HontaiTec Quadcopters (HT F801, HT F803,...)

More
11 Jun 2015 20:33 #33872 by Durete
Just captured maximum rudder+trim values.
One file for every side.

Starting at 50% throttle, neutral rudder/trim values, I move first trim to the maximum value. Next I move rudder stick to these side. Next rudder stick at neutral, and finish moving rudder trim to neutral.
Hope you can see something :dry:

dl.dropboxusercontent.com/u/14941708/Rudder%20Max.zip

Please Log in or Create an account to join the conversation.

More
11 Jun 2015 20:45 #33874 by greenfly
Replied by greenfly on topic HontaiTec Quadcopters (HT F801, HT F803,...)
I was not sure you needed these anymore, but here are the captures of the TX with full RUD trims + full RUD stick.

I did notice that full trim and full stick basically cancel each other out when it opposite directions...

ie... Full Left Rudder Trim + Full Right Rudder Stick = No Yaw movement.

However, when in the same direction, the quad would yaw at a moderate rate.
Attachments:

Please Log in or Create an account to join the conversation.

More
11 Jun 2015 20:53 #33875 by Durete
Greenfly, your csv files are empty. ;)

Please Log in or Create an account to join the conversation.

More
11 Jun 2015 21:06 #33876 by greenfly
Replied by greenfly on topic HontaiTec Quadcopters (HT F801, HT F803,...)
Ehhhhhh! My CSN line pulled loose from the analyzer so all the SPI data was not picked up.

Here they are...
Attachments:

Please Log in or Create an account to join the conversation.

More
12 Jun 2015 00:27 - 12 Jun 2015 00:28 #33890 by hexfet
Both the captures show the same thing. The stock tx is sending max and min values for both channel and trim; it's not limiting them. I'm sure we're sending the correct values, which still leaves some problem at the receiver caused by driving the trims. Any other ideas we could test for? As a co-worker famously said, "I can't think of anything I haven't thought of."

Durete wrote: I setting at your code rudder trim packet as -15, 17. With this values works perfect, about the same yaw rate than with stock TX. The rudder drift is slower but no 0 ;) We are missing something...

If the yaw rate is about the same as the stock tx with the driven trim it's probably not worth chasing further - just leave the trims at zero. The purpose of the driven trims is to get the maximum rates possible. Might want to repeat the experiments with pitch or roll to see if it makes any difference there.

If the range is not symmetric (-15,17) then when the stick is centered some trim value will still be sent. That could be why your drift is slower. Is there any yaw drift when you fly with the stock tx?
Last edit: 12 Jun 2015 00:28 by hexfet.

Please Log in or Create an account to join the conversation.

More
12 Jun 2015 02:59 #33899 by greenfly
Replied by greenfly on topic HontaiTec Quadcopters (HT F801, HT F803,...)
For my HT803C, the yaw is really stable with the stock TX. So I'm not sure what is going on.
I've been flying with my Devo 7e and it's really not that bad right now. Let me know if you think a receiver side capture would help.

Please Log in or Create an account to join the conversation.

More
12 Jun 2015 05:56 #33906 by Durete
My yaw is totally perfect with stock TX also. No drift at all.
I will test with pitch/aileron trim packets what is going.

Please Log in or Create an account to join the conversation.

More
12 Jun 2015 12:17 #33915 by hexfet
Yes, it's time for some receiver captures. Both to verify we're sending the correct data and start on the txid data collection.

The data collection is tedious to do by hand so if the pattern is not recognizable with a few samples we can talk about automation. Let's start with a test build that sets the last two bytes of the txid to the value in the fixed id. I've pushed code for that (which also has the rudder driven trim enabled). The process is:
1) Set the txid (fixed id) to be tested
2) Turn on the aircraft and re-init to bind, capturing the spi data for a few seconds
3) If bind is successful, decode the spi data and record the rx address set at the receiver, and the channels it listens on. (or note if bind fails)
4) Increment the fixed id and repeat

Please Log in or Create an account to join the conversation.

More
12 Jun 2015 14:29 #33922 by Durete
Thanks Hexfet :)
Very good explanation, even I understood :silly:
So we need to capture some samples at the receiver board using stock TX (better every channel/function), and later start using latest code version to find successful binding bytes.
I'm right?

@Greenfly, Here is the latest version of the code compiled:
dl.dropboxusercontent.com/u/14941708/dev...n-devo7e-Unknown.zip

I will disassemble my quadcopter to start captures ASAP.

Please Log in or Create an account to join the conversation.

More
12 Jun 2015 15:13 #33923 by hexfet
Yes, you're right. Since we already know the channel and flags in the packets just a couple of captures should be enough to verify. Maybe one with bind and stick movements, and another for the special functions. With both stock tx and deviation.

Look in the repo utils/spi directory for format_spi.pl. It will take the hex csv file and interpret it according to the radio chip (default is nrf24l01). Run it with 'perl format_spi.pl <csv filename>' and it will print to stdout. When we get rx captures it will be easier to explain what to look for in the output.

Please Log in or Create an account to join the conversation.

More
12 Jun 2015 17:01 #33935 by Durete
A quick/easy question to understand completely how works the Fixed ID method. What should be my Fixed ID to be the same as my stock TX?
My stock tx id is 4C 4B 3A F2 04. So my question is, what fixed id should I select to match F2 04 value?

Thanks :oops:

Please Log in or Create an account to join the conversation.

More
12 Jun 2015 17:32 - 12 Jun 2015 17:44 #33939 by greenfly
Replied by greenfly on topic HontaiTec Quadcopters (HT F801, HT F803,...)
Here are my captures for the receiver messages.

Let me know if they look valid. I want to make sure we get everything we need before I put the quad back together.

Attachments:
Last edit: 12 Jun 2015 17:44 by greenfly. Reason: Add Pic

Please Log in or Create an account to join the conversation.

More
12 Jun 2015 17:55 #33941 by hexfet

Durete wrote: A quick/easy question to understand completely how works the Fixed ID method. What should be my Fixed ID to be the same as my stock TX?
My stock tx id is 4C 4B 3A F2 04. So my question is, what fixed id should I select to match F2 04 value?

Thanks :oops:

It's the last two bytes in decimal - 61956

Please Log in or Create an account to join the conversation.

More
12 Jun 2015 18:01 #33942 by greenfly
Replied by greenfly on topic HontaiTec Quadcopters (HT F801, HT F803,...)
Just in case....

Here are captures with the rudder and full trims on the stock tx.
Attachments:

Please Log in or Create an account to join the conversation.

More
12 Jun 2015 18:46 #33946 by hexfet
Seem to be some glitches in the power and bind data. The throttle capture seems fine.

Also I forgot the --full option to the perl script is required to see the miso data, which is important on the rx :)

Please make the sticks and switches captures with the devo bound (use fixed id 31242).

For the stick and switch captures it's easier for me to work with the csv file. I have a pre-deviation script that can filter and collapse repeat packets so it's easier to spot changes (attached). Doesn't help with the bind phase.
Attachments:

Please Log in or Create an account to join the conversation.

More
12 Jun 2015 19:23 #33953 by greenfly
Replied by greenfly on topic HontaiTec Quadcopters (HT F801, HT F803,...)
OK. Here are my captures of the bind again.

Something is not right because I still cannot get the Devo to bind... even with the correct fixed id. I even flashed the previous build by Durete that I thought could bind and it is having the same problem. The lights flash as though it is binding... but go back to blinking instead of remaining solid (bound).
Attachments:

Please Log in or Create an account to join the conversation.

More
12 Jun 2015 20:07 - 12 Jun 2015 20:08 #33960 by greenfly
Replied by greenfly on topic HontaiTec Quadcopters (HT F801, HT F803,...)
@hexfet
I went ahead and captured the stick and switch commands from the stock TX since that is all I can do. Maybe you can use them to compare to the Devo captures when I can get them.

I just zipped up the CSV files, since you said they were better for this.
Attachments:
Last edit: 12 Jun 2015 20:08 by greenfly.

Please Log in or Create an account to join the conversation.

More
12 Jun 2015 20:16 #33962 by hexfet

greenfly wrote: OK. Here are my captures of the bind again.

Something is not right because I still cannot get the Devo to bind... even with the correct fixed id. I even flashed the previous build by Durete that I thought could bind and it is having the same problem. The lights flash as though it is binding... but go back to blinking instead of remaining solid (bound).

There's nothing in the devo captures but reads of the status register, and occasional errors (missing bytes). The rx is not trying to bind. Maybe a bad connection?

The txid from the stock tx is 4c 4b 3a 7a 14, which differs in the last byte from your original captures. Maybe there's only a small set of valid numbers that are used randomly (always hope for easy :) The rf channels are same as before.

Please Log in or Create an account to join the conversation.

More
12 Jun 2015 20:50 #33963 by hexfet

greenfly wrote: @hexfet
I went ahead and captured the stick and switch commands from the stock TX since that is all I can do. Maybe you can use them to compare to the Devo captures when I can get them.

I just zipped up the CSV files, since you said they were better for this.

The majority of the data is correct, but there are a lot of errors too. It's strange - the channel bits are dancing, and the first byte is wrong a couple times, but other bytes always seem correct. Might want to capture at the highest capture rate and look at the spi timing to see if the sample rate is fast enough.

What could explain received packets like this from the flip-left capture?
0B 00 00 20 20 20 20 00 00 00 19 7F
0B 00 00 20 20 40 40 00 00 00 32 EF
0B 00 00 20 40 40 40 00 00 00 32 EF 
0B 00 00 20 20 20 20 00 00 00 19 77
0B 00 00 20 20 20 20 00 00 00 19 77
16 00 00 40 40 40 40 00 00 00 32 EF 
0B 00 00 20 20 20 20 00 00 00 19 77
16 00 00 40 40 40 40 00 00 00 32 EF
0B 00 00 20 20 20 20 00 00 00 19 77
0B 00 00 20 20 20 20 00 00 00 19 77
0B 00 00 40 40 40 40 00 00 00 32 EF
0B 00 00 20 20 20 20 00 00 00 19 77
0B 00 00 20 20 20 20 00 00 00 19 77
0B 00 00 20 20 20 20 00 00 00 19 77
0B 00 00 20 20 20 20 00 00 00 12 EF
0B 00 00 20 20 20 20 00 00 00 32 EF

Please Log in or Create an account to join the conversation.

More
12 Jun 2015 20:51 #33964 by greenfly
Replied by greenfly on topic HontaiTec Quadcopters (HT F801, HT F803,...)

hexfet wrote: The rx is not trying to bind. Maybe a bad connection?


Hexfet, if my throttle captures are good, can I assume that I have wired the analyzer correctly? I just want to confirm the wires are cleanly connected.

Please Log in or Create an account to join the conversation.

Time to create page: 0.091 seconds
Powered by Kunena Forum