00:38:32 jmkasunich has joined #emc-devel 00:46:20 jmkasunich is now known as jmk_away 00:52:04 hi john 00:52:06 bye john 02:19:45 LawrenceG has quit 02:19:47 anonimasu has quit 02:23:43 anonimasu has joined #emc-devel 02:48:24 LawrenceG has joined #emc-devel 03:28:18 jmk_away is now known as jmkasunich 03:31:17 hiya jmk 03:31:22 hi 03:31:33 I got an email from JonE today 03:31:44 * jmkasunich is happy - just finished a machining job, first one in a while that I didn't underestimate 03:31:50 cool! 03:32:02 I think I'll actually make about $55/hr (for 2.5 hrs of work) 03:32:05 so you made more than $2.50/hr? 03:32:09 woohoo! 03:32:26 my goal when I quote 'em is about $40/hr 03:32:37 when you get to $55/hr for 25 hours of work, (per week), you can quit your day job :) 03:32:45 heh 03:33:20 I tend to gross between $2K and $4K per year on the side jobs... my boss at the day job has nothing to worry about 03:33:31 bummer 03:33:40 but a good bonus nonetheless 03:33:49 (and it lets you write off the fun toys) 03:33:55 yes 03:34:07 heh, it pays for the toys 03:34:16 well - that, too 03:34:36 I started my shop in 1998 with $1500 of "family" money for the Shoptask 03:34:41 by 2000 I was in the black 03:35:05 and now I'm about 4-5K up (and have significantly more machinery and toys) 03:35:09 I bought the stuff for my company, so I was able to deduct the cost of everything (Bridgeport, Johnson bandsaw, tooling, measurement equipment, CNC conversion parts ...) 03:35:12 that's cool 03:35:27 one day, I'll make the first $1 off these machine tools ;) 03:35:52 the trick is to find the right customers 03:36:04 no - the trick is to find customers at all ;) 03:36:07 (I'm sure you know that from your other business) 03:36:11 yep 03:36:32 my main customer is a friend (went to college with them and kept in touch since) 03:36:53 that's cool 03:37:00 she works in a lab where they occaisionally need some strange test jig or something 03:37:12 most of my customers are holdovers from my previous business - since I was the software department ... 03:37:19 the test procedure will have a napkin quality sketch, so normal shops aren't interested 03:37:22 that's a great gig 03:37:41 plus you can actually do some electronics in the things 03:37:54 not to date - this stuff is pretty simple 03:38:14 ah - I was thinking of bed-o-nails testers and the like 03:38:34 more like "hold this piece of concrete block while we push on the tile thats glued to it to test the bond strengty 03:38:38 strength 03:38:49 ah - "stress testing" 03:39:02 all kinds of differnet stuff 03:39:25 one batch was molds to make grout samples that would later be squished or bent or something 03:39:28 cool stuff nonetheless - most engineering (and breaking things) is pretty fun to do 03:39:33 heh 03:39:58 so - she works for a tile manufacturer? :) 03:40:06 this last one was little tubes 50mm long, 20mm ID, they set them on the table fill with wet grout, then lift the tube and see how much it "slumps" 03:40:12 grout maker 03:40:18 (chem eng) 03:40:22 ok - tile bondig stuff 03:40:25 bonding 03:40:31 cool 03:41:23 the biggest things I ever did were "crack isolation membrane test jigs" 03:41:38 ? 03:41:46 I think I've built 80% of the world's supply of those 03:41:51 all 6? 03:41:51 ;-) 03:41:55 8 03:41:59 ah 03:42:04 but they were $2K each 03:42:10 well - that helps 03:42:38 spread out over several years, as different companies adopted the ANSI test that requires the jig 03:42:57 gotta love ANSI 03:43:03 unless you're trying to meet a spec 03:43:11 or even get one 03:43:47 the original test jig in the spec was horribly wrong... so I helped design the replacement (my friend is on the ANSI committee for that spec) 03:44:27 they must have about 6000 committees 03:44:32 maybe 60,000 03:44:42 one for every spec, probably the latter 03:45:02 she doesn't mind, the committee meetings are in Vegas 03:45:09 heh - that's always fun 03:45:13 for the first 3 days 03:45:33 Embedded Systems is in San Jose next year - kind of a bummer 03:45:45 but less expensive than San Francisco 03:50:36 you can write it off, right? 03:50:38 ;-) 03:50:42 yep 03:50:55 I'll just have to renta car so I can go to the House of Prime Rib ;) 03:51:12 or get Jymmm or peteve to drive me to SF 03:52:05 sounds like a plan 03:52:43 yep - can't go to the other coast without visiting the HoPR - it's damned tasty food there 03:57:45 so - the email from JonE - kinda funny 03:57:57 he said he thinks that the latch pulses go out to all the cards when you strobe any unit - it doesn't have to be a master (the master just does a periodic latch) 03:58:21 he thinks.... not "he knows" 03:58:27 that's the funny part ;) 03:58:36 you know, sometimes it seems like he didn't design the thing 03:58:49 sorta 03:59:03 granted, its been a long time (probably 4-5 yrs) since he first cooked it up... 03:59:30 true. though there have been tweaks in the last year or so 03:59:40 else our two USC boards would be the same 04:00:58 yeah 04:18:44 oh yeah - on the timedelay thing ... 04:19:27 I think the delays need to either be floats or longs in milliseconds 04:19:37 floats in seconds 04:20:35 floats would be my choice 04:20:54 ok. how about the resolution? 04:21:07 obviously, the real resolution will be dependent on the thread period 04:21:14 resolution is the same as the thread period 04:21:28 you can ask for 123.34562343644 seconds if you want 04:21:54 hmm - I suppose a float conversion isn't too costly 04:22:06 no 04:22:21 dT = period * 0.000000001 04:22:29 delay -= dT 04:22:35 yep 04:22:43 if ( delay < 0 ) { delay = 0; set output } 04:23:17 I think I'll do an up-counter though - it allows you to change the timeout and have an immediate reaction if the specified time has already elapsed 04:23:27 good point 04:24:12 oh, and make the actual counter a double, not a float 04:24:13 how many do you think I should allow (before the person should be using ladder anyway ;) )? 04:24:32 I was going to limit it to 8 or 16 04:24:36 8? 04:24:44 ok -8 should be fine 04:24:50 not -8, 8 04:25:18 someday I hope that kind of limit goes away, and you can just ask for one more whenever you want it 04:25:18 doubles aren't atomic - I hope there's no problem there 04:25:24 yep 04:25:38 the double is internal, only the component code can access it 04:25:42 the delay is a float 04:25:47 yep 04:26:03 I dunno if the double is actually needed 04:26:20 I was just thinking of the number of seconds you'd have to delay before it becomes necessary 04:26:25 if delay >= 2^24 * dT, you would have a problem 04:26:25 probably a couple of years 04:26:32 or centuries 04:26:56 2^24 is only 16 million, if dT is 1mS, thats 16,000 seconds 04:26:59 a few hours 04:27:05 dT would be at least 10 microseconds (assuming a pretty fast machine, and a step generation thread) 04:27:10 ok 04:27:43 though a few hours is also pretty long 04:27:57 with 10uS dT, it would fail after 160 seconds 04:28:05 and get inaccurate after 10s of seconds 04:28:13 true 04:28:30 with a double, now you are talking centuries 04:28:52 yep - do you know how bad the double -> float conversion time is? 04:29:02 I'd like to have a parameter that shows the timer 04:29:06 not enought to worry about 04:29:08 ok 04:29:23 I always think about cycles that way (when writing close to the hardware) 04:29:30 yeah 04:29:59 in the PC tho, unless you are in a loop so everything is already in cache, you are most likely memory limited 04:30:13 true enough 04:30:29 fetch time > execute time for anything except divides and such 04:30:57 that what sucks about periodic RT code 04:31:10 yep 04:31:36 the entire RT thread gets executed 1000 times per second, but when its not running, the web browser, or GUI, or whatever, is flushing it out of cache 04:31:40 it always amazes me that gigahert machines have multi-microsecond latencies 04:31:48 +z 04:32:17 yet a 10MHz microcontroller has sub-microsecond latency (an AVR has 0.4-0.5 at 10Mz) 04:32:21 took me a moment to figure out what "+z" was about 04:32:29 heh - it hurtz 04:33:03 the AVR doesn't have a virtual memory 04:33:11 it has virtually no memory 04:33:31 even tho RT doesn't swap, the MMU still needs to be loaded for every context switch 04:34:24 yeah - it's just hard to see where the 6000 cycles go 04:37:32 save state, switch to kernel space (which involves changing from ring3 to ring0, and saving and reloading the MMU regs), invoke the scheduler, figure out what task needs to run, if the task uses floating point, save the interrupted code's FPU state, restore the tasks's FPU state, restore the rest of the tasks state.... 04:37:45 ok - I see that ;) 04:38:41 the last time I did anything like that was before the PC had FPUs... 04:39:00 and before flat memory models 04:39:17 so it was just push all regs, change to new stack, pop all regs, done 04:39:41 yep - it's amazing how much state needs to be saved these days, especially when you start getting into peripherals (which teh MMU anf FPU are, in x86-land) 04:39:58 yep 04:40:12 I miss doing stuff with micros 04:40:13 I haven't done much low level code on the x86 - most of it has been on microcontrollers 04:41:18 the last time I did low level code in x86 it was on DOS 04:41:27 heh - OK, me too 04:41:33 did some crude multitasking, just because I wanted to 04:41:45 when the 286 was just getting into the PC/104 space 04:42:18 we actually made an environment controller that used a Z80 or a '186 CPU card (user swappable) 04:42:19 back in the day I did a RT/multi-tasking kernel on 8051 (that was for real, at my day job) 04:42:37 * SWPadnos dislikes the 8051 almost as much as the PIC 04:42:48 it isn't a nice arch 04:42:53 nope 04:43:08 but in the late 80's it was one of few choices 04:43:14 kinda like a bastard child of an x86 and a PIC, actually 04:43:32 except it predates both of those I think 04:43:33 at least it knows what to do with a carry 04:44:02 yep - I think it was based on the 4040 or 8080, just like the 8088 and the Z-80 04:44:03 more like the crazy uncle of the PIC and x86, that nobody talks about ;-) 04:44:07 yeah 04:45:43 http://www.avtron.com/meters.htm 04:45:51 heh, they're still selling that product 04:46:22 was that a design you did way back when? 04:46:32 I left there in 1991, and the design was several years old then 04:46:33 yes 04:46:56 well - it's probably got a 386 in it now ;) 04:47:02 or a high-flying DSP 04:47:06 doubt it 04:47:14 or just 2 resistors and an FPGA ;) 04:47:46 nothing modern in those products, they're built like brick shithouses and sold to people who care more about that than anything else 04:48:02 the good old days 04:48:19 you can hose that thing down (in front of and behind the panel its mounted on) 04:48:20 they'll probably still work long after the last Pentium 4 craps out 04:49:08 they replaced a product that used gas discharge displays (neon technology, not nixey tubes, but only a little more modern) 04:49:43 we had a few special versions too (sw changes) 04:50:03 one could measure the thickness of steel sheet, without any kind of thickness sensor 04:50:12 cool 04:50:34 two encoders, one on a conventional roll at a nip (where the steel passes between two rollers) and one on the coil that is getting wound up 04:50:48 the first measures length, the second counts wraps 04:50:57 length per wrap gives you diameter 04:51:09 change in diameter from one wrap to the next gives you thickness 04:51:24 yep - cool 04:51:38 the meter could display feet, roll diameter, or average thickness of the last few wraps 04:53:01 heh, of the three product lines listed under "speed and length measurement" only one is from after I left in 91 04:54:47 http://www.avtron.com/dc_drives.htm is what I worked on the last few years I was there 04:55:45 looks like the SW has advanced since then, but the power section and packaging looks exactly the same 04:55:56 those are fairly big drives 04:56:13 oh wait - it looke dlike it was on wheels :) 04:56:42 the smaller part on the front has the controller board, we called it the briefcase 04:56:54 it is about 9" wide by 12" hi x 3" deep 04:57:06 the larger part in back has the power section, that varies in size 04:57:15 the one in the pic is probably 50HP or so 04:57:44 ok - still bigger than anything I expect to need, but not as big as I've seen on eBay (though the 2500 HP ones would be) 04:58:11 for the really big ones, the armature bridge was separate, and the "power section" of the pictured unit contained the field supply only 04:58:36 when I left I had just finished the 540HP frame, and the 800HP was on the drawing board 04:59:19 correction, 540A (about 300HP) and 800A 04:59:50 their target market is paper and steel mills 05:00:07 interesting 05:00:25 I know the large printing houses also would need stuff like that 05:00:50 for paper, 30HP is small, 100-500HP is medium, and a few motors on a machine may be 1000+HP 05:00:58 for steel, 100HP is small 05:01:14 heh - harder to move the heavy stuff, I guess 05:01:23 biggest installation I've been in myself was a steel rolling mill with 6 roll stands, 9000HP each 05:01:35 that's a bunch 05:01:52 yes - too much to directly control with semiconductors 05:02:14 they use three stages 05:02:37 there were four 23KV AV motors (synchronous, constant speed) 05:02:51 each motor drove three 700V DC generators 05:03:01 (total 12) 05:03:14 two generators were paralleled to drive each roll stand 05:03:32 and each stand had three 3000HP 700V DC motors 05:03:52 the power electronics controlled the field currents of the generators and the motors 05:04:09 the armatures were directly connected (thru big honkin' contactors) 05:04:48 we would control 2-300A of generator field current at a 300V or so... 05:05:09 and that controlled the actual armature current, which was 0-15 kiloamps at 700V 05:05:45 the motors would probably not fit in my garage, especially with the drive system 05:05:55 no 05:06:08 motor shafts were 2ft + in diameter 05:06:47 it was funny to look at the 4 MW power supply I did the DSP work for - it would have taken up about 1/4 of my house 05:06:48 the shafts are about 3' off the floor, and the motor dips down into the basement 05:06:58 heh - like Hoover Dam 05:07:03 yep 05:07:23 4MW is pretty hefty 05:07:29 yep 05:07:31 5000+HP 05:07:43 I'd love to have seen the motor and generator they were going to drive with it 05:07:58 it was for a motor? 05:08:01 yep 05:08:08 a 40MW DC servo ;) 05:08:19 40? not 4? 05:08:25 yes, 40, with the 0 05:08:30 that's what the fenerator was for 05:08:33 generator 05:08:45 they just wanted to test the viability of a 40MW motor 05:09:28 the steel stuff is "servo" too... we did precise speed control 05:09:29 so they were going to couple a generator to the motor 05:09:32 even some position control 05:09:38 and feed back the power from the generator 05:09:47 using the 4MW as the "charger" 05:09:59 and to help with the losses at full power 05:10:10 ok 05:10:25 I think the motor was for turning battleships quickly, or something like that 05:10:26 AC or DC motors (and generators)? 05:11:03 I know the supply was DC, so I'd assume DC for the rest as well 05:11:14 turning, as in rudder actuator? or bow thrustor? 05:11:29 rudder shouldn't take anywhere near that power 05:12:04 like - actually making the ship turn, not directing the thrust of the normal motors 05:12:09 I guess a bow thruster 05:12:10 thrusters might (we've quoted bow-thruster drives for smaller ships, generally 500-1500HP) 05:12:27 yeah - battleships and carriers are friggin huge vessels 05:12:43 you wouldn't believe the way a carrier turns 05:12:44 carrier probably has 100,000 total shaft HP 05:13:17 well - I was on a day cruise on the Harry S Truman, and it was damned amazing 05:13:44 cool 05:13:44 apparently that class of carrier has been clocked at 60 knots 05:13:57 and that's when they let someone close enough to see it 05:14:05 that seems too high, but about 25 knots 05:14:11 s/but/by 05:14:20 yep, but nonetheless, that's what I've heard 05:14:33 I just don't believe it... 05:14:33 I'd bet the top speed is classified 05:14:44 why, it's only 97 million tons displacement ;) 05:15:05 power needed to move a displacement hull ship (non-planing) goes up as the cube of the speed or even higher once you reach hull speed 05:15:05 or was that 97000 tons (can't remember) 05:15:17 wasn't 97million 05:15:37 but, but, it's nukular - just ask the prezzident 05:16:08 nimitz class carrier (the latest I think) - 102,000 tons, 280,000HP 05:16:15 (thanks google) 05:16:43 max speed "over 30 knots" 05:16:54 I can believe low 40's maybe 05:16:59 ok - 97000 tons - I had partly remembered the poundage 05:17:00 but 60, NFW 05:17:41 I've got a friend who's a sailing and navy buff, he mentioned that a carrier with a broken screw was clocked at over 30, and he also mentioned a top speed of 60 kts 05:17:50 I'm not sure the context though 05:18:54 I can buy 30 on one screw 05:19:02 again, that high power law 05:19:21 ok - the Harry S Truman was the last Nimitz clss, apparently 05:19:27 50% power will give you 80% speed (or something like that) 05:19:43 sure - for a cube law 05:21:36 one of the coolest things on the cruise was going into a turn 05:21:51 we were in the hangar, and I noticed an orange rolling across the table 05:22:11 looked out the hangar bay door, and the ocean was pointing up at about a 30 degree angle 05:22:21 I hadn't even noticed that we had tilted! 05:22:33 heh 05:23:22 I got a great photo of a fighter jet taking off - the front wheel was just off the ship, and I was about 50 feet off the wing 05:23:43 nice 05:23:48 maybe 100, come to think of it ;) 05:24:03 how did you manage to get aboard? 05:24:35 BTW, theoretical "hull speed" for a ship that is 1000ft long at the waterline is about 42 knots 05:24:37 my brother in law was stationed on the ship at the time 05:24:49 above that speed power requirement increases very quickly 05:24:55 ok 05:25:03 "tiger cruise"? 05:25:07 it's probably only 800-900 or so at the waterline 05:25:15 "dependent's day cruise" 05:25:21 ah 05:25:51 my stepson is in the Navy (just entered in Sept).... maybe someday I'll get a chance to do something like that 05:25:51 I would liike to have known him when he was stationed on the submarine 05:26:04 if you get the chance - do it, for sure 05:26:13 take off work, buy the plane tickets, go! 05:26:18 I will 05:26:28 although its very unlikely to be a carrier 05:26:41 he's training for gas turbine engines 05:26:53 my wife hates military stuff, and it was one of the best days she's had (even though we were on the ship from 3:45 AM until 9:30 PM) 05:26:53 that means cruiser, destroyer class ships 05:27:01 ah 05:27:14 it should be fun anyway 05:27:18 yep 05:27:25 military hardware is really interesting 05:27:36 I was at Aberdeen Proving Ground in MD for my AIT 05:27:48 there's an Ordinance museum there that's pretty cool 05:28:07 they have everything from wooden tanks to Pershing missiles 05:28:14 nice 05:28:30 I think some of my co-workers were just at Aberdeen 05:28:39 I had guard duty once or twice, so I took a few walks around the museum ;) 05:28:54 we're working with Oshkosh Truck Co. to make a diesel electric heavy duty truck for the army 05:29:06 ok - to replace their HEMTT? 05:29:17 yes - it is a HEMTT 05:29:28 I'd love to see a diesel electric hybrid car 05:29:45 especially since you can then make it into a greasecar-electric hybrid ;) 05:29:47 the hybrid HEMTT can start on and climb a 60% grade 05:29:58 that's a lotta torque 05:30:02 yep 05:30:40 neat feature of the hybird - get where you are going, flip some switches, and it can become a 200KW genset to run a field hospital or whatever 05:30:42 wow - they're thinking of making buildings to hold the museum pieces 05:30:50 cool 05:31:10 that would be good for the film industry 05:31:59 except they would rarely need the off-road performance 05:32:04 (or would they?) 05:32:08 unless they're on location 05:32:35 I've shot on top of mountains, and people I've worked with have been in some really bad spots ;) 05:34:08 http://www.hybrid-vehicle.org/hybrid-truck-hemtt.html 05:34:49 hmmm - I think I've seen that before 05:35:06 I've ridden in a HEMTT (not sure I ever drove one though) 05:35:13 I don't work directly on that project, but it is the same small group (liquid cooled drives group) 05:35:24 were you in the service? 05:35:27 yep 05:35:33 national guard for 6 years 05:35:46 I see 05:36:01 I used to repair the fire control systems on M1 tanks 05:36:19 (if we had any, that is - I actually never did a repair the entire time I was in) 05:36:28 heh 05:36:32 except helping some other guys change brakes and stuff 05:36:38 on trucks 05:37:03 that's another interesting thing - the M1 can go about 90 MPH 05:37:14 it's governed down to 45 05:39:12 dang, after midnight again 05:39:21 yep - was just thinking that 05:39:47 I gues sit's time to turn in - I'm trying to get back on a normal schedule (and you never left one) 05:40:03 re: the governer.... would you give a who-knows-how-many ton vehicle that can do 90 mph to a 20 yr old driver? 05:40:12 f*ck no 05:40:18 66 tons 05:40:38 the turret is 16 tons, and can do a full 360 degree traverse in 5.6 seconds 05:40:42 did you ever read the blog "armorgeddon" 05:40:45 nope 05:40:58 written (quite well) but a guy names Neil Prakash 05:41:10 he's a Lt, commands an M1 in Iraq 05:41:24 hm 05:41:34 s/but a guy names/by a guy named/ ;-) 05:41:52 funny - I didn't notice ;) 05:42:21 http://www.avengerredsix.blogspot.com/ 05:43:20 he tells some wild stories 05:43:29 they hit a mine once, knocked a track off 05:43:30 interesting - thanks for the link 05:43:39 that *really* sucks 05:43:44 they did'nt realise it was a mine, thought they were hit by artillary 05:44:21 until they were walking around getting ready to repair, and one guy kicked this thing that was next to the track... then he realized it was another mine 05:44:34 oops 05:47:01 ok, bedtime 05:47:32 I was running around all day, I had hoped to talk to Alex today about run script stuff... hopefully tomorrow will be better 05:48:00 yep -tomorrow is the meeting after all ;) 05:48:12 time to rest after cleaning the garage today 05:48:20 night 05:48:24