mxk

Just playing with my Singer 401A and I'm having a bit of trouble finding the right setting for the bottom thread tension. I understand that the goal is to sew with the lowest tension that still gives you a good stitch. Am I correct in assuming that a good stitch, among other things, is one where you cannot tell the top side from the bottom? I don't have any problems adjusting the top thread tension so that the stitch locks in the middle of the material. However, I can easily tell the two sides apart because the bottom thread has a noticeable S pattern to it, while the top is completely straight. This seems to indicate that the bottom tension is insufficient, but the pattern remains even when I tighten the bobbin spring to the point where it feels too tight. At this setting I can easily lift the bobbin case by the thread and bounce it up and down without the case dropping at all. I've attached photos of what my top and bottom stitches look like with the loose and tight bobbin thread tension. I'm using E thread with size 18 needle, the material is two layers of 1000 denier cordura. Would appreciate any tips you may have on proper tension adjustment.

I got my V3 six months ago. Unless they changed it again in that time, and there is nothing about it on the website, the current design is split for about 4".

I'd like to understand this a bit better. By the old design do you mean the one for which the Split RSL-Vector 3 Modification was published? Thinking about the reserve deployment sequence with the Skyhook, in the old version the red lanyard would apply ~8 lbs of force (enough the break the safety thread) via the black tape directly to the edge of the Velcro, and possibly pass some of that to the left cutaway cable. I can see how that might be enough to pull the cable a few inches and release the left non-RSL riser. What the new version does is shift the load further up the Velcro and away from the cutaway cable. If the Velcro releases, the new version might still pull the cutway cable, but I'm assuming that it holds with sufficient force that the Skyhook safety thread would break first. Seems like this would only be a problem if you deploy the reserve with a somewhat functional main still attached, but still an interesting consideration.

I don't think so. First, if you only have one parachute, then it's also your emergency chute and 91.307(a) kicks in: "No pilot of a civil aircraft may allow a parachute that is available for emergency use to be carried in that aircraft unless it is an approved type and has been packed by a certificated and appropriately rated parachute rigger." Second, part 105 has specific definitions of a "parachute operation" and "parachutist," which is "a person who intends to exit an aircraft while in flight using a single-harness, dual parachute system to descend to the surface." If you don't meet that definition, then you're not doing a part 105 jump, which puts you in violation of 91.307(b): "Except in an emergency, no pilot in command may allow, and no person may conduct, a parachute operation from an aircraft within the United States except in accordance with part 105 of this chapter." But hey, if you can find a pilot who is willing to risk his certificate on this supposed loophole, have fun

"Will not kill you" is not the right criteria for selecting a canopy. A Spectre 190 puts him into the expert category, according to PD's wing loading chart. For what it's worth, I'm about the same size as the OP, just over 200 jumps, and very happy with my Spectre 230. No plans to downsize any time soon and wouldn't even think of jumping a 190 at this stage. Incidentally, I was demoing the Spectre about 140 jumps ago, flared too late as the canopy started sinking due to a wind gust, and this resulted in my only (very minor) skydiving injury. I was limping for a couple of days afterwards, but glad that I was on a 230 and not anything smaller.

Here's what it says on page 25: With the new limits, I would take the first of 10 years, 1,500 jumps, or Bat warning message. The date of the manual is 10/4/2006. You can get that from the PDF properties.

Depends on the canopy model and size. I can't stall my Spectre in full brakes even after shortening the lines a bit. From PD's Sabre2 Flight Characteristics: And from Getting the Best Performance from Your Canopy: I found this to be the best advice when adjusting the brakes on my Spectre. About an inch shorter than the factory setting resulted in a noticeable improvement in the flare while still keeping sufficient slack in the lines.

Same. I started with an Altimaster Galaxy on my wrist during AFF, switched to AltiTrack at ~80 jumps, added a Quattro audible at ~180, and moved the Galaxy to my chest strap at 200 for wingsuiting. I tried comparing the AltiTrack and Galaxy in flight this past weekend, but couldn't see the former well enough. During climb, the Galaxy seems to indicate 2-300 ft lower than the AltiTrack and the audible.

+1

Do you by any chance have the contact information of the person who did this presentation? I would like to ask them if they would send me some raw sensor data to play with. I wrote an AAD simulation in MATLAB that allows me to generate fake sensor data and then pass it through any imaginable algorithm to see what the theoretical best performance of a pressure-based AAD could be. Unfortunately, without having actual data or even knowing the exact performance characteristics of the sensor, it's hard to generate a realistic noisy signal. For example, I have no idea what a flip looks like to the sensor. If I could get my hands on some real, unprocessed, and preferably annotated data, we could have a bit of fun with the algorithms.

How do we know? And what additional parameters are those and how are they used? Temperature data is available, but that is useful mainly to adjust pressure sensor data. Time/timing is available of course. What else, and how is it used? We need to make a distinction between external information available to the AAD and information used to make the firing decision. Current AADs have just two input signals - pressure and temperature. You can think of time as a third. From those, the device extracts multiple features, which are used to update the internal system state. That state contains many variables, altitude and descent rate among them, which are functions of current and past sensor readings. All of these variables are fed into the firing logic. Thus, it's fair to say that there are many other firing parameters. For example, the armed/disarmed state is another input. However, all of them are derived from the history of a single pressure/temperature sensor.

Without real data to show that they do open too slow it's really hard to say. PD has probably done more testing than all their competitors combined. The same PDR design has stayed in production so long because of careful design and extensive testing. The incident filmed isn't even a smoking gun against the optimum. How was it packed? Did the PC get a good launch? Is the exact cypres firing altitude known? In a single incident there are too many unknowns. You need a careful statistical analysis of multiple reserve openings to draw _ANY_ conclusion. Not 1, not 2. Dozens. Opening softer can simply mean the peak opening force is lower as the force is better distributed over the entire opening time. I've had a 800' snivel on my katana that ended in a bang. -Michael The recent Mirage Trap video shows multiple Optimum and PDR deployments. If you compare the Level Flight test (Optimum) with Right Spin (PDR), I think it's pretty clear that the Optimum takes about 200 ft extra to open in these low speed malfunction tests. Yes, still one or two non-identical examples, but at this point I think there are enough of those to make some valid inferences.

The last jump in the video shows that the trap will disconnect if the RPC is generating more drag than the main. Hard to see, but I've attached the relevant frames.

Photos are attached. The Louie loops on the front risers tend to keep the grommet in contact with the slink, so you can see two marks there. On the rears, the grommet slides past the slink, so you can see one mark near the bottom and another one on the webbing.

What canopy was that? Full name and size? Spectre 230.

I put about 25 jumps on my new V3. It has standard risers, PD reserve slinks, and Dacron lines. PD recommended that I jump without the slink riser covers (SRCs) initially to see if there are any problems. I'll take some photos when I get home today, but I think there are definite signs of heat damage on the slinks (also discussed here). I jumped this rig without gloves for the first time a week ago and made it a point to feel the slider grommets immediately after deployment. They were rather warm, almost hot, and I obviously wasn't feeling the temperature right at the line contact points. I knew that Dacron lines would generate more friction and heat, so this is just a confirmation for me that some sort of protection is needed. I'm not really sure what to use. The ideal solution would keep the slinks and risers protected, but without constricting the lines or preventing the slider from coming down past the slinks. Some people put the SRCs upside down with standard risers, but that still leaves the slider in the contact with the lines and sitting higher than it would be otherwise.

Reserves bigger than 250 ft^2 and certified after 1994, that is, certified under TSO-C23d or C23f, are allowed to open slower. Some examples are the PD-281, OP-253, and the TR-425 I recently packed in a big-boy rig. How much slower? The short version is 0.01 seconds for every additional square foot over 250, or 1 foot of altitude loss for every square foot over 250. The TR-425 could open in 3 seconds + ((425-250) * 0.01) = 3 seconds + (175 * 0.01) seconds = 3 + 1.75 seconds = 4.75 seconds The owner of the rig was comfortable at belly-fly speed of 160mph. The additional 1.75 seconds means he would need an additional 400 feet or so of altitude in case of an AAD fire. Mark This is why I mentioned that usable flight time after opening is the ideal objective for determining when to fire, but then the AAD needs to know what reserve it is deploying. It will have to contain a database of known reserves and probably an option to manually specify the maximum deployment time and distance. Just to add a correction, the increase in time or distance is according to the maximum operating weight (MOW), not parachute size. TSO-C23f also factors in maximum pack opening speed (MPOS):

As long as we're adding a micro USB port, put it on the control unit so that it actually recharges the battery while you download your log files over the media transfer protocol with the reserve still packed.

A few thoughts on sensors and software, though hopefully you've already considered some of these issues: Your collections of sensors is good, and pretty much what I would expect from a newly developed AAD. What I don't like is that there is only one of each type (yea, battery/processing power and all that...). There is a reason why airplane autoland systems have three computers that are constantly voting for the right answer. A broken sensor can still be giving you data, but you have no way of measuring the quality of that data and nothing to cross-check it with. What does the AAD do if it suddenly starts getting bad or no data from the magnetometer? Your software is already complicated, but now you have to handle every possible combination of one or more failed sensors, which you may not even be able to detect without having more than one. This is especially true for the pressure sensor. I don't get why manufacturers do not put a second sensor in the control unit, or even better, route a small probe next to the cutaway housing to the front of the harness. Measuring the difference in pressures between two separate location, ideally one in the front and one in the back, can go a long way toward eliminating that 200+ ft error. It would also provide you with some information about the orientation of the jumper. It doesn't surprise me at all that Vigil, and probably CYPRES, are using very primitive (to put it mildly) algorithms for filtering and analyzing the data from the pressure sensor. Looking at the last five data points, if Lee's description is correct, is not the right way to do this. A Kalman filter should be the starting point for any such computations and there are other techniques for figuring out the "true" value from noisy measurements. Furthermore, you have to employ statistical tests to estimate the error for each calculated data point. It makes no sense to feed raw sensor readings directly into the algorithm that makes the firing decision. Going back to my autoland analogy, it's like building a system that tells you where the airplane is in relation to the runway, but doesn't bother figuring out whether the error on that answer is 5 ft or 50... That might be important. As for the firing algorithm, it's awesome that you're using "canopy over head" as your target. I've thought about this problem myself a while ago and came to the same conclusion. However, it's a well-known fact that some reserves open faster than others. Now your AAD needs to know whether it's making calculations for a PDR, which should open in ~2 sec, or an Optimum that will take closer to 3 sec (getting these numbers from UPT's BPA 2015 presentation). Size is another factor, as is the relationship between the rate of descent and canopy inflation time. On top of this, it sounds like you're not arming the AAD until it detects the exit from the airplane, trying to figure out the actual "mode" of flight, feeding it more data at a higher sampling rate, and greatly complicating the firing logic. At least for the mode of flight problem, I know I would be looking at machine learning algorithms, assuming there is enough processing power. Get enough examples of free fall, normal canopy flight, swoop, cutaway, etc., and train a neural net or to decide what is happening. The trouble with all of this - an absolute nightmare to validate and prove correct operation in the general case. That might be a problem when you try to convince people that your AAD will do the Right Thing(TM) more often than the (much simpler) existing products. Good luck! I, for one, am really interested in seeing what you come up with

At BPA 2015, Mark Procos from UPT mentioned that they were doing reserve deployment testing with different containers to check for AAD or any other compatibility issues: https://www.youtube.com/watch?v=1khn7DY9OPk#t=2112 The report was supposed to be published at the end of March in time for the symposium. Is there any more information about this?

Also, PIA TS-100 for terminology and diagrams: http://www.pia.com/piapubs/TSDocuments/TS-100.pdf http://www.pia.com/piapubs/TSDocuments/TS100-AT.pdf

For pre-stretching, do you mean that you use the leverage tool before installing the loop in the container? What do you use as an anchor? I also noticed that pre-stretching by hand (and I put about as much force into it as I can) doesn't seem to be enough for a stable loop length. For my last repack, I measured the force for pulling the temporary pin out after closing flaps 4 & 5, which was 18 lbs - higher than the manual recommends (8 to 12 lbs, page 37). I had to use the leverage tool to close the container and the pull force at the pin was ~28 lbs, if I remember correctly. Tested it again 24 hours later and got 14 lbs. Verified this again just now. Had I followed the suggestion for 12 lbs after flaps 4 & 5, my loop would have been too long. So much conflicting information... Kenneth Gajda is giving a Vector 3 packing demo at PIA next week. Anyone care to record a video for those of us who can't attend?

I've attached three photos of how the plate looks now. I also took hundreds of photos just after receiving the rig so that I know what various pieces looked like when they were brand new. One of these original photos happens to capture just a bit of the anchor plate. To my eye, it looks like the plate came bent at the sides from the factory. I can't understand how such a bend could be created from too much force applied at the grommet. The middle part is perfectly flat, only the sides are bent. On the other hand, this is exactly the type of bend I would use to give the plate additional strength.

Are you sure this is true? My understanding is that you can do what you want to your own main. Legally, the only thing that you can do if you're not a rigger or under the supervision of a rigger is pack the main if you're the next person jumping it. AC 105-2E also contains the following text, which is wide open to interpretation and probably will not carry the same weight as the CFRs in a court room:

Thanks for the information. I checked with UPT and they told me that the plate is supposed to be flat, but minor bends are normal and can be fixed by hand as long the grommets aren't damaged. The uniformity of the bend and the axis is what made me wonder if it came from the factory like that. As you say, if you pull too hard, you would expect the bend to form along the shortest dimension, not the longest one. I was going to repack everything this weekend anyway, so I'll try to take some photos from both sides of the pack tray before I do anything else to it. Another thing I found out during my discussion with UPT is that the staging loop was never needed for sport containers, and is only "recommended" for student and tandem rigs. Since this thread has morphed into complaints about the manuals, I'm curious how many other riggers were lead to believe that the loop must be installed with a skyhook rig?