AAD Fatality Thought

[df8m1 19]

One of the projects I am working on is instrumenting up a rig to measure reserve deployment sequence times, vertical distances, and forces.

A thought occurred to me the other day while thinking about jumpers impacting with the reserve at line stretch after an AAD fire. As far as I can recall there have been at least three instances, I may be incorrect on the exact number, but in one case it happened to two jumpers at the same time. Both AADs had fired and neither reserve was useful in time.

It is that instance that really shakes up the argument of if it is a rig & reserve problem or an AAD activation problem, as they were two different rigs, and I do not think they had one of the reserves that may or may not tend to open slower than it should, as in, require more vertical distance to arrest the jumpers descent rate to a survivable one.

Although I do not think that the type of AAD in either rig was identified, that would be the only thing that two different rigs could have common which would have an influence on reserve deployment initiated by an AAD, other than possibly a common rigger with very bad rigging practices, which is not impossible.

The fall rate of the two jumpers was higher than that of a stable neutral body positioned jumper as one was tumbling and the other was in pursuit (student and instructor), which could have an affect on the actual activation altitude depending on the firing protocol of the AAD. If a capacitor has to be charged prior to firing, then that takes time, and if the jumpers are falling faster then the vertical distance fallen will be grater during the time it takes to charge the capacitor. I have seen and heard of head down jumpers being saved by AAD, so if the fall rate is a factor, then perhaps their AADs were not working as fast as the majority of the other AADs?? But two AADs having that same obscure problem at the exact same time??, I guess anything is possible..

I am thinking that a more likely reason that both AADs would fire too deep (at the same time), would be that they, the AADs, were not adjusting the AGL altitude as the barometric pressure changed during the day. Even if they turned on the AADs at a lower elevation and then drove to the DZ, unless they immediately jumped on the plane, the AADs should have adjusted for the new “zero” after a period of time at the new elevation. That is why it is not good to fly at a constant low altitude for very long.

I would love to see the barometric pressure readings in that area during that day. If the pressure gradually drops throughout the day, and the AADs did not correct their “zero” then they will think they are higher above the ground than they really are during free fall, and the onboard data would indicate that the AAD fired at the proper AGL altitude.

We zero our altimeters every jump before we get on the plane, how do we know that our AAD has also made the adjustment? It is not uncommon for the barometric pressure to swing + - 250ft during a nice weather day. Again both AADs would have had to fail to make the proper pressure correction, but then again, why do service centers have to send some AADs back to the factory?? The good news is, now that the AAD firing altitude is starting to be accepted at a higher altitude there is more room for error in regards to fall rate and “zero”.

Full disclosure, although I do not make a civilian AAD, I am designing military AADs, and I think I am going to have them indicate when they make a zero correction and how much it was. It would be nice to be able to track it with my altimeter. Personally I do not agree with the “set it and forget it” philosophy or set it and trust that it is doing what it is supposed to do for that matter, I am more of a “trust but verify” kind of guy.

Again, just a thought, I don’t think that this possibility was ever discussed while those incidents were in the forefront.

[skydiverek 55]

These two skydivers were Islandic skydivers who when in in Z-Hills in Florida.

Both had Cypres.

Florida is flat, no change in elevation more than then 15 feet probably (around Z Hills).

Cypres adjusts to changes in weather throught a day.

If you do not fall flat, Cypres will fire at around 1000 feet, not 750 feet.

[obelixtim 100]

df8m1

One of the projects I am working on is instrumenting up a rig to measure reserve deployment sequence times, vertical distances, and forces.

A thought occurred to me the other day while thinking about jumpers impacting with the reserve at line stretch after an AAD fire. As far as I can recall there have been at least three instances, I may be incorrect on the exact number, but in one case it happened to two jumpers at the same time. Both AADs had fired and neither reserve was useful in time.

It is that instance that really shakes up the argument of if it is a rig & reserve problem or an AAD activation problem, as they were two different rigs,

The answer is neither.

It is a jumper problem.

While interesting, I think you are embarking on an exercise in futility. Unless you can pinpoint absolutely a problem with the AAD itself, the variables are just too great, because the circumstances of no two incidents would ever be exactly the same.

In the end everyone must understand that an AAD is a backup device, and nothing is guaranteed when margins of survival left by the jumper are left so slim. At that point it is either your lucky day......or not.

My computer beat me at chess, It was no match for me at kickboxing....

[Southern_Man 0]

There are a lot more than 3 instances. I believe iy is in double digits. USPA and PIA have a list but good luck getting them to share it. I've made my own list of the ones I know about buy I don't pretend it is complete.

"What if there were no hypothetical questions?"

[df8m1 19]

skydiverek

These two skydivers were Islandic skydivers who when in in Z-Hills in Florida.

Both had Cypres. .

Good to know they both had the same AAD. I also believe the other instance I seem to recall was a Cypres as well.

Please do not misinterpret any mention of a specific AAD as bashing them. Just looking for commonality in equipment and what happened.

skydiverek

Florida is flat, no change in elevation more than then 15 feet probably (around Z Hills)..

So that takes the "turned it on at home" possibility off the table in this case.

skydiverek

Cypres adjusts to changes in weather throught a day.

If you do not fall flat, Cypres will fire at around 1000 feet, not 750 feet.

Every electronic AAD is supposed to adjust for barometric pressure changes throughout the day, and I think that all the electronic AADs will fire at a min of 750 AGL. That is if it is working and calibrated properly.

I cant imagine that the student rig was a tight little one with one of those new reserves that is said to possibly take more time. If the instructor had a perfect storm so to speak of tight reserve container, and slow reserve, (not saying it is or isn't.. I have not measured it yet), then I would have not thought twice about it if the instructor went in and the student's reserve deployment was as other AAD deployed reserves generally are.

But in this case they both went in wearing two different container and canopy combinations and both with the same AAD. Both individually rigged jumpers hit the ground before either reserve could inflate.

The Cypres 1 used an analog pressure transducer and I have not had a Cypres 2 apart to know if they went digital or stayed with the Analog, both have pluses and minuses. One would think that the sensor would put out a steady reading if it is on the ground, but they do not, there is always noise to contend with. The sensors are actually very close to the bottom of their scale when at sea level. I have used both Analog and Digital baro transducers and neither will put out a solid number without a lot of averaging. If the sensors in both AADs were part of the same batch, and that batch of sensors was noisy, or became noisy after some use, (I have had to reject a batch of sensors for that reason),(remember every 4 years they have to be checked, and some have to be sent to the manufacturer, and there was a batch of them recently that had issues with a component, Again I’m not bashing the Cypres). It would be nice to know when both AADs were made and their service history.

The noise is a ranging output frequency and the actual location is some where in the middle usually. If the noise band is wider then the program is ready to handle, then the gradual changes in ground level barometric pressure could easily be lost in the noise and not compensated for.

With out knowing what every jumper was using, how it was maintained, and the conditions they were jumping in, it is almost impossible to find a thread of commonality gear wise, let alone the conditions. This case offers a little more potential as there were two different rigs in the same conditions, and both jumpers died from impact with out an open reserve after an AAD deployment. The only difference between the two jumpers deployment processes is I think the instructor had pulled his reserve handle, and I do not know if his loop was cut or not, but both AADs reportedly had fired.

[pchapman 261]

You've got a lot of speculation in there...
but on the other hand, nobody has a good handle on exactly how the various factors add up to cause failures to save a jumper.

Given that the FAA mandates that AAD's must be used and maintained according to the manufacturer, wouldn't it be nice if they mandate that after an incident, AAD internal data must be sent to them and made public? In the absence of certification standards administered on behalf of the public, the public gets access to more data in order to create more feedback and interaction between designers and users.

(I know that's a thought for the ideal world and not the real world of the FAA, who still think a Lap parachute rigging rating is an important distinction.)

If one had data, one could see for example if the jumper ended up at what the AAD thought of as height zero (or close to it), putting to rest any fears that the AAD didn't recalibrate itself "to zero" for air pressure changes.

I don't seem to recall seeing Cypres data graphs, but a few from Vigil have been published.

df8m1

If a capacitor has to be charged prior to firing, then that takes time

But isn't the idea that the capacitor is always charged up, so that a higher current discharge is available extremely quickly to fire the cutter?

The extra battery thing in the Vigil is sort of a hybrid battery / capacitor. One statement on the web about them: "Pulses Plus™ batteries were developed at Tadiran and introduced into the market few years ago. These batteries combine a primary high-energy bobbin type Li/SOCl2 cell with a hybrid layer capacitor (HLC). The HLC is a battery-like capacitor consisting of lithium intercalation compounds as electrodes with pseudo capacitance of 785F for standard AA size".

But what does the Cypres use? From opening up the Cypres 1, there is no big capacitor in there. My photos show only what look like a couple small ones on the circuit board, no more than 1 cm long and 1/2 cm dia. Whatever their design, it normally seems to work.

AAD's do tend to have poor feedback about the mode that they are in, what they are thinking. (Although the Vigil does have the flash 3 times thing at activation altitude. Yet it traditionally has also had the problem of thinking it is airborne if driven away from the DZ, now mitigated by the new 14 hr turnoff feature.)

[df8m1 19]

obelixtim

***One of the projects I am working on is instrumenting up a rig to measure reserve deployment sequence times, vertical distances, and forces.

It is that instance that really shakes up the argument of if it is a rig & reserve problem or an AAD activation problem, as they were two different rigs,

The answer is neither.

It is a jumper problem.

While interesting, I think you are embarking on an exercise in futility. Unless you can pinpoint absolutely a problem with the AAD itself, the variables are just too great, because the circumstances of no two incidents would ever be exactly the same.

In the end everyone must understand that an AAD is a backup device, and nothing is guaranteed when margins of survival left by the jumper are left so slim. At that point it is either your lucky day......or not.

This is a good point. Too many jumpers these days put way to much responsibility on the AAD to save them instead of them selves. Additionally, the general lack of equipment working knowledge puts them at even more or a disadvantage so to speak.

I design and build parachute data collection systems and have discussed developing a process to measure every critical aspect of the rig and reserve. One way to look at it is, it is business..

I often am told that there is video of everything happening and they know how long it takes for X to happen, but I then say, how much force was the pilot chute generating behind the jumper during that test? How long did it take from when the pin was extracted to when the pilot cute left the container, or when the free bag left the container? This is something that John Sherman has been talking about in regards to using a shorter bridle to save some time, but what does that do how much pull force the pilot chute can generate in the burble behind the jumper?

If there was a standard test method then next time there is an incident, the gear could be tested, granted it is not as simple as that, no, not even close.. What I have struggled with is, what if there is a combination of equipment that in certain conditions fails to operate with in TSO standards? Then what?? As I told one of my partners, “you can’t put the shit back in the horse”. And at the same time, given that this could be done, to, at the very least, allow us as jumpers to say that we have done everything that we can do to keep this from happening again, would it not be prudent? Just as Consumer Reports does independent crash tests, that the OEMs always contest, unless it makes them look good that is..

There is a lot of politics involved, as well as several different countries, languages, and cultures with such a project, which can make it difficult if one whishes to involve the manufacturers in the process.

But back to your thought, I agree that it is the jumper who is ultimately responsible for saving their own life, so long as their equipment will allow them to do that.

[df8m1 19]

pchapman

You've got a lot of speculation in there...
but on the other hand, nobody has a good handle on exactly how the various factors add up to cause failures to save a jumper.

Given that the FAA mandates that AAD's must be used and maintained according to the manufacturer, wouldn't it be nice if they mandate that after an incident, AAD internal data must be sent to them and made public? In the absence of certification standards administered on behalf of the public, the public gets access to more data in order to create more feedback and interaction between designers and users.

If one had data, one could see for example if the jumper ended up at what the AAD thought of as height zero (or close to it), putting to rest any fears that the AAD didn't recalibrate itself "to zero" for air pressure changes.

I don't seem to recall seeing Cypres data graphs, but a few from Vigil have been published.

Absolutely, that data would be very valuable, but to more than just jumpers, the layers would be all over that as well. Something happened to both jumpers equipment that caused them both to fail to operate with in any of the manufacturer’s specks. I know that is another bold statement, but it is not like they both were using the exact same container, container size, or reserve size. I’m not even sure if both reserves were of the same manufacture..

Now one could argue that one can not make any speculative statements given that one does not have all the facts, but none of us do, that is, none of the actual people who chose to expose themselves to great risk based on the trust / belief that their equipment meets, and will perform at, the basic requirements, (provided that it has been maintained and was operated with in the parameters accepted by the manufacturers, i.e., if the reserve fails to open because it was activated too late, then it is not the reserves fault as it did not have enough vertical distance to operate with in the manufactures speck).

It is up to us as jumpers to save our selves, that includes post incident, arm chare, investigation, in the absence of anything else..

pchapman

***If a capacitor has to be charged prior to firing, then that takes time

But isn't the idea that the capacitor is always charged up, so that a higher current discharge is available extremely quickly to fire the cutter?

The extra battery thing in the Vigil is sort of a hybrid battery / capacitor. One statement on the web about them: "Pulses Plus™ batteries were developed at Tadiran and introduced into the market few years ago. These batteries combine a primary high-energy bobbin type Li/SOCl2 cell with a hybrid layer capacitor (HLC). The HLC is a battery-like capacitor consisting of lithium intercalation compounds as electrodes with pseudo capacitance of 785F for standard AA size".

But what does the Cypres use? From opening up the Cypres 1, there is no big capacitor in there. My photos show only what look like a couple small ones on the circuit board, no more than 1 cm long and 1/2 cm dia. Whatever their design, it normally seems to work.

The batteries used in AADs are generally made to put out a small about of currant over a vary long period of time, as apposed to a car battery that is made to put out a lager amount of currant for a short period of time. The pulse battery you mentioned is just like a car battery, it is made to put out a lot of power for a short period of time, so it is great for firing a cutter, but would not be good in your walk man..

Super Caps are used to fire cutters and they are most commonly available in flat wafer type design. This makes them great for stacked circuit boards as they are thin, but have a larger foot print.

The main benefit of using a capacitor to fire the cutter is that a slow discharge battery can be used for the AAD, and the cap is slowly (relatively speaking) charged as to not collapse the battery voltage by drawing too much power. Once the Cap has been charged, then it can discharge a lot of power very quickly. It is not uncommon to also boost the cutter voltage to a higher voltage than battery voltage which reduces the amount of currant required to fire the cutter. There is a balance between battery output capacity, boost regulator output, and capacitor size.

It is a waste of battery power to charge the cap every time the AAD is turned on, so when they are used, they generally are not charged until the conditions closely approach firing conditions.

I do not know if the Cypres 1 or 2 use a cap, if they do not use one, then the required charge time is taken out of the equation.

pchapman

AAD's do tend to have poor feedback about the mode that they are in, what they are thinking. (Although the Vigil does have the flash 3 times thing at activation altitude. Yet it traditionally has also had the problem of thinking it is airborne if driven away from the DZ, now mitigated by the new 14 hr turnoff feature.)

Electronics are funny in that we can not see what is going on in there.. The more I work with electronics and code, the more I am amazed that it works lol.. then again... I have seen the same code in 10 prototypes and some work fine and some have problems that pop us once in a while..

[RiggerLee 61]

This is speculation but it seems to fit with problems I've seen in the past.

Tried to use cypresses for some drop test. They would never fire on time. It had a higher fall rate then a normal jumper. This is straight out of my ass, I can't prove how it works, but our theory was that it made it's decision to fire at about 1,200 ft. It then charged up a capacitor and fired the cutter at 750. But with our higher fall rate it didn't have the time to charge up the cap and could not fire on time. The guys at Airtech were impossible to work with so these guys just built their own firing circuit.

I'm not saying that they don't function as designed under normal circumstances. However with a higher fall rate you can exceeded their limitations and when you do you get these late firings.

Vigil has made the software available so that any one can down load the data from their units and the data is excellent. Cypress has always kept it a secret. Every time there is a fatality they send the unit in and airtech goes over it. Almost with out exception they come back with a reply that the unit fired on time but I've never seen any of the raw data. You could look at some of their action over the last few years such as campaigning for higher opening requirements in the BSR's and make out that they were covering up for failings and inadequacies in their system but I'm not convinced. I still think that the issues lie in the container design and the PC launch in most of these cases. And depending on which ones you want to count, there are way more then three.

Lee

Lee
[email protected]
www.velocitysportswear.com

[RiggerLee 61]

There are right now combinations of equipment which do not meet, or at least have failed to meet, TSO standards. It's on video. Right there. Watch the reserve PC tow. Tows like a fucking bitch. Every bodies seen it. Oh look theirs another one, watch it tow too.

There it is... big steaming pile of horse shit. The thing is nobody really wants to clean it up. It's messy. It's smelly. Yeeu, nasty. Did I mention expensive? And what if their has already been deaths? If you put out a change now isn't that like admitting guilt? I would think that they would be more invested in fixing problems or potential problems but some times functionality seems to take a back seat. I fear that some day some one will get their ass in a sling in a real law suit.

Lee

Lee
[email protected]
www.velocitysportswear.com

[councilman24 36]

df8m1

One of the projects I am working on is instrumenting up a rig to measure reserve deployment sequence times, vertical distances, and .

This is being done by PIA. I don't know the current status of the project.

There are about 15 instances know and I have reports on most but frankly haven't even studied them in detail. No common factor to find.

Remember tso standard is 3 seconds and up to.4. 300' if 3 seconds can't be meant for cutaways. 3 sec at terminal with firing alt. of 750 gives little toon for delay. 4 secs may put you in the dirt at terminal from 750' 750 was always the ragged edge but worked because most were faster when introduced. Now with new rig designs and reserves designed to be closer to 3 sec we've pushed into the dirt. That's why PIA risk management comm and others pushed for higher activation altitudes.

For what little it's worth based on getting to know the principles I trust AirTec implicitly. I am neither a dealer or have other.affiliation with any gear other than a couple.of.pilot rig.manufacturers.

I'm old for my age.
Terry Urban
D-8631
FAA DPRE

[erdnarob 1]

About capacitors to be charged when AAD triggering conditions exist (Speed and altitude), this is the case for Vigil competitors and it takes time. Not too much but enough for the jumper to keep on falling for precious fractions of a second.
OTOH, Vigil 2 AAD is equiped with a PulsePlus device which is sort of between a battery and a capacitor. The big difference with competitors is that Vigil 2 PulsePlus doesn't have to be charged when triggering conditions exist because it is already charged and can keep that charge for a very very long time. Vigil 2 takes 2/1000th of a second to fire when needed which is considerably less than other AAD makers. Even better, that technological approach saves a lot of energy and is one of the reasons why Vigil 2 batteries last that long. Few fractions of a second can and do make the difference since the free fall speed is about 174 ft/sec.

Learn from others mistakes, you will never live long enough to make them all.

[nigel99 143]

Good and interesting discussion. One thing I really don't like is that the AAD manufacturers don't publish the change logs on their software. Vigil do to some extent. I would want to know what is being changed when my unit is serviced.

With algorithms as you tighten them up to avoid false positives (unwanted firing of an AAD) you get to a point where you miss the real event. I wouldn't be surprised to see that Cypres algorithm is so robust against false fires that it fails to fire on time.

The anecdotal evidence from the florida fatalities sure puts the spotlight on the AAD. BUT we also don't want unwanted AAD firing at the same time. Seems like raising AAD firing altitudes is a good compromise to keep algorithms robust.

Experienced jumper - someone who has made mistakes more often than I have and lived.

[df8m1 19]

councilman24

***One of the projects I am working on is instrumenting up a rig to measure reserve deployment sequence times, vertical distances, and .

This is being done by PIA. I don't know the current status of the project.

There are about 15 instances know and I have reports on most but frankly haven't even studied them in detail. No common factor to find.

Remember tso standard is 3 seconds and up to.4. 300' if 3 seconds can't be meant for cutaways. 3 sec at terminal with firing alt. of 750 gives little toon for delay. 4 secs may put you in the dirt at terminal from 750' 750 was always the ragged edge but worked because most were faster when introduced. Now with new rig designs and reserves designed to be closer to 3 sec we've pushed into the dirt. That's why PIA risk management comm and others pushed for higher activation altitudes.

For what little it's worth based on getting to know the principles I trust AirTec implicitly. I am neither a dealer or have other.affiliation with any gear other than a couple.of.pilot rig.manufacturers.

LOL... Yeah... Someone mentions that every time there is a fatality... IMOP they are to closely associated with the manufacturers to do something like that without a cloud over it, if you know what I mean. I'm not saying that anyone would go out of the way to skew the results, but this is business and human nature.

There could be a benefit from two or more methods to test as if the data matches (to with in a reasonable degree), then all is good, and if they are way off it would be interesting to see where the differences are and how the came to be. In other words, more than one effort would tend to keep all in check so to speak.

I am very pleased that this has not turned into a specific manufacturer bashing or extreme defense of a manufacture thread, but the night is still young lol...

As was correctly pointed out by another poster, there is a balance or compromise in order to avoid un wanted activations and still get them to activate when we need them to. There are several different ways to go about doing this, and the results can be seen with the history of the different AADs on the market.

Also, as one poster pointed out, there are several instances of reserve pilot chutes in tow, which scares me almost as much as a broken pin. My rigger checks how much it takes to remove the free bag from all of my rigs, and not because I ask him to.

I was thinking today that there could be several issues, each alone being deadly, and each occurring very rarely so to speak. So as was pointed out, if it appeared that a manufacturer was to “hypothetically” make a change to their design in order to help avoid a “hypothetical” equipment malfunction, then it could appear that the manufacturer was admitting guilt. But, if they all pointed their fingers at each other, and in the mean time, raised the min reserve deployment up a little as a overall good idea that in no way is to imply that there is a problem, but rather just good regulation, then after a while it will be forgotten. That’s my conspiracy theory for today lol..

Personally I think that 750 ft is too low for a last ditch reserve effort, I do like 1000 ft better. (NOW I’m not saying I agree with the higher main pack opening altitude, just a higher reserve opening altitude)

You can’t bring back the dead, so as long as another jumper does not go in after an AAD fire, or if they pulled the handle at an appropriate altitude, then all is good. BUT… That being said, I personally think that we owe it to our fallen brothers and sisters, as well as their loved ones to find out what happened, even if that means they screwed up, which I bet would be more the rule. It is when the jumper did everything right that I am really interested in. Yes, a jumper should not put their life in the hands of an AAD, but, if they are going to take that extra step, and add an AAD to their gear, then the chose of which AAD to buy “could” be the difference between life and death, or at least it appears that may be the case, and the same may be the case for container and reserve combinations as well.

I hope I am not coming off as a finger pointer, or looking to cause trouble as that is not the case. We as jumpers have been allowed the freedom to self regulate, and IMOP we have done a piss poor job of it. I think that this type of problem can be handled internally with out tremendous fan fair if everyone is open to look into the deep dark corners of each instances regardless of what implications may be drawn , weather proven actual or not.

Certainly a test standard for AADs is a must. This could be simply a standard method of confirming that the AAD performs as the manufacturer says it will. As long as “it does what they say it will”, with in a reasonable tolerance, (to be determined by we the people), then all is good. And when there is an incident, the gear, all of it, undergoes this standard test, and its performance is evaluated independent of the manufactures. This way each AAD manufacturer has the freedom to have their AAD perform as they like, as in, activate 10 ft above the ground, as long as it’s performance is confirmed, then it is up to the consumer to decide if they like that performance feature in their AAD.

And I too like the idea of having a log of what exactly was done to an AAD throughout its life, this is a requirement for an airplane is it not. It would be interesting to know exactly why some AADs have to go back to the factory.. this would imply replacement of parts, as software is easy enough to down load is it not? A log would provide a history that could hold valuable information in a fatality investigation.

[pchapman 261]

To speculate more, like most of us here:

CURRENT OUTPUT TO CUTTER

It is so hard to argue things because the manufacturers give us so little to go on.

So we can say, hey, the Tadiran Pulses Plus hybrid battery/capacitor in the Vigil 2 is really neat, and works fast. Heck I found a Vigil document where they claim an even faster "under 1/1000th of a second" from fire instruction to cutting the loop. But they don't say how the competition works, and neither does Airtec about the Cypres 1 or 2.

And how fast is the discharge from a capacitor vs. a battery? What current can it achieve while maintaining voltage? I don't know. Pulses Plus battery/capacitors of size similar to that in the Vigil are in product literature listed as being able to have 1A pulses (up to 1 s, maintaining 3.0 V). A battery similar to one of the two in a Cypres 1 is similarly quoted as being able to sustain 2A pulses (quoted for 0.1 sec, maintaining above 3V). (But I don't know for sure what type is in the Cypres 1).

So just on published data, it makes it look like Cypres batteries could well provide a high current too, although rate of onset is still not known to me.

Someone with knowledge of pyrotechnic line cutters might have some insight into what kinds of electrical flows are needed to set off the charge.

FIRING DELAYS

Now if the Cypres 1 and/or Cypres 2 do rely on a capacitor (which we're unsure of), they will obviously have taken charging time into account. Really, how long does it take? I don't know. Is it really more than a small fraction of a second, and either way, wouldn't it be done before the decision to fire, getting ready for a possible firing, and not as some afterthought?

If one argues about firing time, then one should also look at the delays induced by the overall algorithm, which might be a lot more significant than the delay in getting the firing command executed.

From the little I know, any traditional smoothing algorithm that in some way averages results will be a little "behind the times". Any sudden change in one measurement period can't be taken on face value; it has to be interpreted in terms of prior data, and so delays any decision. To get more complicated, one can try to add some intelligence and do some prediction of what future values are likely to be reasonable, and compare that the incoming data. (Very roughly that's getting into Kalman filtering.)

The Cypres patent from ~1988 doesn't give any clues, and I haven't tried looking for any of the Vigil patents.

VIGIL ALGORITHMS

Sometimes bits of data slip out. Consider this in a document from Vigil about the firings of Vigils in a pressurized C-130 at the 2006 World Team event, where the Vigil measured an apparent speed from the pressure change, that had a peak above 78 mph:

Quote

The [...] peak was below the activation altitude of 256m (840 ft) during 7/8 of a second, which created the conditions to activate the Vigil cutter.

The Vigil when in airborne mode does 8 readings a second.

In another document, Vigil wrote:

Quote

That's why the Vigil only needs 5/8 sec to register it is in freefall and if the next measurement point is still in freefall it will activate in 125 msec and the loop will be cut within 2 msec.

Also:

Quote

In the US activations, the linked virtual speed stayed only 0.875 sec above the trigger speed value (7 points) and the system detected the freefall after 0.625 sec (5 points), so it justifies the activation.

These quotes are about the "door open" accidental Vigil firings, where it had to both decide the jumper was out of the plane and then decide there was a high freefall speed.

The quotes can be found in "Official_AAD_Statement_Jun_2010[1].pdf ", which is on dz.com. I actually have them also in a longer document which was called "19sep2010_Vigil_PlaneCrash". Not sure where I got it from - not even showing up on google - but it internally claims to be a draft report for internal use only at Vigil. (It even quotes a section of a dz.com thread.)

So the quotes make it sound like their algorithm takes 7/8ths of a second of data to decide that an apparent firing speed situation is true.

What does that imply? That even with perfectly smooth data they always fire 7/8 sec late? They can't be that stupid, so do they basically start the count higher than activation altitude? But in the first quote they talked about 7/8 second WHILE below 840 ft, NOT that it counted activation speed for at least 7/8 second and then by the end noticed it had reached 840 ft.

Their statements are all very confusing.

Vigil also talks about their "time remaining" calculations as part of the algorithm process, perhaps predicting when the firing altitude will be reached, rather then depending entirely on noisy pressure readings. I haven't tried to read up or analyze their claims on that.

Anyway, whatever the issues are with completely hidden Cypres algorithms, Vigil either has a pretty slow and crude firing algorithm -- or gives really confusing descriptions of its firing process. Despite Vigil crowing about super duper fast firing circuits.

[df8m1 19]

Batteries and Capasators

Aside of the time required to charge a super capacitor, once the firing circuit is triggered, the power will reach the cutter as fast as an electron moves. Once power reaches the cutter initiator the standard max activation time is 10ms.

I have had a custom cutter initiator designed for my cutters and they have dual ignition bridge wires for redundancy, over kill I know, but it the engineer thought it would be easy to do given the design, so we went with it. Anyway…

A capacitors ability to maintain a voltage level at a given discharge rate depends on the capacity of the capacitor. One of my cargo AADs has 8 individual firing circuits and there are several capacitors in parallel to be able to maintain voltage and currant levels with all 8 circuits fired. Yes it takes time to charge a capacitor, but as long as that time is factored in then it is a non issue. Of course, exceeding the descent rate that the charge time is factored for will result in a “late” activation.

When deciding if one wants to / can, fire a cutter directly from a battery, one needs to look at the worse case scenario, an old battery with only ¼ charge left that is cold soaked to the min temperature speck. In this case it takes a pretty stout battery to be able to put out power on demand with out collapsing.

There are two common designs for hi-density batteries, Spiral and Bobbin core construction. Bobbin core are designed for long term low currant draw applications, and Spiral core are for high amp output and pulse applications. AADs have a requirement for both technologies so, in the case of Vigil they have both types of batteries, one to power the processor, and one for the high demand of the cutter. Another way to manage the different power demands is to slowly charge a high output capacitor just before it is needed.

I looked at a capacitor firing circuit and given the nature of the application (exits as low as 500 ft AGL), it would have had to charge every time the AAD was turned on so it was ready to go given the little time there is at that altitude. That would waste a lot of battery power, so I went with a dual battery design, but different than what Vigil has.

Sample rates and filtering

You are correct that the more you average the later the response is. 6 to 8 sps seems to be the range for the AADs on the market, looking at maybe 4 channels? The data recorders in mine are recording 10 channels at 20 sps in contrast, but the AAD is still a traditional AAD that is barometrically referenced, so really inside the box is two separate devices, a data recorder and an AAD.

You mentioned some in aircraft firing instances, which every electronic AAD has done except for the M2 given it is new and is just starting to get some seat time so to speak. Personally I do not think that it is acceptable for a processor driven AAD to fire at it’s command while in an aircraft, regardless of the aircraft location or descent rate. I have spent several years working on an “intelligent” design for Military cargo and personal applications that is situational aware and knows if it has “cleared” the aircraft. I say “cleared” because one of the malfunctions is a static line jumper in tow, in that instance, an AAD fire can be disastrous, as any in aircraft activation has the potential to be.

This intelligence is handled by the data recorder and requires a lot of processing power which requires more battery currant and capacity compared to the traditional electronic AADs on the market. I certainly can not claim a battery will last 15 years or 5000 jumps lol. But I can and do claim that it will not fire in an aircraft period,… well if a wing were to brake off, then maybe… one of the tests we are going to do next summer is put one of every AAD next to mine, in an aerobatic plane, and see if the pilot can get them to fire.

In the lack of intelligence, the logic and settings have to be very tight as was mentioned, and that tightness is dependent on the consistency of the instrumentation out put quality and accuracy, as well as the speed of the sample rate. Also battery life is dependent on the processing demands and the civilian market seems to be demanding extreme battery life for the power requirements that this type of application “would” demand given more intelligence was programmed into the AAD.

Thoughts so far..

This has been a great experience. The ability to discuss possible technical issues, albeit hypothetical as we do not have, and will never see the data that would allow us to prove or disprove any theories, is great, and rare from my experience with all types of forums, props to everyone!

I definitely like the idea of displaying the zero calibration changes throughout the day so the jumper knows what their equipment is doing, none of us would board a plane without zeroing our altimeter, so why would we not do so with our AADs?

I also think a log book of the AAD from the initial test results at the time of validation, every maintenance test result, part replacement and why, till it has reached its end of life should be a requirement. As my partner pointed out, that is required for a reserve, so why not the device that “might” activate it? I think the Military will like that.

This summer I will have the Free Fall AAD ready for testing and will start using the built in data recorder to measure burble affects on the indicated altitude as bodies close in on each other, like an AFF jump would be. I will publish all the data that I collect so we all can have a better understanding of what is going on in that little black box. From there I will start to add sensor input from the container, such as pin extraction detection, pilot chute acceleration, bridle pull force, free bag acceleration, time from cutter fire to pilot chute acceleration, etc..

Once we develop a method to measure the performance of a complete “off the shelf” rig, then the next time there is an incident, there is no excuse not to be able to test that combination, should it be in question, remember that 99% of the time it is jumper error that is responsible, but when you see a video of a reserve pilot chute in tow, or have two separate jumpers impact with their reserves at line stretch (at the same time)then that equipment, including the AADs, should be instrumented up.

[df8m1 19]

pchapman

Anyway, whatever the issues are with completely hidden Cypres algorithms, Vigil either has a pretty slow and crude firing algorithm -- or gives really confusing descriptions of its firing process. Despite Vigil crowing about super duper fast firing circuits.

I was thinking about what you said and I would say you are correct in that the actual firing algorithm for any AAD is crude, in that it is a altitude and descent rate function. The magic is in how the raw data is processed. There are lots of different ways to get an AAD to fire with in a window of altitude at with in a range of speed.

Perhaps there is a language barrier issue when trying to describe how the Vigil works? Personally I agree with not making any logic or algorithms public because they can not help but evolve over time. I protect my logic as a Trade Secret.

Hopefully the latest Vigil update will lessen the hair trigger tendency they seem to have and from what the memos you have sighted seem to back up. Perhaps they are that way because of the delayed data due to averaging that you also pointed out as being possible from how they describe how it works? Really, if you think about it, from the time that the processor authorizes the and executes the firing, to the time the pyro agent is ignited should take no longer than 20ms, which at 200 ft/s that still covers some distance.

[df8m1 19]

councilman24

There are about 15 instances know and I have reports on most but frankly haven't even studied them in detail. No common factor to find.

Is there any reason why those reports are not available to jumpers????

[hackish 8]

I'm not sure what sensor they use but I played with a MEMS unit and it can only do a limited number of samples per second. I don't recall the exact number but it was somewhere around 8hz. Since I'm not involved in any AAD R&D I have no idea what the innards of any of the AADs look like. It would be interesting to take someone with a chamber and a "fake" cutter circuit (timer) to actually instrument the 2 brands.

Another variable to consider is that a number of riggers have tweaked packing to their own taste which sometimes deviates significantly from the manufacturer published method. I tend to be the stubborn published way or not at all type but I often worry how differences really affect the opening time.

If I am involved in doing drop testing in an upcoming project then I'll see if I can't process the data in such a fashion to extract some information from this. I'm especially interested to know if my packjobs would yield a statistically significant difference as compared to some other rigger's work.

-Michael

[RiggerLee 61]

I'm impressed that there even are reports. I've seen the whole range of responses over the years. Some drop zones have no interest in gathering any meaningful data and at times actually oppose it. The thought is that if there is no information then there is nothing for a lawyer to spin. So good reports are some times hard to find even within the industry. The quality of the information released to the larger public tends to be even lower unless some one opens their mouth and starts blabbing. I actually fall into that category. You can call me nieve or idealistic but I believe in openness, honesty, the free exchange of information... the truth. I believe that there is a Truth and that we ultimately can not hide from and should not fear it. It's a very unpopular attitude these days and pretty much categorizes me as a fool. But the point is that because there is no out side party investigating these accidents there isn't a lot of good information on some of them and there is no real motivation to change that.

Lee

Lee
[email protected]
www.velocitysportswear.com

[df8m1 19]

hackish

I'm not sure what sensor they use but I played with a MEMS unit and it can only do a limited number of samples per second. I don't recall the exact number but it was somewhere around 8hz. Since I'm not involved in any AAD R&D I have no idea what the innards of any of the AADs look like. It would be interesting to take someone with a chamber and a "fake" cutter circuit (timer) to actually instrument the 2 brands.

Another variable to consider is that a number of riggers have tweaked packing to their own taste which sometimes deviates significantly from the manufacturer published method. I tend to be the stubborn published way or not at all type but I often worry how differences really affect the opening time.

If I am involved in doing drop testing in an upcoming project then I'll see if I can't process the data in such a fashion to extract some information from this. I'm especially interested to know if my packjobs would yield a statistically significant difference as compared to some other rigger's work.

-Michael

The options for relatively high speed digital absolute pressure sensors has been very limited until a few years ago. The reason I was still using Analog pressure transducers was because the digital ones had such a slow refresh rate. Today there are a couple of different digital (MEMS) sensors with refresh rates available over 100 sps, but there is usually a trade off with the resolution as is the norm with many other digital instruments.

If you are looking for one PM me and I will send you the part numbers to a one that I have used and the one that I am currently using, the only difference is features like on-board processing, alert thresholds etc..

It scares me when I see some of the pack jobs that some riggers do, and the gear owner dose not question it because the rigger is their friend or the cool guy on the DZ.. The reserve is our last ditch chance to save ourselves, and it's maintenance has nothing to do with a popularity contest lol...

The more data that is out there the better!!

[df8m1 19]

RiggerLee

I'm impressed that there even are reports. I've seen the whole range of responses over the years. Some drop zones have no interest in gathering any meaningful data and at times actually oppose it. The thought is that if there is no information then there is nothing for a lawyer to spin. So good reports are some times hard to find even within the industry. The quality of the information released to the larger public tends to be even lower unless some one opens their mouth and starts blabbing. I actually fall into that category. You can call me nieve or idealistic but I believe in openness, honesty, the free exchange of information... the truth. I believe that there is a Truth and that we ultimately can not hide from and should not fear it. It's a very unpopular attitude these days and pretty much categorizes me as a fool. But the point is that because there is no out side party investigating these accidents there isn't a lot of good information on some of them and there is no real motivation to change that.

Lee

It does seem like the thinking, "no body, no crime", doesn't it. As, if there is not any information the container type and age, reserve make, age, and maintenance, if there was an AAD or not, and if so, the make, age, and maintenance history, let alone any data that they "may" have in memory, then all will be good in time...

In time 90% just forget about the fact that someones reserve did not open in time (assuming it was activated at an appropriate altitude) but again, with videos of reserve pilot chutes in tow, it is hard to argue that the system was operating with in TSO standards.

I by no means mean to imply or create the idea that the manufacturers are knowingly producing gear that will not work as they advertise. I too have had the opportunity to get to know some of the founding fathers so to speak and I know that they are deeply troubled when there is an incident.

The truth is nothing to fear as it is the cover up that is always worse. I think is looks very shady to a woofo that we do not have detailed incident reports available, like "we" are attempting to hide a bad safety record or something.

If a piece of equipment did indeed fail to operate properly, which happens ever day in real life, (just look at the most recent air bag recall), then that information needs to be studied and changes made. Now granted that if there are several similar incidents that involved the same equipment, then there is a marketing problem for those manufacturers.

There are recalls on our gear all the time, the idea of hiding the details of an incident is just silly IMOP.

[Southern_Man 0]

RiggerLee

I'm impressed that there even are reports. I've seen the whole range of responses over the years. Some drop zones have no interest in gathering any meaningful data and at times actually oppose it. The thought is that if there is no information then there is nothing for a lawyer to spin. So good reports are some times hard to find even within the industry. The quality of the information released to the larger public tends to be even lower unless some one opens their mouth and starts blabbing. I actually fall into that category. You can call me nieve or idealistic but I believe in openness, honesty, the free exchange of information... the truth. I believe that there is a Truth and that we ultimately can not hide from and should not fear it. It's a very unpopular attitude these days and pretty much categorizes me as a fool. But the point is that because there is no out side party investigating these accidents there isn't a lot of good information on some of them and there is no real motivation to change that.

Lee

I wish everybody shared that philosophy. I think it is particularly bad that USPA, which has a bunch of this information in compiled form, will not release that information to its members. I've asked. At least two other members and riggers have also told me that they have asked with no success. I think if there as enough outcry from the members they would release it.

I understand that some of these reports get released in parachutist. I don't think that is all the information they have. I've compiled my own list from parachutist and from dropzone. Often there is pretty significant information missing from those reports (details on equipment, most prominently).

"What if there were no hypothetical questions?"

[Divalent 89]

df8m1

... This summer I will ... start using the built in data recorder to measure burble affects on the indicated altitude as bodies close in on each other, like an AFF jump would be.

Something I thought about 2 years ago after the Z-hills double fatality is how an AAD's triggering algorithm would handle unusual data.

For example, where the jumper is tumbling, so that the sensor would be alternately in and out of the burble. Or where a person in a fast steep dive (so maybe AAD not so much in burble) decellerated and flattened out (so now falling slower, and AAD fully in burble). Or where one jumper is docked and tumbling with a non-responsive jumper trying to find a reserve handle.

Would the AAD be able to make sense of things and still trigger at the right altitude, or would it need some time to adjust. (Given that going from out the burble to in the burble would appear to the AAD as if the person had suddenly accelerated *upwards*.)

Sure wish we had the data from the AADs in that incident.

[RiggerLee 61]

They're trying to do it with just a static pressure sensor. Depending on how smart you tried to make it you could try to do some interesting things. When you tumble or change position you basically get discontinuities in the graph. With enough brains you could try to peace that togather to extrapolate where you actually are at. The people at cypress have at times implied that they are doing some thing like that. I can also see it getting fooled like when you set up on opening under a snivelly canopy. Every body know that it will fire high that way. What I've always wondered about is why they don't have a three axis accelerometer in there. They've gotten pretty cheep and it's not like you need a fancy super accurate one. Even just the real simple questions like are you on your belly or on your back would allow you to correct a lot of your burble error. I'd really like to see how much correlation you could get out of the data. It's fixed pretty well to the rig and your back. How would it relate as you did transitions? And you could actually look at the total acceleration relative to the rate of change in pressure. It might let you distinguish between a sudden roll and a sudden deceleration. It could show you the difference between the pressure increase from setting up under canopy and the increase from a continued fast fall rate. Basically I think you could peace the discontinuities in the pressure curve together with the acceleration data. First with a burble correction to the data it self from the angle. Second with a correction at the points of discontinuity based on the total acceleration.

Lee

Lee
[email protected]
www.velocitysportswear.com