The PANTHER platform!

[Gandolf]

It's getting that time of year where I start to get that "wild hair" to do something 'new'!! 

I drive an 88 Ford Crown Vic.  I was trying convince my 14yo niece that someday I might give it to her    She could not "see" the tires on it are far better than the cops have.  She could not "see", the $400 MP3 Stereo in it.  She could not "see" the mustang engine...  It gets 18 city 27 highway...  You can ram it into a mountain at 40 MPH and walk away.  It can get back ended at a 65 MPH delta and you will survive... 

http://en.wikipedia.org/wiki/Ford_Panther_platform

All she could see is that is was not "new looking"...

But the "concept" is very 'technically' clear!  A rock solid "frame" you can "stick anything" onto!  In her youth, she could not see what was needed far before she was born.  What it really takes to "make the machine"...

So I predict that airsoft needs the same "very basic platform"...  Nothing fancy in itself.   But a "power plant".  A "work horse" that can take any load and deliver!!  Where all that "motor driver" s*** is taken care of...

Probably based on this MOSFET for both direct and breaking currents:

http://www.irf.com/product-info/datasheets/data/irf1324s-7ppbf.pdf

That's a GOOD FET    We will just run it in the high drive configuration like everyone else.  That P-FET stuff is getting TOO old...

Lots of folks can make uC controls now to do wonderful things.  But the world of the insane voltage spikes, currents, magnum motors...  Eakkk   

So I am thinking of an H-bridge uC controlled FET configuration.  Good to go "stand alone" with intelligent drive and can adapt to the situation in stride.  Intelligent current control, speed control, and such all "figured out" by the uC brain already "on board"  But plenty of local uC space left.  It can be driven by a BASIC STAMP and all after that to run your gun to your bidding 

A machine really made to do the bidding of 'other uC machines'...  Data logging and failure reporting too!  But hopefully the engine will disregard commands that might do harm, and report back to the "master" why this is not a good idea"...

I note that many can program, but the need for that controller to take those commands without busting the gun all up is great 

Gandolf

[utc_pyro]

If you make this, I'll be buying one. I've got code ready on two different platforms to control an AEG, but all of my DIY drive circuits kept dieing, leading me to buy SW-Evolutions.

Even if you could design a small cost effective platform that could respond to +5/Ground logic levels while eating up the insane current and voltage spikes associated with some AEG's (my m4 comes to mind), it could be useful to those designing their own controllers.

What type of interfaces are you planning on using between this new drive platform and host controllers by the way? I know with a basic stamp and some other controlers it's easy to do serial commands, but on a PSoC it's easier to write to an I2C slave.

[Gandolf]

The design is really up in the air, but it will be able to be driven by any logic.  If it has it's own CPU it will also be able to supply 5V power.  The FETs are driven by real gate drive ICs now.  It might be very modular going from just the drive FETs and basic support circuit, a logic controlled one that has over current protection and a few basic tricks, and someday perhaps its own full blown, ah, make that 'power' CPU controller. 

The size might be much better too since the smaller parts can be separated and stuffed around the gun where there is room.  I'll try to get some rough schematics up soon.

Gandolf

[Gandolf]

Hi,

Here is the schematic for the basic engine:



http://extreme-fire.com/Panther/out.jpeg

The battery and motor on the left are just shown for reference.

There are now 18V 1500W TVSs across each FET.  BIG TVSs that can eat 20,000 watt spikes    One 'pico' second response in case the normal anti-parallel diodes are too slow which might be a factor on other designs.

The IRF4905S is a 55V 20mOhm P-FET.  A nice strong FET used on the SW-AB-LONG.  Nice low RDSon for low heating and that high voltage makes it pretty hard to spike even without any protection.

The IRF1324S is a MONSTER MOSFET!!!  Only 24V so have to be real careful of spikes, but it can pump 1640 amps peak and 430 amps continuous!!  Only 0.8mOhms.  On a 5W heat sink, it can pump an astounding 80 amps RMS!  Sort the the "answer to Ringsted" MOSFET    Once I laid eyes upon the IRF1324, I knew I must someday build a machine around it 

http://www.irf.com/product-info/datasheets/data/irf1324s-7ppbf.pdf

But now...  We must move up to "big" gate drive voltages!!!  Thus we step up to FET driver ICs...

The usual gate drive limiting resistors give a switching time of about 1.5uS.  7700nF of gate capacitance!  But in our very low frequency case, we don't care at all...    Just have to think for a short time about the power dissipation in the gate resistors, but it is 8mW in the 4000Hz speed control case.  I jumped up to the 1/4W 1206 size resistors just to aid in assembly though.  I do everything at 1% now too just to make testing and all more precise...  The added cost there is nothing.

A small poly resettable fuse, 1N5819 blocking diode and a 150uF 16V low ESR cap runs the show through voltage droops.  It all draws only 1.5uA so it will not run the battery down over time.  Note, for sure, that the poly fuse has some significant "resistance".  That limits the inrush current through the diode and cap to safe limits...

Nothing too special about the FET driver.  A nice performer that should be free of problems.  There are dozens that do the exact some thing.  Ground turns it off and almost any voltage from anything will control it.  High impedance inputs so have to watch that.

The main drive FET is pulled low (off) by the 15K resistor, but all other control is external. All the power and signals needs for external control are available.

Connecting +V and BREAK and a say 220 ohm load resistor between DRIVE and GND makes it into a Standard MOSFET configuration (probably a zener diode for low voltage lock out too).  Adding the break control/charging circuit of say the SW-AB-LONG turns it into a breaking MOSFET.  Something like a basic stamp that has it's own voltage regulator would probably plug right in.  All the hooks there to off any uC are easy.  The DRAIN can do current sensing and out of saturation detection.  100 amps of drive only gives an 80mV signal 

So that is at least the "idea".  Nothing is tested but it is pulling off all that I know to make a very basic engine using best of breed parts and technology.  The power components are simply the most powerful available so not much rocket science there    No current limiting unless a uC or something is monitoring the current.  An external poly fuse is easy and a super good idea if you are playing with the control signals.  This thing could turn your battery and wires into a fuse real fast!  If something goes wrong, it won't just automatically turn off!

It's no secret that this technology could very easily end up in the Evolution    It only adds two more parts and the top of the board is wide open    Only the RDSon coefficient would change in the software.    But the max limit too...  Hmmmm...  like 1000 amps amps or something...  Have to think about that 

This is all Open Source / Public Domain so feel free to copy, steal, and plagiarize at your pleasure.  But help out with ideas too!

Gandolf

[Sinewave]

That looks really interesting!  I'd love to plug something into a platform like this.

I really do like the idea of separating the "work" platform from the "brains".

[utc_pyro]

Wow, now that's a hardened fet driver! Hopefully that will handle any systema magnum +++ they might come out with in the future.

One thing to also consider if you want to aim this at people using "beginners" controllers: Most don't have a capability or are fast enough to do PWM outputs while doing other things or analog voltage inputs needed to handle testing the current. I know both of these are true for the Basic Stamp line that many start with, and many of it's clones. The basic stamps have little to no analog voltage reading capability, and the PWM function does not output a nice steady square wave and thus can lead to some strange things when driving Mosfets with high gate capacitance (like them going ohmic, even if going through a fet driver).

So, in addition to the actual driver here, you might want to also develop a "Panther ECU" co-processor that just does the bidding of the user developed controller. Essentially turning into an SW-Evolution, but with a serial connection instead of a trigger and AUX ports.

I'm not in any way saying that should be incorporated into every one of these, but as possibly an advanced version (think SW-Computer and SW-Evolution). The beauty of this current setup is that it can be used for everything from an SW-AB to an output for a high end micro controller, so they could possibly be made in higher volumes decreasing your unit cost.

You can think of it like this: Currently this is your basic carburetor engine. In order to reach more potential developers, you might want to make an optional modern electronically control unit for it with a CAN bus people can plug into and pass commands.

With a fairly fast data rate it wont add to any delay, and it can free up the main user developed controller to do other functions like monitor "eyes" in the hopup, additional user inputs, and talk to other controllers like ones monitoring the battery.

Oh, and an interesting note... My first car was a '89 Grand Marquis, so I'm familure with the platform that you're naming this after.

[Gandolf]

Hi,

I have often thought about a dual processor.  The control processor could run the Panther and would do all that A/D, PWM, current/power monitoring, and safety stuff.  Another processor would then be free to do the higher level things without having to worry about that.  The control processor board would also supply the +5 volts.  The SW-COMPUTER and EVOLUTION are running the processors very hard.  It would be very nice to split that load between two processors.

Of course that is a "future" thing, but I think it is a very good idea.

Gandolf 

[Bench]

It would be very nice to split that load between two processors.

Multi-processors... I like it! 

[Gandolf]

Hi,

Here is what the added PCU might look like:



http://extreme-fire.com/Panther/out1.jpeg

Pretty basic and the pinouts and such are almost identical to the SW-Computers.

A 15k resistor pulls up the break line to the +V.  And a conventional +5.00V power supply.

The +Battery and Drain voltage are divided and filtered and fed to the Attiny 85.  That allows high speed battery voltage and current/saturation monitoring.  The computer has the time now to check for current in the breaking FET too.  There are many tricks this allows like looking for a weak battery, calculating actual motor power and the actual power being dissipated in the FETs.  That allows a very high level of protection for all.  If on the same board, the uC can also watch for actual FET temperature.

The inputs can be hooked to anything.  For the simplest operation just to the trigger for control or active breaking.  In a more complex case to another uC to do all kinds of fancy stuff.

The PCU can monitor everything the drive and gun is actually doing.  All the currents, voltages, powers, even energies are now known.  It still would have vibration error warnings (pretty sophisticated now) and the black box functions.  It would basically take ON/OFF commands but it would do everything possible to prevent errors or damage.  Much of the software is just cut and paste from what the SW-Computers use now. 

The PCU really would not have any external programming.  It would self adapt and just figure everything out.  As long as the FET current, power and temperatures are safe, it does not care.  It would calculate the same for the motor and determine the difference between a Magnum motor and a flaming arc    It would monitor that continually now in case a happily running motor suddenly jams.  It would look for high battery droop or poor recovery indicating a discharged pack as well as just low voltage.  It can actually track the long term voltage/ current profile of a battery and figure out the mAh rating if that would be useful to it.  As long as the available voltage can keep the FET drive up to supply the needed current it would be good to go.  A weenie gun running 50 amps might only need 5.5V of battery, while a monster might need to maintain 8V and it would know that.

What it would not have are the higher functions like 3-round burst, pre-cocking, semi-only, and all that messy stuff.  That would be entirely handled by a second CPU.  So while this CPU is constantly racing in a loop checking all kinds of numbers and conditions, another CPU would be totally freed up to do smart stuff.  That could be really anything from a basic stamp on up.

So that's the idea this moment. 

Gandolf

[Gandolf]

Hi again,

I actually tried the circuit today and I am VERY pleased! 

Here is a picture of the 'finely' built prototype (file size is sort of big so I will just link them):

http://extreme-fire.com/Panther/PantherProto.JPG

No twisted wires or good connections there!  I run 60 amp pulses through the proto-board and you can watch the noise increase as the little connectors burn on the scope 

Extreme-Fire's work area is an engineering area tonight 

http://extreme-fire.com/Panther/P1010013.JPG

The only thing I would change on the drive schematic is to add a 15k pullup resistor for the Break too.  I sketched out a bunch of drive circuits and they all need that anyway.  That also makes the bare drive circuit "fully stable" as is.

It works beautifully!!!  Everything is stone cold so far.  Even though I messed up the original wiring greatly and let a lot of smoke out of a few things like a TVS, nothing broke!  I even got that BIG FET super hot but it just "liked" that!  After I finally managed to get all the wires right it just did what it was supposed to do.  I could not get anything hot with the testing so far and I could not break anything.  Not even close!

I started out kicking some big inductive spikes around at various voltages up to 18V with a "little" inductor I have 

http://extreme-fire.com/Panther/Coil.JPG

Just a 170mH dead short like the Systema motors, but this one won't break    You can easily hear the current "clicking" it.  No weenie AEG has THAT much inductive wiring in it!!!

That baby could hold those TVS on for like 15uS with 300mJ spikes!!  Those monster TVS didn't even get warm...  The voltage profile during conduction showed they were under no stress at all!!

One thing I did note is that the monster FET's breakdown is 26V.  It is rated at 24V so there is NO headroom there.  Here it is running without the TVS into a fairly nasty spike (just ~10k gif files) :



It ate it right up with no problem, but it shows that the TVS voltage does indeed have to be pretty low.  The blue line is current coming off 60 amps into that coil at 13.1V battery 

With the TVS in place and cranking a really nasty 18V battery spike!



A normal FET would die there.  Our big say IRL1404Z'z would "probably" live.  The monster IRF1324 would not laugh at it, but should not care much.  I have seen enough breaking FETs with holes blown through them to know what they would do    But the big TVSs seem to love this stuff!!!  The mighty IRF4905 with tons of gate drive is the most resistant breaking FET I have.  "NONE" have ever failed (wildly knocking wood and throwing salt...).

So early testing is all super good!!!

There was one "OS!" thing the scope caught as some connections were going bad...



A 35V (!!!!) 2nS spike caused by lead inductance into a bad connection...  Low energy and very fast.  But crap like that can punch holes in FETs that fail sometime...  That will pass through all normal protection circuits due to its super high speed...  But my new giant TVSs are still 2000X faster according to the quantum guys (nobody can really 'measure' things that fast  ).  But that scope shot is a warning to keep all connections very solid with high current electronics!

I did a rough cost:

The basic driver (here) seems to run at about $45.  The PCU is about $30.  A second CPU thing about $25.  My cost model is 3X parts +$5 shipping...  Not cheap, but it pays for the "stuff" and keeps things going.  It's all open source so as always, if you can do it cheaper please do!!  I seem to need a little more engineering time, rather than building and shipping time these days 

So next I will try to get a PC board for it.  Maybe try to stuff the PCU on that panel too to save R&D cost...  A lot of time there so things will slow down now.  But when it all comes together, it will strike like a... Panther 

So I end with a little dull video of the driver beating up a motor this night...

http://www.youtube.com/watch?v=PPZ6EpAncIs

Mostly trying to bust the breaking FET into an open loaded motor.  The FET did not notice.  None of the FETs where on any heat sinks "yet" either 

Gandolf

[Gandolf]

Hi,

Not anything really new    I have been mostly goofing off and sending MOSFETs out 

But I think these will be a "SUPER COOL" thing for the 'Panther Platform'. 

http://dkc3.digikey.com/PDF/T082/P0316.pdf

These wire screw down connectors are industry standard.  They are sort of big, but they can solve many problems for us!!!  No need for all those extra connectors or soldering with them.  You just screw clamp the wires down now and take them on and off at your whim.  You might have to get a set of "small" screw drivers, but this seems like a giant help to those that were not born with a soldering iron in their hand like "I" was    I note that kids now days watch Bugs Bunny cartoons on their moms I-Phones in the line at the post office....  A loooong story... 

I really like the Fig. 3 version since it has dual solder connections and looks real tough while still not being too big.  I am thinking these part numbers:

ED1516-ND
ED1518-ND

Might be good for low power connections.

And the ED1556-ND might be real good for motor and battery connections.  Hopefully there would be room to stuff a full soldered PC board hole in there too for #14 wire if the connector just did not cut it for the super power folks.  Probably a full big lap solder pad in there too...

Maybe I will put the Dean's connectors out of business    But after seeing many soldering "jobs" by so many of my dear customers  , these simple screw down connectors seem like a real life saver!!

This also simply "eliminates" the need for "pre-wired" MOSFETs!!  You can use you own existing wire, or your fancy silicon wire as you please.  The MOSFET does not care about all those other wires now!  And you don't have to worry about how to make the connections!

The down side is that these things take up more room.  But I feel that most folks have the little extra room needed given the advantages (or they will carve it out  ).

Certainly the early versions of the Panther Platform will use these.  It makes it just too easy to play around with!

Gandolf

[Gandolf]

Hi,

I have most of the parts and have been playing with stuffing them on prototype PC Boards.  Here is pretty close to the layout.



Pretty big but nice for a prototype where I may have to cut and paste.  Everything is easy to get to and mess with too.

The computer controller for the monitor looks like this:



Also giant in size but all the parts are on top and such.  That same board can also be used in the dual CPU system.  Of course any computer chip or whatever could be the master controller.  But this allows me to play using the existing SW-COMPUTER software almost as is.

It has been going sort of slow but all is looking very good!

Gandolf

[utc_pyro]

Do you have any sort of reference on the actual size of these boards?

[Gandolf]

Hi,

I got a ton done today on this 

Here is the final prototype drive board:

http://extreme-fire.com/Panther/Panther-105-1.gif

That is roughly 1.8 x 1.2 x 0.6 inches.  Big, but it is a prototype made to test stuff.  The connectors are the BIG parts, but they make connecting stuff up SUPER easy!

The Schematic is here:

http://extreme-fire.com/Panther/Panther-DriveUnit-105.jpeg

http://extreme-fire.com/Panther/Panther-DriveUnit-105.sch

Note that I have gone to the Fairchild. FAN3227T now since it has hysteresis on the inputs for simple drive, no metal back that would short things, and is very inexpensive    But the IXDN404SIA or IR4427S would drop in there too.

The CPU controller is here:

http://extreme-fire.com/Panther/Panther-105-2.gif

http://extreme-fire.com/Panther/Panther-MonitorControl-105.jpeg

http://extreme-fire.com/Panther/Panther-MonitorControl-105.sch

Very versatile!  It can be easily loaded with SW-COMPUTER 3.2 software.  Change R2 and R3 to 100 ohms and it takes anything on the ports like the WolfDragon battery monitor board.  Or, it can be a dedicated board driver and protector monitoring all systems while taking commands from a "second" CPU which could be just about anything!  Basic stamp, PIC, laptop, Iphone    It does not care in that case.  This same board can also be used as the second CPU controller too!

2.0 x 1.3 x 0.5 in this case.  I made no attempt to make it small for the prototype 

The expressPCB file is here:

http://extreme-fire.com/Panther/Panther-105.pcb

But be VERY aware this is all untested prototype stuff!  If you want boards yourself, I will probably have extra 

The parts placement is this:

http://extreme-fire.com/Panther/Panther-Placement-105-a.jpeg

Lots of room and big parts.   Easy to cut, saw, grind, weld, hammer, etc. to fix anything 

A first look at the cost and parts list is here:

http://extreme-fire.com/Panther/Panther-Costt-105.jpeg

Again beware that there may be errors there!  I did it sort of fast 

Pricey little thing really!  But it is mostly a research project to really base other stuff off of.  But still cheap enough for those that need the "ultimate" drive system that is hopefully unbreakable 

This puppy will pump out 6X the peak current of the Evolution.  3X the RMS current.  That's 1500 amps peak and 75 amps sustained there baby!!  A very high level of protection against all that can burn it up.  And it can be hooked up to "anything" to run it.  There is no other real practical solution that can carry as much beef as this can while still having a chance to realistically fit in a gun.  But this should insure everything else in the gun will burn up before it does   

We went from crawling, to walking, to running...  This type of stuff should get us airborn!!

BTW - Yes!  The main drive FET can be "piggy backed" for 3000 amps 

BBTW - This is all Public Domain, Open Source, etc. so party hardy!!

Gandolf

[Sinewave]

I've been playing a little with the screw terminals you specified -- they are tall but I think they maybe could be used like this too.  (PCB used is just for illustration purposes)

Basically it is simply edge-mounted, and the wires installed and bent 90 degrees then a small heatshrink over it to secure the wires and prevent any possible shorts.  (A separate heatshrink than the one covering the main board, that is.)  Still big but it at least shifts the size - I think "long" is easier to manage than "high" for installations.