

That's already the case. There is no power plane anywhere near the RF traces:

[image: Screenshot from 2020-10-03 20-56-25] https://user-images.githubusercontent.com/1565933/95006583-ffe22a00-05ba-11eb-860c-8cfedc02ecb0.png

BTW, I don't see any of the attachments in your messages.

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/peterantypas/ais_transponder/issues/13#issuecomment-703198169, or unsubscribe https://github.com/notifications/unsubscribe-auth/ALFTBRFRDS5SYO2AJDQXYPLSI7XCJANCNFSM4MBGCNFQ .

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub, or unsubscribe.

[peterantypas]

All latest designs and firmware have been pushed to the master branch. Closing this ticket now.