Warthog is not 16-bit

[+streakeagle 871]

Despite all the marketing ballyhoo, for the Warthog to be 16-bit resolution, it would have to have 65,536 possible positions (as represented by an integer ranging from 0-65,535 in increments of 1).

 

It would appear that the Warthog is 15-bit + 1:

In either the X or Y axis, you can reach any even value from 0-65,534 and 65,535.

15-bit is 0-32,767, which is the same as using only the even numbered positions of 0-65,535, or 32,768 possible values.

Adding 1 for the one odd value of 65,535 equals 32,769 possible values, which is very much the same as 15-bit rather than the claimed 16-bit resolution.

 

Maybe Warthog marketing read and engineering spec that said 15-bit + 1 and decided that 15+1 = 16-bit?

 

I thought maybe I was crazy. I had noticed this problem years ago, but never researched it. I know that when you first plug in a USB stick, Windows starts out with some default values that can be wrong and need to be reset. Saitek gave specific instructions on how to clear the registry to get Windows to re-learn the min, max, and center of their sticks. So, expecting that maybe there was a driver problem with the Warthog, I "Googled" " Warthog 15-bit resolution" and found people that not only saw the same 15-bit plus 1 values that I observed, but researched the black silicon chip involved in converting the analog to digital. Apparently, the black chip only supports 14-bit and/or 15-bit, not 16-bit. So, Thrustmaster marketing is wrong by a factor of 2.

 

Marketing inflation aside, 15-bit is still nothing to laugh at. Most A-D chipsets are 12-bit or 8-bit. But, despite the sensitivity of the hall sensor and the theoretical 15-bit A-to-D chip, can you really get 15-bit resolution with the Warthog?

 

This whole topic rises to the forefront when you experience Warthog "stiction". That is, the stock grease combined with the platstic ball/plastic socket gimbal makes it almost impossible to move the stick only one count (in this case, 2 counts since the stick reports only even numbered counts over the entire range). i.e. if you are at the physical center (no force applied, ideally 32,768 in both the pitch axis and roll axis), if you try to move just a tiny bit, the value won't change until you have applied so much force that it changes by more than 2 (i.e. less than 32766 or more than 23,770). I have found that if I monitor the raw position values using either the Windows game controller calibration tool or JoyTester2, I can move exactly 2 counts with a lot of patience and practice, but typically the value changes by 8 or more. 8 is 2^3, which is two bits more than you would want with 15-bit precision. So, in practice, the actual precision of the Warthog with "stiction" is 12-bit or 13-bit, little or no better than my home made F-4/B-8 stick using a 12-bit BU0836X USB joystick card. Keep in mind, this is after I added a 15 cm extension to the Warthog that greatly reduced the stiction problem by doubling the lever force.

 

Yet, I can see a very noticeable difference in precision between those two sticks in DCS World. I use noisy pots for my BU0836X which cause spiking and generally prevents achieving smooth, true 12-bit precision. I may not feel stiction with my homemade stick. In fact it feels smooth as butter. But if you graph its roll and pitch values in JoyTester2, it moves all over the place depending on when I last cleaned the pots with alcohol. Recent cleaning gave very smooth, almost noise free 12-bit performance. After a week or two, accurate gunnery was impossible and stick fluctuations were readily visible on screen in the virtual cockpit. This is one of the factors (besides the limited number of hats/switches on a B-8), that made me decide to mount the Warthog in place of my otherwise great home made F-4/B-8 USB stick.

 

So, I see the solution to Warthog stiction is to disassemble the stick, clean off the factory grease, and lightly lube the ball/socket with a better grade of grease, preferably a high-end PTFE damping grease typically used with microscopes and binoculars to provide smooth movement with some resistance (damping). Unfortunately, the recommended grease is hard to come by in the USA without spending some money ($90 for a grease sample kit for testing out several grades of grease. I decided on a cheaper, more readily available substitute with similar properties, though far less damping effects: Super Lube. The Super Lube arrived today. I looked up the procedure for disassembly and discovered that it is a relatively high-risk procedure that may damage some wire connections.

 

I think I will live with stiction a little while longer before performing grease removal/replacement on my Thrustmaster stick.

 

 

[+mue 715]

I also have mixed feelings about the warthog stick. I really like the haptic (metal grip) of the stick and the possibility to add an extension. But the stiction of the stick is quite annoying.

I, too, regreased the stick and I found out that the stiction DON'T come from the ball/socket joint but from the disc riding up and down on the 4 shafts. So if you want to regrease the stick you don't have to do the high-risk procedure of disassembling the ball/socket joint. You only need to regrease the 4 shafts. Still the problem is to find the proper grease. With the grease I use (Tamiya Cera-Grease HG) the stiction returns after some time.

[+streakeagle 871]

I partially disassembled the stick today. Just enough to lift the main socket. I used q-tips to remove as much of the OEM red grease as I could, then used q-tips to jam some of the Super Lube into any place I had been able to reach with the first q-tips.

 

I re-assembled the stick, verified that I had not damaged anything, and found the stiction to be pretty much gone. Aside from the centering detents, it now feels as smooth as my F-4 stick. I would like to add some kind of spring tension so I can feel how far I have pulled the stick back.

 

What I am discovering is that the signal processing that gives the stick such noise free performance makes it almost impossible to move 2 raw counts at a time. You generally have to move the stick more than 2 counts to get the value to change. Once you are moving, you can see the values changing 2 counts at a time if you can move slowly enough.

 

Of course, the amount of movement that it takes to register 2 counts, even with the 15 cm extension, is incredibly small. All that resolution is completely wasted without an extension. I would use an even longer extension, but the range of motion measured with a ruler at the top of the stick matches my real F-4/B-8, which is exactly what I wanted.

 

With the stiction fixed for the moment and the centering detents barely noticeable, I am very happy with the results. I need to save some money to get the KG13 grip for flying Fw190s and Bf109s I saw at SimHq. I will probably get the F/A-18 grip I have seen on DCS forums if it is ever finished and released. I would like the Black Mamba Su-27/MiG-29 style grip as well.