|
shovelbum posted:Vibration, heat (commonly 50C+, infinite salt/SOx/humidity), even the air is filthy with oil and particulates, the requirement to build to an incredibly tight price point, shoddy shipyard wiring and installation, 400 coats of paint, etc. The fact that sometimes your bilges flood with fuel (don't forget it's solid at temps up to 38C but will be 100C+ when it hits the bilge on a bad day) and not water. Or acid (used to clean everything). Don't forget trash and rats!
|
#
?
Oct 15, 2019 14:46
|
|
|
|
| # ? May 9, 2024 19:56 |
|
|
shovelbum posted:No offense but I'm a marine engineer and every aspect of this project gives me cost, regulatory, and maintenance nightmares. If this is for a large fully unmanned ship your client needs to be looking much more at ground up reconceptualization of their plant design than a gigantic mess of wiring. I was hoping nobody was gonna latch onto the bilge/marine use part b/c it’s distinctly possible this is not actually a system for bilge pumps specifically but instead a terrestrial application with broadly-similar use conditions  (i am exhaustively paranoid about posting anything halfway-identifiable about ongoing projects, sorry for the wasted well-intentioned “this seems ill-advised for a bilge” guidance posts)
|
|
#
?
Oct 15, 2019 16:25
|
|
|
Foxfire_ posted:Imagine holding the top of a tube still with your hand and tracing a circle with the bottom. That's the motion a vortex mixer makes. If you spin the tube around its central axis, you've built an axial centrifuge instead and it'll separate the contents instead of mixing (densest stuff out on the walls of the tube). Yup, you're 100% correct. I was looking at DIY builds like this and got bad ideas. https://www.youtube.com/watch?v=YR40KgQAnK4 Thanks (for Dairy Days too!) for the heads up and for saving me a lot of effort, because I don't really need a centrifuge right now 
|
|
#
?
Oct 15, 2019 18:08
|
|
|
On the mechanical side of things, I think you could generate the vortex mixer motion fairly easily, with just a bearing where the inner diameter is larger than your motor shaft. Then you cram a puck in the center of the bearing, drill a motor shaft sized hole off-center to the puck, and put a rubber cap on that only touches the outer race. Done!
|
|
#
?
Oct 15, 2019 18:15
|
|
|
Ambrose Burnside posted:I was hoping nobody was gonna latch onto the bilge/marine use part b/c it’s distinctly possible this is not actually a system for bilge pumps specifically but instead a terrestrial application with broadly-similar use conditions If it's not for a bilge all the advice you've been given is good to go. The marine environment and especially the large-ship environment where some poor bastard might think this level of retrofit desirable are so unique bc of the oily saline air, heavy regulatory burden on any bilge systems (bc they're where oil pollution control meets flooding control), vibration, lack of maintenance/tenuous parts availability, conservative customer base, and lack of basically consistent local gravity bc you might be rolling 30 degrees to each side on a very horrible day. If it's for terrestrial heavy industry and you want to incorporate flow meters or whatever be aware a lot of like ore/wood pulp/etc processing environments are going to have incredibly abrasive wastewater.
|
|
#
?
Oct 15, 2019 20:03
|
|
|
shovelbum posted:If it's not for a bilge all the advice you've been given is good to go. The marine environment and especially the large-ship environment where some poor bastard might think this level of retrofit desirable are so unique bc of the oily saline air, heavy regulatory burden on any bilge systems (bc they're where oil pollution control meets flooding control), vibration, lack of maintenance/tenuous parts availability, conservative customer base, and lack of basically consistent local gravity bc you might be rolling 30 degrees to each side on a very horrible day. I regret going with the bilge thing b/c it's several magnitudes more challenging an environment than the reality (stable temps/conditions, fluid is water w low suspended solids, and it's wastewater so no potability/human-use considerations irt part selection) and yeah, jesus, everybody's gonna make sure there isn't some dude about to chuck a Hobbyist Sensor Array (ft. Poorly-3D-Printed Custom Enclosure) into an actual bilge of an actual ship. of course. thats a good thing, but like. man i gotta think before i post, some times
|
|
#
?
Oct 15, 2019 21:06
|
|
|
Ambrose Burnside posted:I was hoping nobody was gonna latch onto the bilge/marine use part b/c it’s distinctly possible this is not actually a system for bilge pumps specifically but instead a terrestrial application with broadly-similar use conditions  ??
|
|
#
?
Oct 16, 2019 03:45
|
|
|
gotta be
|
|
#
?
Oct 16, 2019 04:23
|
|
|
move to a state where it's legal so we can give better advice on your Bilge Weed setup tia
|
|
#
?
Oct 16, 2019 06:19
|
|
|
Yeah hopefully it's this and he's not building a nuclear reactor cooling system like um excuse me.
|
|
#
?
Oct 16, 2019 14:36
|
|
|
*sweats in Roentgen*
um excuse me fucked around with this message at 14:51 on Oct 16, 2019 |
|
#
?
Oct 16, 2019 14:44
|
|
|
Hmm, not great; not terrible.
|
|
#
?
Oct 16, 2019 15:10
|
|
|
mobby_6kl posted:Yeah hopefully it's this and he's not building a nuclear reactor cooling system like um excuse me. Going for your merit badge?
|
|
#
?
Oct 17, 2019 01:51
|
|
|
Splode posted:You can also use the aircraft safety approach and use triple redundancy. If one sensor goes bad, you'll know because it'll disagree with the other two, allowing it's data to be discarded until you replace it. Obviously this approach is best suited to cheap and easily replaceable sensors, but it does let you get away with less reliable parts When doing redundant sensors for critical systems it's also a good idea to design each sensor set to work on completely different physical principals to lessen the chances that some odd edge case 'tricks' one kind of sensor into reading a false positive since that same 'trick' could affect multiple sensors of the same type. For the case of a pump that needs to produce a known flow rate you could make sensor set one monitor pump current and a dP across a Bernoulli gauge, while sensor set two could be a set of load cells in the pump mounts that show torque is actually being created and a sonic flow sensor tube to ensure that stuff is moving through the pipe. The two systems need to agree that everything is working within some reasonable error bars otherwise flag a warning/shutdown/etc as appropriate.
|
|
#
?
Oct 17, 2019 04:16
|
|
|
Yes, I should've mentioned that the approach requires good risk analysis to work. If all three sensors share a power supply and it malfunctions you're out of luck, for example.
|
|
#
?
Oct 17, 2019 07:35
|
|
|
i wish, at least that'd have a sweet payoff at project end. it isn't really that exciting or demanding a project, either. i'm just overcautious about anything involving NDAs and non-competes even though they're largely unenforceable here and none of this stuff would be particularly actionable even if i wasnt beating around the bush, mostly b/c my current industry is cozy enough that even a dumbass aggrieved client who just doesnt understand their own contracts is enough to get you blackballed locally  PDP-1 posted:When doing redundant sensors for critical systems it's also a good idea to design each sensor set to work on completely different physical principals to lessen the chances that some odd edge case 'tricks' one kind of sensor into reading a false positive since that same 'trick' could affect multiple sensors of the same type. This is probably a factor or two more rigorous a setup than we'll need, but this is still the sort of overall implementation we'd like to be possible for full-trim-package implementations, yeah. Anything too invasive re: the pump, like the motor mount load cell deal, is probably out because we're not touching the pumps themselves, sensors have to be flexible in application and simple enough to be installed by the client's technicians but largely maintained and operated by the end-user company's own employees. That means stuff compatible with standard NPS fittings, non-contact sensors in housings that attach to pipe exteriors with hose clamps and generally aren't picky about pipe OD or exact placement, preferably sensors with fairly intuitive operating principles like thermocouples looking for a sudden temp drop from cold water filling the pipe after the pump is signalled to begin pumping, that sort of thing. Ambrose Burnside fucked around with this message at 20:50 on Oct 17, 2019 |
|
#
?
Oct 17, 2019 20:43
|
|
|
Can anyone tell me wtf the point of tantalum caps are Like, those orange smd dealies are $2 each on Digikey, their failure mode is awful, their ESR is not really any better than some electrolytics, but with lower capacitance, and they're barely smaller than the electrolytics, too
|
|
#
?
Oct 17, 2019 23:14
|
|
|
They used to be the only way you could get capacitance higher than a few uF in a small package. Their failure mode is amazingly bad (they fail short and may start on fire), and they don't have very much headroom. You can usually abuse a ceramic or electrylytic a little but, but tantalums are toast right after their rated voltage. Ceramic and electrolytic caps have gotten better and basically made tantalum pointless afaik. And Tantalum is a conflict mineral, so that is a good reason not to use it without the other considerations.
taqueso fucked around with this message at 23:30 on Oct 17, 2019 |
|
#
?
Oct 17, 2019 23:25
|
|
|
They still edge out the other types in compactness for some combinations of spec, and while the over voltage failure mode is not ideal they also aren't as sensitive as mlcc or aluminum electrolytic are mechanically or thermally. For example if you're designing the worlds thinnest phone or some gizmo what lives in a drilling head you're probably including tantalum caps in it. Niobium caps are also a thing that exist and are nearly the same thing except from what I've seen they're also more expensive despite being based on a material that is more abundant.
|
|
#
?
Oct 18, 2019 00:50
|
|
|
taqueso posted:They used to be the only way you could get capacitance higher than a few uF in a small package. Their failure mode is amazingly bad (they fail short and may start on fire), and they don't have very much headroom. You can usually abuse a ceramic or electrylytic a little but, but tantalums are toast right after their rated voltage. Ceramic and electrolytic caps have gotten better and basically made tantalum pointless afaik. And Tantalum is a conflict mineral, so that is a good reason not to use it without the other considerations. Ceramics, especially large ones have cracking problems that cause catastrophic failures. Ceramics also have horrific capacitance vs voltage curves - they typically lose 50-80% of their capacitance by the time voltage reaches their rating (with some exceptions like COG/NPO).  I do see tantalum requested specifically in datasheets for regulators, opamps and ADC's. I havn't dug in entirely to figure out why that is but generally it seems in the 12-36V area Tantalums are a good compromise for ESR and energy density. Above that other electrolytics are the only game in town in terms of energy density. But honestly caps are a surprising bitch if you try to dig into them. In my experience the most 'honest' reliable caps are film. They have few parasitics, good performance and very good reliability. Plus they're most likely to have an actual datasheet with curves and specs like ESR and ripple (most cap datsheets are a joke). But they're big. COG/NPO ceramic are also great in terms of performance but not size or cost.
|
|
#
?
Oct 18, 2019 01:07
|
|
|
There's multiple generations of tantalums too, the older ones were MnO2 based so they could self sustain a small fire if they overheated. The newer equivalent is the Polymer cap, the tantalum branding is often not used, but to my knowledge they all are tantalum based. They have around the same capacitance per volume unit, but they generally have hugely better ESR and current ratings. According to Kemet they are also better about derating the operating voltage (previous spec was 50%, now they're more like 60-70%). Here's an example of a common 7343 220µF 10V cap suitable for bypassing anything below 5V: http://ksim.kemet.com/?pn=T520D227M010ATE040 Note that the ESR of this cap is around 20 mOhm instead of 1 Ohm for the same general spec. K-SIM is a pretty good tool worth knowing about, they have very detailed specs for pretty much all their capacitors. For ceramics they even model the actual DC bias effect for specific caps so you can check what you'll actually be getting.
|
|
#
?
Oct 18, 2019 06:53
|
|
|
Tantalums and their hybrid/polymer descendants don't really have better characteristics (ESR/ripple current/stability) than MLCCs and electrolytics. But they do still have an advantage in energy density, apparently, so they still pop up in situations where you need very low profile. Recently in the depths of the great MLCC famine, there were articles flying around suggesting that we start replacing large MLCCs with tanalum/film caps (lmao gently caress you). Fortunately we're starting to climb out of that bullshit. longview posted:K-SIM is a pretty good tool worth knowing about, they have very detailed specs for pretty much all their capacitors. ANIME AKBAR fucked around with this message at 13:21 on Oct 18, 2019 |
|
#
?
Oct 18, 2019 13:11
|
|
|
I want to buy one of those thru-hole PCB assembly... box... things, you know the things that are like a frame you can clamp the circuit board on, and then you put the components in, and you can clamp this "cover" over them that has soft foam on it, and flip the whole thing over, and the soft foam holds everything in place while you solder it. I cannot for the life of me seem to conjure up the correct search term though, do those have some special name I'm not aware of?
|
|
#
?
Oct 18, 2019 14:31
|
|
|
Shame Boy posted:I want to buy one of those thru-hole PCB assembly... box... things, you know the things that are like a frame you can clamp the circuit board on, and then you put the components in, and you can clamp this "cover" over them that has soft foam on it, and flip the whole thing over, and the soft foam holds everything in place while you solder it. I cannot for the life of me seem to conjure up the correct search term though, do those have some special name I'm not aware of? Man, you're not kidding. These aren't common. Here's a company that sells them: https://www.manixmfg.com/flip-over-pcb-assembly-system.html
|
|
#
?
Oct 18, 2019 14:41
|
|
|
KnifeWrench posted:Man, you're not kidding. These aren't common. Here's a company that sells them: Thanks for the tip, even with "flip jig" and "flip over assembly" there's not that many goog results, huh. I'd think this would be the kind of thing China would be turning out a million of on Amazon for like $20, like every other variation of clamp or helping hands tool out there, it's not exactly a complicated or expensive design and I'd think hobbyists would be all over it... I did find this guide to just building one with what appears to be some 80/20, I have some lying around so I might just do that: https://www.nutsvolts.com/magazine/article/september2011_Collier e: I finally found some actually being sold to end-users and the smallest one costs over $200, what the hell. Yeah I'm definitely building one instead, guess I know what I'm doing this weekend
Shame Boy fucked around with this message at 15:08 on Oct 18, 2019 |
|
#
?
Oct 18, 2019 15:04
|
|
|
ANIME AKBAR posted:Tantalums and their hybrid/polymer descendants don't really have better characteristics (ESR/ripple current/stability) than MLCCs and electrolytics. But they do still have an advantage in energy density, apparently, so they still pop up in situations where you need very low profile. Do you mind linking that TDK search? On a quick look I didn’t find it. I haven’t gone to the trouble of comparing Jouls per volume for caps (I’ve done it for inductors) but I have the impression that the best films (see Paktron) are in the ballpark (maybe 2-3x volume?) in terms of size after you factor in DC bias derating with no cracking issues.
|
|
#
?
Oct 18, 2019 15:25
|
|
|
asdf32 posted:Do you mind linking that TDK search? On a quick look I didn’t find it. Scroll to the bottom, check the box next to "DC-Bias Characteristic" quote:I haven’t gone to the trouble of comparing Jouls per volume for caps (I’ve done it for inductors) but I have the impression that the best films (see Paktron) are in the ballpark (maybe 2-3x volume?) in terms of size after you factor in DC bias derating with no cracking issues.
|
|
#
?
Oct 19, 2019 00:50
|
|
|
I want to build an arcade stick for the sega saturn. I found this page with a guy who did the same thing, but I'm struggling to understand the circuit diagram: Are those 4.7K resistors on every single button input? That's kinda what's tripping me up right now. He posted a picture of his completed protoboard design as well:  Are the long yellow things a resistor bank or something? It's too bad he didn't post the other side. Anyway this is what my (pathetic) breadboard looks like right now, I want to use header strips because I've got a crimper for dupont connectors: 
|
|
#
?
Oct 19, 2019 01:47
|
|
|
Yeah, that looks like a 4.7k resistor pack and they're being used as pullups. Note that the buttons are wired so that they connect to ground when pressed. This means that pressing the button reads a logic low, but when you release the button the circuit is connected to nothing, so it reads a logic ??? . Putting a pullup resistor on the input allows a small amount of current to flow from Vcc to the input when the button is floating, biasing the input high so it reliably reads as such. The common lead of the resistor pack is connected to the circuit's Vcc. The 74HC153s are multiplexers, which convert the parallel button inputs into a sort of serial packet (basically the opposite of a shift register). Sagebrush fucked around with this message at 02:05 on Oct 19, 2019 |
|
#
?
Oct 19, 2019 02:03
|
|
|
Those resistor arrays are really nice. There's a dot on one of the pins that to indicate which is common to all the other pins. You just tie that to the 5V coming from the controller port. Edit: Though he must be doing some voodoo on the other side of that prototyping board. You won't be able to do what he did with it right next to the chip like that on a breadboard. With a breadboard you'll have to either do individual resistors to the 5V rails of your board, or set a resistor array off to the side and run wires to the correct pins. Cojawfee fucked around with this message at 02:27 on Oct 19, 2019 |
|
#
?
Oct 19, 2019 02:21
|
|
|
I guess I'm using 14 individual resistors instead of a couple resistor banks 
|
|
#
?
Oct 19, 2019 03:10
|
|
|
Wait why? Couldn't you just tie the common to the horizontal rail directly above your chip and then jumper that over to the power rail? Am I misunderstanding the array topology? Edit: I was misunderstanding the chip pinout KnifeWrench fucked around with this message at 03:56 on Oct 19, 2019 |
|
#
?
Oct 19, 2019 03:48
|
|
|
Do the multiplexers get vcc from anything??
|
|
#
?
Oct 19, 2019 04:02
|
|
|
If you are sticking with the breadboard, run the 5V from the controller port to the red rails and run the GND from the controller port to the blue rails. Then wire the VCC and ground pins from the chips to these rails. This also allows you to just connect the resistors straight into the 5V rail and the other side of the buttons to the ground rail.
|
|
#
?
Oct 19, 2019 04:08
|
|
|
Cojawfee posted:If you are sticking with the breadboard, run the 5V from the controller port to the red rails and run the GND from the controller port to the blue rails. Then wire the VCC and ground pins from the chips to these rails. This also allows you to just connect the resistors straight into the 5V rail and the other side of the buttons to the ground rail. OK this helped a lot, I see how to wire the pullups now
|
|
#
?
Oct 19, 2019 04:16
|
|
|
I got myself one of those chinese 1000W zvs induction heaters with the water-cooled coil just to gently caress around with; I do a lot of delicate forging of v small titanium rod (1/4" dia and under) and i'd love love love to replace compressed gas + torches w something flameless and electric, and the stock i work has so little mass that 1000 watts should actually be more than enough for anything I could want to work on, assuming titanium heats up even a quarter as fast as a ferrous metal will. anyways: What are my power supply options if I'd like to avoid the expense of a UL-approved 24-48V 1500W+ switching supply without also burning my house down in the bargain? The youtube experimenter set seems to often use multiple 12V tractor/marine lead-acid batteries, but I'd have to buy those and the charger brand-new so I wouldn't save much. Maybe I can steal the transformer out of a microwave or sth like that
|
|
#
?
Oct 19, 2019 06:24
|
|
|
Ambrose Burnside posted:I got myself one of those chinese 1000W zvs induction heaters with the water-cooled coil just to gently caress around with; I do a lot of delicate forging of v small titanium rod (1/4" dia and under) and i'd love love love to replace compressed gas + torches w something flameless and electric, and the stock i work has so little mass that 1000 watts should actually be more than enough for anything I could want to work on, assuming titanium heats up even a quarter as fast as a ferrous metal will. TiG welder unit?
|
|
#
?
Oct 19, 2019 06:27
|
|
|
What kind of bargain are you looking for? You can find a lot of weirdly high output power supplies in the industrial section of ebay for not a whole lot (used), I think they use them for driving motors in factories or something. Here's one I found in about 3 minutes of searching good for 24 volts and 38 amps for $53: https://www.ebay.com/itm/Acopian-W24GT38-Regulated-Power-Supply-24v-dc/292512240929
|
|
#
?
Oct 19, 2019 07:23
|
|
|
Rewire a microwave oven transformer. Super cheap 60 hz high wattage iron core transformers. https://www.instructables.com/id/Build-a-Microwave-Transformer-Homemade-Welder/
|
|
#
?
Oct 19, 2019 08:45
|
|
|
|
| # ? May 9, 2024 19:56 |
|
|
ANIME AKBAR posted:https://product.tdk.com/en/search/capacitor/ceramic/mlcc/characteristic/ Yeah I missed it at the bottom. Thanks. I had completely written off cap manufacturer websites but this will be very helpful. Ambrose Burnside posted:I got myself one of those chinese 1000W zvs induction heaters with the water-cooled coil just to gently caress around with; I do a lot of delicate forging of v small titanium rod (1/4" dia and under) and i'd love love love to replace compressed gas + torches w something flameless and electric, and the stock i work has so little mass that 1000 watts should actually be more than enough for anything I could want to work on, assuming titanium heats up even a quarter as fast as a ferrous metal will. $265 on digikey. Not a ton more than batteries + charger. Obviously cheap ebay stuff will be 1/3 of that. A used server/telecom/datacenter power supply is a good option. asdf32 fucked around with this message at 15:35 on Oct 19, 2019 |
|
#
?
Oct 19, 2019 15:27
|
|