Update 3/14/2014

posted Mar 15, 2014, 11:12 AM by Danny Park

So, attempting to keep most of the design points in tact while bringing the design back into the realm of reality, I've changed the design somewhat. I've added a boost stage to compensate for the lack of diameter inside the buffer tube, as the swash is limited to ~1" diameter and the most you might be able to get is 1" of travel... assumedly... the first design was around 60% swash angle, which bothered me because it seems like the piston feet won't be very stable. I'm convinced that going beyond 45 degrees causes some issue, but I'm not sure what yet, so for the first second attempt, I've reduced the swash angle to something more reasonable like 30 degrees and added another stage. The stages, ignoring compressibility and other effects like that, should behave like gearing for compression ratios... So if the first stage has a 15:1 ratio (I'll call it boost or low pressure stage), will raise the pressure from atmospheric (~15psi) to ~225. The second stage, let's say 5:1, will raise the pressure again from 225 to 1125psi for a final compression ratio of ~75:1. Quick stress analysis gives me ~1500psi as the maximum  pressure I can get with a 1" tube with a 1/32" wall thickness (inventor gives me a safety factor of 1.3 with 6061... probably T6) but by the time I start back up, I'll hopefully be able to afford "real" machines that will laugh at steel rather than struggle with aluminum... 

I don't believe that this axial piston design puts much load on the motor (but I could be wrong), so my current design point for the motor is ~15,000rpm. My motor has a 2370rpm/kv rating, which puts me beyond 25k unloaded, so maybe 20-22.5krpm isn't that far outside of the realm of possibility, assuming that I can sink the heat into the receiver, casing, and main reservoir.

Because there is an increase in pressure, I'm considering removing the second pressure sensor or moving it somewhere else where feedback might be useful. After the initial testing and curve fitting, I believe we'll be able to extrapolate the pressure from the torque/current draw of the motor, or I could monitor it a little more directly if I were to put in a rpm sensor... As long as I have a reasonably accurate (I believe my sensors are +/-5psi at the moment, but with the design change, I'm guessing I can get it down to +/- 1-2psi as I'll need to drop the pressure anyways and the accuracy of the sensor is proportional to the full scale value... that and the arduino micro I plan on using only has a 12bit ADC... 

The current iteration has a ~1"x3" DxH cylindrical reservoir which can store 18 shots before it is empty at 800psi... and takes 4.46s to fill and can sustain an embarassing 4rps rof... if I increase the rpm of the motor to 20k, that increases the rof to roughly 6rps, and that is with the motor running continously... I have the torque code somewhere that I can run for the latest iteration, but I'm guessing full auto is going to be a big power drain (I think last time I estimated something like 1 shot per mah, and that was with more optimistic numbers). What I think I want to explore next is replacing the outer barrel (since I've already taken it upon myself to build the "loader" into the hopup unit and have the barrel secured into the hopup (maybe threads or a bayonet lug), I'm considering also replacing the outer barrel and having a sort of tubular (ID/OD) main reservoir with the barrel running through it and then have the compressor in the stock. 

The ROF is largely limited by the "constant-on" compression/stroke and rpm. The compression is limited by the number of stages if we set the swash angle at 30 degrees (or ~.5" stroke) and the piston diameter, which is limited by the diameter of the buffer tube (I'm also considering replacing the buffer tube with a motor if I can move the reservoir to the front, so I can add more stages... maybe a total of 3-4... The RPM is going to be limited by the W-Hr of the battery and by heat dissipation... 

So... I guess we're back to whether or not we're willing to give up on full auto for now... I keep feeling like if I spend another day or week trying to come up with a scheme to get a workable rof (~15rps) but, aside from manufacturing and weather issues... it's beginning to hold back the finalization of the second iteration prototype... (I sort of gave up on the first one when I found I couldn't hold parts to within a few thou...) So that's where we are now...

AHDS Introduction - Compressor and System Overview

posted Jan 4, 2014, 6:51 PM by Danny Park   [ updated Jan 4, 2014, 7:43 PM ]

Hey Guys,

If you're following me on Facebook, you should be more or less up to date on the state of things. I've decided to move the major portion of the design of AHDS into the public domain, so this video outlines the current stage of development of the compressor-regulator subsystem and an broader system-level design. 


If you have any questions feel free to comment here, there, or in PM/email. 



posted Dec 7, 2013, 11:14 PM by Danny Park   [ updated Dec 7, 2013, 11:14 PM ]

So, I've been writing up the code in Matlab and improving the model from about a week ago. I found a few errors, and it seems like I have a good idea where to start on Monday. I ran 30 cases, and it seems that everything improves as you move down the page and it seems like P3, P5, and P7 are around the sweet spot; however, I'm hesitant to post the results because some of them seem a bit wacky (such as 13.3 shots/mAH or 31rps, or the power consumption for the D67.5/75 block), which might indicate there is another error somewhere. But anyways, understanding that I still need to verify my results, here is the excel file:


This still does not include loading energy or powering anything but the bb. 

One of the things I'm considering in the final design is to use one of the WA WOC's as the platform for ADHS, so, for the moment, I am designing around those dimensions. Right now I'm deciding between a functional trigger break and full auto. The D60 block seems to be a safe bet, and I'd imagine that ~5-10rps is about as fast as one can pull the trigger. Personally, I prefer semi instead of full auto, maybe burst if it works well, so I'd choose the trigger break instead of FA capability, but I'd like your thoughts on this.

This guy still runs on propane, but sometime next year, I'll end up gutting it and installing my internals:

Update 11-30-2013

posted Nov 30, 2013, 1:55 PM by Danny Park

So, I decided to run some quick calculations, and here's what I came up with. The calculations take into account the motor power, pump torque, pump/reservoir volumes, a conservative lb-in value for each shot and a 1000mah 3s battery. I used the torque equation to get me around to the maximum rpm I can run without exceeding the motor's specs, and this is probably the place where the calculated values will deviate the most from the actual results. 

But anyways, here's what I got (this calculation doesn't take into account friction/resistance or loading/regulating power, so this is a ballpark upper limit estimation):

- 12.87rps
- 9.127 shots/tank
- 0.709s/fill
- 2286.84 shots/battery (or 2.287 shots/maH for a 3S)
- Full Auto Runtime ~3 mins

I don't have a good intuition about how much power/flow the air nozzle mech will consume, but even the power consumption is doubled, we're still running at >1 shot/mah, which is still pretty darn good. The full auto rps is a little puny, but I guess it's close to the actual rof of an m4/m16, and this was supposed to be a semi-auto focused build. 

Also note, that the tank will refill before the ~9 shots are used up, probably between shot 1 and 5, depending on how much the regulating circuit can compensate and how weights are balanced between efficiency and accuracy. 

Update 11/12/2013

posted Nov 12, 2013, 4:08 AM by Danny Park   [ updated Nov 12, 2013, 4:10 AM ]

(meh, wall of text)

The first compressor iteration is pretty close to being finalized, I just need to decide on a few things like stroke and bore diameter. The broad design for the controller is coming along. The plan for the initial board is to tack my modifications onto an arduino micro (pressure sensors, adc, mosfets, solenoids etc). 

Since each component is dependent on other components, to save some time, the plan is to develop the regulator, controller and compressor separately. The regulator (solenoids, valves, air nozzle, etc) will be tested off of a 200psi shop compressor and a adc connected to a computer. The compressor will be tested with a servo tester, battery, speed controller, motor, as well as the adc/computer. In the meantime, the controller will be developed in parallel, and likely last. 

Here is the tentative schedule:

Mid-Late November - Finish initial valve assembly, begin testing

Early December - Finalize compressor design (depending on performance, may subdivide into different "tiers")

Mid-Late December - Begin testing control laws

Mid-Late January - Finalize board design, begin integration

Late February - Complete internal testing phase and begin sending out test models.

Late March - Complete first Beta trial

Late April - Complete second Beta trial

May - Buffer trial period/soft release

July - Official Release

At some point I'll need to test the viability of the modular barrel concept and would like to close the control loop with either a doppler radar or chronograph. I'd also like to implement a hopup concept that I've been working on. The other thing I'll have to do is work on implementing a fire selector/trigger and make it compatible with some model of aeg.

By the looks of things at the moment, the pricing for the entire system will be comparable to a P* setup, but I can't say for sure until I get a source for externals. The first model is planned to be a semi-auto DMR-type gun. While the system might be compatible with multiple platforms, I'd like to standardize the externals for the first 6-12 months to ease troubleshooting and compatibility issues. 

If anybody has any externals they'd prefer, I'd like to know. Around the February/March time, I'll need to start integrating the internals and externals, so we need to plan ahead. 

1-5 of 5