Understanding er9x / Ersky9x

[jhsa]

Hi Guys, I'm starting this thread because people (me included) sometimes find difficult to understand the way the firmware works and how to fully use it's features.
So, this thread is intended to explain and discuss how to work with the firmware, and help us all to understand it better and perhaps make it even easier in the future..

I would like to start by posting a link to the 9XR-PRO manuals (written by Steve and Nigel), that are no more no less than er9x and ersky9x manuals. Er9x might not have all the features but some (The basics) of it also applies to Er9x. You can find the manuals here:

viewtopic.php?f=122&t=5575

I will be updating this post and the next one with information like tutorials, videos, etc that might help us to master the firmware. So, if you have questions on how things work, and/or are not very clear in the manual, please don't hesitate in asking..someone will answer I hope
It will also help the guys that write the manuals
João

[jhsa]

Reserved... Just in case

[jhsa]

Ok, There is something that always bugged me in the open source firmwares, and despite using it sometimes in mixes that I get from the forum, I don't really fully understand it.

Multiplex: Multiply..

I have been playing with it today in eepskye and it doesn't look like multiply is the right name for it?? Or am I completely south??
When I use it, it seems more that we are applying a "Percentage" of some source to the mix above.
For example:

CH1 80% RUD
__* 10% FULL
The result will be that CH1 will travel only 10% of it's weight value (80%), right? -8 to +8

If we replace the 10% in the multiply mix with 125%, the channel will go from -100% to 100%, because 100 is 125% of 80, correct?

The manual says:

"Multiply: Use this to multiply the previous values in the same channel."

This is a bit.. Hhhhmmm... vague??

Would someone please be kind enough to explain it in a way that the common mortal would understand?

Thank you

João

[ReSt]

Several examples of what one might do with multiply mixes were given in the past. ... And I collected most of the tips, that I found in the posts. So I could add here multiply mix examples for:

What are the different Mix options (Quick lesson on the three types, ADD, Multiply and Replace.)
shift a curve up and down by a pot (possible through gvar)
Steering wheel servo thrust depending on throttle position
Scaling a channel with a pot
Adjust throttle idle with a pot
Inverse proportional channels

Reinhard

[jhsa]

Thank you Reinhard, you're very welcome to add documentation or a link to it.. But we're not looking only for examples, but to understand why it works like it does
To me it looks like the name is a bit misleading because I see it applying a percentage to the weight and not some multiplying, or am I wrong and there's more to it?

Thanks

João

[ReSt]

This are the examples and may be they help a little bit more to understand the effect of the multiply mix.

What are the different Mix options
==================================

Quick lesson on the three types, ADD, Multiply and Replace.

An Add mix is added to the result of all mixer lines above it.

A Multiply mix works on the result of all mixer lines above it.

If several replace mixes are active, the last active mix replaces everything above


Generally you use ADD,
Example

CH4 100% RUD
+ 10% AIL Switch(RUD)

would move the rudder servo on ch4 with the rudder stick and mix in 10% from the Aileron stick when the Rudder switch is on.

Change it to

CH4 100% RUD
R 100% AIL Switch(RUD)
This would control the rudder with the rudder stick only with the rudder switch off. With the switch on the rudder stick would do nothing, but the aileron stick would move the rudder.

Change it to

CH4 100% AIL
* 5% P1 Offset 100
+ 100% RUD

And ch4 would move with the rudder stick, and by adjusting the pot you would have variable control of the amount of the aileron to rudder mixing.

Pat MacKenzie

shift a curve up and down by a pot (possible through gvar)
==========================================================


Set up another curve (e.g. C2) 0, 100, 100, 100, 0.
On an unused channel e.g. 16

CH16: +100% THR Curve(c2) (or whichever curve you just set up)
CH16: * 50% P1 Offset(100%) (MULTIPLY multiplex NOT ADD)

now on you original throttle channel add:
CH3: +10% CH16

This will add 0 to 10% to your throttle output controlled by P1, in the middle range, but the addition will be steadily reduced as you go towards the ends of the throttle setting.

Mike.

Steering wheel servo thrust depending on throttle position
==========================================================

I have a Cessna 180 200mm with a independent Nose Wheel stering servo. So now i want to do a mix in combination with throttle.
(Same like cars have for Steering and Speed).

So, as more throttle i have, as slower (weight) i want to have reaction to my Nose Wheel steering servo.

For Example:

Throttle Steering Wheel Servo
0% 100%
10% 90%
20% 80%
30% 70%

and so on...

by MikeB » Mon Sep 30, 2013 10:23 am

Add a mix line to your steering servo output like (note a MULTIPLY multiplex mix):

* -50% THR Offset(-100%)

Mike.

Scaling a channel with a pot
============================

For a scaling operation you need to use an offset of +100 and and weight of 50%. THe offset changes -100 to + 100 into 0 to +200, then the weight scales it back to 0 to 100.

Using the same +100% mix on one channel is the same as having a single mix with a 5 point curve set to -100, -100, 0, 100, 100.

An intersting way of increasing the rate of response is:
+100% AIL
* +50% P1 offset 100 (multiply mix)
+100% AIL

Now, if the pot P1 is fully one way you get the normal movement, and if the pot is fully the other way you the same as having the +100% AIL in twice. But you can now choose, while flying, the amount of increased movement by turning P1.

Mike.

Adjust throttle idle with a pot
===============================

Help with pot for idle adjust and logic definitions


OK, the T-TRIM effect may be obtained, but using a pot for the trim, like this:
CH03: -50% THR Offset (-100)
CH03: * +20%P1 Offset(100%) -- MULTIPLY multiplex
CH03: +100% THR

This allows P1 to adjust the throttle at the low end, but does not affect to top end.

Mike.

Inverse proportional channels
=============================

Is is possible to have an inversely proportional mix between throttle and ailerons? ie idle throttle having 90-100% aileron movement 100% throttle having 10-20% aileron movement.

Aileron channel:
CH01: +100% Ail
* -40% THR Offset(60%) Late Offset (MULTIPLY multiplex)

Mike.

Reinhard

[MikeB]

Its multiply because we are multiplying percentages!
80% is 80/100
60% is 60/100
So if they are multiplied we get
(80*60)/(100*100) = 4800/10000 = 48/100 = 48%.

Mike.

[pmullen503]

An updated, detailed flow chart would be nice. Sometimes it's not clear in the stream from input to output where one operation takes place relative to another.

[jhsa]

Ok, for the people like me that aren't very good at math..
Could we think about it as I wrote it above? That means that when we apply a multiply mix, we are applying a percentage to the resulting weight of the mixes above it?

João

[MikeB]

Yes!

Mike.

[jhsa]

Thanks Mike. I think we got to understand the Multiply mix a little bit better. Now the manual could be updated

João

[Daedalus66]

Thanks Mike. I think we got to understand the Multiply mix a little bit better. Now the manual could be updated

João

Will take another look at the 9XR Pro manual Ersky9x Explained in light of this discussion.
Nigel

[jhsa]

Thank you Nigel.. This is something we NEED to understand in order to program the mixer.. This is the basis of programming the mixer, so it has to be very well explained in my opinion.. Please find attached to the first post a .Doc file with all the examples Reinhard posted above.
Also I think that here I found an easier way to explain it?

When I use it, it seems more that we are applying a "Percentage" of some source to the mix above.
For example:

CH1 80% RUD
__* 10% FULL
The result will be that CH1 will travel only 10% of it's weight value (80%), right? -8 to +8

If we replace the 10% in the multiply mix with 125%, the channel will go from -100% to 100%, because 100 is 125% of 80, correct?

The manual says:

"Multiply: Use this to multiply the previous values in the same channel."

This is a bit.. Hhhhmmm... vague??

Would someone please be kind enough to explain it in a way that the common mortal would understand?

Thank you

João

and alsoa bit techical but a good explanation why it is a Multiply:

Its multiply because we are multiplying percentages!
80% is 80/100
60% is 60/100
So if they are multiplied we get
(80*60)/(100*100) = 4800/10000 = 48/100 = 48%.

Mike.

And still:

Ok, for the people like me that aren't very good at math..
Could we think about it as I wrote it above? That means that when we apply a multiply mix, we are applying a percentage to the resulting weight of the mixes above it?

João

And Mike's answer:

Yes!

Mike.

I hope this info helps you to find some easy words for the manual

Thank so much for taking your time to write it.. You and Steve..

João

EDIT: Ahh, and including those examples in the manual would also help.. It'sjust cdopy and paste mainly.. You could included after the Mix multiplex chapter as it includes examples for ADD, MULTIPLY and REPLACE..

[jhsa]

I've been playing with offset and late offset. I can understand the late offset quite well now, but can't really understand the normal offset
I understand that by applying 100% offset to a mix we shift it by half of the complete travel. That is it will now work from 0% to 200% if that was physically possible. Now if we apply 50% of weight, it will move from 0 to 100%, that I can also understand. But if I start inserting other values for the offset and weight the values I get just don't make sense to me..
Could someone please be kind enough to explain this to us, as I'm sure there are many more of us that are in the same situation? Again, the manuals, and therefore the firmware and all of us would benefit from an easy to understand explanation. I do thank you in advance.

The late offset is easier to understand. The way I think of it is:

The center of the travel is shifted by the value of the offset, and then we have the value of the weight to each side of the travel. So for example. If we want an offset of +25% with a weight of 75% we get a travel from:
Shifted center position +- 75%.
That means:
+25% - 75% = -50%
to
+25% + 75% = +100%

So, we have the center of the travel at +25% and 75% (weight) to each side of the travel. Is my thinking correct here?
I think this is so much easier to understand because you get the real offset and the real weight. Do we really need the earlier offset still?

Thank you for your answers.

João

[MikeB]

I agree that using the offset, rather than the "late offset" is not really needed, however, to avoid existing mixes changing I'm leaving it in.
Maybe I need to turn "late offset" on when you create a new mix (edited the source file so that will happen in the next version).

The order of processing of a mix is:
1. Obtain the source value, with expo/dual rates applied to sticks.
2. Add offset (if "late offset" is not enabled).
3. Delay and Slow.
4. Curve/differential.
5. Weight.
6. "late offset" if enabled.

With "late offset" NOT enabled, then the offset value is affected by curve/differential and weight which is, as you are finding, quite confusing. With a curve or differential applied the resulting offset value is almost unpredictable.
Quite simply, any offset value you enter is scaled by the weight (and curve/diff) if you don't enable "late offset".

Mike.

[jhsa]

Thank you Mike, that is exactly what I thought, that means unless there is any special situation I would want to use the normal offset instead of the late offset, I won't give it much thought anymore and even recommend others to use the late offset instead. Meanwhile your "order of processing of a mix" would be a good addition to the manuals.

The order of processing of a mix is:
1. Obtain the source value, with expo/dual rates applied to sticks.
2. Add offset (if "late offset" is not enabled).
3. Delay and Slow.
4. Curve/differential.
5. Weight.
6. "late offset" if enabled.

I don't know if Nigel or Steve have been taking notes from this thread but I will do it also just in case. There is already some excellent information here. And more to come for sure

Thank you

João

[jhsa]

Maybe I need to turn "late offset" on when you create a new mix (edited the source file so that will happen in the next version).

I think that is a good idea. so we won't even notice the old offset. Eepe and eepskye would have to be updated to avoid confusion I guess?

João

EDIT: I've been playing with the late offset together with some multiply mixes and slow, and woow, that is exactly how a dumb brain like mine would expect things to work.. Very nice.
One more step in the right direction towards a more user friendly, easy to work with firmware, but still keeping all the power of it.. ME LIKE IT..

[jhsa]

The order of processing of a mix is:
1. Obtain the source value, with expo/dual rates applied to sticks.
2. Add offset (if "late offset" is not enabled).
3. Delay and Slow.
4. Curve/differential.
5. Weight.
6. "late offset" if enabled.

So Mike, what comes next on the list just for completeness?
I know these two don't belong to the Mixer, they are applied after, but....

7. Safety Switches??
8. Limits??

Thanks

João

[MikeB]

The order of processing of a mix is:
1. Obtain the source value, with expo/dual rates applied to sticks.
2. Add offset (if "late offset" is not enabled).
3. Delay and Slow.
4. Curve/differential.
5. Weight.
6. "late offset" if enabled.
After all mixes:
7. Apply Sub-trim.
8. Apply Limits.
9. Apply Safety Switches.

Just changed eepe and eepskye to set "late offset" when creating a new mix (including the default simple 4-channel template when creating a new model).

Mike.

[jhsa]

Thanks Mike. Very useful information.

João

[jhsa]

I hope it helps to understand the mixer a little better.. The video is a bit long. There is an index in the description.

https://www.youtube.com/watch?v=Vn07XDdesmQ

João

[jhsa]

Ok, got a challenge here
Here is what I'm trying to do..
I have a voltage sensor connected to my receiver's A1 telemetry analog port.
I need that my radio reduce the throttle proportionally if the voltage goes under 10V when I throttle up. But only enough to get it above 10V, and not cut the throttle completely
I know that I must use scalers,and I know I must ADD the result to the throttle channel. But I don't know how to calculate the dividing and multiplying thing.. Some help would be wonderful..
Let's see if the scalers are up to the task they were designed for Joking

João

[MikeB]

That sort of operation is not actually straightforward. You actually have a feedback circuit involved. If the voltage is under 10V, then you need to reduce the throttle, but you then need to continually monitor the 10V and reduce the throttle some more if it stays below 10V, but allow the throttle to increase if it goes back above 10V. There will also be a time delay between adjusting the throttle and seeing what the response to the voltage is. The delay can lead to an oscillation in the control operation.

I think, with your example, there is an added complication where A1 (and A2) used as input to a scaler have the RAW A2D value sent by the receiver, rather than the value converted by the range setting.
You can check this on ersky9x by setting a scaler to use A1 as the input, then looking at the value on the top line. I'll need to check this before I can fully answer your question.

Mike.

[MikeB]

I can confirm that using A1 and A2 as inputs to scalers results in the raw A2D value being used as the source value. It could be argued that this is not correct, and they should be scaled first, thinking about that now!

Mike.

[jhsa]

That sort of operation is not actually straightforward. You actually have a feedback circuit involved. If the voltage is under 10V, then you need to reduce the throttle, but you then need to continually monitor the 10V and reduce the throttle some more if it stays below 10V, but allow the throttle to increase if it goes back above 10V.

Yes, that 's the idea

There will also be a time delay between adjusting the throttle and seeing what the response to the voltage is. The delay can lead to an oscillation in the control operation.

It could. Isn't there a way to compensate for the delay? some kind of sensitivity adjustment?

I think, with your example, there is an added complication where A1 (and A2) used as input to a scaler have the RAW A2D value sent by the receiver, rather than the value converted by the range setting.
You can check this on ersky9x by setting a scaler to use A1 as the input, then looking at the value on the top line. I'll need to check this before I can fully answer your question.

Mike.

Thank you

João

[MikeB]

I'm looking into using the set ratio to scale A1 and A2 before they are used as an input to a scaler. While looking into this, I think I can see some rationalisation of the use of these ratios that may well save some flash, particularly on er9x. This will, however, take a little longer to sort out.

Mike.

[jhsa]

Thanks Mike. So as far as I understand for now I can forget about using the scaler in the mixer as it won't probably work the way I wanted.. I know that some ESCs can do it.. it would be a super killer feature if we could use the telemetry to control the mixer.

João

[jhsa]

Ok, is there a way to scale it this way?
A range from 10V to 9V, scaled to a range from 200% to 0% (-100 to 100 = 200% of movement)

This might not be the right way to go but I would like to try it anyway. I don't like to believe something is impossible before proving it impossible
My problem to achieve what I want to do above is that I haven't a clue about how to work the math to get those values.. But I wil learn as I learnt other stuff about the fw This is just a bit more complicated.
And I would like to find some good applications for this feature and create some good examples. As far as I remember from reading on this forum a while ago, the main reason they were implemented was that they could be used in the mixer and control it.. And for that we will always have some kind of feedback circuit in my opinion.
The custom switches can also work using a feedback circuit. the difference is that they just switch, trigger an alarm, or even control the mixer by turning it on or off, like on my glider's automatic motor control, where the motor stops automagically at 70 meters of height, and starts again when going below 20 meters, to just stop again at 70m. I don't touch the throttle at all. The scalers are meant to work in a proportional way?

Thanks..

João

[Kilrah]

To properly implement what you want would need a PI controller... The P can be done with what you're discussing, but the I part needs a decent time reference and a way to accumulate past values and is not really possible using the mixer system. Maybe a dirty hack is possible with gvars, but... Likely not precise/clean and/or safe.

Sent via mobile

[jhsa]

I'm not giving up
I will try delay and slow also to avoid oscillations. We also have a one shot timer. That might help. I have some ideas..
I can't see where it won't be safe.. I have the batetry alarms anyway. everything will be on a switch for testing, and even if it is working.. One thing I know, if were not crazy ideas like this one, and if we didn't try them, we didn't have half of the features we have now with the firmware. Voice, haptic, GVARs, etc..
So, I'm going for it and will give all I can.. If a fricking ESC with an atmega8 inside can do it, a radio with an ARM chip inside must be able to do it as well, right??

Anyway, I have been playing with the mixer:

CH03 -50% Thr Late Offset (-50%)
___* -100% P1 Late Offset (100%)
____+100% THR

P1 will be replaced by the scaller:
I would take this a starting point and deal with oscillations later

João