00:00:00 --- log: started forth/13.06.22 00:20:45 --- quit: DocPlatypus (Ping timeout: 240 seconds) 00:51:40 --- quit: Bahman (Quit: ave atque vale) 01:06:15 --- join: ErhardtMundt (~quassel@host226-1-dynamic.52-79-r.retail.telecomitalia.it) joined #forth 02:38:28 --- join: yunfan (~root@unaffiliated/yunfan) joined #forth 02:53:59 --- quit: Anarch (Ping timeout: 248 seconds) 03:13:37 --- join: obobo (~obobo@dyn-76-75-122-30.nexicom.net) joined #forth 03:13:53 --- join: dto (~user@pool-96-252-62-13.bstnma.fios.verizon.net) joined #forth 03:16:03 KipIngram: well, I'm not an embedded guy... 03:17:27 * ttmrichter tries to picture a use case for Forth outside of embedded space. Fails. 03:18:29 oh, just for kicks. 03:18:38 proving that one tool isnt the solution for everything! 03:18:41 OK, that works. 03:19:02 To clarify: "just for kicks" is what I was commenting on. :) 03:20:32 --- quit: itsy (Ping timeout: 268 seconds) 03:24:32 Though built-in help would work for tethered development too... 03:26:49 I'm considering tethering the Forth in my STM32F3DISCOVERY board. 03:26:59 I haven't quite broken that barrier yet, but it's tempting. 03:27:30 I have a user USB port to communicate through and the STM32F3 has good USB support baked right in. 03:37:50 cool 03:38:53 ttmrichter: so one could make a single chip usb Forth dongle with the STM32F3 ? 03:39:18 Not the F3 I've got, no. That's a pretty big chip. 03:39:22 Let me check the other form factors. 03:39:53 (It's a 100-pin quad.) 03:40:29 thats just that version 03:40:45 theyd make it in a 32 pin package I'm sure 03:40:45 Yeah, I'm checking to see what other form factors there are for the STM32F3. 03:41:04 Honestly, though, I think the F4 might be the better chip for something like that. 03:41:43 I think a 32 pin UFQPNEP is the smallest they do 03:41:49 Smallest F3 is LQFP 48 7x7x1.4 03:42:02 thats small enough 03:42:14 I'm assuming the triad at the end is in mm? 03:42:18 to a smt process, it doesnt really matter 03:42:25 would be 03:42:54 the 100 pin job tensa to be most pricey due to the material used to make it 03:43:05 and the 32 pin, the cheapest 03:43:16 The LQFP 100 14x14x1.4 is just a bit big for a USB stick. 03:43:32 yeah definitely 03:43:39 and too pricey 03:44:16 Let me check that actually. 03:44:42 Given that I got the F3DISCOVERY as a whole (it's got TWO STM32s on board, plus two MEMS devices) for about $14... 03:45:01 What's the largest size in mm for something feasible as a proper USB stick? 03:45:11 (Not the wide USB sticks that suck so terribly, but a proper one.) 03:45:24 Oh, wait. 03:45:31 UFBGA 100 7x7x0.6 03:45:35 That's ... impressive. 03:45:40 But I don't know what UFBGA means. 03:46:14 it's a bal grid array 03:46:17 ball 03:46:34 Oh. That's the ones with the massive grid on the back that if you fuck up once you're screwed? 03:46:39 all the pins are underneath, so they tend to be cheap 03:46:42 yeah 03:46:56 can be done by hand, but it's not easy 03:47:20 Well those are $3.56 per (in bulk). 03:47:28 yeah, amazing 03:47:38 but the wrong chip 03:47:47 What's wrong with that one? 03:47:50 44 pin would be good 03:47:54 too many pins 03:47:56 100 pins, 7x7x0.6 03:48:03 Too many pins is a PROBLEM? :-o 03:48:20 but youd only use about 5 of those pins, and couldnt bring the rest out, no room 03:48:28 Oh, I see. 03:48:35 Because of the confines of the board size. 03:48:51 OK, so let's price the best of the 48-pin ones. 03:49:01 it's not a problem only slightly too much chip for the use 03:49:15 but sure, you could use it 03:49:35 just that most of it would be wasted 03:49:54 OK, the 48-pin version is $3.06 (in bulk). 03:50:05 cheaper 03:50:20 and planty of pins, plus it's hand solderable 03:50:38 if it's a LQFP, that is 03:51:38 That's LQFP 48 7x7x1.4 72MHz, 256KB Flash, 32KB RAM, 9x16-bit timers, 2x32-bit timers, SysTick, 2xWDG, RTC, 8x16-bitADC, 9x12-bitADC, 36 I/O pins 03:51:46 and one could bring 20 odd pins out the end for projects 03:52:00 perfect 03:52:12 thats one kick arse embedded chip 03:52:23 And so on and so on. 3x12-bit DAC, 3xSPI, 3xI2S half-duplex, 2xI2C, 3xUSART, etc. etc. etc. 03:52:28 um I don't see usb ? 03:52:39 I just got bored with typing it all. :) 03:53:00 Ah, this one doesn't do USB. It doesn't have integrated op-amps either. 03:53:03 your pdf reader doesnt copy 03:53:15 I'm getting this off the web page. 03:53:31 It's a stupid web app done up as a spreadsheet. 03:53:36 no, the opamps are in the 160MHz version of the M4 ? 03:54:05 No, I have op-amps in my F3 which is 72MHz only. 03:54:14 a decent usb to serial is only $2-3 anyway 03:54:15 But that's the 100-pin LQFP. 03:54:21 oh 03:54:28 Cheaper than that! 03:54:34 why not use a M0 ? 03:54:38 I could probably get USB/Serial chips for under a buck. 03:54:40 mine were 0.50 03:54:43 amd 32 pin 03:54:54 yeah, that makes sense 03:55:19 oh wait! 03:55:24 It *DOES* support USB! :D 03:55:30 http://www.st.com/web/catalog/mmc/FM141/SC1169/SS1576/LN10/PF253090# 03:55:37 Man, this seriously IS a kick-ass chip! 03:55:44 if you write it to run on one of these :- STM32F051K8U6, I'll make you a pcb and post it to you 03:55:49 For three bucks you're getting ... loads of stuff. 03:56:07 I don't think I have one of those. 03:56:07 :) 03:56:12 I can get 'em though. Probably for free. :D 03:56:58 I have them already 03:57:08 I mean for my purposes. I can't write code to a machine I don't have. ;) 03:57:09 http://www.st.com/web/catalog/mmc/FM141/SC1169/SS1576/LN1531/PF252054 03:57:11 400 of them :) 03:57:17 That's the F3 on my discovery board. 03:57:22 ohh 03:57:46 4 operational amplifiers. 03:57:53 7 rail-to-rail analog comparators. 03:58:02 I DON'T EVEN KNOW WHAT THAT MEANS!?!!! 03:58:09 * ttmrichter whistles. 03:58:24 I do 03:58:33 Less sarcastically, I'm not sure what use case there would be for on-chip op-amps. 03:58:45 rail-to-rail. simply means they will work over the full power supply range 03:58:57 No, I know what it means. :) I was being sarcastic. 03:59:00 which is awesome really 03:59:03 oh 03:59:11 I just can't figure the use case for some of this stuff. 03:59:18 Especially the baked-in op-amps. 03:59:20 sure there is a case for on chip opamps! 03:59:35 they have a 8Hmz bandwith if I recall correctly 03:59:43 oops 8MHz 03:59:57 So what's the use case? 04:00:06 lets imagine you want to build a set of scales for weighing stuff 04:00:18 Oooh! That F0 of yours has baked-in motor control. 04:00:19 Nice. 04:00:47 you have load cells, which have 100 ohm platinum resistors that stretch when weight is applied 04:01:16 and you'll need opamps to amplify the small change in resistance per gram etc 04:01:47 or you might want to make a ECG, for your dog ... 04:02:02 a lightning detector 04:02:08 Why on-chip though? How often do you need op-amps such that baking it on-chip is a thing? 04:02:14 there are a zillion resons 04:02:24 anytime you want low cost 04:02:33 and isnt that *all the time* ? 04:02:36 Well, hard to argue with $3.56. :) 04:02:40 absolutely 04:02:45 For a seriously kick-ass chip. 04:02:49 (You read the specs there?) 04:02:55 separate opamps means a LOT more cost 04:02:58 yep 04:03:09 Now let's look at the F4 line. :) 04:03:22 That's probably a lot pricier. 04:03:26 But ... much more capable. 04:03:38 it wasnt until 'LSI' or larrge scale integration became available, that electronics stuff became cheap 04:03:53 and even the F4 is as cheap as 04:04:38 but as we go up the ARM scale, the peripherals begin to change, A-Ds are traded for HDMI etc 04:04:44 Let's look at one that could fit in a USB dongle again. 04:04:52 10x10x1.4 OK for that? 04:04:57 we go from embedded, to computing 04:05:06 sure 04:05:14 10x10 is still small 04:05:25 but is it a BGA ? 04:05:29 OK, I've got an LQFP 64 10x10x1.4 in the F4 line. 04:05:41 thats cool 04:05:53 168MHz. 04:06:00 1024 Flash. 04:06:02 fast too 04:06:03 192 SRAM. 04:06:22 --- join: Anarch (~olaf@c-67-183-64-49.hsd1.wa.comcast.net) joined #forth 04:06:31 that one is seriously complex 04:06:52 Well, if you want to give people a Forth toy, might as well make it complicated. :) 04:06:58 http://www.st.com/web/catalog/mmc/FM141/SC1169/SS1577/LN1035/PF252143 04:07:11 That's $6.31 in bulk. 04:07:58 you just want that cryptographic acceleration cell ! 04:08:12 As I said, if you're going to give people a Forth toy, make it a nice one. :) 04:08:18 What do you think? Good target for that? 04:08:57 but who will use 1MB flash for a Forth app ? 04:09:16 well, it wont be obsoleted too quickly 04:09:25 You can always find a use for 1MB Flash. :) 04:09:39 http://www.st.com/web/catalog/mmc/FM141/SC1169/SS1577/LN11/PF252140 04:09:48 That's the F4 I've actually got right now (on the discovery board). 04:09:52 but, I was assured 640 K was plenty for anyone! 04:10:18 It doesn't have the crypto interface. :( 04:10:41 no good then! 04:10:48 So what do you think? Is that F4 a decent target for a Forth dongle? 04:10:52 one problem with the crypto interface .... 04:11:01 USA export controls 04:11:09 STM is an Italian company. 04:11:11 Fuck the USA. 04:11:14 hahah 04:11:27 I can get these chips right here. :) 04:11:35 PRISM has you now, is that a red dot I see shining thru the window of your study ? 04:11:51 Yes. It's my own laser, though. 04:12:06 (One project I keep dreaming of is a laser harp.) 04:12:20 yes, a F4 is the latest hot shit, it's a brilliant Forth target imho 04:12:26 lol 04:12:47 you cannot fail if you use the latest hot shit ARM chip 04:12:53 Not the Jarre-style with the 1W+ lasers that need hand and eye protection to use. 04:12:57 I figure with a decent 32-bit embedded processor and a grocery store checkout scanner prism I could make one that works by having optics scanning for where the dot is seen and selecting the note based on that. 04:13:16 I really am somewhat of an STMicro fanboi these days. 04:13:23 me too 04:13:32 They do a few things that drive me bonkers (like STLink), but overall I quite like them. 04:13:42 the prices and capabilities are a embedded designers dream 04:14:04 Exactly. 04:14:07 They're nailing both. 04:14:09 yeah, stlink sucks, my quad core i5 is still down because I listened to you :) 04:14:21 *AND* they're making the hobbyists work too. 04:14:32 The DISCOVERY boards are a hobbyist's dream. 04:14:41 (Even with the sucky STLink.) 04:14:47 as a result, I just *had* to go and put a small i3/16GB ram running FreeBsd together as a spare machine 04:15:25 root@gronk-13:/boot # uptime 04:15:25 6:03PM up 1 day, 4:40, 4 users, 04:15:36 FreeBSD gronk-13 9.1-RELEASE-p4 FreeBSD 9.1-RELEASE-p4 #0: Mon Jun 17 11:42:37 UTC 2013 04:15:56 I'm pulling the F3's of my STlink boards 04:16:05 and anything else thats useful 04:16:27 JTAG is fine for S3's 04:17:04 I rekon a usb F4 dongle with jtag on the far end would be hot! 04:17:10 running Forth 04:18:11 The Discovery boards also have F0s on board. 04:18:33 So, such a dongle would have to have: 04:18:35 1. The F4. 04:18:47 2. The USB glue (very light given the F4.) 04:18:58 3. Some kind of serial memory to fire up the F4? 04:19:05 Or should that just be in Flash? 04:19:42 flash! 04:19:43 (I kind of would want a fall-back of being able to load a vanilla system from non-rewritable memory because I'm a paranoid git.) 04:19:51 no extra memory needed 04:20:00 The problem with putting it in Flash is that you can brick it then. 04:20:08 almost true 04:20:23 but if you have a JTAG connector, it's impossible to brick 04:20:33 Fair point. 04:20:47 as JTAG allows uploading your latest Forth FW in a flash 04:20:52 As long as there's an image file lying around for a core piece of functionality. 04:21:16 just like Riscy Pygness does it ? 04:21:38 you're only going to copy that anyway .... 04:21:43 04:21:53 Nah, I want to make my own Forth for something like this. 04:22:09 I was kidding, just stirring 04:22:26 I know you will do your own, it's the doing that interests you most 04:22:29 * ttmrichter takes everything seriously. Had a humourectomy at age 3. 04:22:34 hahah 04:22:36 Got it in one! 04:22:37 sure 04:23:10 you're a NBH 04:23:15 natural born hacker 04:23:41 me, Id get pleasure *using* it, I'm just a user 04:24:23 my design fun comes from applying the chip with Forth to a hw project 04:24:43 Im working on a GPL security system at the moment 04:25:06 Oh, that reminds me. 04:25:09 We need LEDs. 04:25:12 Lots of them. 04:25:12 I actually havent started yet, I'm only at the planning stage 04:25:17 Like an 8x8 array of them in different colours. 04:25:28 YOU CANNOT HAVE ENOUGH LEDS IN A FORTH TOY! 04:25:29 SMT leds are fairly cheap 04:25:38 I know. :) 04:25:50 Can you buy LED arrays, though? 04:25:53 not as cheap as say, NPN bipolar transistors 04:26:01 Like a 1x8 pack? 04:26:06 I'm not sure 04:26:13 Im not into BLING 04:26:22 This would be a Forth toy. 04:26:27 Bling would be part of the point. 04:26:49 it wouild be SMT, so why not just put the array on a pcb using individual SMT leds ? 04:27:04 And if you give it some useful MEMS sensors (gyro, accel, compass) you could even have them do something useful and/or interesting. 04:27:24 I figure an array would be cheaper and/or easier than doing it individually. :) 04:27:28 a PNP (pick and place) machine doesnt care what it places 04:27:31 nah 04:27:47 arrays would be bulky, hard to handle 04:27:55 smt leds are like grains of sand 04:28:25 could do a 8x8 really easy 04:28:29 I'd think grains of sand would be harder to handle than something with some actual dimensions. :) 04:28:38 not to a smt machine 04:28:45 by hand they would be easier 04:28:45 What about making them two- or three-colour LEDs? 04:28:54 http://www.avagotech.com/pages/en/leds/surface_mount_chipleds/multi_color_series/ 04:29:00 I have tiny red and green smt leds 04:29:23 but the tricolor are bigger, at least the ones I've seen 04:29:28 http://www.avagotech.com/pages/en/leds/surface_mount_chipleds/multi_color_series/hsmf-c118/ 04:29:37 These are 3.2x2.7x1.1mm 04:29:43 if you really want h res tricolor, just fit a small OLEd display ? 04:29:51 OLED would be pricier, no? 04:29:57 I'm not sure 04:30:21 And OLED would take up more IO pins. 04:30:26 I have a 1" oled with touch screen, the whole board was $90 about 5 years ago 04:30:57 So those chips I've got up there, how do they look to you? 04:31:00 the biggest problem here is deciding when to STOP adding stuff 04:31:06 Well, yeah. 04:31:11 But LEDs are not negotiable! :D 04:31:12 the S4 is perfect 04:31:15 hahah 04:31:27 tathi: good idea. thanks! 04:31:28 SMT leds are easily done, even on a dongle 04:31:30 F4 you mean? 04:31:36 oops 04:31:40 Just checking. 04:31:41 yeag F4 04:31:45 But yeah, that F4 is hot. 04:31:49 absolutely 04:31:51 64-pin quad. 04:31:55 And six bucks. 04:31:59 Holy. Shit. 04:32:02 yep, thats 32 spare io pins! 04:32:15 Which we can make accessible on a bus, right? 04:32:28 So USB dongle with a 16-pin edge connector? 04:32:29 maybe, connectors are big 04:32:48 probably a 20pin small connector 04:32:51 smt 04:33:07 So we'd be wasting 8 pins of that? 04:33:11 they are really small, far smaller than you can etch a pcb edge connector 04:33:28 Oh, you mean 20 actual pins, not 20 pin pairs. 04:33:35 could fit more sockets 04:33:41 OK, fair enough. 04:33:50 Again, the more I/O pins accessible the better, generally. 04:33:51 yeah you can get them in fixed sizes 04:34:12 problem the pitch is so fine, you need special mating plugs 04:34:28 Sell them together with the dongle. 04:34:29 and the price goes up 04:34:52 yeah sure 04:35:01 it would be a shit hot dongle 04:35:11 We'd have to balance price against utility of course. 04:35:12 powered from the USB plug 04:35:15 yes 04:35:24 The endless engineering issue. :) 04:35:30 plug in, run a terminal and you're doing Forth 04:35:36 not so endless 04:35:47 Im good at establishing realistic limits 04:36:11 Hah! According to Avago you're in Asia. :D 04:36:34 youd have to make a 'extender' board also, one that plugs into the dongle and makes all those spare pins accessable 04:36:57 australia is a little isolated island of rednecks and racists all on it's own 04:37:19 at the arse end of the world, next to asia 04:38:01 ttmrichter: your little Forth board, produced cheaply, would rival the Arduino easily 04:38:28 I can pretty much guarantee the "cheap production" part of this. :) 04:38:32 handle it right, and youd never need to work again ;-) 04:38:55 the big problem, is UL certification 04:39:18 Sell it as a "kit" and that goes away for a bit. :) 04:39:24 many countries get upset if electronics gear isnt certified, it's enough to put a person off 04:39:34 Basically sell it as a "kit" until you get enough money to pay off the UL shakedown. 04:40:06 OK, those tri-colour LEDs are $2.17 unit, $8.64 in bulk. 04:40:10 I don't thing that worked for the Rasberry Pi folks ? 04:40:14 OK, those tri-colour LEDs are $2.17 unit, $0.64 in bulk. 04:40:36 There was no assembly for the RPi. 04:40:37 and theyre big 04:40:48 nor would there be for your board 04:41:02 no one will hand solder a 64 pin chip 04:41:19 or a usb smt connector 04:41:22 If you sell them from overseas, too, like I am (and you are), UL has less sway. 04:41:29 the dongle would need to be complete 04:41:34 true 04:41:45 You'd lose a few market opportunities at the start. 04:42:01 just have a low freq crystal as possible, and use the PLL in the ARM 04:42:23 that will keep interference well down 04:42:26 But you'd get enough customers I think that you could eventually pay off the UL shakedown (and their equivalents like CSA and DIN and such). 04:42:33 yeah 04:42:46 these days you need a EM and a RF cert 04:43:01 plus a no lead cert 04:43:15 You could probably also give the first customers a deal where they buy now and they get a certified device for free (or for some nominal cost) later when they become available. 04:43:31 Be up-front with your non-compliance. 04:43:33 (we had to go to no lead solder cause the yanks need all the worlds lead for their bullets) 04:43:38 Hobbyists WON'T GIVE A SHIT. 04:43:52 thats true 04:43:56 And it's not like you're going to start a fire with a 5VDC0.5A max device. 04:43:58 they just want it NOW 04:44:10 true 04:44:25 The worst thing that could possibly do is break someone's laptop. 04:44:38 And that's not very likely. You'd have to be a concussed designer for that to happen. 04:44:45 just make sure it uses less than 0.3 A 04:44:56 and the arm chip can do that 04:45:00 What happens past 0.3A? 04:45:09 the extender board would have it's own psu 04:45:10 (I thought USB did up to 0.5?) 04:45:22 I think the rating for USB is 0.3 A 04:45:24 ? 04:45:31 No, I'm pretty sure it's 500mA. 04:45:32 it's a complex thing 04:45:46 initially, it has to be less than 50mA I think 04:46:06 and it has to be able to sleep at under 1mA etc 04:46:22 there is a whole bunch of USB specs 04:46:38 Yeah, but power management is something the STM32 line is very good at in general. 04:46:39 I've never made anything USB until soon 04:46:42 sure 04:46:44 That part of it isn't a problem. 04:46:52 agreed 04:46:58 I was just looking at the maximum draw. 04:47:00 the STM chips are perfect 04:47:06 yeah I know 04:47:27 anyway, specs are easy to sort out 04:47:39 (unless you have to pay $300 to get them) 04:47:55 * ttmrichter doesn't have to pay... 04:47:56 ;) 04:48:01 haha 04:48:04 From Wackypedia: "The USB 1.x and 2.0 specifications provide a 5 V supply on a single wire to power connected USB devices. The specification provides for no more than 5.25 V and no less than 4.75 V (5 VÂ±5%) between the positive and negative bus power lines. For USB 3.0, the voltage supplied by low-powered hub ports is 4.45â5.25 V.[49]" 04:48:17 (Take a grain of salt when reading that. Anybody can edit Wackypedia.) 04:48:21 yeah 04:48:39 "A unit load is defined as 100 mA in USB 2.0, and 150 mA in USB 3.0. A device may draw a maximum of 5 unit loads (500 mA) from a port in USB 2.0; 6 (900 mA) in USB 3.0. There are two types of devices: low-power and high-power. A low-power device draws at most 1 unit load, with minimum operating voltage of 4.4 V in USB 2.0, and 4 V in USB 3.0. A high-power device draws, at most, the maximum number of unit loads the standard permits. Every device functions i 04:48:40 nitially as low-powerâbut may request high-power, and gets it if it's available on the providing bus.[50]" 04:48:54 use a FTDI usb to serial chip, and it has it's own USB id etc 04:49:05 So that's 500mA in USB 2.0 (our target, I think) and 900mA for USB 3.0. 04:49:13 We can EASILY fit in that power spectrum. 04:49:26 remember it's 500mA MAX 04:49:37 people often have other USB devicxes 04:49:50 That's 500mA max per port, no? 04:49:53 anyway, this could run on 10mA 04:50:08 um, you could be right 04:50:20 but depends on many things 04:50:32 * ttmrichter is trying to find the maximum current draw period from an STMF4. 04:50:49 as a Forth device, current draw could be quite low 04:50:55 I'm just going to take the shit-hottest F4 (not even the one we're looking at here) and see what pops up. 04:51:00 current drain will be hard to work out 04:51:24 Oh, JESUS! 04:51:28 it depends on clock rate, cpu load ?, how many LEDS, etc 04:51:37 1/2W ? 04:51:43 The hottest one has a 260ÂµA per MHz. 04:52:15 that's neglecting the 3000 LEDS you want to run 04:52:26 I'm looking at the STM32's top draw right now. :P 04:52:41 So the hottest of the hot in the F4 line will take up 46.8mA. 04:52:48 Out of our alloted 500. :) 04:52:49 perfect! 04:52:57 the lower the better 04:53:08 So let's call it 50mA. 04:53:13 why not 04:53:29 that's at 168Mhz ? 04:53:31