#robotics Logs

Mar 15 2015

02:26 Flynnn The worst bit is that it isn't a perfect half circle, it's like 75% of a circle, so that rules out my magnetometer x axis not registering negatives
02:28 Flynnn Image of the data: http://imgur.com/XqjNAo9
02:44 Anniepoo could it have a dead spot when it's parallel to the lines of force?
02:46 Flynnn AnniePoo: I don't think so, I would expect to see a straight line somewhere in there (like at x=0, or x=max) but instead, all data is within the semicircle
02:46 Flynnn AnniePoo: or am I misguided in thinking that?
02:47 Anniepoo suggestion - you might try repeating your experiment inside a set of Helmholtz coils
02:47 Flynnn AnniePoo: ahh, to get a good, strong magnetic field?
02:47 Anniepoo and a uniform one
02:48 Flynnn AnniePoo: fair enough. Though that might max out the sensor
02:48 Anniepoo I wouldn't make it too strong, yes
02:49 Anniepoo helmholtz coils are two coils wrapped all the turns together (not a solenoid) with, IIRC 1 D spacing between them
02:49 Flynnn Anniepoo: So, if I get the same result with the hemholz coil, then what?
02:50 Anniepoo I'm sure that's on the internet somewhere
02:50 Flynnn What is 1 D spacing?
02:50 Anniepoo the coils are like hula hoops
02:50 Flynnn Right, but, what is 1 D spacing?
02:50 Anniepoo they're on the same axis, and their centers are separated by one 1diameter
02:50 Flynnn oh, one diameter
02:50 Flynnn ahh, I see
02:52 Anniepoo https://en.wikipedia.org/wiki/Helmholtz_coil
02:52 Anniepoo it's 1R
02:53 Flynnn I see
02:53 Flynnn So, what if I end up with the same C-shape?
02:56 Anniepoo well, you at least know what direction the field is
02:56 Anniepoo and you can try varying the field
02:57 Anniepoo and looking at the pattern
02:57 Anniepoo if your data's confusing, improve your data
02:57 Anniepoo that's what I'm saying
03:00 Flynnn I see, fair enough
03:00 Anniepoo it'd be very useful to know the relationship between angle of sensor and the points
03:01 Anniepoo that is, do you know which angles aren't producing data?
03:01 Anniepoo or are they flipping?
03:01 Anniepoo rerun experiment, turning through small angles and adding to the data,
03:06 Flynnn ahh, yeah, I have no good way to do that at the moment
03:06 Flynnn but I can tell you that it appears, from the graph of my data, that, basically, the datapoints sweep back and fourth across the data with angle
03:07 Flynnn squinting, you can see things that resemble "lines" and it does appear that the graph bounces and doubles back on itself when it hits the opening of the C
03:07 Anniepoo ok, so it's overflowing
03:08 Anniepoo is there some 'reset to zero' command you can send the device?
03:09 Anniepoo you might also want to display the actual values and see, eg, if you're having an overflow as you approach that point
03:09 Anniepoo print the values in hex, decimal, anything else taht might make sense
03:17 Flynnn anniepoo: hmm -- have you seen the picture I sent of the data?
03:17 Anniepoo yes
03:18 Flynnn huh -- I would expect an overflow to reset the data to the opposite end of the range
03:18 Anniepoo not necessarily
03:18 Flynnn oh interesting
03:18 Anniepoo depends on the representation
03:18 Flynnn ahhh, you make a very good point
03:18 Flynnn now, I find that perplexing
03:18 Anniepoo you also might be 'tumbling gyros'
03:18 Flynnn it overflows at a small negative value, but it does not overflow at large positive values?
03:19 Anniepoo well, print out the values
03:19 Anniepoo mount your board on asomething you can slowly move in a controlled manner
03:19 Anniepoo like a tripod
03:19 Flynnn alright, will do when I get a chance -- I'm at home right now, can't access the physics lab for equipment
03:19 Anniepoo and figure out, e.g. what's the biggest value before it starts going the other way?
03:20 Anniepoo right
03:20 Flynnn I do know the min value after which it does not return data (the apparent overflow point)
03:20 Flynnn see, the thign is, I would expect to get two arcs, one smaller, and one larger, unless the overflow happens at zero
03:21 Flynnn but the overflow doesn't happen at zero, it happens at about -10
03:21 Flynnn and yet the arc is fairly consistent in position
03:22 Flynnn I'll run all of these tests you suggest -- thank you for the suggestions. In the absence of equipment all I can do right now is speculate on the geometry of it.
03:22 Flynnn could behavior like this occur if I messed up soldering, by any chance?
03:23 Anniepoo I'd also get data after rotating the device 90 deg through the other axes
03:23 Flynnn ooh, yeah, see, it works perfectly fine on the other axes
03:23 Flynnn Here, I'll show you some other data
03:23 Anniepoo well, if you do it slowly, you might see it happens when you reach a point of pulling a cable
03:23 Anniepoo that happened on apollo 12, IIRC
03:24 Flynnn oh, fascinating!
03:25 Flynnn though I'm using i2c to read the data over a fairly heavy duty GPIO ribbon cable for raspberry pi
03:25 Flynnn also, what is gyro tumbling?
03:26 Flynnn Other data: http://imgur.com/7n3CiTV as you can see, it works fine on all axes but the x axis, which for some reason hates to go past -10, even though the sphere is not centered on x = -10
03:26 Anniepoo gyro tumbling is when the axes of the gyro end up aligned, so the stable member matrix becomes degenerate...
03:26 Flynnn it's almost like one of the adcs is screwed up
03:27 Flynnn oh dear, I'm not sure what the stable member matrix is
03:27 Anniepoo grrr... watch apollow 13 (the tom hanks movie) - right after the explosion they're trying to get the spacecraft lined up
03:27 Anniepoo and have to avoid a 'red dot' on the 8 ball display
03:28 Anniepoo hmm... hang on, let me see if I can find a clear explanation
03:28 Anniepoo that doesn't involve a bunch of tensor math
03:29 Flynnn yes indeed xD I'm not sure if I made this clear or not, but I'm not running this through any kind of filtration or calibration -- it's all raw data
03:29 Flynnn probably not relevant, but just in case it is
03:30 Anniepoo ok, so maybe it's just that your angle scales are somehow off
03:30 Anniepoo (which gives the 270 degree thing)
03:31 Flynnn oh, I just went around in a circle two turns, I don't deal with angles at all in the actual code or sensor readout, it's all just result vectors
03:31 Flynnn sorry, I probably should not have said 270 degrees, that makes it sound like I knew to the degree what my rotation was
03:31 Anniepoo I'm just looking at the C
03:31 Anniepoo but yah, you need a more controlled experiment
03:31 Flynnn fair enough
03:33 Anniepoo gyro tumbling is another name for gimbal lock
03:33 Anniepoo https://en.wikipedia.org/wiki/Gimbal_lock
03:33 Flynnn ahh, gimbal lock
03:33 Anniepoo it may be what's happening to you
03:33 Flynnn yes, I know this. but I'm not dealing with angles at all
03:33 Flynnn that's only going to happen in an euler rotational coordinate system, I thought
03:34 Anniepoo but mostly you need to measure the actual angles, and plot them vs the actual output
03:34 Anniepoo right, but you probably have a set of euler angles here
03:34 Flynnn absolutely not
03:34 Flynnn I literally take the x, y, z outputs form my gyro and plot them to dots in 3D space with the same coordinates
03:34 Anniepoo no? you're not getting tait bryant angles?
03:34 Flynnn I have no idea what those even are
03:35 Anniepoo yah, sorry, this stuff is a mess'o math
03:35 Flynnn I'm fine with math, I just not super knowledgeable
03:35 Flynnn but I know for a fact no angles are involved here
03:35 Anniepoo tait bryant angles are 'roll pitch and yaw'
03:35 Flynnn ah, yeah, I am not dealing with that at all
03:36 Anniepoo it could be there's a true dead spot in your sensor's field
03:36 Flynnn my plots are just raw sensor data in local space, I don't even account for potentially skewed measurement vectors
03:36 Flynnn hmm, but a deadspot would cause the sensor to go to a default value, wouldn't it?
03:36 Anniepoo I'd get this data for various settings of all angles of the sensor
03:36 Anniepoo no, who knows what it'd do
03:36 Flynnn and we'd expect to see this dead spot happen on only one axis, thus, we'd see it temporarily collapse into a 2D plot
03:37 Anniepoo your 'C' might actually be a pac man
03:37 Flynnn as in goes around the c, jumps to zero, jumps back?
03:39 Anniepoo or stays at 0 during some dead band, yes
03:39 rue_shop3 oh I'm doing a bad job of being present
03:39 Flynnn that makes sense -- but since this problem happens on only one axis, I'd expect the deadspot to be on only one axis
03:40 Flynnn thus x would go to zero (or some other value, maybe random), but not y or z
03:40 Flynnn so we'd see the data temporarily flatten into a plane perpendicular to x
03:40 Flynnn but the data does not do this
03:40 Flynnn or at least, I can't tell if it does this
03:41 Flynnn like you say, I need to do more controlled tests
03:41 Flynnn it could be that it does this, and I can't see it
03:42 Anniepoo yes, the deadspot's probably in one axis
03:43 Anniepoo I've seen sensors go to zero, rather than max value, when overloaded
03:43 Flynnn zero, max, or otherwise, I don't see *any* kind of flattening -- that's why I'm skeptical of the deadspot theory
03:43 Anniepoo but yes, you need more controlled tests
03:43 Anniepoo I'd agree with you
03:44 Flynnn I'll perform the experiment one last time, and this time have the program draw the lines between each consecutive datapoint
03:44 Flynnn I can't do that with the current data because it's not all in order
03:44 Flynnn but I can do that if I do the experiment one more imte
03:45 Flynnn the one frustrating part is I don't have live readout on my data
03:45 Flynnn I have to do the experiment, and then process the results into a 3d graph
03:45 Flynnn I'll need to set up live readouts at some point, clearly
03:46 Flynnn anyways, thank you for the help, Annie! I appreciate your patience with me
03:48 rue_shop3 gimbal lock?
03:48 rue_shop3 oh you already covered that
03:48 rue_shop3 right then
03:48 Anniepoo I was a physics student - you've wandered into a fun area of math, 8cD
03:49 rue_shop3 I'm waiting, I want to make a peice on the printer and I have to wait another hour before the printer is done the current job
03:50 Jak_o_Shadows I oughta figure out my heated bed thermistor values
03:50 rue_shop3 serves me for a) printing frivilios things, b) not finishing the second printer
03:50 Jak_o_Shadows I oughta connect the heated bed thermistor
03:50 rue_shop3 yea I got the new controllers I ahve to load up with firmware
03:50 rue_shop3 I was gonna use my ramps board, but I really want to put the controllers AT the motors
03:50 rue_shop3 just run out signal and power
03:50 Flynnn oh wow
03:51 Flynnn that "inner c" I was talking about actually is happening -- it got way more pronounced this time
03:51 rue_shop3 so, I'm feeling I may never use my ramps baord now
03:51 Jak_o_Shadows Why? Wouldn't one place to do active cooling of the motor drivers be better?
03:51 Anniepoo is it by any chance rate driven?
03:51 Anniepoo that is, is it dependent on the speed you move it at?
03:52 Anniepoo how repeatable is it?
03:52 Flynnn Anniepoo: are you talking to me?
03:52 Anniepoo yes
03:52 rue_shop3 Jak_o_Shadows, mmm, I like the fact the noisy power lines are short
03:52 Flynnn Anniepoo: well, it's a magnetometer, so theoretically it shouldn't be rate driven -- it's a measure of absolute orientation
03:52 rue_shop3 and so far cooling isn't an issue
03:52 Jak_o_Shadows fair enough
03:52 Anniepoo youre not in the land of theory
03:52 rue_shop3 damn, what am I gonna do for an hour
03:52 Jak_o_Shadows signal would be steps and direction?
03:53 rue_shop3 Jak_o_Shadows, oh intresting new problem,
03:53 Jak_o_Shadows and motor on/off?
03:53 Flynnn Anniepoo: But it measures the magnetic field vector, how can that be rate driven?
03:53 Anniepoo imagine you mount your dectector inside a motor where the mag field is changing thousands of times a second
03:53 rue_shop3 if your motors ahve too much votlage drop, the microstepping cant work, casue the votlage saturates before the winding is up to current
03:53 Anniepoo every sensor has some frequency response
03:54 Anniepoo if you vary the pressure on my bathroom scale between 0 and 100lbs at a rate to 100,000 cycles/sec I assure you the needle won't move that fast
03:54 Anniepoo the cure is to move your sensor slowly
03:54 Flynnn Anniepoo: I sincerely doubt I'm dealing with magnetic fields at such frequencies -- I'm barely managing 0.1 rotation per sec
03:55 Anniepoo ok, I'm just speculating causes
03:55 Flynnn Anniepoo: fair enough
03:55 rue_shop3 'a watched print always toils'?
03:55 rue_shop3 zzzz
03:55 Anniepoo and yes, the way to test is to run the experiment slowly
03:55 Flynnn Anniepoo: I am awkardly spinning in place holding a raspberry pie chained to my laptop xD
03:55 Flynnn anniepoo: and the laptop is on the ground so that slows me down significantly
03:56 Anniepoo and do you know the actual direction of the mag field in your lab?
03:56 Flynnn I've ruled out it being on any of my axes of rotation
03:57 Anniepoo (I used to work in a physical measurements lab as a student, and there was a diagram on the wall giving it 8cD )
03:57 Flynnn nice
03:57 Flynnn Yeah, I don't know what the magnetic field is at my lab, though I'm sure it's fairly similar to the field here
03:57 Anniepoo well, you can find the axis with a coil and a scope
03:57 Flynnn which, just based on my coordinates, should be a downward facing thing facing, of course, north
03:57 Flynnn or south
03:57 Flynnn I always gti tmixed up
03:58 Anniepoo hold it out and shake it, and it'll have min voltage when the coil axis is parallel to B field
03:58 Flynnn Yeah, it won't be too hard, I just don't have those things on me right now
03:58 Flynnn here's my latest data: http://imgur.com/rq5mbnM
03:59 Anniepoo ah, interesting
03:59 Flynnn so as you can see, there is an inner C thing
03:59 Flynnn which I thought didn't exist
03:59 Flynnn so that supports the theory that hte values bounce back
03:59 Flynnn and flip around after going past -10
03:59 Flynnn which is just weird
03:59 Anniepoo well, the interesting pat is that it flips several times
03:59 Flynnn that's just because I'm going 2 full rotations one way, and then back the other way
04:00 Anniepoo well, quit doing that
04:00 Flynnn and then actually I did some more rotation after that
04:00 Flynnn lol
04:00 Flynnn very well, I'll do a single 360 rotation real quick
04:00 Anniepoo ok
04:00 Flynnn I'l ltry to get it as close to 0.1 rps as I can
04:01 Anniepoo I'd just say 'go slow'
04:01 Anniepoo cause any speed effect is just 'can't keep up'
04:01 Anniepoo and common sense says it's got to be in the sub 30msec read time range to be a viable commercial product
04:03 Anniepoo oh! this is a MEMS based gyro?
04:03 Flynnn I am fairly certain it is
04:03 Flynnn I haven;t honestly looked at the spec sheets in a while
04:03 Flynnn here, one sec
04:03 Anniepoo ok, it could be that it only operates with one side more or less 'up'
04:04 Flynnn oh god
04:04 Flynnn I hope that is not true
04:04 Flynnn though, actually, that would be the z axis
04:04 Anniepoo because MEMS gyros use the gyroscopic precession of a vibrating read
04:04 Flynnn x axis is what's giving us issues
04:04 Anniepoo reed
04:04 Flynnn wait hold on
04:04 Flynnn this isn't a gyro
04:04 Anniepoo who knows which axis is 'up'
04:04 Flynnn it's a magnetometer
04:04 Anniepoo oh, yes
04:05 Anniepoo ok, I'm an imbecile
04:05 Flynnn that's fine, I kept misreading your gyro as magnetometer
04:05 Flynnn that's why you were mentining angles, isn't it?
04:05 Flynnn ** mentioning
04:05 Anniepoo well, my comments about gyro tumbling etc still apply
04:05 Anniepoo no, I was clear it was a magnetometer back then
04:05 Flynnn ah I see
04:05 Anniepoo I just got confused in last 5 mins
04:06 Flynnn the chip is HMC5883L
04:06 Anniepoo at 01:41
04:06 Flynnn I still don't see how gimbal lock can apply to magnetometers
04:06 Flynnn they return values as vectors, not angles, if you get the 3D ones
04:07 Anniepoo if it's a 'digital compass' and you point it with it's 'up' north, you can see how that'd fail?
04:08 Flynnn Anniepoo: yes, but it is not a digital compass, it is a magnetometer
04:08 Anniepoo right
04:08 Flynnn Anniepoo: it will return valid results no matter how it is oriented! You just cannot map this back to an orientation -- only an infnite set of orientations, all rotations of eachother about the magnetic axis
04:09 Flynnn or at least, that's how it should work, if not for this weird x-axis thing
04:09 Anniepoo yah, this is odd
04:10 Anniepoo and yes, if it's just reporting the component, that doesn't have a degeneracy
04:10 Flynnn http://imgur.com/Vt9K42W
04:10 Flynnn There's the single rotation data
04:10 Anniepoo ok, the read time is 6msec so your speeds should be ok
04:10 Anniepoo ah!
04:10 Anniepoo so the X axis has a degeneracy
04:11 Flynnn wait, it does? in the spec sheet?
04:11 Anniepoo the sensor can't tell 'north' from 'south'
04:11 Anniepoo no, looking at your data
04:11 Flynnn why would that happen perfectly on the x axis?
04:11 Anniepoo imagine we're playing a game where you're blindfolded, I talk and you try to guess my location
04:12 Flynnn it can tell north from south just fine, if I point my x axis straight up
04:12 Anniepoo hmm
04:12 Anniepoo if red is X axis
04:12 Flynnn (indeed it is)
04:12 Flynnn I am suspicious that I fried one of the adcs
04:12 Anniepoo could be
04:12 Flynnn though how the hell I managed to only partially fry an adc is a mystery to me
04:13 Flynnn usually these things either fail catastrophically, or not at all
04:13 Anniepoo oh - your chip could be defective - suppose you didn't fry it, but there's a bad transistor on the chip
04:13 Anniepoo I agree
04:13 Flynnn hmm, that could explain it, I suppose -- I don;t know enough aobut how adcs work
04:14 Flynnn oh wow, I just noticed something
04:15 Flynnn god damnit, it's a dead zone
04:15 Flynnn my x has a dead zone
04:15 Flynnn look closely at the data
04:15 Flynnn when it goes past +10, it jumps suddenly to -10, and then continues in opposite directly
04:15 Flynnn ** direction
04:16 Flynnn you were right -- but how the hell does this sensor have such a huge dead zone, and why would it flip directions on the opposite end? and why only on the x axis?
04:16 Anniepoo hmm
04:16 Anniepoo is it the way you mounted it?
04:17 Flynnn how do you mean?
04:17 Flynnn What I actually got was this:
04:17 Flynnn http://www.vetco.net/catalog/product_info.php?cPath=640&products_id=13835
04:17 Flynnn so the chip came pre-mounted
04:17 Flynnn all I had to solder was scl, sda, ground, and vcc
04:19 Anniepoo so probably it's properly mounted
04:19 Anniepoo (I was thinking maybe there's a big bolt next to it, etc)
04:21 Anniepoo yes, I wonder if you have a defective device
04:24 Flynnn oh, yeah, that's a good point
04:24 Flynnn Yeah, I suppose I might
04:24 Flynnn I've had the thing for a year, only just now had a chance to use it
04:25 Flynnn I think you're right, the part is definitely damaged in some way
04:25 Flynnn either defective, or something stupid I did a year ago
04:25 Anniepoo I bet it's defective
04:25 Anniepoo not you
04:25 Flynnn haha
04:25 Anniepoo you'd fry it and it'd die
04:25 Flynnn yeah, that's a good point
04:26 Flynnn unless I shorted something
04:26 Anniepoo you might look with a microscope for a broken trace
04:26 Anniepoo and see if the caps are OK
04:26 Flynnn hmm! how can I test the caps?
04:26 Anniepoo some multimeters have a cap test setting
04:26 Flynnn oh, so just test the values
04:26 Flynnn gotcha
04:27 Anniepoo but it's pobably as easy to just replace
04:27 Flynnn honestly, we're eyeing a newer sensor that's not much more expensive, single body, and has built in motion processing algorithms
04:27 Anniepoo well, test with the cap test setting, and test with ohmmeter for dc short
04:27 Flynnn ooh, good idea
04:27 Flynnn We have no equipment for dealing with surface mount
04:28 Flynnn any tips on how I can do those measurements by hand?
04:29 Flynnn oh, shoot, the adcs are built into the compass
04:29 Flynnn duh
04:29 Flynnn there's no way I could have damaged the adcs without damaging the chip
04:29 Flynnn I see what you mean now
04:29 Anniepoo but ther's two external caps in the data sheet
04:29 Flynnn a good point
04:29 Anniepoo yah, I'd say get a replacement board
04:30 Flynnn greaaat
04:30 Flynnn alright, thanks for the help
04:30 Flynnn sounds reasonable to me
04:30 Flynnn though
04:30 Flynnn is it possible this is software?
04:30 Flynnn perhaps the thing is on the wrong setting?
04:34 Flynnn Yeah, both the x and y axes share the capacitors, so I would expect both x and y to operate teh same way
04:34 Flynnn that rules out faulty caps
04:34 Flynnn it must be an internal fault
04:34 Flynnn or software
04:35 rue_shop3 half an hour left
04:35 rue_shop3 its 2am
04:35 Flynnn rue_shop3, what are you doing?
04:35 Anniepoo you mean your interpretation software?
04:36 Anniepoo could be - you could be mixing unsigned and signed
04:36 Anniepoo but if 'twer me, I'd get a replacement board'
04:36 Flynnn Well, I'm using someone else's library to interface with the sensor
04:36 Flynnn Yeah, in the interest of time, we will
04:37 Flynnn hmm! it has a builtin test function...
04:40 niepoo wonders if that's a contradiction in t
04:41 ynnn is not sure what Anniepoo m
04:42 Anniepoo 'builtin test function'
04:42 Flynnn Well, it induces a magnetic field on itself
04:43 Anniepoo ah
04:52 rue_shop3 Flynnn, I'm playing with robotic arms
04:52 rue_shop3 kinda, I'm starting to have to circle while I wait for the printer to finish making
04:52 rue_shop3 http://www.thingiverse.com/thing:303469/
04:53 rue_shop3 parts of that silly thing
04:53 rue_shop3 I need to make a part and it wont even take long
04:53 Flynnn I see, very cool!
04:53 rue_shop3 I should work on my second printer, but I'm having a debate with myself about the type of plate to make for the bed
04:53 Flynnn 3d printer speeds can be the worse
04:53 Jak_o_Shadows I saw a 5-speed gearbox today
04:53 Jak_o_Shadows was impressive
04:54 rue_shop3 heh, no link?
04:54 rue_shop3 "print speeds can be the worse" "I saw a 5 speed gearbox today"
04:54 rue_shop3 heh
04:57 Jak_o_Shadows http://www.thingiverse.com/thing:713815/#instructions
04:59 Anniepoo Flynnn, good luck with the magnetometer
05:00 rue_shop3 oh dear god thats got to be 2kg of plastic
05:01 Jak_o_Shadows the housing is overkill I think
05:03 rue_shop3 its done! yay! its done!
05:15 rue_shop3 funny hwo a part can look so small till you print it
05:38 e_shop3 laughs as his printer makes an accidental 20cm br
05:41 rue_shop3 pretty good, from last part position to home
12:18 veverak nah
12:18 veverak nah nah
12:19 veverak autoamtic generation of parts
12:19 veverak kills my brain
13:25 rue_bed I realy like being able to sit back in my chair and doze off while making parts
13:29 rue_bed veverak, its great for partswith perpendicular edges that if you wanted to make out of one piece would mean machine the hell out of a llarge block
13:31 veverak :)
13:31 verak got the
13:31 veverak that if I have 'inners' of robot
13:31 veverak I can easily generate frame around it
13:32 veverak but
13:32 veverak it's....
13:32 veverak 'meh'
13:41 rue_bed its not for everything
13:41 rue_bed I have to write a cam package and use my cnc more
16:20 stoopkid anybody know of anything similar to a PLC, but using microprocessors instead of ladder logic?
16:21 Hyratel what's a PLC?
16:22 stoopkid a programmable logic controller
16:22 stoopkid it has an LCD touch screen, and accepts physical analog and digital inputs from sensors and control machines using 24 dc output
16:22 Hyratel you could use a http://www.cypress.com/?rID=92146
16:23 Hyratel PSoC
16:23 Hyratel microcontroller paired with CPLD fabric
16:23 chris_99 can't PLCs actually use any hardware internally?
16:23 stoopkid chris_99: not without violating the warranty
16:24 chris_99 i mean, can't they actually be made from, microprocessors, FPGAs etc.?
16:24 chris_99 or are they actually made from something in particular
16:24 stoopkid chris_99: yea, it's a system made out of microprocessors
16:25 Hyratel stoopkid, are you looking to avoid high-priced units?
16:25 stoopkid chris_99: but i'm essentially given a black-box unit and can only write applications in the application interface
16:25 Hyratel or roll your own control system
16:26 stoopkid Hyratel: well, i have the limitation of not being able to construct the hardware myself due to the expenses of getting it officially tested by engineers, apparently
16:27 stoopkid Hyratel: at least in the control unit
16:27 Hyratel so... what's the question? forgivve my puzzlement
16:28 stoopkid Hyratel: i guess i have the answer, there are many alternatives to PLCs
16:32 stoopkid Hyratel: at my job i've been presented with a situation where they have a PLC and a control panel box that the PLC goes into in order to interface with peripherals, like sensors for input and machines for output
16:32 stoopkid Hyratel: the PLC has a touchscreen interface and runs the user-interface and all the command logic, the panel box is just making the wiring nice and neat
16:34 stoopkid Hyratel: but this PLC has many limitations, and in order to write an application i have to write code using ladder-logic, which is basically wiring diagrams
16:34 stoopkid it's extremely tedious, error-prone, hard to make it modular and reusable, etc.
16:35 stoopkid it's nice for extremely simple things, but i've been given a very complex application to build including a reasonably complex user interface (which ladder logic really isn't made to handle)
16:35 stoopkid i've built the application to a working degree
16:36 stoopkid and it's actually installed several places running water-treatment systems
16:37 stoopkid but this i've pretty much maxed out what i can do with this thing, and any major modifications to the application really aren't worth it because it's so hard to make even small changes to this very complex system
16:37 stoopkid (in ladder logic)
16:38 stoopkid if i was writing something like C code, then everything would be modular, reusable, easy to write
16:38 stoopkid but my boss doesn't know what to look for in a control unit to find that, and unfortunately neither do i, because i really got thrown into this position with no experience whatsoever
16:39 stoopkid and apparently my PLC program is more complex than any other PLC developer that my boss has talked to has ever seen, let alone built
16:41 Hyratel stoopkid, so you're looking for an alternative controller?
16:41 Hyratel can you get a board custom fab?
16:41 stoopkid Hyratel: perhaps, i really don't know how expensive that would be
16:42 stoopkid Hyratel: if i really knew the details then perhaps
16:42 Hyratel what kind of interface are you needing?
16:43 Hyratel what kind of IO, what kind of buttons, what kind of screen?
16:44 stoopkid Hyratel: here's the ideal situation (which i accept i may only be able to approximate): essentially a similar control unit box that comes with an LCD touch-screen interface, but with code written in something like C, so that i can build an application allowing the user to configure the sensors, the machine outputs, and the command logic (the application logic is layed out already)
16:44 Hyratel so you need a UI-panel?
16:44 stoopkid yea
16:45 Hyratel the System interface is already there with machine IO for a UI panel?
16:45 stoopkid i have some leeway as to what type of UI panel since changing systems means changing all the UI anyway
16:45 stoopkid no it's not
16:45 stoopkid but it could be made that way
16:45 stoopkid i would prefer to rebuild it that way
16:46 Hyratel what kind of IO do you need to the big-system?
16:46 stoopkid right now it's all just one application in ladder-logic, so everything would have to be rewritten
16:46 Hyratel and how broad is your budget
16:46 Hyratel you could use a http://www.cypress.com/?rID=92146
16:46 Hyratel I've played with one of those ^ and it's really quite powerful
16:47 Hyratel it's a C-like language
16:47 Hyratel https://www.sparkfun.com/products/710
16:48 Hyratel https://www.sparkfun.com/products/8653
16:49 stoopkid i'm not sure the exact budget but it's at least a few thousand
16:49 stoopkid probably more
16:49 Hyratel how much I/O do you need?
16:49 Hyratel what kind of sensors are you interfacing
16:50 stoopkid only about a dozen sensors
16:50 ecstaticpessimst Any general but concise robotics book recommendations?
16:50 stoopkid same with the outputs
16:50 Hyratel what /kind/ though
16:50 stoopkid maybe even less
16:50 stoopkid generally 24 DC relay outputs, 4-20mA analog input sensors
16:51 stoopkid 24 V*
16:51 deshipu stoopkid: recommended book: http://www.dspguide.com/pdfbook.htm
16:52 Hyratel stoopkid, https://dl.dropboxusercontent.com/u/33593534/images/Fritzing/RelayBoard%20for%20StigChristian%20R2A_pcb.png
16:52 Hyratel I designed this as a "my first PCB" design for learning purposes
16:52 stoopkid deshipu: i've learned about a lot of that stuff, where i lack knowledge is about the hardware
16:53 stoopkid deshipu: like, specific hardware that i can actually purchase
16:53 Hyratel stoopkid, psst
16:54 Hyratel 8 relay outputs, 8 optoisolated inputs
16:55 stoopkid Hyratel: which one was that
16:56 Hyratel see link
16:56 Hyratel also - get an irc client
16:56 stoopkid Hyratel: why's that
16:57 Hyratel because webclients are janky
16:57 stoopkid what does janky mean
16:58 Hyratel crappy, unreliable, inconsistent,
16:58 Hyratel they also don't notify worth a damn
16:58 stoopkid Hyratel: what do you use
16:59 Hyratel HexChat (windows)
17:00 deshipu stoopkid: hardware really depends a lot on what you are doing, and changes fast
17:01 deshipu at least for the hobby market
17:01 Hyratel tthere's no real way to futureproof a design like what you need
17:01 Hyratel either you make it so it can be easily replaced, or you don't
17:02 stoopkid deshipu: of course, i'm just not up to speed on what's out there right now, i gotta do some research
17:03 Hyratel stoopkid, ask around. in here, ##electronics , #sparkfun
17:07 deshipu there was a book about mechanics of robot design recommended to me some time ago, but I can't remember what it was
17:07 deshipu it was mostly about wheeled robots, I think
17:16 Hyratel stoopkid, so how many system-side I/Os do you need (tops) and what kind of User Interface do you want to have?
17:17 Hyratel write down your UI option tree
17:17 Hyratel your UI options will inform the interface design
18:18 stoopkid Hyratel: for the projects we're working on, six 4-20mA analog input sensors, 6 digital inputs (some industry standard i can find), twelve 24 V DC digital relay/transistor/other outputs, and potentially MODBUS and BACNET I/O
18:19 Hyratel http://www.freemodbus.org/index.php?idx=32
18:19 Hyratel http://bacnet.sourceforge.net/
18:19 Hyratel first things first
18:20 Hyratel get yourself an arduino or a bare AVR, some jumper wires, and a small screen, and some buttons
18:20 Hyratel learn how they interact
18:20 Hyratel by the sounds of it, you could probably convince your project lead to bankroll such a purchase
18:21 Hyratel stoopkid,
18:21 stoopkid for sure
18:21 Hyratel what *do* you know
18:21 Hyratel so I can help you find a good start point
18:22 Hyratel you know ladder logic
18:22 stoopkid i also know standard programming, like C
18:22 Hyratel aha ok
18:22 Hyratel so you'll be comfortable in an AVR
18:23 stoopkid yea i understand microprocessor logic and i'd much rather be using that
18:23 Hyratel so your problem is "I need a control box to fit my I/Os and a decent user interface"
18:23 Hyratel which can probably be accomplished with an AVR or two
18:23 Hyratel how real-time is your system?
18:23 Hyratel does it have to be low-latency?
18:24 stoopkid Hyratel: some things should happen within a second
18:24 Hyratel so "no"
18:24 stoopkid Hyratel: it runs a water control-system
18:24 Hyratel unless you mean microsecond
18:24 stoopkid with treatment processes
18:25 stoopkid no, definitely not microsecond
18:25 Hyratel so not reatime
18:25 Hyratel but time conscious
18:25 Hyratel you can't have it chew on a process for five seconds and then get to a toggle
18:26 stoopkid yea, there are just a few mechanical operations which need to trigger in certain situations but they must happen pretty quickly so some of the logic must all be being done within about a second
18:26 Hyratel I don't think you're following what realtime means
18:26 stoopkid perhaps not
18:26 Hyratel "reaction output must happen withing N<100 millisec
18:27 Hyratel for ex
18:27 stoopkid so within a tenth of a second?
18:27 Hyratel a robotic arm needs realtime for its limit switches
18:27 Hyratel otherwise it'll shred the gearboxes or ruin the motors
18:27 Hyratel or crash the effector into a delicate glass wall
18:28 stoopkid no, not real-time then, though perhaps interfacing with real-time systems that are out of my control
18:28 Hyratel so you'll probably need buffers which can operate in realtime
18:28 stoopkid but, no my application doesn't need to be real-time, just "fast enough"
18:29 Hyratel ok
18:29 Hyratel that opens up options then
18:30 stoopkid how real-time would like, a standard home water-heater be
18:30 Hyratel not at all
18:30 stoopkid that's about where i'm at
18:30 Hyratel it's a large thermal mass, with fairly lenient hysteresis on a bang-bang control
18:31 stoopkid how about, a dish-washer
18:31 Hyratel N<5sec
18:32 stoopkid that's the dishwasher?
18:32 Hyratel yeah. time tolerance
18:33 Hyratel mayhbe down to Tol<1sec
18:33 Hyratel but
18:33 stoopkid that's pretty much what i'm working with
18:33 Hyratel different segments of a system can have different realtime tolerance
18:33 stoopkid yes of course
18:34 Hyratel so your communication might be Tol<Real
18:34 Hyratel but your heavy IO might be Tol < 0.5sec
18:35 stoopkid i think it's safe to assume that none of it is real time
18:36 Hyratel hmk
18:36 stoopkid i'm sure any solution would be fast enough
18:36 Hyratel I do reccomend hitting up sparkfun for a starter kit
18:36 Hyratel https://www.sparkfun.com/products/12060
18:41 stoopkid apparently the main concern is whether or not the hardware meets or would need to meet some engineer approval. for example the PLC hardware is engineer approved and my boss doesn't have to pay for testing before he installs this for water-treatment projects
18:42 stoopkid so apparently some configuration that doesn't need to meet engineer approval because it is already engineer-approved somehow would be preferable
18:42 Hyratel not going to exist
18:42 Hyratel especially if you're rolling your own
18:43 Hyratel you're going to have to design, troubleshoot and stress test hardware
18:44 stoopkid well of course, and i consider myself an engineer, so if i do all that does that make it "engineer approved"?
18:44 stoopkid i'm not even necessarily sure what authority we're answering to here
18:46 Hyratel what kind of environment is this going in?
18:46 Hyratel wet? salty? mildly corrosive? acidic? water spary? risk of flooding?
18:47 stoopkid it's a water treatment system, so possible risk of flooding
18:47 stoopkid but
18:47 stoopkid i'm not even sure the PLC's we're using are safeguarded against that
18:48 stoopkid actually i can almost guarantee they aren't
18:49 stoopkid but i would definitely build a system that was water-resistant
18:50 stoopkid although things were f'd up so bad that the control unit was flooded over i think that means it's f'd up beyond repair
18:50 stoopkid although if**
18:50 Hyratel yeah probably
18:51 Hyratel it's fairly easy to build splash tolerant
18:51 Hyratel immersion tolerant is another matter entirely
18:53 stoopkid i think it's safe to assume that it's in an environment safe from external threats
18:53 stoopkid and that any external threats are beyond anything i need to think about this point
18:53 Hyratel that simplifies things
18:53 stoopkid we can build a basic panel box that has apparently been sufficient for everything else we've done
18:54 Hyratel alrighty
18:55 Hyratel so yeah
18:55 Hyratel I'd say get yourself a SIK
18:55 Hyratel https://www.sparkfun.com/products/12060
18:55 stoopkid as long as the internal wiring doesn't make itself explode i think we'll be good
18:56 stoopkid i think the limitation is that we can't have our own PCB's fabricated
18:57 stoopkid well, we are at least that limited
18:57 Hyratel O_o why not
18:57 Hyratel is that "we can't fab pcbs in house"
18:57 Hyratel or is it "We can't send off to fab PCBs"?
18:58 stoopkid i think that is one of the things that requires engineer testing or is out of the budget
18:58 ratel VERY LOUD EYEROLL at it being 'out of bud
18:59 Hyratel https://oshpark.com/pricing
18:59 stoopkid well, budget looks good
18:59 Hyratel whoever says "we can drop $5000 on standard industry boards" but "we can't afford $$300 for custom PCBs" is talking out the ass
18:59 stoopkid i think the board design still requires engineer testing though
19:01 Hyratel "engineer testing"
19:01 Hyratel that is so un-usefully vague I don't know where to start
19:03 stoopkid yea i know, i think the main concern is like eletrical codes, like i can't have them print a circuit board that's going to be a fire-hazard
19:03 Hyratel you can look up UL requirements for safety
19:04 Hyratel and design with them in mind
19:06 stoopkid Hyratel: are you from the US?
19:06 Hyratel yeah
19:07 stoopkid well, i essentially don't want to build anything that violates any US legal codes, apparently what we're building now doesn't
19:08 stoopkid the PLC itself is also UL tested i think, so i can't exactly compete with that without doing the same thing of course
19:09 Hyratel stoopkid, you'd know what the relevant codes would be better than I would. look them up and build your boards to the codes
19:10 stoopkid Hyratel: haha well, i didn't necessarily know that, you know way more about everything other aspect of this than i do lol
19:11 Hyratel I'm here to ask questions of, but I don't want to design it for you (I'd expect to get paid if I did that
19:13 stoopkid well i was just wondering if you knew about any US legal standards concerning commercial electronics like this
19:13 Hyratel I do not
19:14 Hyratel I'm not even in the industry
19:14 Hyratel I'm just a hobbyist
19:14 stoopkid what industry are you in?
19:15 Hyratel I just said
19:17 stoopkid i meant as opposed to commercial electronics
19:17 Hyratel eh?
19:17 Hyratel I'm a *hobbyist*
19:18 stoopkid yea i meant what kind of work do you do then
19:18 Hyratel i just tinker
19:18 stoopkid i meant for $
19:19 Hyratel > Implying I have a job in this shitass economy
19:20 stoopkid well you would certainly have my job if you were to apply for it you could probably do way better than i am
19:20 Hyratel what's your job definition?
19:21 stoopkid water control systems *inventor*
19:21 stoopkid i somehow magically stumbled into this position with no experience whatsoever
19:21 stoopkid and somehow i made it work
19:22 stoopkid you would have made it work better and probably have built a better control system instead of just making an application on the one i was given
19:23 stoopkid now i'm kind of tied to this job because i'm the only person who really understands how to work with the internals of the application i built
19:23 Hyratel an awkward spot to be sure
19:23 Hyratel you've never done any hardware design before?
19:25 stoopkid nope, i used to database work and i did some database work a few years ago for my boss, that's how he found me, but i was interested in PLCs so he ended up contacting me asking me to work on this PLC application
19:25 stoopkid i told him i had never actually done it before but i could see what i could do, and i ended up making it work
19:25 stoopkid that's the closest i've gotten to hardware design
19:26 Hyratel well, you're in a good place to ask the right questions
19:33 Hyratel stoopkid, do you have a maximum box size?
19:33 Hyratel also, how much of a workshop do you have onsite?
19:33 Hyratel how custom-order can you make the panel
19:35 stoopkid Hyratel: they have the panels custom-built
19:36 Hyratel that takes a lod off you then
19:36 stoopkid Hyratel: idk how flexible this supplier is in what they can build, but i assume i can build something that functions similar enough to interface to this control panel
19:38 stoopkid Hyratel: this is what they're using now http://www.unitronics.com/plc-hmi/plc-vision-enhanced/v350-
19:38 stoopkid and others in the same series
19:40 Hyratel what interfaces do they have in-use right now?
19:41 Hyratel features listed modbus and TCP/IP
19:41 Hyratel what do you NEED for this
19:42 stoopkid necessities are simple analog and digital (on/off) I/O
19:42 Hyratel is it interacting with a network?
19:42 stoopkid at this point no, that is only a future desire and it can remain a future desire
19:43 stoopkid just as long as i can connect to it enough to install the software from the PC that i write the code on
19:46 stoopkid i guess if i'm designing PCB then it would be up to me how exactly to interface
19:46 stoopkid if i'm using prebuilt PCBs then i guess i'm given something that i have to interface with no matter what
19:47 stoopkid so any method of being able to install the software onto the system is fine
19:48 stoopkid ethernet, serial, usb, anything will do
19:48 Hyratel http://www.andymark.com/product-p/am-3000.htm
19:50 Hyratel also I guarrantee that if you designed your own UI for the PLCs from the ground up
19:50 Hyratel you can do it cheaper, better and more powerful with a hand-rolled system
19:53 Hyratel stoopkid, http://www.trossenrobotics.com/p/arbotix-robot-controller.aspx
19:53 Hyratel and write your own display control
19:54 Hyratel you can seriously do everything it sounds like you need with a smartly-designed avr (arduino)s tack
20:07 Hyratel stoopkid, beep
20:59 Hyratel stoopkid, you still there?
22:41 stoopkid Hyratel: sorry i was working on the application we're already doing
22:41 stoopkid Hyratel: but it looks i've made my PLC non-functional until i get technical support so i guess im done with that for the day lol
22:41 Hyratel lolwoops
22:42 Hyratel time to start reading up on basic electronics then
22:42 stoopkid well i've got this handy dandy book here called basic digital electronics, looks promising
22:43 stoopkid although it's a little too basic
22:44 stoopkid as a start i can disassemble the PLC and model my design off the components it uses
22:49 stoopkid it looks like there are 3 major components, touch-screen, a CPU component for control, and a component which runs the I/O in parallel to the CPU
22:50 stoopkid so it looks like all the application logic, computations, main memory, data is on the CPU component and this simply initializes and controls the touch-screen component and the I/O component
22:52 stoopkid there are probably 3 products out there that i can purchase which have been tested and verified such that i'm not going to breaking any codes by simply putting them together a certain way
22:57 Hyratel I don't understand what the aversion is to rolling custom boards
22:58 stoopkid i don't think i have enough experience at this point to take on the project at that level
23:00 Hyratel then start learning
23:00 Hyratel start reading sparfkfun turtorialsx
23:00 Hyratel take a wiki walk on electronics
23:02 stoopkid of course, what i mean though is there's no way i'll be able to get proficient enough fast enough to get my own PCB's made for this project
23:02 Loshki stoopkid: one really well written book is worth a 100 disorganized wikis...
23:03 Hyratel stoopkid, https://dl.dropboxusercontent.com/u/33593534/images/Fritzing/RelayBoard%20for%20StigChristian%20R2A_pcb.png
23:03 Hyratel this was my first-ever try at designing a PCB
23:03 Hyratel two sided, completely through-hole
23:03 stoopkid Loshki: i've also found that books provide a better foundation because they're immutable
23:04 Loshki stoopkid: and there's no shame in buying off the shelf, if it meets your needs.
23:35 rue_house Hyratel, at a glance it dosn't look too bad
23:36 Hyratel rue_house, I got a lot of feedback from sparkfun while making it
23:36 rue_house your optos usually need to go somewhere
23:36 Hyratel I do spot one hard corner though
23:36 Hyratel ground plane not laid
23:36 rue_house and your driver transistors
23:36 Hyratel in the software i was using, it's the last-last thing to do
23:36 rue_house Hyratel, 4 layer board?
23:37 Hyratel anything with a 'dead end' pin, it's probably going to ground plane
23:37 Hyratel 2
23:37 rue_house if you add a ground plane, thats 3...
23:37 Hyratel no
23:38 Hyratel ground pour
23:38 rue_house I think your relays are wired wrong
23:38 rue_house usually common is behind the coil and NC/NO are on the front
23:38 Hyratel nah, they're just weird
23:38 rue_house wahts 4014?
23:38 Hyratel shiftreg-input
23:39 rue_house yea, one of the 3 lines is a strobe?
23:40 rue_house why do you havea crossout thru the relay section?
23:40 Hyratel yeah
23:40 Hyratel that's "no not pour" blockout
23:40 rue_house no fill zone?
23:40 Hyratel do not pour
23:40 rue_house k
23:40 Hyratel yeah
23:41 rue_house why are the opto inputs ont eh same ground plane as the outputs?
23:41 Hyratel I don't really know
23:41 rue_house ah
23:41 Hyratel but I could address that
23:42 rue_house might be good to have them in a no fill, and make each input isolated
23:42 rue_house so you can use different supplies all togethor to trigger inputs
23:42 Hyratel yeah
23:42 Hyratel but this was something I did as a "my first PCB"
23:42 rue_house sometimes having a common line on the isolated inputs is a pita
23:42 rue_house its pretty good
23:42 Hyratel thanks
23:42 rue_house if the ground fill works...
23:43 rue_house this is a circuit you already know works?
23:43 Hyratel I made it all-PTH because I-don't-know
23:43 Hyratel nope
23:43 rue_house ah
23:43 Hyratel also Fritzing needs Grouping
23:44 rue_house I suggest changine r3, thru R10 with 1206 smt
23:44 rue_house or 080....4?
23:44 Hyratel if I redux this deaign, I'd start over completely in a more powerful EDA
23:44 rue_house then again, your not drilling the holes
23:45 rue_house I ahve to drill my own holes :)
23:45 rue_house 1404 pin 1, keep in on the orange layer
23:45 rue_house (!?!?)
23:46 Hyratel ehwaht
23:46 rue_house I suspect you know about the rouge push button
23:46 Hyratel yeah
23:46 rue_house 4014, sory
23:46 rue_house pin1
23:46 Hyratel how's it on the wrong layer
23:47 Hyratel yellow is top/near
23:47 rue_house pin 1 goes from orange to yellow, back to orange
23:47 rue_house but it dosn't cross any other orange
23:47 rue_house so, no need to change layers
23:47 Hyratel preventing ground plane discontinuity
23:47 rue_house oh, fair enough
23:48 rue_house R21 looks small?
23:49 rue_house is it standing?
23:50 ratel squints a
23:50 Hyratel I don't know
23:50 Hyratel R13-R20 are standing
23:50 rue_house yup
23:50 rue_house pitch looks short
23:50 Hyratel weird
23:50 Hyratel maybe it's a 1/8W ?
23:51 Hyratel it's the Rest pullup
23:51 Hyratel reset
23:52 Hyratel little secret - I did this without fully doing the Schematic
23:52 rue_house my board editor is dos, its schematic capture is a joke, I do it from my head
23:52 rue_house verry few errors ever
23:53 Hyratel dos?
23:53 Hyratel oh, DOSmode?
23:53 rue_house nope, real dos
23:53 rue_house runs an XY plotter with a pen
23:54 Hyratel oh
23:54 rue_house https://cdn.hackaday.io/images/6153531421889702965.jpg
23:54 Hyratel nice
23:54 rue_house https://cdn.hackaday.io/images/6584381415935964522.jpg
23:54 rue_house https://cdn.hackaday.io/images/5051671415935986091.jpg
23:54 rue_house https://cdn.hackaday.io/images/479731415936005742.jpg
23:55 rue_house http://hackaday.io/project/296/logs?page=2 <-- for what its worth
23:55 Hyratel oh, so you describe the traces directly to the plotter?
23:55 Hyratel and go from that to an etchable pattern?
23:55 rue_house see the image with the black ink, its resist, from there into tank
23:56 rue_house less heating up the tank, its about 4-6mins/board
23:57 mrdata cool
23:57 armyofevilrobots @rue_house; haveta get you on altium :)
23:57 mrdata i'm thinking i can use a sharpie
23:57 rue_house what that then!
23:58 armyofevilrobots Way too expensive, but full of win.
23:58 rue_house mrdata, you have to go really light on the pen, the ink layer needs to be thick and consistant
23:58 rue_house http://ruemohr.org/~ircjunk/tutorials/elex/etch/pcb_etching.html
23:58 rue_house I should rewrite that page...