Whats The Forums Opinion Of Mad Dog?

I think we have strayed a little from the second issue, that being if n= 2 is a good value.

It's obvious that the original question has been more than answered in here. Edge, enjoy your ATAK. Test it to your liking. If it doesn't fail your standards, then you should be happy with it.

Now, back to the second issue. N value for a given population varies with the factors you are looking for. Is n=100 enough to give you a good idea of how many humans have genetic mutations? Is n=5 enough to give you an idea of how many humans have two arms, two legs and one head? Isn't it important to know what you're looking for within that population in order to determine what n value will be needed or at the very least get an idea of what n value you will need.

So for a bell curve of the general population, you would eliminate the ends of this curve and still get a diverse group but without the extremes. Lets start with the cheap pakistani butter or taiwan castings. Investment or other casting methods for steel has a high internal porosity factor and a high chance of voids developing. So for this you will want a high n factor compared to your production population. So now, you use cold or hot rolled steel and machine from billets thereof. Now you have improoved the quality and moved yourself up the bell curve a little towards the higher end.

Now for heat treatment: The pakistani/taiwan blades again offer a miriad of heat treats completely unintentionally. you can get brittle and ductile in the same knife. This is a large population factor. The more samples you test the better idea that you will have junk. High quality knives have set heat treats that are usually done to high standards. This again moves your knife to the higher end of the curve.

Quality control. Not much for Pakistani and Taiwan stuff. Quality knives? you bet. Very good quality control. Again moving you up the curve.

Continuing with these factors brings you to a very small population size at one extreme of the bell curve.

So now you have(assumption) a population size of 600 high quality knives, with high production standards, high QC standards, and high quality of materials used, with their own high QC standards. We know that Rc is 63 at the edge. In O-1 this should make for a fairly brittle edge. We also know that the spine is kept softer for some ductility and toughness. We also have a flat ground blade that gives considerably less material very quickly as you travel from spine to edge. Another factor is the very long blade which will definitely be a factor in bend strength. From this you can postulate that the knife may fracture if high lateral stresses are imposed on the knife and especially on or near the final temper line transition to Rc=63. You test one knife and it fails prematurely. It is said that it had a flaw. Lets for now assume that it is true that the first one had a flaw. So now we test a second example and it fractures at or near the temper transition with very little lateral force being applied. Again another premature failure.

Conclusion 1: If you assume that the first one was flawed than you have an n=1 and need at least one other data point to assure accuracy, or in other words another test.

Conclusion 2: If you assume that the first one was not flawed and was a typical sample, then your n=2 should be enough to tell you what you already knew; that the knife was to brittle and the differential temper did not help in increasing the toughness of the knife. The more data points you have, obviously will give you more conclusive results. But who is willing to subject another $900 knife to that.

Will an ATAK do the same. It is much stouter due to it's being shorter for the same thickness. A different knife needs separate tests.

I have to go to work now, sweeping the halls of my local office building, seeya.
 
Hoodoo,
This is really comparing apples to apples.
In all of the threads that Cliff has posted I have never found him to be arrogant but rather temperate during conversations where I would have lost my cool.

Concerning Cliff's testing, he has earned the respect of many on these boards and he has remained objective through many conversations where others have resorted to insulting him.

Concerning Cliff's aims he summed it up when he stated:

There is currently a group of people that are working to try an develop a comprehensive series of tests for knife evaluation. These will include materials properties as well as field work and will correlate the two. Assuming this comes to pass will such work on a regular basis be published ?


Hoodoo, if you know of the existence of such a work already published I will be interested in the reference.


[This message has been edited by Donna Barnas (edited 05-02-2000).]
 
On topic:

1. What do you think of Mad Dog knives? I like them and I do own quite a few. I think the warranty sucks for such a highly prized and priced knife.

2. Are they all hype? There are some hype that can not be verified and some that can be disspelled. Check the archive.

3. And can you get a more superior knife for the buck? Overall, yes you can.

Off topic:

n=2 is indeed a good number. My company sold thousands of parts (electronic components) to the DOD. We normally take 2 production parts randomly from the assembly line and run Reliability Development and Growth Test (RDGT) on them to establish the Mean Time Between Failure (MTBF) number.
 
NamViet

You stated that you normally take to parts from the assembly line.

This sounds to me like you are taking two parts per batch or production run. If this is the case then your n-number is actually quite high because you total the number of samples you take and it will become much larger than 2, if you have run more tha one batch. If you are running two shifts per day and 2 to 3 batcher per shift that would be 12 samples in one day. multiply that by 365 days and in one year that is a large n number.

Now if you mean that your company has only take two samples for testing, ever, since the begining of production of your product, then that is scary. I have not seen this done anywere in the manufacturing industry and I have seen a lot including 6 inch and 8 inch wafer manufacturing, gage manufacturing, hydraulic system mfg, injection molding, pump manufacturing, gun manufacturing, aircraft industry, etc.
 
Walt,

You question was rhetoriacal and don't call me surely
smile.gif


Can you guys beleive I have managed to stay out of this
smile.gif


BTW I have a Mad Dog ATAK Bayou for sale!
No really I have one!

------------------
Best Regards,
Mike Turber
BladeForums Site Owner and Administrator
Do it! Do it right! Do it right NOW!
Show Your Support -Visit Our Sponsors - Click On The Banners!
Visit www.onestopknifeshop.com
All sales from 1 Stop Knife Shop help support this site!
 
Cobalt, we take two and only two from the production run. Our products are proven in our material selection and manufacturing processes. This way, we can produce toilet seat for $2 each instead of $200.
 
Nope, I got it from another dealer and I will continue to get products when available. I have no desire to become an official Mad Dog dealer. Should the need arise in the future it would be nice if Mad Dog had a dealer and consumer policy that is conducive to doing business.

------------------
Best Regards,
Mike Turber
BladeForums Site Owner and Administrator
Do it! Do it right! Do it right NOW!
Show Your Support -Visit Our Sponsors - Click On The Banners!
Visit www.onestopknifeshop.com
All sales from 1 Stop Knife Shop help support this site!
 
Namvet, I understand what you mean. In low quality parts you really don't need much checking. I saw this at an iron ball casting place that casts them for ball mills in the tertiary crushing process of a copper mine. Really didn't need to test them much.

But using your quote on the fact that the materials were tested so you don't really need to test that many parts, then almost all steels available have stress/strain diagrams from testing as well as charpy values from impact testing. All you need to know about any material is there so why test them. Because of Imperfections, the human factor, innovative processes that are not completely proven, etc. That's why.

Look at the american car industry in the late 70's. They were absolute junk. No one cared, quality control was low, testing was non existant, etc.
 
Cobalt :

Isn't it important to know what you're looking for within that population in order to determine what n value will be needed or at the very least get an idea of what n value you will need.
Click to expand...

Of course, many of the statistical tests being referring to in the above are used when you are trying to make a difficult decision. For example if you took two world class 100 m sprinters and wanted to decide which one was faster would one run be a good method? No. The difference in performance from one run to the next could easily be of similar magnitude as the difference in performance between the two runners. However if you wanted to know if I was a world class sprinter one run would tell you that I am not immediately.

The sampling statistics are used, in the way being described in the above to allow you to estimate the population mean *and* variance, and thus make some kind of deduction about it - does it satisify a particular claim or similar. In a lot of cases you can estimate the variance independently, if this is the case a sample size of one can be used to draw stable conclusions from. Getting specific to knives how to you estimate the variance in a makers abilities? Well you ask the maker for one thing he should have a very good idea assuming he is making using knives. You talk to others who have used his products and you reason based on experience with other knives that you have had.

For example, lets assume that you want to have your fault predictions at the 5% p value which HooDoo referred to (which I don't think is reasonable because of the size of the Type I error but anwyay). For this to happen all that you need is for the makers fault rate to be 5% or less. So you have to decide if this is reasonable. Could a maker be screwing up one out of every 20 blades. I don't think this is in anyway reasonable. If it was it would quickly become known that his QC was very poor. However lets assume that you think that it might be possible. What to do then? Well you decide that in order to actually make a decision you would want two faults in a row, that is to say you would always allow a maker to give a replacement before making a decision. If you do this then your judgements will pass the 5% p value unless the makers QC is so low that he faults about 25% of his blades. Any one can conclude that this is completely insensible.

This does however bring up an important point, you don't want to do reviews completely in ignorance of everything else that goes on around you. Read the other posts discussing the product, discuss the results with the maker, and other makers/users, and use all this information to make your decision. The more information you bring in the more stable a conclusion you will make.

Donna, if you could, could you glance over the following thread :

http://www.bladeforums.com/ubb/Forum39/HTML/000228.html

And give me your opinion on the technique described. I realize that the particular blade in question is not one you would prefer but I would like your perspective on the kind of strike that would be optimum for that particular cutting test.

-Cliff
 
Originally posted by Cobalt:
Look at the american car industry in the late 70's. They were absolute junk. No one cared, quality control was low, testing was non existant, etc.
Click to expand...

My old man got 280,000 miles out of a '78 Ford Fairmont station wagon, and it never went into the shop! He also uses the same Camillus hunting knife he got when he was a kid. And he kept a Craftsman chainsaw running every fall for 25 years. That ol' buzzard didn't throw anything away.



------------------
David

Hooray for Hollywood (Maryland)
 
David, Out of 100's of 50 cent cigars, I have had a few that were really good. The rest were all crap! I had a 1993 Ford Van whose transmission started slipping and burning fluid at 54,000 miles. I had a 1997 Ford Taurus whose transmission would slam into gear, with some sort of slippage at 44,000 miles. A friend had a Ford 1988 ranger, that had to have it's engine rebuilt at 100,000 miles. He also got a valve job after another 77,000 miles on the same truck. Another friend had a 1993 Ford F-350 whose rear drive shaft broke while towing a 17 ft little boat to the river. The drive shaft pogoed and tore up the bed of the truck and sent the boat into the weeds. The same truck had to have it's rotors replaced at under 40,000 miles. Another friend whose father worked for Ford, always had problems with their 1990 mustang, always in the shop. Another family friend had to replace the water pump in the 1992 thunderbird at 60,000 miles. Shall I go on.

I know several people who own toyota's, honda's, nissan's and mercedes, that have gone over 200,000 miles without a problem. I have personally had a Jetta go to 180,000 miles and the engine still ran like new. I have a japanese luxo car with no engine problems and it has 150,000 miles. When I went to take it in to the dealer for service at 100,000 miles the dealer told me that it was just getting warmed up. The car looks like new. As does the paint.

Do you think you would use the same n number for these two different examples? nonononononononononono.

Don't get me wrong I want american made cars, but not at my expense. I bought an american full size truck in the last 2 years and it has not given me any problems. So far so good. I knew several people that had the same truck with no problems before I bought it. So I took a chance and lucked out.
 
You must log in or register to reply here.
Back
Top