Wicked Air Sportz: Turbo Rev

[314159]

Re: A little challenge...

Originally posted by WickedAirSportz
Some of the posts here are dead on the money about how it works, and others are so far off in left field.

would thoes be my posts?

and a reply to everyone that is saying that the picmicro microcontrollers are not state of the art. microcontrollers are reducing circuits to themselves, and input and output components, no external ram/rom/bus logic, they reduce the area and thus the cost of a pcb, making electronics cheaper.

they are powerfull because of their limits. their reduced instruction set allows them to have their instructions and data on the same word of memory. they process the instruction and data in the same processor cycle (except for instructions that write to the program counter). they do not have to wait to retrive information from a disk.

for everyone that is calling them a $4 dollar chip, i can get them for 1.58 in quanity of 1. (if you wish to buy them at $4 a chip, contact me about being your distributor ) http://www.digikey.com/scripts/us/dk...575&Row=241037

i would urge a mod to do some weeding and clean up this thread a little. some people are litterly holding down the wickedairsportz guy while the other one works the gut.

i have heard of a bug that will allow you to dump the code of a pic with the protection bit set. note: you do risk ruining the pic in the process of dooing this. i read this a while ago, so if it has been fixed, or if my memory no longer serves me correctly... it involves raising vcc line of the picmicro to around +14v instead of +5v while reading the pic.

[Butterfingers]

Originally posted by WickedAirSportz
I'll do you better than that... this test was conducted by a respected author for APG (Paul "Doc" Koch), using his Tippman Model 98 with an experimental e-bolt.

Oh yeah... that should be really easy to verify dosent webby write for APG too... err...

[314159]

this might pertain to other pics than the 16c84, you run a good chance of DESTROYING the pic you are trying to read. one of the followup posters to this got 3 fizzles before 1 good read.

> 1. I use the PIC16 programmer from Farnell Components (part no. 459-471).
>
> 2. The standard programming software supplied is ASLPIC from Application
> Solutions Ltd.
>
> 3. Install the 16C84 into a turned pin socket with pin 14 (VDD)
> cut off. Attach a flying lead to the stub of pin 14 and
> connect this to a power supply (0V to +14V) sharing a common
> ground with the programmer.
>
> 4. Run ASLPIC.
> Insert the PIC+socket into the ZIF on the programmer board and switch
> VDD to 5V.
> From the menu set the CP configuration fuse to OFF.
> Now set VDD to VPP-0.5V (approx 13.5 volts).
> Program the configuration fuses. (Reply on screen saying
> error invalid?? Ignore this error and set VDD back to 5V.)
> Switch VDD supply off at the power supply.
> Switch off programmer supply.
> Wait 10 to 20 secs.
> Switch on programmer supply.
> Switch the VDD supply to 5V.
> Read PIC.
>
> What may be confusing to people is the error message displayed
> when programming the configuration fuses, and next not waiting for
> the charge on the cells to fall back to 5 volts after setting the
> fuses. This is why I say switch off for 10 to 20 secs, but don't
> forget to reset the VDD supply to 5 volts first.

[WickedAirSportz]

This bug was found only the first 16C84 PICs, and was immediately corrected by Microchip. In fact, because of this bug, extra security measures were added to make sure something like this would never happen ever again.

By the way, YOU basically have the entire TurboRev logic figured out. 512 bytes (words if you like that phrase better) of code space is plenty of space to handle the algorithm. It was a challenge to get that along with the LED control, battery voltage comparison, push button switch debounce, and a few other misc control routines.

There are no interrupts in the 12C508/509 PICs. You have to simulate one by using the timer.

The TurboRev II will use the latest state of the art PIC micro. How lastest? Well, it gets released this month. It is the same package size (8 pin SOIC), has interrupts, EEPROM, 1K of code space, 3 times the RAM, an A/D converter, and is flashable. Not bad for a part that is $1.12 each.

Some people question that a motor's speed and torque can be controlled via a PIC micro. You will find numerous motor speed control circuits using the 12C50x family parts on the internet. This is where I got the example source code for handling the duty cycle and frequency output to the motor driver.

Since there is an eye, you can "clock" the amount of time between balls passing by the eye. By varying the motor speed, you can easily determine if the current speed is working good or bad, and increasing or decreasing the speed can be compared against a speed that is known to be the fastest feeding. Taking this to another level, you can deliberately stutter the rotation of the paddles with the feeding to determine if you can in fact "smack" the stack of balls to improve the feeding, all by adjusting the torque as well as motor speed. You can easily build patterns that work, and stick with them. When paint stops flowing, the table pointers are cleared and the process starts over. Due to the limited RAM and lack of EEPROM in the 12C508A part, there is no way to permanently store what has been learned for further comparison. With the new PIC part, that will be possible so that the feeding algorithm will function even quicker.

[314159]

Originally posted by WickedAirSportz
This bug was found only the first 16C84 PICs, and was immediately corrected by Microchip. In fact, because of this bug, extra security measures were added to make sure something like this would never happen ever again.

heh, oh well. life goes on i suppose.

[Miscue]

Question: Why does a revy need to be able to learn? Does it play chess too?

[WickedAirSportz]

Do you want the hopper to feed fast? If so, you need to determine the optimal speed/torque for the motor (at the very least).

A very simplified form of hopper AI can be viewed like this:

The time between balls is clocked in, and an average over a period of time is recorded. Now, increasing and decreasing the motor speed is going to affect that average. So, the software can determine what the best possible speed is to allow the balls to flow the quickest.

This method alone works great, but it can not deliberately accelerate the stack of balls, although this does happen on occassion as you see a spike in the timing chart.


Miscue, you and others asked for an explanation (proof) of how the TurboRev technology works. I am providing that to you. If you intend to make smart *** remarks, then I won't waste my time.

[manike]

Originally posted by WickedAirSportz
This method alone works great, but it can not deliberately accelerate the stack of balls, although this does happen on occassion as you see a spike in the timing chart.

Exactly! It works great as an agitating system, which I have said all along, but there is no way with such a paddle system you will be able to synchronise the falling of the paintballs into an empty breech in order to reliably feed faster than any other agitating system.

The chance of a paddle hitting a ball is just too randomn, and the chance of that ball falling into the feed tube at the right time within a few ms, is unreal.

And each ball would have to fall down an empty feed tube to make a difference or be exactly timed to work in conjunction with when the bolt moves (not possible when it's a random paddle rotation and collision with paintball). Since the hopper is not linked to the gun's firing, nor does it know how far away the eye is from the breech it's just a pipe dream to think that your hypothesis on how it works is even plausible.

It's a great agitating hopper, it's not a force feed system and it can not run a gun set at 18bps.

We are still waiting for a video of it running a gun at 18bps full auto reliably... Still waiting for any proof whatsoever on the claims of feed rates you make for it, other than 'word' of mouth Which just doesn't cut it.

manike

[marley618]

This is a little off topic but:I've heard people say you use a Halo B when you are giving a demo of your Timmy Board. Why don't you use a turbo rev instead? It should feed fast enough, right?

[WickedAirSportz]

... there is no way with such a paddle system you will be able to synchronise the falling of the paintballs into an empty breech in order to reliably feed faster than any other agitating system

Wrong!

How can YOU be so bold as to say this? Have you done months worth of research on this technology to determine if it is possible or not... I didn't think so.

This IS entirely possible, and this IS how the TurboRev works. Like it or not... Nothing else is faster in the agitated hopper arena.

Let's break this down into steps: Are saying that it is absolutely impossible to cause the motor to oscillate with the falling of the balls? (a simple yes or no question)

[WickedAirSportz]

Originally posted by marley618
This is a little off topic but:I've heard people say you use a Halo B when you are giving a demo of your Timmy Board. Why don't you use a turbo rev instead? It should feed fast enough, right?

The TurboRev won't feed the 22+ bps that we need for the demos.

[manike]

Originally posted by WickedAirSportz
Wrong!

How can YOU be so bold as to say this? Have you done months worth of research on this technology to determine if it is possible or not... I didn't think so.

I haven't done months of research no... I've done about 8 years of it... That's how long I have been testing and looking at paintball loader systems. I have conducted tests like you say and I have looked at the synchronisation of feeding balls out of a loader to how the bolt cycles. I've done this with guns without ACE and guns with ACE.

via the randomn knocking of balls down a tube I've never achieved over 16bps and certainly not 18bps. By actively feeding them down at specific times with intervals to allow the bolt to cycle in relation to the bolt and firing of the gun I was able to increase the feed rates, but not reliably. And this is with a release system that purposely dropped a ball to be in time with the open breech. Not a randomn paddle rotation as you use.

That's one of the reasons I still think you are full of crap. I'm still waiting to see any theory of yours that isn't absolute codswallop.

Just show us some proof. Where is a video of a Turbo Rev on a gun shooting 18bps? that's all we want to see!

It is not possible. I watched a Turbo rev in a clear hopper and all it does is reliably keep the feed tube full. It has no synchronisation to the opening of the breech or the gun firing. It's obvious from just watching it keep the feed tube full that it can't synchronise the randomn smacking and dropping of paintballs into the breech.

It's easy to make the motor oscillate with the falling of the balls. It's impossible with such a paddle to ensure it knocks a ball down when required.

manike

[hitech]

It looks like Miscue had this figured out from the start.

Originally posted by WickedAirSportz
The TurboRev II will use the latest state of the art PIC micro... Not bad for a part that is $1.12 each.

Originally posted by Miscue
Now, I'm looking at this picture of the TurboRev Board. I see some capacitors, a diode, a switch, some other stuff, and a PIC 12C508A microcontroller. I'm still looking for the "latest state-of-the-art microprocessor with artificial intelligence software," that's supposed to be on there.

A $1.12 "computer" that is capable of artificial intelligence software. I never would have believed...

Originally posted by WickedAirSportz
512 bytes (words if you like that phrase better) of code space is plenty of space to handle the algorithm. It was a challenge to get that along with the LED control, battery voltage comparison, push button switch debounce, and a few other misc control routines.

Originally posted by Miscue
The 12C508A has 33 single word instructions, 512 words of program memory and 25 bytes of data RAM. Out of these 33 instructions, the ability to multiply or divide numbers is not present.

It was a challenge to get the simple routines in there, but not the AI code. I wonder why we haven't seen any "robots" using this "computer"? AI software would be great for "robots". And BTW, a byte is 8 bits, period.

Originally posted by WickedAirSportz
You can easily build patterns that work, and stick with them. When paint stops flowing, the table pointers are cleared and the process starts over. Due to the limited RAM and lack of EEPROM in the 12C508A part, there is no way to permanently store what has been learned for further comparison.

Originally posted by Miscue
All they have is a SINGLE trial in which they mention average feed rates. Seems to me that they produced numbers AFTER the board was made, and did NOTHING before it was made.

Since it can't store anything permanently, and it clears the tables after paint stops flowing the best it can do is adjust to a single string of shots. Artificial intelligence software? Really?

[manike]

Originally posted by WickedAirSportz
Nothing else is faster in the agitated hopper arena.

Agitated? exactly. To get it to feed at the rates you claim it would have to be force feed

BTW I have found that the re-load kit works better than the turbo rev.

manike

[hitech]

Jeez, I gotta learn to type faster...