|
Post by runner42 on Jun 24, 2016 8:49:46 GMT
I am posting this more in frustration than soliciting advice. I am querying the efficacy of producing a stainless steel ball on a Viton O ring Clack Valve, this configuration has been advised as a sure method of obtaining a positive seal. I have produced two for a boiler and these were tested using an air compressor. After a momentary leak they both sealed perfectly at 90 PSI, which I thought was indicative of a useful clack valve. But no when fitted they initially stopped water leaking from the boiler but as soon as heating was applied they immediately started dripping and continued until about 30 PSI was indicated on the pressure gauge, which appears to be the minimum pressure required to force the ball into the O ring for a perfect seal.
Looking at the stainless steel ball and O ring interface shows that the O ring cross section is circular and the stainless steel ball is also circular (apologies for stating the obvious) so it lacks the same interface as a stainless steel ball resting on a D bit finished metal face. The latter relies solely on the weight of the stainless steel ball to provide the closing pressure and the interface having no compression is not pressure sensitive to variations in boiler pressure. It seals from 0 to maximum pressure. The O ring seating does not provide the same characteristics as a D bit finished metal face. A useful clack valve should provide sealing for 0 to maximum PSI at both increasing and decreasing pressures.
For the record I used a 1/4" OD, 1/8" ID O ring with a 3/16" stainless steel ball.
Any comments would be appreciated.
Brian
|
|
|
Post by keith6233 on Jun 24, 2016 9:28:08 GMT
|
|
|
Post by Roger on Jun 24, 2016 18:09:10 GMT
It seems likely to me that the leak isn't between the ball and the 'O' ring, both of which will have been produced with exceptionally close tolerances and excellent surface finishes. I suspect it's small irregularities in the machined face that are causing the leak. Presumably the 'O' ring has clearance in it's pocket, so the 'O' ring can be sitting across any ridges on the 'flat' face if it's not sitting concentrically. The flat face needs to be flat and smooth to get a seal. If the face is a shallow taper because the 'D' bit wasn't truly square to the axis of rotation, an 'O' ring sitting out of centre would form a slightly elliptical shape for the ball to sit on.
|
|
pault
Elder Statesman
Posts: 1,496
|
Post by pault on Jun 24, 2016 19:06:46 GMT
When I made some oring/ball clacks I found the oring needed to be a snug fit in a counterbore. This arrangement did sometimes leak a little when cold but sealed with a few psi. If the oring was not a snug fit the water flow could displace it causing the clack to blow by big time can you post a picture of the internals of the clack?
|
|
|
Post by runner42 on Jun 25, 2016 5:51:59 GMT
This is a sketch of what I produced. Let me say that I did not D bit the face that the O ring abuts to because I assumed that because it is circular in cross section it is of no consequence to have this so, it is only to capture the O ring and prevent movement. The O ring is sitting squarely on the bottom fitting.
Brian
|
|
|
Post by runner42 on Jun 25, 2016 6:25:05 GMT
Thanks for the link. I have in fact purchased a number of the shuttle O ring type clack valves from the UK. Unfortunately I have experienced similar problems with these as I have with the ball O ring type.
Brian
|
|
|
Post by Roger on Jun 25, 2016 6:50:47 GMT
Hi Brian, Just a few thoughts on the sketch. I presume that there's a small axial clearance on the 'O' ring when it's assembled so that the ring isn't being influenced by any irregularities on the bottom of the pocket by being trapped between the two parts?
It looks to me like the ball is a very close fit in the bore, and that it has to lift past the exit hole for any water to pass. If the axis of the ball hole isn't concentric with the one constraining the 'O' ring, you won't get a seal until the 'O' ring deforms. I would have thought that there would be a significant clearance around the ball.
I'd be inclined to polish the flat face of the bottom fitting where it touches the 'O' ring since this has to form a seal. I'd use a piece of 1500-2000 grit wet & dry on a surface table and use small circular movements to get that end really flat and smooth.
I also presume that there's more lift than is implied by the drawing.
My last suggestion is to ditch the Stainless Steel ball and replace it with a Silicon Nitride Ceramic one which is much harder and less likely to have surface blemishes.
|
|
|
Post by runner42 on Jun 26, 2016 4:43:17 GMT
Hi Brian, Just a few thoughts on the sketch. I presume that there's a small axial clearance on the 'O' ring when it's assembled so that the ring isn't being influenced by any irregularities on the bottom of the pocket by being trapped between the two parts? It looks to me like the ball is a very close fit in the bore, and that it has to lift past the exit hole for any water to pass. If the axis of the ball hole isn't concentric with the one constraining the 'O' ring, you won't get a seal until the 'O' ring deforms. I would have thought that there would be a significant clearance around the ball. I'd be inclined to polish the flat face of the bottom fitting where it touches the 'O' ring since this has to form a seal. I'd use a piece of 1500-2000 grit wet & dry on a surface table and use small circular movements to get that end really flat and smooth. I also presume that there's more lift than is implied by the drawing. My last suggestion is to ditch the Stainless Steel ball and replace it with a Silicon Nitride Ceramic one which is much harder and less likely to have surface blemishes. Hi Roger,
it is only by measurement using a digital vernier as a depth gauge have I established that a small clearance exists between the bottom fitting and the O ring. I don't think the O ring is being compressed, I have eased the bottom fitting a little further out just to make sure. No change in performance.
I increased the 13/64" measurement to 7/32". No change in performance.
The sketch is only that not a scale drawing the actual clearance for the SS ball is 1/32".
I shall try improving the bottom fitting face on which the O ring sits as you have suggested. I didn't think that it was an issue because the clack valve does seal. I think your final point in using a Silicon Nitride ball has more relevance because although you indicated in an earlier post that the SS ball is made to high tolerances, I am not so sure that this is so and could be the reason for my protracted struggle with all the clack valves I have made is perhaps due to BAD BALLS.
Brian
|
|
|
Post by Roger on Jun 26, 2016 7:24:45 GMT
Hi Brian, Contrary to what most people will tell you, Stainless Steel balls are not that hard and can be scratched or deformed without difficulty. Hit one with a hammer and you'll see. I wouldn't recommend you do that with a Silicon Nitride ball though, in case it shatters. They are as different as chalk and cheese. It's also hard to gauge the quality of Stainless Steel ball bearings, they won't be used in anything like as arduous applications because they're too soft. Silicon Nitride ball bearings, on the other hand, are usually selected for the most difficult applications so are likely to be made to higher tolerances with that in mind.
I would be surprised if a Viton 'O' ring would seal at near zero pressure though, the assumption is that it's a perfect toroid in all dimensions and a perfect surface finish. Obviously the closer all these parts are to the true geometry and perfect surface finishes, the lower the pressure it will seal at. I guess this is one advantage of the old style of metal ball on a metal seat if they are lapped in. Again, this is easier to achieve with Silicon Nitride balls which will retain their shape and resist scratches more readily.
|
|
|
Post by runner42 on Jun 26, 2016 23:33:48 GMT
Roger I have ordered Silicon Nitride balls but they are coming from the USA so delivery will take up to 3 weeks. If I go back to the original ball/D bit machined interface type of clack valve, using Silicon Nitride balls, how does one establish the final finishing of the face if one cannot hit the Silicon Nitride ball with a hammer as is recommended for SS balls.
Brian
|
|
|
Post by Roger on Jun 27, 2016 7:10:18 GMT
Hi Brian, I'm surprised they're coming from the US, I get mine from eBay and I haven't seen a supplier who's not in China. They normally take that sort of time to come to the UK though, I think they ship them by sea.
There's nothing to stop you using a hammer onto the seat, you won't shatter one like that. Just don't put one on an anvil and try to flatten it! I think you've better off crushing the ball into the seat rather than using a hammer.
Grinding/Lapping the ball into the seat is probably a better option. I've used a length of rod that's smaller than the ball diameter, centred the end deeply, and then used Araldite to bond a ball onto the end of it. You can then use it in the same way you would grind in a valve on an internal combustion engine. The finer of the two grinding compounds they use for that is still too coarse for this job in my opinion, but you can get any grit you want from eBay.
|
|
|
Post by Jim on Jun 27, 2016 7:55:18 GMT
I must have missed something but I can't see any mention of the amount of lift you have in the ball Brian.
|
|
|
Post by runner42 on Jun 27, 2016 23:48:41 GMT
Roger thanks for the grinding procedure. I selected a USA source on Ebay for the Silicon Nitride balls because I am cautious about using balls that may be not quite the best. I need everything to be spot on so I can achieve a useful clack valve. I hope the USA supplier didn't get the balls from China.
Jim the lift for the ball is 1/32".
Brian
|
|
|
Post by Jim on Jun 28, 2016 2:04:22 GMT
According to Everett Clem's seating data for ball check valves your 3/16" (.1875") SS ball would need a seat diameter of .132"< .156" to seal properly. The amount of lift depends on whether you are using the check valve with a reciprocating pump where it should be 1/8th ball diameter, 1/6th diam for hand pump and 1/3 for an injector. No 'O' rings are used in these calculations. I've made quite a few check valves using this data and all work perfectly. The only problem I encountered was early on when I had too much lift in the check valve to the boiler which was fed by a mechanical pump with the result I had back flow as the ball didn't reseat between pulses.
Hope this helps, Jim
|
|
|
Post by cnc3d49 on Jun 28, 2016 7:10:26 GMT
Gentlemen
I've read this thread twice yet see no mention of the grade of balls used, nor the effects of the coefficient of expansion of O rings.
Having produced many clack valves, both conventional ball on bronze/brass seat and O ring shuttle valves, I have found the following. For conventional ball on metal seat, the grade of ball is important, grade 100 or better is essential. No ball is perfectly spherical and grade 100, 100/millionths of an inch out of true, seem satisfactory. Most model engineering suppliers are unable to confirm the grade of ball supplied and many appear to be unfit for the application.
Regarding O ring valves, the design outlined in Gentoo's journal work very reliably and machined accurately are dry at zero pressure difference. The designs where the O ring forms the seat, whether trapped or contained seem to suffer leaking until there is a significant pressure difference which I suspect is due to deformation as the O ring expands at a faster rate than the valve body. This is likely to cause the O ring to deform because it is trapped in the valve body.
I hope my observations are of value.
GeoffV
|
|
|
Post by runner42 on Jun 28, 2016 7:12:43 GMT
Yes I think that helps. If we are persisting with a SS ball/O ring check valve are you suggesting that say an 1/8" ID O ring would require a smaller than 3/16" SS ball? Say 5/32"? It's been advised on this forum a 5/32" ball in a 1/8" hole achieves better results than a 3/16". Your ball lift criteria is new to me I hadn't realised that what is feeding the check valve influences the amount of lift required, so thanks for that.
Brian
|
|
|
Post by runner42 on Jun 28, 2016 7:30:11 GMT
Hi Geoff,
you posted whilst I was compiling a reply so was unaware of your post. All inputs to increase the understanding of how such devices work is of value. I take on board your criterion for ball accuracy and hope that the Silicon Nitride balls meets this. I also note your comments on the coefficient of expansion of a Viton O ring is different from say the brass housing in which it sits. I assume that axial (across the OD) expansion will be the greatest and some allowance is required for this to occur. Without going into details would say a 1/64" (0.015") be sufficient?
Brian
|
|
|
Post by Jim on Jun 28, 2016 7:32:59 GMT
Brian the data I've supplied is for your 3/16" SS ball sitting on a metal orifice of between .132" and .156" without any 'O' ring. It isn't needed as the hole to ball ratio will seal perfectly when the ball reseats after each pulse. This is why they are called clack valves as it is possible to hear the 'clack' on bigger valves as it reseats. On my Burrell I can check the clack valve is working as it will be cold to touch if it is and b@##dy hot if not. Jim
|
|
|
Post by runner42 on Jun 28, 2016 7:54:16 GMT
Jim thanks again. An interesting observation in using touch to determine if the clack valve is working, I often wondered how in a closed system how one would determine correct operation. It seems as an operator we must use all our senses to evaluate the correct operation and touch particularly in a steam environment is just as important as sight and sound.
Brian
|
|
|
Post by Roger on Jun 28, 2016 8:26:06 GMT
Gentlemen I've read this thread twice yet see no mention of the grade of balls used, nor the effects of the coefficient of expansion of O rings. Having produced many clack valves, both conventional ball on bronze/brass seat and O ring shuttle valves, I have found the following. For conventional ball on metal seat, the grade of ball is important, grade 100 or better is essential. No ball is perfectly spherical and grade 100, 100/millionths of an inch out of true, seem satisfactory. Most model engineering suppliers are unable to confirm the grade of ball supplied and many appear to be unfit for the application. Regarding O ring valves, the design outlined in Gentoo's journal work very reliably and machined accurately are dry at zero pressure difference. The designs where the O ring forms the seat, whether trapped or contained seem to suffer leaking until there is a significant pressure difference which I suspect is due to deformation as the O ring expands at a faster rate than the valve body. This is likely to cause the O ring to deform because it is trapped in the valve body. I hope my observations are of value. GeoffV Hi Geoff, Those are very interesting points. Just taking the bearing grade one, the Silicon Nitride balls I get from China are listed as grade 5 which is what are used in some precision high speed bearings, but you can find grade 3 ones if you want. In other words, these are 5/1000,000 accuracy compared to the 100/1000,000 you say is the minimum. I don't share Brian's worries when it comes to precision items made in China, they produce superb Carbide products for bearings and cutting tools and I've been really impressed with the quality. Cheap goods are another matter. Here's a link to some interesting information about ceramic balls and bearings.
|
|