Introducing: TelemetrEZ

[gohsthb]

I've never seen it happen. And there should be no reason that data changes at all. The Tez simply captures a full packet to a receive buffer. When the whole packet is in, it then copies it to a transmit buffer, and sends it on it's way. If there is another packet (either switch or telemetry) already in progress it will wait to send the next one. The only thing I could see happening is a packet getting dropped, not changed. I'll still have a look though.
-Gohst

[MikeB]

I think there is something not quite right. I've spent some time monitoring the serial data from the TEZ board. I'm graphing the last 200 height values received, and also I have a 'packet capture' set up where any user data packet longer than 12 bytes is displayed. Slightly odd as it isn't just looking at the '7E' values to get the size.
What I'm seeing is occasional jumps in the graph data, one by over 50m.
Over a period of several minutes I got 7 'long' packets. 6 of these look OK, except 5 had 4 zero bytes at the end, and one had 10 zero bytes at the end.
The 7th packet was wrong - 7E FD 00 00 00 5E 00 00 00 00 00 00 00 00 00.
I'm re-running this test on the data straight from the Tx module, no jumps in the height and no 'long' packets. (Must have been for 10 minutes as the inactivity alarm just went off!)

BTW should I be seeing encoder bytes in the 'private data'? I'm only getting 3 bytes - 1B 01 40, for the switches, or have I not got r60 properly? I'll double check re r60 being loaded.

While typing I reverted to monitoring the TEZ output, I'm seeing 'jumps' in the height again, and another 'long' packet - 7E FD 00 00 00 7E 00 00 00 00 00 00 00 00 00.
Now, it has given me another 'long' packet - 7E FD 04 15 06 00 5E 30 5E 21 1B A6 1B 01 00 7E, then stopped forwarding telemetry packets, but still sending switch data.

Mike.

[s_mack]

Without looking at the code... what you're describing sounds like two values are getting appended together inadvertently.

[gohsthb]

The Tez won't send the encoder part of the packet unless you have an encoder and you move it. In that way it auto detects if the encoder is attached.
-Gohst

[barneyg]

Am I right in saying the only reason this doesn't work for the 9XR is a lack of space in the back casing ?

[gohsthb]

I've worked with Mike on the packet problems he was seeing. I think it is fixed now. I sent him a new version of the Tez firmware to test, if that comes back ok I will update the code site for this new version.

I don't know enough about the 9xr to comment for sure. I understand there is much less space in the 9xr radio. I also understand there may be some hardware changes also, I don't know if they affect the telemetry link though.
-Gohst

[maninvan]

While typing I reverted to monitoring the TEZ output, I'm seeing 'jumps' in the height again, and another 'long' packet - 7E FD 00 00 00 7E 00 00 00 00 00 00 00 00 00.
Now, it has given me another 'long' packet - 7E FD 04 15 06 00 5E 30 5E 21 1B A6 1B 01 00 7E, then stopped forwarding telemetry packets, but still sending switch data.

That's interesting... I was also trying to get my APM to send GPS data using TEZ and it wasn't working properly as the data was all messed up. it also a loooong packet.

@gohsthb:I just checked the repo and I see you have found the issue. I'll flash r75 later today, and test to see if that fixes my GPS coord corruption issues. Thx for your efforts.

Cheers

Bill

[s_mack]

Am I right in saying the only reason this doesn't work for the 9XR is a lack of space in the back casing ?

No, its more than just that. But its all just physical problems. The connectors are completely different, so each and every connection has to be made manually. If you're going to bother doing that, then its probably easier to do the standard solder-in FrSky mod.

I am entirely comfortable on a design specific to the 9xr... but frankly, to date only 4 people have mentioned it to me (and I'm counting you). Not exactly overwhelming demand. If there's a community of people needing it, I'm not aware of them. If they exist, point them to me or me to them... if there's call for it, I'll make a 9xr version. Its the same circuitry (mostly) and same firmware... but completely different physical layout and connectors.

- Steven

[Buckmaster]

I would assume it would take quite a while for there to be enough 9xrs out there before there is enough demand. There are how many 9xs out there and there is finally enough demand to make it financially responsible to take on the required investment. Steven has been great about offering products that have a demand (and gotten them to market pretty quick). I'm guessing it is a fairly long time period to allow enough "market soak" on something like an rc tx to have enough of them in in the hands of people who are willing and able to modify them. Unfortunately its not like building the transmitter that any one can buy, we also need people o buy the tx, want to mod it, and find the product. It will get there at some point if its as mod friendly as the 9x is

[s_mack]

It will get there at some point if its as mod friendly as the 9x is

Its not. Not even close. That's kind of the irony... they make it specifically easier to upgrade the firmware, but make it nearly impossible to internally add anything. There's just no space!

[pmullen503]

Its not. Not even close. That's kind of the irony... they make it specifically easier to upgrade the firmware, but make it nearly impossible to internally add anything. There's just no space!

Too bad, it could have been so much better. I'm waiting for someone to come up with a spacer to bolt between the two halves to increase the internal space (and relocate those unreachable top switches, maybe replace a pot with a slider too.)

[Kilrah]

Just dump the damn thing already

[Buckmaster]

Just a bit off topic, but.... I've read numerous complaints about the lack of internal space. Is the case smaller than the 9x, or does the 9xr just have more inside it to take up space? Seems from what I have seen that the general consensus is that people ( at least the ones who frequent the forums) are by and large disappointed with the 9xr. Too bad, everyone had looked forward to it reaching the market and waited and waited....

[s_mack]

While I hate to say "I told you so"... people may remember that I predicted pretty early on that the 9xr would be a bust. There was so much hype, it couldn't possibly be anything but a disappointment! There was just no way they were going to live up to expectations. So while it was the consumers' own fault for building it up so high.. it was HK's fault for not releasing appropriate truthful information as they went along. They let people's imaginations run rampant. With that said... for the price, its not half bad. Its just no better than the 9x and in some ways its worse.

As for why there's no room... its the physical shape of the case. Its more ergonomical, which is why it fits better in the hands in my opinion, but that shaping leaves no room internally. Also, they provided a HUGE battery bay which eats at the preciously little internal space. There are some interesting case features it does have that I'd like to explore, so I'm not ruling it out. I'm sure I can fit a TelemetrEZ design in there.

- Steven

[gohsthb]

I did some of my own testing between r60 and r75. r75 is no longer changing packets at random. So I have changed the code page to reflect r75 as the latest stable version.
-Gohst

[s_mack]

That's a little... umm... precarious, if you haven't also identified the difference that caused it, no??

[gohsthb]

Did I say we didn't find anything causing it? Mike sent me some PM's. It had to do with some variables that weren't protected from changing in interrupts. You can look up diff's for the sources on the code page to see exactly what was changed.
-Gohst

[s_mack]

Guess I mis-read your last post.

[maninvan]

Ok, I have done some testing, it didn't fix my issue, but that is probably unrelated. When I get some time I will do some debugging of the MAVLink to FrSky protocol decoding.

[s_mack]

I did it!

Those that have been following this project from the beginning will remember that I wanted to make it so if the customer had an SP board that they could use it to update the TelemetrEZ without having to open the case. I struggled a lot with that and eventually scrapped it because there just was no way to do it. Well, no way with the existing SP design anyway. Now that I'm almost sold out of SP 2.3's... I took another look at it. So starting with SP 2.4, the TelemetrEZ (existing and future) will be able to be updated internally by the SP with no need for an external USBasp programmer or opening the case at all.

It really wasn't that difficult (once I found an appropriate chip) and if the timing of the SP 2.3 release had meshed better with the conception of TZ, we'd already be there. Basically, a demultiplexer is used to "decide" whether the MISO/MOSI/SCK/RST signals should go to the 9x (for updating the 9x's main firmware) or to the TelemetrEZ (for updating its firmware). When both battery power and USB power are present, the SP board understand that the user is wanting to update the 9x. When only USB is present, it understands that the TelemetrEZ is to be updated.

So can ambitious pre-2.4 owners convert to 2.4 without buying a new board? Yeah, its possible (I just did it)... but, keeping in mind my customers tend to buy my products to AVOID soldering... its unlikely that any will. Considering just that the necessary demultiplexer only comes in surface mount design... right there its really not a worthwhile endevour, I wouldn't think. But should anyone seriously want to do it, I can provide links to the required products (a demultiplexer, an inverter gate, a Schottky diode, a Zener diode, plus any adapters for surface mounting) as well as the procedure.

- Steven

[Buckmaster]

Nice work Steven! Way to go. So....... how difficult is the conversion from v2.3? I am a repeat SP customer ( 2 SP programmers and a TelemetrEZ), but I didn't buy your products to avoid soldering. I was more interested in a clean installation, an easily undoable installation, keeping things a little more tidy inside, ability to easily add to a new radio if the current one frys, future expandability ( with TelemetrEZ), so on and so on. And with the enhancements that use the TelemetrEZ as the interface, the need to occasionally flash the TelemetrEZ is certainly there. Thanks again for all you have done !

[s_mack]

Cool. I'll try to post the procedure later tonight.

[LTMNO]

Hi Steven, I am with BuckMaster... Congrats on doing this... and I would also be interested in perhaps upgrading my board. I remember the dialog and frustration you, Mike and Gohst were having... Great work!

[s_mack]

As I try to organize the steps into something coherent... I start to wonder if maybe instead I should do a run of daughter boards to make the conversion easier? Due to the surface mount design of the chips, its easier said than done otherwise. Plus there's certain economies to be realized by producing them in "bulk" (though the term hardly fits... I'd probably do a run of 50, maybe 100). For example, the DEMUX chip costs some $0.60 if bought single but its only $0.11 in volume (etc, etc, etc.). And then the hard-ish soldering is done. There would still be some soldering that I can't do but I can cut out about 80% of it for you.

I've got a volleyball game to get to now. Later tonight I'll describe the steps and I'd appreciate some feedback as to whether I should just leave people to do the mod themselves or if it makes sense to produce little add-ons.

Stay tuned (but don't wait up... good chance It'll have to wait till tomorrow)

- Steven

[s_mack]

Step 1... buy the components.

• You'll need a 74CB*3257 demultiplexer (referenced as "Demux" from here on). Unfortunately they are all surface mount, so you'll also have to pick up an adapter to solder them to (main reason I'm suggesting perhaps a small run of daughter-boards). Make sure its a 4.5v to 5.5v version (example)
• A common cathode Schottky diode (referenced as "Diode" from here on) is necessary for isolating 9x power from USB power. If you're OK with surface mount, this example is the one I tested with. For through hole, I suspect this one should be fine.
• A 4.7V to 5.0V Zener diode (referenced as "Zener" from here on) to act as a basic voltage regulator. The specifications aren't all that important as the current draw is really very low. This is the one I tested with (again, surface mount). This cheap through-hole one should be good too.
• A 390 Ohm resistor (or close) with a minimum 1/4W rating
• Finally, an inverter is needed to change the high USB signal to low, as per the requirements for the Demux chip. Any inverter (that operates in the 4.5v to 5.5v range anyway) will do. I used this surface mount one. It seems there's no such thing as a single-circuit inverter in through hole design at digikey. You can make your own using a transistor and a couple of resistors... I leave you to your own devices if you wish to go that route (hint).

Step 2... remove the three-pin MOSFET chips labelled FET2, FET3, FET4, and FET5 on the Rev 2.3 SP board.

Step 3... make the connections! In [brackets] I'm giving the pin numbers assuming you're using the surface mount chips I mention above that I tested with. If you're using other components, you'll have to check the datasheets carefully to be sure you're connecting the right things. Note that I give pin NAMES (not numbers) for the ISP connectors (ISP1 and ISP2 are labelled on the SP board. TelemetrEZ has its own ISP as well). They follow the standard pin-out for ISP 10 pin connectors.

• Demux [1] connects to Diode [1] (either one of the annodes) as well as Zener [3] (the cathode) and one end of the Resistor
• Demux [2] connects to MISO of TelemetrEZ's ISP (TZ_ISP).
• Demux [3] connects to MISO of ISP2
• Demux [4] to MISO of ISP1
• Demux [5] to MOSI of TZ_ISP
• Demux [6] to MOSI of ISP2
• Demux [7] to MOSI of ISP1
• Demux [8] to ground (use any one of the 4 ground holes in either ISP1 or ISP2 as convenient for any ground connection)
• Demux [9] to SCK of ISP1
• Demux [10] to SCK of ISP2
• Demux [11] to SCK of TZ_ISP
• Demux [12] is special. It needs to connect to a specific pin on the SP's chip. Refer to image below (click to enlarge) and solder to any of the highlighted spots you fine easiest.

  

  Demux [12] connects to any one of these spots.

• Demux [13] to RESET of ISP2
• Demux [14] to RESET of TZ_ISP
• Demux [15] to Gate [4] (the inverter's output)
• Demux [16] and Gate [5] (Inverter's VCC) both connect to Diode [3] (the cathode)
• Gate [3] to ground (inverter's ground connection)
• Gate [2] (inverter's input) connects to Diode [2] (the other cathode that you haven't already connected) as well as the VCC hole of either ISP1 or ISP2
• Zener [1] (anode) to ground
• Now the only thing that should be left unconnected (other than pins/holes marked n/c) after all the above is the other end of your resistor. This connects to the switched battery voltage. See image below for suggested possibilities (use whichever suits you. I personally think the "A" pin of the LED connector is easiest).

  

  resistor connects to any of these spots

Now wasn't that freakin complex??!! Its probably not as bad as it sounds... but wow. Wouldn't that be a lot easier with a little board? Let's see... here's what those steps would look like:
Step 1... buy the board.
Step 2... remove the three MOSFET chips (same as before)
Step 3... make the connections!

• Solder ISP headers (included) to ISP1 and ISP 2.
• Drop the board onto those two headers (this board would be designed to mirror the exact position of the two headers) and solder (easy)
• Solder a wire... one end to a nice convenient hole on the board. The other end to one of the points in the first picture above
• Solder another wire... one end to another convenient hole on the board. The other end is the same as the second picture above
• Plug an included cable into the board and into TelemetrEZ's ISP existing header

Now how will it look for SP Rev 2.4 future owners? Dead simple:

• Plug an included cable into the board and into TelemetrEZ's ISP existing header

That's it.

I hope the 90 minutes it took me to author the above provides SOME benefit to someone somewhere

- Steven

ps. Note that software hasn't been mentioned at all. This is entirely about the hardware side of things. For software, you'd flash the 9x using eepe or companion or however you do it now. For the TelemetrEZ, you'd follow the steps I outlined back on page 21 or otherwise however you've decided to do it. There won't be any one-software-for-both-jobs (unless someone wants to make eepe/companion do it. It wouldn't be too hard. An updated avrdude.conf and a few lines of code should do it). As I mentioned before... how the hardware "decides" whether its going to program the 9x or TelemetrEZ is simply a matter of having the 9x powered on by battery (for 9x) or not (for tEZ).

[s_mack]

incidentally, if I do make that daughter board (let me know if there's interest!)... the idea would be for it to be very low cost, like $2 to $4 plus shipping.

[Rob Thomson]

Definite interest from me

Sent from my GT-I9300 using Tapatalk 2

[w4kac]

Steven, I would be interested in the daughter board also.

Ken

[Buckmaster]

Holy cow Steven, great write up! Once again it would be simple (!? ) enough to to follow your excellent procedure, but for neatness , if a daughter board became available in that price range I would jump on it! Thanks again!

[mpjf01]

I have had some difficulty fitting my sp board and my Tez and my Frsky Diy in. It is a squeeze and hard to stop wires fouling gymbal movement. How about a version where sp and tez and daughter board all integrated?

Sent from my GT-I9100 using Tapatalk 2