Page 1 of 1

Hitec, we have a problem

PostPosted: Tue May 04, 2010 6:51 am
by RCModelReviews
As you can see, Hitec has a small problem with the Aurora 9

Fortunately it's not a biggie but it would be annoying if it affected your radio (does it?)

However, Hitec isn't the only brand of computer radio to suffer from "dead zone" at the extremes of stick movement.

Why not go check yours now and tell us if you've observed the same problem with your radio. If your radio is affected, tell us what brand/model.

Re: Hitec, we have a problem

PostPosted: Sat May 08, 2010 9:32 pm
by Gordks
I have an E-Fly (Art-Tech) ETC62 2.4 GHz transmitter. It does have dead spots at the top, but not the bottom, of both the throttle and the elevator pots. The servos seem to have the normal travel but it is reached while there is still a little travel left on the stick. Have not yet looked into it to see if there is anything that can be done. Neither aileron nor rudder are effected by this problem.

BTW, it seems like a perfectly functional system for a beginner. The Rx says it is DSSS and they also mention frequency hopping elsewhere. The Tx, Rx and three micro servos all for about $110 US. Digital trim, 12 model memory. The only problem I see is that I can not see anywhere where I can get additional receivers. The shop I got it from lists the Rx, but it is out of stock and has been on order from Art-Tech for months. Also, the manual states that the range is equal to or greater than 350m. I have seen a post that said it was really 800m on the ground and 1200m in the air, but that was on the Art-Tech site so I don't know if I believe it.


Re: Hitec, we have a problem

PostPosted: Sun May 09, 2010 4:06 am
by RCModelReviews
Unless you can rotate the throttle pot very slightly you will have to add a resistor to eliminate the "dead zone" on your stick.

Hitec have confirmed that what I wrote in this article is 100% correct and repairing the Aurora 9 radios with dead-zone *will* require the addition of a small-value resistor.

Re: Hitec, we have a problem

PostPosted: Mon May 10, 2010 2:05 am
by F111 John
It seems to me that there should be a better way than just soldering on a fixed resistor to get rid of this "dead zone." I have a Flysky CT-6A (also sold under the Hobby King, Turborix, and Exceed brands, among others) and it is programmed with a program called "T6Config." There is also a second program, "T6Recalibrate," that does all the mix, endpoint, travel, etc. programming that T6Config does, but it includes another mode to recalibrate the pots.

My day job is as an elecronics technician at a television station, and our video production switcher also has a joystick and several fader bars that can also be recalibrated through a maintenance subroutine. This, I believe, is what these Hitec radios also require, a recalibration.

The proceedure is simple enough. In the recalibrate mode, the joysticks are moved to all of their extreme ranges, up, down, left, right, corner to corner, etc. The microprocessor reads the pot positions through it's Analog to Digital converter (A-D), keeping track of the highest and lowest values read for each pot. When you tell the microprocessor that the calibrate proceedure is finished, i.e. that you have moved the pots to all of their extremes, then the program knows what A-D values represented the extremes, and by inference what the center of travel was as well.

I'm fairly certain that the factory performs this joystick calibration proceedure during manufacture, and that there must be a way for Hitec service to also invoke this recalibrate mode. It may be that it cannot be invoked by simple combinations of key presses during power on, but must be initiated by attaching a computer to a test port internally.

In any case, that would the proper proceedure to eliminate this dead band in radio joysticks.

Re: Hitec, we have a problem

PostPosted: Mon May 10, 2010 2:41 am
by RCModelReviews
The problem is that the ADC will only accept a certain voltage range.

Because the pots don't move the entire 270 degrees of sweep (in fact just a tiny fraction of that) and we want to get the entire 1024 steps (10-bit) resolution, the pots have to be precisely aligned so that at one end of the travel, they're returning the minimum readable value (the ADC returns 0) and at the other end of the travel they're returning the maximum value (1023).

Now, if the pots aren't aligned perfectly then what will happen is that the min or max voltage returned will fall outside the readable range and therefore the ADC will be "pegged" at 0 or 1023 once the voltage goes out of range.

I've spoken with Hitec and they agree that my diagnosis was correct, the fix to the A9 will require the use of a resistor.

Re: Hitec, we have a problem

PostPosted: Mon May 10, 2010 3:50 am
by F111 John
Now you've stumbled upon the best kept secret in RC. Yes, the A-D converter circutry may have 10 bits of resolution, but the joystick pots are tied to +V and Ground, and there are no divider network before the wiper voltage is presented to the microprocessor's A-D converter. You are absolutely correct that the joystick only rotates the pot approximately 90 degrees out of the pot's 270 degree rotation. The microprocessor's A-D codes ground as zero, +V as 1023. Therefore, the actual A-D numbers being used for the extremes of the joystick are nominally 341 to 683, with 512 as the center. If the pot is mounted slightly out of center, then those numbers will shift accordingly. The actual values for each extreme are determined during the joystick calibrate process, and the center value is the midpoint between the two extreme values.

You don't have to take my word for it, you can look for yourself. Here is the schematic from the FCC website as part of their FCC Type Acceptance filing for the FlySky CT-6A. You will note that the joystick pots are tied directly to +5 and Ground, and the wiper is presented directly to one of the 8 A-D converter channels of the STC12C5410AD microprocessor, which has 10 bit A-Ds: ... or_pdf=pdf

Perhaps Hitec does use all 10 bits of the A-D to code just the middle 90 degrees of pot rotation, I don't know. However, I do know for a fact that none of the FlySky products; the 4 channel, 6 channel, or 8(9) channel 9X do on their four joystick pots. (They do, ironically, use all 1024 counts on the VR knobs, since they do rotate the entire 270 degrees, making them more precise than the joysticks!) All of those radios have their schematics on the FCC website, and can be found if you search the FCC ID number at the FCC website.

They may use 10 bit internal calulations to take the raw A-D number from a pot, factor in the digital trim number, factor in the programmed endpoint percentage, factor in exponential rate, then generate a 0 to 1024 count to send to the receiver as a servo position, but the FlySky radios do not use all 1024 counts from joystick min to joystick max.

I think a return phone call to Hitec would be in order, and a schematic would tell all. However, I will bet you a $20 USD donation to your fantastic RC Model Reviews site that they won't release the schematics. I'll double it to $40 USD if they do release the schematics, and it does turn out that they condtion the joystick pot signals so that the 90 degrees of rotation are coded in a full 0 to 1023 count, or something very close.

Edit: Another possibility is that 12 bit A-Ds are very common. If Hitec used 12 bit rather than 10 bit A-Ds, then the middle 90 degrees of joystick pot rotation would code to a minimum count of nominally 1365, a maximum count of 2730, with a 2048 for joystick center. That would give you 1365 counts to code the entire joystick range, more than enough to meet the ultimate 1024 count resolution. However, the same joystick calibration proceedure would be done to determine the actual joystick end counts, which would vary from radio to radio. Any "deadband" near a joystick extreme would be eliminated by a recalibration proceedure.


Re: Hitec, we have a problem

PostPosted: Mon May 10, 2010 5:38 am
by RCModelReviews
Depending on the microcontroller being used it may not be as bad as you suggest.

I'm not too familiar with Atmel's products but with the PIC line of MCs and most others I've used, you can apply two reference voltages (Vref- and Vref+) to two of the otherwise unused ADC pins and they will determine the lowest and highest voltage able to be read by the ADC.

So, if you put 0V on one end of the pot and +5 on the other end then, assuming the pot traversed 30% of its total range (a delta of 1.66V), you could set the Vref- to 1.66V and the Vref+ to 3.33V, which would result in the full 10 bits of ADC resolution being applied to the range between these two voltages (1.66-3.33V) which would also be the full range of stick movement.

The problem is (as witnessed by the Hitec situation), if any of the pots aren't properly centered on 2.5V then, when the wiper of that pot returns a voltage that exceeds Vref+ or is less than Vref-, the ADC output value will not change -- hence the "dead zone".

So you can' t just assume that because there's VSS on one end of the pot and VDD on the other, the ADC is only going to see 1/3 of the total range, it should actually be seeing the whole 10-bit range providing Vref- and Vref+ are set correctly and the pots are all physically calibrated the same.

Re: Hitec, we have a problem

PostPosted: Mon May 10, 2010 11:36 am
by F111 John
Facinating about the two Vrefs. That's why you are such a respected font of knowledge in several fields! Looking as the schematics for the CT6-A, I see no reference voltages from a precision divider on any of the ADC pins, so those radios are most likely using zero and +5 as their references. Which, given their cost and intended application, is perfectly acceptable to me.

Which simply leaves the question why my lowly FlySky CT-6A can run a software routine like T6Recalibrate, have all knob and joystick pots exercised (including the mechanical trim knobs for this particular radio) and the pots are magically reclaibrated with no dead band, while the pricey, capable, and sophisicitated Hitec Aurora 9 requires warranty-voiding surgery and custom trimming resistors installed to recenter it's joysticks.

As I said, the Ross Synergy 3 video production switchers (perhaps for you they're known by the British term 'vision mixer') also has joystick and fader pot recalibrate routines in software. There simply must be a recalibrate routine for the Aurora 9 as well, just perhaps only accessible during factory assembly, leaving the end user with no other option than the surgical one.

Thanks for the education!