Hi all,
When cutting yesterday, I had an idea for a project: How about I create an app which will listen to LowRider and show you if it’s doing ok. I obviously lack experience with cutting to know what is good sound and what is bad sound - hell - I don’t even know if there really is possibility to run router by ear? What I mean exactly can be compared to car and manual gears: you just hear when to change gear.
Possible?
If you need an app for it, you shouldn’t use a CNC. 
It’s similar but still different for every machine, depends on DOC as well. So it sounding right is not the only indicator for it not destroying your endmills. 
It might be a really good idea. I bet you could pretty easily tell when a bit was getting dull.
Might be easy to use something directly touching the router though. So it can feel the vibrations. Too high you know it is getting close to failure.
that is not very friendly to newcomers…
I’m still new here and I just am not sure which sound means what. Some of them are really easy to catch I imagine.
thank you! vibrations would work as well, but then you need hardware, and with sound you just need something you already have in your pocket.
It wasn’t meant disrespectful, you really need to get to know the sounds your CNC makes and it’s not that hard. An app could lead to the user not thinking about it themselves, blindly accepting something an app tells them that was trained on the sound of a different machine.
Again, I don’t really think so. It makes a difference whether you are using a LowRider or mpcnc, Makita, AMB, cobalt etc.
But maybe I’m just too old and can’t go with the times. 
I see both sides here but I do lean towards what @Tokoloshe is saying. First, I think it would be difficult to create something that could do this. On top of different routers/machines, the sounds can be different in different material/feedrates/router speed/etc. Sometimes it will sound different just because you hit a knot in the wood or a funky spot in the plywood. It might even sound different depending on Lowrider gantry length. By the time you created it, you would have already learned what the different sounds are and would no longer need the app.
I obviously hate adding anything. In the end it is always adding cost or complexity. I would love for new users to more easily “know” if they have the right settings. Currently, for wood and plastic we have, too fast you get a bit breaking, slightly too fast we get poor dims and surface quality, wide “good margin”, slightly too slow we get burning/melting, way too slow burning melting bit breaking.
An app to know if you are cutting in the ideal range by sound, sounds extremely difficult if it is possible.
As for the bit-minder, I guess I just immediately imagine and little thing you strap onto your router hit a button during first cut to set zero or something and if that vibration increases by a certain amount, 80% for example, you get a bit warning. Now that I say it, swapping tools, putting it back in at a slightly different length would be impossible to track. I guess that is why big machine usually just track cut time per bit.
Isn’t it still just easiest to look at the chips it’s making? If you have chips, good. If it’s dust, it’s not and you’re killing your bit. I don’t think the issue here is when the bit gets dull from proper feeds and speeds, it’s that we’re killing bits early due to poor feeds and speeds?
I would be inclined to just have some gcode you can use to test a bit. It could cut a certain shape, then change the feedrate and cut the same shape again. Or do the same thing with router speed with pauses in between.
THIS 
It wasn’t until I read a post by someone here about chips vs dust that I started understanding how to make better cuts. I was running my router on too high of a speed setting. When I started slowing it down and watching the chips start to form, instead of dust, my cuts really started to improve. Also, I haven’t burned any of my bits since then either.
I understand dust vs chips, and I think I make proper chips?
I usually start makita at 1 and go up if router is struggling. Or, like with hardwoods, I start around 2 - just in case, and then slow down. I don’t know if it’s good or not, but whenever I measure temperature of the bit, it’s in 20s C… but I still made endmill completely dull while cutting out just two trays from hardwood ( Christmas gifts! Trays! ). I buy only cheap bits for now, cos I expect them to live short, but not that short, tbh?
Can anyone tell me how many hours of milling cheap/expensive bit should last?
That really depends. I once destroyed a 45€ 2-flute with extra hardening in about 20 minutes by going too slow, not deep enough. I went through quite a few 2-flutes before finally accepting Ryan‘s gospel about 1-flutes, 1800mm/min, 4.5mm DOC works fine in oak and ash and the 1-flutes last forever compared to the 2-flutes.
Does settings differ for 3.175mm or 6mm bits?
And “bit last forever” is 10h of work, or 100 hours of work?
In MDF I was getting about 5-7 hours of cutting time in a single flute 1/8" before they were dull enough to show a difference in tear out. I suspect natural woods would easily last double that provided you are not chewing through a bunch of knots.
Die a 3.175mm endmill it differs a bit because the diameter is less, so faster RPM are needed to rotate the same distance. Since you are running them more slowly, it‘s effectively similar. 
I get a few hours out of the 1-flutes before I can see the stepdown lines even though using a finishing pass. That‘s usually my indication the endmill is done for.
It might be a really good idea. I bet you could pretty easily tell when a bit was getting dull.
I know this thread is stale, but I just saw it and wanted to add a thought in case someone has the proper skills to do some testing.
Years ago, when I worked in a semiconductor fab, folks were using neural networks (aka AI) to monitor equipment sounds to determine when maintenance was needed. Typically, it would be a vacuum pump, but could be other items as well. The idea was that as parts became worn, subtle, typically inaudible to humans, sounds would occur that could be detected and by analyzing the audio spectrum, those inaudible components of noise could be used to alert to the possible need for maintenance.
As I said, I don’t have the skills to do this, but I’m betting someone does.
