Jump to content

Perfect Mechanical Filter


Recommended Posts

Many of you probably know the Mechanical Pipe Filter as originally invented/discovered by PVD. It has a major advantage that it doesn't use any power to filter, and a few disadvantages - particularly that it spills (when a full packet comes to its input, small part of the packet continues to the output) and overflows (when the filtered output blocks, all the incoming pipe contents continues towards the output, including the contents that should be filtered out).

Here's a little more complicated solution that works exactly like powered filter, i.e. it blocks if either of its outputs blocks, and does not let any of the filtered medium to the unfiltered output.

It consists of two mechanical filters, the second of the two taking care of the spills of the first one. Then it continues into merger, where large packets split by the two filters are again recombined into single packets. And then the output is followed by a backfeed where any filtered medium that did not make it into the output goes back to the input, eventually blocking the input pipe.

The schema looks like this:

F7PZx9W.jpg

And there are many ways to "compress" the device into less space. It can then look e.g. like this:

yr2ER1W.jpg

Note the filtered output is the bottom one in the bottom image. It could be made even smaller using valves instead of bridges for merging/priority joints.

Priming is relatively easy - set both valves to 1 g/s and send a continuous stream of at least six (or whatever is the total length of filtering loops) full packets of filtered material to the input. Both valves will fill their packet loops while already making sure nothing of it will get to the unfiltered output. Once the device is primed, you can send mixed input to it and it will work just like normal filter.

A simple variant can be made with a single filtering loop if you put a valve in the way that will reduce the input rate to the maximum rate the filter can filter out without spilling:

1xBA3mz.jpg

I wouldn't consider it a "perfect mechanical filter" since its flow is slightly reduced. It suffices when placed behind e.g. a gas pump, though.

Disclaimer: I've been out of the loop lately so please forgive me if someone else came with the design earlier. I am not intentionally copying anybody, the solution just occurred to me and I wanted to share.

Link to comment
Share on other sites

Welcome back.

Generally it's assumed not to be a good solution to have a direct feedback loop of unfiltered gases/liquids because there is always a reason for why it wasn't filtered. Usually it is because one or more of the filtered output lines is blocked, or it is simply not a gas/liquid the filter can filter. So it is generally considered better to have a filter overflow room that can warn yo that the filter is experiencing overflow, and you can do something about it.

Like illustrated in this variant of your original double bridge full packet variant of PVDs design.

There's also no real need for the repacketer to be part of the filter. If the specific line needs full packets, most people would just use a packet stacker on that specific output.

 

Link to comment
Share on other sites

25 minutes ago, Saturnus said:

Generally it's assumed not to be a good solution to have a direct feedback loop of unfiltered gases/liquids because there is always a reason for why it wasn't filtered.

Thank you for your concern but that's not how this design works.

25 minutes ago, Saturnus said:

There's also no real need for the repacketer to be part of the filter.

There are two good reasons. One, it makes sure neither of the two filter loops ever block. And two, it delivers single packets on input as single packets on output, just like the powered mechanical filter. Besides other, it may also save on processing cycles on machinery downstream.

But feel free to consider it optional in your implementations.

Link to comment
Share on other sites

2 minutes ago, Kasuha said:

Thank you for your concern but that's not how this design works.

I appreciate that but just making you aware that it is not considered a good design choice to let filters block when you can use overflow buffers instead.

6 minutes ago, Kasuha said:

There are two good reasons. One, it makes sure neither of the two filter loops ever block. And two, it delivers single packets on input as single packets on output, just like the powered mechanical filter. Besides other, it may also save on processing cycles on machinery downstream.

There could be a few instances where this could be useful but in such cases most would as I noted use a repacketer to make sure the output is full packets, or no packets.

Link to comment
Share on other sites

7 hours ago, Saturnus said:

I appreciate that but just making you aware that it is not considered a good design choice to let filters block when you can use overflow buffers instead.

There could be a few instances where this could be useful but in such cases most would as I noted use a repacketer to make sure the output is full packets, or no packets.

I don't really care what's 'considered a good design choice'. My criteria involve if the thing works, if it fits the purpose, and if I can save power by building it. Standard filter doesn't come with overflow buffers, either, so I'm not sure why are you even bringing it up.

I guess you just have urge to say something negative even though you don't have any rational arguments. Well, you may not need my design, I'm fine with it. I did not post it for you, then. I posted it for those that may have use for it.

 

Link to comment
Share on other sites

So guys, I've been running yalls old school variant with the bridge loop overlock and I stuff it slam full of too much of everything and give it a cycle and it clears.... image.thumb.png.9d9f9998cc64e8fd90eb6b055d7f0bdf.png

So if what Kasuha is saying is true then you could bypass the need for that round table refilter by having each gas get refiltered :)

Link to comment
Share on other sites

Ah yes, wouldn't be a real ONI design thread without Saturnus talking down to people in the replies.

21 hours ago, Saturnus said:

Where was there anything negative in my comments?

Kasuha explicitly said he didn't want his design to overflow and you went out of your way to tell him it was a poor design choice.

22 hours ago, Kasuha said:

I posted it for those that may have use for it.

Ignore him, thanks for the design, I'll definitely find a use for this.

Link to comment
Share on other sites

Archived

This topic is now archived and is closed to further replies.

Please be aware that the content of this thread may be outdated and no longer applicable.

×
×
  • Create New...