**** BEGIN LOGGING AT Thu Jun 09 23:59:57 2005 Jun 10 03:10:50 <[g2]> ep1220-away, hope you got me e-mail about the success with redboot :) Jun 10 04:01:56 [g2]: You can open the bottle of champaign :-) Jun 10 04:02:27 Cool Jun 10 04:02:31 I can I open it for him? Jun 10 04:02:39 Your welcome Jun 10 04:03:19 I cant wait! Jun 10 04:03:41 Seems g2 is missing the party. Jun 10 04:03:58 He did help a lot Jun 10 04:04:50 what FT2232 device are you using? Did you make something special for your testing?> Jun 10 04:05:05 The FT2232C. Jun 10 04:05:29 Right, but are you using a devkit, or... ?? Jun 10 04:05:36 I used an eval-module and added a voltage regulator and a pinheader Jun 10 04:06:11 no PCB. just an experimantal setup Jun 10 04:06:14 I'm chomping at the bit to get my hands on a FT2232. Jun 10 04:06:38 Donot know if ka6sox made progress on the PCB. Jun 10 04:06:50 and I'm pretty interested to see your re-written library functions too! Jun 10 04:07:34 A few are still missing (like set JTAG Clockrate) Jun 10 04:07:40 currently i run at 6MHz Jun 10 04:07:54 thats *fast* compared to what I'm used to. Jun 10 04:08:27 The actual rate is slower. the clock is gated. Jun 10 04:08:45 how fast it is depends on what You do. Jun 10 04:09:05 e.g. if You write to JTAG only, You can get to 5MHz effective rate Jun 10 04:09:17 If You have to read inbetween it will be a lot slower Jun 10 04:11:17 * [g2] is ready to party ! Jun 10 04:11:51 great Jun 10 04:11:57 <[g2]> CONGRATS! Jun 10 04:12:05 Thx. Jun 10 04:12:10 and Thx for Your help Jun 10 04:12:18 the 3 LEDs are on. Jun 10 04:12:20 <[g2]> hey NP Jun 10 04:12:43 I also added code to communicate with the CPU. Jun 10 04:12:59 am getting used to ARM asm ;-) Jun 10 04:13:06 <[g2]> me too! Jun 10 04:15:18 I feel so left out. Jun 10 04:15:40 <[g2]> It's not too late to join the party! Jun 10 04:15:48 <[g2]> it's really *just* beginning Jun 10 04:17:07 I need some hardware to be of any real use. Jun 10 04:18:37 we might have to target HW next, so more people can use it. Jun 10 04:19:01 <[g2]> ep1220, what are your plans in that regard ? Jun 10 04:19:56 the FTDI to me is a proof of concept thing Jun 10 04:20:09 and to see if a FPGA makes sense. Jun 10 04:20:38 my conclusion: ther are cases were a FPGA will be siginificantly faster Jun 10 04:20:52 so my interest primarily goes into this direction. Jun 10 04:21:32 For the FTDI I see several possibilities: Jun 10 04:21:54 Someone makes a complete print (as ka6sox seemed to plan) Jun 10 04:23:39 2) Someone makes a PCB using the eval-module (cheaper PCB) Jun 10 04:24:22 3) everyone buys the eval module and adds the needed parts on a protoyping-PCB Jun 10 04:24:49 the question is also which target VCC to support. Jun 10 04:25:10 3.3V only is easy, all others need moer parts Jun 10 04:25:58 not to forget: Some soldering on the NSLU2 is requires as well. Jun 10 04:26:23 <[g2]> I've got re-worked NSLU2s Jun 10 04:26:45 <[g2]> They even brought out the AC13 line Jun 10 04:26:46 Do you have RESETn and TRST on there as well ? Jun 10 04:27:01 <[g2]> I think so Jun 10 04:27:22 Good without these icache load will not work Jun 10 04:27:25 <[g2]> I've got 7 pins brought out on a header Jun 10 04:28:08 <[g2]> ok... I jut be missing TRST Jun 10 04:28:19 You need: TCK,TMS,TDI,TDO,TRST,RESETn and GND; Jun 10 04:29:34 <[g2]> I've got TCK, TMS, TDI, TDO, and RESTn(AC13/Pin 1 on the Q8) and GND Jun 10 04:30:02 <[g2]> I'm missing TRST but that could be brought out pretty easily Jun 10 04:30:25 adding TRST You must remove R135 Jun 10 04:31:07 then bring out one pad of R135 and add a pull-down (TRST to GND) Jun 10 04:32:42 <[g2]> we just need to tell them exactly what to do and they can get it done Jun 10 04:33:44 how many pins does Your Jtag header have ? Jun 10 04:34:10 <[g2]> 7 now, but they are wires brought to a header Jun 10 04:34:59 I used a 14pin "ARM standard" header. Jun 10 04:35:12 I'l post photos a little later Jun 10 04:35:25 <[g2]> I've got one of those on the Avila board Jun 10 04:36:11 <[g2]> The digilent didn't seem to work with that I think it may have been due to the TRST Jun 10 04:36:37 If it is an ARM standard: the layout is different then on the Digilent. Jun 10 04:36:50 Also the Digilent needs VCC from the target. Jun 10 04:37:11 <[g2]> The VCC pin is there. The Digilent does 2.8 to 5V Jun 10 04:37:30 <[g2]> and the VCC helps set that from my understanding Jun 10 04:37:39 yes. Jun 10 04:38:03 <[g2]> So with or without I get the same results Jun 10 04:38:24 on the gateworks web site the header seems to have 14-pin. Jun 10 04:38:26 <[g2]> the board just boots and runs when the Digilent JTAG cable is attached Jun 10 04:38:42 <[g2]> It does... I made a converter Jun 10 04:39:03 well, the digilent does not have a RESET line (at least mine) Jun 10 04:39:18 so the board will just boot up. Jun 10 04:39:25 <[g2]> right. That's PIN 12 right ? Jun 10 04:39:46 Pin12 on "standard" ARM Jun 10 04:40:00 <[g2]> I think the gateworks is the "standard" pinout Jun 10 04:40:06 very likely- Jun 10 04:40:36 <[g2]> When I first connected I had GND connected to that PIN and the board was held in reset :) Jun 10 04:40:47 heh Jun 10 04:40:56 no wonder :-) Jun 10 04:41:27 <[g2]> So I then knew I had the odds and evens right :) Jun 10 04:42:00 But anyway You should be able to do basic JTAG like read Device ID, once the CPU is out of reset Jun 10 04:42:18 <[g2]> I'm getting nothing Jun 10 04:42:34 MAybe You have TDI, TDO crossed ? Jun 10 04:42:35 <[g2]> however, the same cable attached to the S3 board works Jun 10 04:43:08 <[g2]> I'll double check again on the NSLU2 Jun 10 04:43:12 e.g at FTDI chip the pin named TDI is the output (!!) while it is the input to the XScale Jun 10 04:43:29 Do not know how Digilent names its pins. Jun 10 04:43:51 <[g2]> hmmm Jun 10 04:44:06 if You now have TDI->TDI and TDO-> TDO amybe try TDI->TDO and TDO->TDI or the other way round Jun 10 04:44:07 <[g2]> I'll have to double check stuff and compare with the wiki Jun 10 04:44:43 <[g2]> nod. I think right now it's TDI->TDI and TDO->TDO Jun 10 04:44:59 YOu also could check the S3 schematic. Jun 10 04:45:00 <[g2]> I'll trace the lines back to the R Jun 10 04:46:01 I can confirm that the WIKI is correct. Jun 10 04:46:37 <[g2]> So TDI is R133 Jun 10 04:46:52 yes Jun 10 04:46:52 <[g2]> The used the wiki as the instructions so it should match Jun 10 04:47:07 likely the header is correct. Jun 10 04:47:37 but You might not know for sure how Digilent chooses its names. Jun 10 04:47:52 It could be TDI is the ouput (so needs to go to R133) Jun 10 04:48:16 but also could be TDI is the input (as its name suggests) -> then it must go to TDO (R137) Jun 10 04:49:02 <[g2]> dyoung, jacques when you used the Digilent cable which way did you hookup those lines ? Jun 10 04:49:15 <[g2]> both dyoung and jacques used the Digilent cable with the slug Jun 10 04:49:49 <[g2]> jacques, has the honor of the the *only* needed slug CPR in the history of nslu2-linux Jun 10 04:50:21 CPR = ? Jun 10 04:50:55 I plugged the lines to the labelled stuff, ie: TDI-TDI TDO-TDO Jun 10 04:50:56 <[g2]> Cardio-Pulmonary Recessation (sp?) Jun 10 04:52:01 I think theres a 'u' in there someplce. Jun 10 04:52:06 <[g2]> when someone's heart stops and ppl push on it and do mouth-to-mouth to keep them alive Jun 10 04:52:27 <[g2]> www.dictionary.com *is* my friend Jun 10 04:53:06 I try this site, thx Jun 10 04:53:27 <[g2]> If I wasn't having *so* much fun with the slug/JTAG/etc... I'd put a spell checker front-end in xchat Jun 10 04:54:19 it says: Cardiopulmonary Resuscitation (CPR) Jun 10 04:57:00 <[g2]> that's the one! Jun 10 04:57:03 <[g2]> :) Jun 10 04:59:00 <[g2]> ep1220, does the FTDI also support hw flow-control on the RS-232 side ? Jun 10 04:59:27 It does, does the NSLU-2 ? Jun 10 04:59:34 also does Xon-Xoff Jun 10 04:59:50 <[g2]> then NSLU2 just has RX/TX Jun 10 05:00:00 Without any handshake serial is problematic on the FTDI. Jun 10 05:00:29 You could loose data in case the PC is busy and not-serving USB Jun 10 05:01:07 It can buffer 384 chars only. Jun 10 05:01:19 <[g2]> that should be plenty :) Jun 10 05:01:51 at 100 this is roughly 35ms. Jun 10 05:01:58 I meant 110Kbaud Jun 10 05:02:04 <[g2]> nod. Jun 10 05:04:56 and rwhitby has the honour of causing the *only* needed slug CPR :-) Jun 10 05:05:21 <[g2]> :) Jun 10 05:05:43 <[g2]> mtdblock0 mtdblock4 whatever :) Jun 10 05:06:11 <[g2]> ep1220, have you met rwhitby ? Jun 10 05:06:22 not so far. Hi rwhitby ! Jun 10 05:06:30 hi ep1220 Jun 10 05:06:59 [g2]: actually it was offset 0, offset 0x50000000, whatever :-) Jun 10 05:07:31 <[g2]> rwhitby, ep1220 successfully loaded the mini I cache with and adapted FTDI board Jun 10 05:07:50 <[g2]> via JTAG and then enabled the processor and the program ran Jun 10 05:07:53 caused by turning a padFF routine into a pad00 routine, which caused the partition table to have a final entry which started at offset 0 Jun 10 05:08:54 <[g2]> it's kinda amazing that in 10 months there's only been 1 of those Jun 10 05:09:14 <[g2]> especially considering the bootloader's replaced Jun 10 05:09:54 * jacques feels special. Jun 10 05:10:06 <[g2]> jacques is special :) Jun 10 05:10:18 special.... ed! bwahh ha ha haaa! Jun 10 05:10:24 (someone had to say it) Jun 10 05:10:50 <[g2]> ep1220, so you put the code in the mini i cache and then you changed the PC to match ? Jun 10 05:11:16 g2: no. I put the Code at address 0, then left the CPU out of reset Jun 10 05:11:35 its true. I'm kind of suprised thata noone else has "accidentally" rendered their nslu2 unbootable to the point of requiring JTAG. All of mine were totally intentional. Jun 10 05:12:06 it found code in the cache and did not bother reading the FLASH Jun 10 05:12:26 <[g2]> ep1220, Ok... The cache lines were pointed to 0 then ? Jun 10 05:12:27 dyoung: no-one else has been stupid enough to make such a programming error Jun 10 05:13:00 You load: ADDRESS, 8 Instrucions (== 1 cache line) Jun 10 05:13:07 rwhitby, it could have happened to anyone :-) Jun 10 05:13:17 <[g2]> jacques, amen Jun 10 05:13:18 So you can put code at any address You like Jun 10 05:13:36 jacques: we were flying by the seat of our pants at that stage .... Jun 10 05:13:44 jacques was so calm after the incient too. I remember exactly what he said. Jun 10 05:13:44 <[g2]> and this is the mini I cache or the full I cache ? Jun 10 05:13:55 mini IC Jun 10 05:13:56 you do? I don't Jun 10 05:14:01 dyoung: what did he say? Jun 10 05:14:07 I recall that time rather fondly. Jun 10 05:14:09 "Why oh why?" Jun 10 05:14:27 I said that? :-) Jun 10 05:14:30 "why oh why did you have to change (something) to (something else)" Jun 10 05:14:36 lol Jun 10 05:14:45 0xff to 0x00 Jun 10 05:14:45 "I think it's broken" Jun 10 05:14:58 I was out and about at the time, and had to apologise by IRC while driving ... Jun 10 05:15:02 me: "Why dont I just send you a cable...." Jun 10 05:15:30 and for a few days I just osed my other slug - no muss no fuss Jun 10 05:15:34 used Jun 10 05:16:02 I thought you only had one at that stage Jun 10 05:16:15 <[g2]> ep1220, the mini I cache is 2K bytes large right ? Jun 10 05:16:19 no, I always buy at least two of my hack toys Jun 10 05:16:28 for reasons like that Jun 10 05:16:47 (well, when I can afford it - I couldn't afford two ipaqs for example, but I bought four tuxscreens) Jun 10 05:18:09 g2: I did not check. the rule is: Jun 10 05:18:22 jacques: how many 770s are you going to buy? Jun 10 05:18:32 rwhitby, two Jun 10 05:18:33 If the main IC is 32K the mini IC is 2K, if the main IC is 16K the mini IC is 1K Jun 10 05:18:40 ASAHP Jun 10 05:18:52 so depends on the specific Xscale CPU Jun 10 05:19:09 <[g2]> The main IC is 32K iirc Jun 10 05:20:27 You can also load the main IC. Jun 10 05:20:32 I'm gonna get a couple of 770's too. Jun 10 05:20:45 I have a feeling thats the next Toy for me. Jun 10 05:20:55 i really really want some Jun 10 05:20:58 Intel a bit cryptically says one has to be carefull with this; but no specifics Jun 10 05:21:01 <[g2]> ep1220, with LDIC ? Jun 10 05:21:15 Even if I have to ask my european or australian friend(s) to buy/ship one to me. Jun 10 05:21:22 g2: Yes, LDIC has 4 variants. Jun 10 05:21:36 dyoung: what makes you think australia will get them early? we never get anything early Jun 10 05:21:56 g2: Invalidate one main IC cacheline, invalidate the entire mini IC load 1 main/mini IC cache Line Jun 10 05:22:03 the only thing missing from the 770 for me is a keyboard. my treo650 has a keyboard, and it's much smaller Jun 10 05:22:51 but, 800x480 :-) Jun 10 05:23:07 dyoung: but if they do appear here first, you can be sure that I will get you a couple Jun 10 05:23:13 <[g2]> ep1220, I'd be way cool to be able to load APEX directly into the main IC Jun 10 05:23:25 802.11b/g and BT Jun 10 05:24:26 g2: agree Jun 10 05:25:08 g2: but I will not copy the opcodes manually into a C array :-) Jun 10 05:25:20 <[g2]> lol :) Jun 10 05:25:40 <[g2]> I think bin2c is our friend Jun 10 05:26:10 what format U use for images binary, intel.hex,S3, ... Jun 10 05:26:29 <[g2]> ELF :) Jun 10 05:26:51 already relocated I asume ? Jun 10 05:27:25 <[g2]> all the code is PIC Jun 10 05:27:32 the better Jun 10 05:27:51 <[g2]> so I hexdump'ed the object and matched up the first instruction Jun 10 05:28:01 <[g2]> which happened to be at offset 0x34 Jun 10 05:28:19 thats some real hacker work there. Jun 10 05:28:33 <[g2]> I did a load .... cache off, exec 0x...34 Jun 10 05:29:24 Does JTAGtool read the ELF ? Jun 10 05:29:44 <[g2]> I don't know Jun 10 05:29:50 <[g2]> but it doesn't matter much Jun 10 05:30:13 <[g2]> We could easily make the linker do our bidding Jun 10 05:30:26 yea, likely the linker can output a Hex format as well. Jun 10 05:30:41 ~nefarious Jun 10 05:30:43 nefarious is probably to be infamous by way of nastyness Jun 10 05:30:51 ~botsnack Jun 10 05:30:51 dyoung: aw, gee Jun 10 05:32:50 I did work with Intel-HEX and S3 a lot. Jun 10 05:33:12 <[g2]> cool. Jun 10 05:33:21 Do conversion from this is primarily copy/paste for me Jun 10 05:33:55 <[g2]> I'd wipe up a simple perl script probably Jun 10 05:34:06 <[g2]> perl objectfile routine_name Jun 10 05:34:24 Maybe You can try to compile APEX into one of these formats and then mail to me :-) Jun 10 05:34:30 perl is an option also Jun 10 05:34:37 <[g2]> it could run object dump find the offset lenght of a given routine and then convert the binary to the C array Jun 10 05:35:06 <[g2]> beewoolie is an ELF master Jun 10 05:35:53 <[g2]> I just need to look at the .map file and possibly with the linker we can move some sections around to enable it to run out of the main IC Jun 10 05:36:05 <[g2]> sections as in linker sections Jun 10 05:36:22 <[g2]> it's most certainly doable Jun 10 05:36:33 You and/or beewoolie are the right people for this. Jun 10 05:37:11 If ELF processing becomes complicated. Using a HEX file is OK as well. Jun 10 05:37:15 <[g2]> I'll talk to beewoolie about this, he's quite keen on the whole thing Jun 10 05:37:37 HEX is standard for FLASH/EPROM programming Jun 10 05:38:17 <[g2]> I'm sure converting to other formats will be trivial Jun 10 05:38:30 <[g2]> SREC, etc... Jun 10 05:38:37 bfd stuff is supposed to be able to do that, right? Jun 10 05:39:09 <[g2]> BDM ? Jun 10 05:39:37 bfd - gnu binary file format handling libraries Jun 10 05:40:10 http://sources.redhat.com/binutils/ Jun 10 05:40:18 <[g2]> it does a bunch of stuff and there are lots of tools in the family Jun 10 05:40:22 Binary File Descriptor library Jun 10 05:40:27 <[g2]> right Jun 10 05:40:39 <[g2]> ELF, COFF, etc... Jun 10 05:40:59 yeah, I think it can do SREC, etc too Jun 10 05:41:21 using objcopy Jun 10 05:51:03 <[g2]> ep1220, which tools were you using for the FTDI ? Jun 10 05:51:09 <[g2]> and which dev board do you have Jun 10 05:51:31 ep1220Tool, eh? Jun 10 05:57:52 g2: VisualStudio.Net /C++ native Jun 10 05:58:08 * dyoung cringes Jun 10 05:58:36 Poor dyoung :-) Jun 10 05:58:48 I had some bad experiences with VS.net I guess. Jun 10 05:59:02 http://www.dlpdesign.com/usb/2232m.shtml Jun 10 05:59:12 I was gonna buy that one myself. Jun 10 05:59:25 Mouser has it for $35 in single quantity. Jun 10 06:00:15 <[g2]> Does that have the 2232m on it ? Jun 10 06:00:21 dyoung: You need to add external parts to be able to use it for JTAG Jun 10 06:00:49 <[g2]> that's pre-dates the 2232C right ? Jun 10 06:00:57 g2: the chip is 2232C. Jun 10 06:01:11 there are 2 modules, both use the 2232C Jun 10 06:01:18 <[g2]> ok... So they moved to the new chips then Jun 10 06:01:30 there was no 2232m chip. Jun 10 06:01:38 i guess the m means module Jun 10 06:01:45 <[g2]> there was a B right ? Jun 10 06:02:02 no to my knowledge. Jun 10 06:02:09 there was a 245B Jun 10 06:02:36 iirc parallel FIFO only Jun 10 06:03:21 just see. DLPdesign has a 2232PB == 2232C + PIC Jun 10 06:03:40 <[g2]> that sounds fun :) Jun 10 06:04:55 <[g2]> actually we could probably adapter the 2232ML right ? Jun 10 06:05:51 yes, if You want to install it onto another PCB and add a case. it makes sense Jun 10 06:06:17 <[g2]> actually.... we should either built that plus the extra parts or get them to build it Jun 10 06:08:49 Using the module means You have not to handle fine-pitch SMD Jun 10 06:09:57 Donot know if they make custom products. Jun 10 06:11:25 a board to hold it with the support bits should be easy. Jun 10 06:11:59 yes, Jun 10 06:12:34 But still, for just a few the boards might get relative expensive Jun 10 06:12:47 <[g2]> Ok... and we cut a hole and put a usb-device connector right in the back of the slug Jun 10 06:13:02 ~ube Jun 10 06:13:05 hmm... ube is Ugly But Effective. Or sometimes Unorthrodox But Effective Jun 10 06:13:26 <[g2]> or... we bring out the usb on a tiny cable Jun 10 06:13:39 g2: You want the 2232C inside the slug ? Jun 10 06:13:49 <[g2]> exactly Jun 10 06:14:35 <[g2]> ep1220, I've got just about everything to start selling IXP420 development kits based on the slug Jun 10 06:14:42 the module has a height of approx. 1cm Jun 10 06:14:53 So I would not see were it fits in Jun 10 06:15:28 <[g2]> I'm saying the LM or a spin of that Jun 10 06:16:13 It is 5,5x2cm (approx 2x0,8 inches) Jun 10 06:16:29 Use the Low-profile version ? Jun 10 06:17:12 <[g2]> doh! I'm saying LM it's ML Jun 10 06:17:18 the low-profile is same sice Jun 10 06:17:30 then the carrier board could have a USB2 style mini connector. Jun 10 06:17:48 its the same size, but it doesnt have the big USB-B connector on it. Jun 10 06:17:54 so it can be made to fit. Jun 10 06:17:58 Ithink.... Jun 10 06:18:15 I have both lying before me. Jun 10 06:18:25 It only would fit over the Xscale CPU Jun 10 06:18:30 and this CPU gets hot Jun 10 06:18:40 well, warm Jun 10 06:19:12 <[g2]> Ok.. so it's better to bring out the wires and a header to the outside Jun 10 06:19:23 think so. Jun 10 06:19:38 <[g2]> except for strain-releif Jun 10 06:20:03 maybe You can glue the header to the NSLU2 case Jun 10 06:20:51 (I glued mine to the plastic leading the LED lights) Jun 10 06:21:20 when i modded mine for JTAG, I epoxied the header to the top of the ethernet jack. Jun 10 06:21:43 mmmm epoxy Jun 10 06:22:01 right up there with duct tape, dremel tool, and paperclips Jun 10 06:22:10 <[g2]> yeah.... that's something I want the kids playing with.... Jun 10 06:22:14 Hey, I resemble that remark...... Jun 10 06:22:49 * dyoung quickly hides his duct tape and dremel tool.... Jun 10 06:24:48 * prpplague didn't sleep well last night and is cranky today Jun 10 06:26:45 I havnt had a good night sleep in 2 weeks, so I've been cranky for around that long. Jun 10 06:26:48 too. Jun 10 06:26:59 <[g2]> sleep what's sleep ? Jun 10 06:27:16 its that thing that normal people do between the hours of 11PM and 7AM. Jun 10 06:27:17 <[g2]> I thought that was something dead people do Jun 10 06:28:05 hehe Jun 10 06:28:13 dyoung: whats your excuse? Jun 10 06:28:31 * prpplague has had a bad case of culture shock after moving Jun 10 06:28:56 <[g2]> it's s-l-o-w-e-r done there ? Jun 10 06:29:07 I think its related to the catastrophic failure of my the car. Jun 10 06:29:10 [g2]: slow isn't the word Jun 10 06:29:16 <[g2]> you're in da islands man Jun 10 06:29:20 and the related stress of finding a suitable replacement Jun 10 06:29:28 [g2]: my main problem is not having anyone to talk tech with Jun 10 06:29:42 dyoung: ahh Jun 10 06:29:49 dyoung: are you in the states? Jun 10 06:29:53 <[g2]> I thought that's what IRC was for :) Jun 10 06:30:02 I live island life too. I'm in Hawaii. Jun 10 06:30:04 [g2]: yea Jun 10 06:30:11 [g2]: thats my only outlet Jun 10 06:30:20 dyoung: really? Jun 10 06:30:36 dyoung: hehe, i have lots of my work there Jun 10 06:30:55 <[g2]> prpplague, sounds like you need the Treo650/Nokia 770 so you can talk tech from anywhere :) Jun 10 06:30:57 Really? Jun 10 06:31:04 in what sort of 'stuff' ? Jun 10 06:31:18 dyoung: x86 thin clients and some linux based servers Jun 10 06:31:27 I'm always looking for "other stuff" to do here. Jun 10 06:31:42 dyoung: i can hook you up with regular work there Jun 10 06:31:50 oh yeah? Jun 10 06:32:05 we should talk later then when its not 3:30AM local time. Jun 10 06:32:14 <[g2]> dyoung, won't sleep well tonight... Jun 10 06:32:15 dyoung: send me you contact info, they are always looking for local ppl to do basic maintaince work Jun 10 06:32:21 dyoung: yea yea Jun 10 06:32:32 dyoung: dave123_aml@yahoo.com Jun 10 06:32:32 <[g2]> he'll be too excited Jun 10 06:32:44 dyoung8888@yahoo.com Jun 10 06:33:39 dyoung: lovely. just send an email Jun 10 06:33:52 Ok, will do. Jun 10 06:33:54 dyoung: there are about 10 stores there Jun 10 06:34:13 really? Theres actually retail stores using linux here? Jun 10 06:34:19 dyoung: probably be mostly just simple stuff, replacing drives, powersupplies Jun 10 06:34:19 I'm amazed. Jun 10 06:34:30 dyoung: yea, i designed the thin clients they use Jun 10 06:34:44 dyoung: they were originally on *cough* sco Jun 10 06:34:47 are these chained stores? Jun 10 06:34:51 dyoung: ported everything to linux Jun 10 06:35:01 dyoung: hardware and lumber stores Jun 10 06:35:05 hehe, Jun 10 06:35:27 In a strange twist of happenstance, I probably know the vendor whose stuff you replaced. Jun 10 06:35:40 :D Jun 10 06:35:42 because theres only one I know of that does POS stuff using SCO. Jun 10 06:36:09 dyoung: well the company i used to work for used sco originally too Jun 10 06:36:09 <[g2]> are they public ? I'd love to short'm Jun 10 06:36:09 and lumber stores was one of their things. :-) Jun 10 06:36:16 dyoung: www.abcsinc.com Jun 10 06:37:06 I see... Jun 10 06:37:34 [g2] not public, privately held. Jun 10 06:37:45 <[g2]> I was just joking :) Jun 10 06:38:18 err... I cant even remember the.. oh yeah. DSI. Jun 10 06:39:24 [g2]: why's that? Jun 10 06:39:36 dyoung: ahh yea Jun 10 06:39:54 <[g2]> why was I joking ? Jun 10 06:40:10 prpplague, we'll talkl later. Honestly, I dont really have an interest in changing power supples and stuff. But I do know someone capable that works in the POS industry that is looking for something new. Jun 10 06:40:40 dyoung: hehe, pay is _really_ good Jun 10 06:40:56 Really *really* good? Jun 10 06:40:58 dyoung: figure, they would normally buy a plane ticket from dallas Jun 10 06:41:03 Hmm Jun 10 06:41:05 dyoung: and looking at 2 dallas Jun 10 06:41:07 point taken. Jun 10 06:41:10 dyoung: 2 days Jun 10 06:41:13 dyoung: down time Jun 10 06:41:19 dyoung: plus expenses Jun 10 06:41:44 dyoung: might be just some easy beer money Jun 10 06:41:46 <[g2]> think of it as a toy budget for dyoung Jun 10 06:42:01 Yeah, my toy budget just got re-tasked to car budget. Jun 10 06:42:04 * dyoung pout Jun 10 06:42:13 <[g2]> toy budget replacment :) Jun 10 06:42:20 dyoung: hehe, its really funny that you live there, hehe, i didn't like hawaii, so i moved to another island Jun 10 06:42:36 hehe Jun 10 06:42:38 <[g2]> that *is* funny Jun 10 06:42:47 It *IS* smallish here. Jun 10 06:43:02 with a typical island attitude. Jun 10 06:43:27 thats probably why I dont talk to anyone. Jun 10 06:43:38 <[g2]> what is "a typical island attitude" ? Jun 10 06:43:53 insular and non-accepting toward change. Jun 10 06:44:14 leading to strange behaviour Jun 10 06:44:43 <[g2]> traditions are probably hard to give up Jun 10 06:44:57 [g2]: when you live in a small place, ppl tend to ignore every one else as a matter of nessity Jun 10 06:45:15 [g2]: look at the japense culture as an example Jun 10 06:45:17 Its not even about tradition though, its just a general attiude issue. Jun 10 06:46:03 <[g2]> hmmmm... interesting Jun 10 06:46:40 I can say with certainty that I am not typical of Joe Hawaii Guy. Jun 10 06:47:33 But that probably comes with exposure to many other places, cultures, etc, etc. Jun 10 06:48:20 brain... turning off.... Jun 10 06:48:23 bzzt. Jun 10 06:48:32 <[g2]> sweet dreams Jun 10 06:48:51 Lumber... is my friend.... Jun 10 06:49:01 <[g2]> yes... your new friend :) Jun 10 06:49:51 dyoung: later Jun 10 06:50:49 night. Jun 10 06:52:17 power supplies are my friend.... Jun 10 06:53:36 Beer is my friend. :-) Jun 10 06:53:42 amen Jun 10 06:54:00 My only friend? hmm, maybe not. Jun 10 06:54:10 that would be bad. Jun 10 09:43:22 Hey doodz Jun 10 09:43:45 hi beewoolie ! Jun 10 09:44:03 I'm back on the jffs2 horse. Jun 10 09:44:47 yes! :-) Jun 10 09:47:05 <[g2]> beewoolie, wb :) Jun 10 09:47:14 Yo Jun 10 09:47:35 beewoolie: yo Jun 10 09:47:37 <[g2]> ep1220-away, ran the LED program from the mini I cache! Jun 10 09:47:39 beewoolie: crack whore Jun 10 09:47:46 [g2]: Most excellent. Jun 10 09:47:56 beewoolie: found my lcd issue Jun 10 09:48:01 beewoolie: freakin sharp docs Jun 10 09:48:08 prpplague: Share Jun 10 09:48:52 beewoolie: sharp left out the fact that one of the register bit _also_ turns off the data lines to the lcd Jun 10 09:49:06 Which bit. Jun 10 09:49:27 bit 11 on the lcdcontrol register Jun 10 09:49:36 * beewoolie reads Jun 10 09:51:02 prpplague: 0xffff401c ? Jun 10 09:51:42 prpplague: The PWR bit? Seems wierd to turn that one off. Jun 10 09:52:20 beewoolie: i'm not using the lcd_vdden able line, so why turn it on? Jun 10 09:52:46 Because that isn't the only thing that that bit controls. ;-) Jun 10 09:52:56 my point exactly Jun 10 09:53:11 beewoolie: but sharp doesn't give any other details as to that fact Jun 10 09:53:19 beewoolie: only the pl110 manual does Jun 10 09:54:01 beewoolie: sharp makes it out as that bit only controls the lcd_vdden gpio Jun 10 09:58:39 <[g2]> beewoolie, ep1220-away can also load into the main icache Jun 10 09:59:32 <[g2]> I think we can load APEX directly into the icache, however we'd need to adjust the memory locations and deal with the data/memory sections Jun 10 10:06:07 <[g2]> beewoolie, I'm thinking we could modify APEX to run from the icache by 1st setting up the memory controller and doing some quick memory tests, then copying APEX to memory and executing from there Jun 10 10:07:15 <[g2]> a more difficult alternative would be to setup the dcache and move the volatile sections there Jun 10 10:28:45 g Jun 10 10:29:22 [g2]: Let meget this straight. You want to write APEX to icache. Start execution, setup SDRAM, and then copy it to SDRAM. Jun 10 10:30:09 [g2]: I don't think we can copy anything from icache. Jun 10 10:30:41 <[g2]> sure I think we can Jun 10 10:30:46 I am thinking that it might be more reasonable to have a small stub that initializes SDRAM. then, we can download APEX into SDRAM and jump to it. Jun 10 10:31:06 What instruction copies from Icache? All data operations will read from dcache. Jun 10 10:31:14 <[g2]> load Jun 10 10:31:20 AFAIK, load reads from dcache. Jun 10 10:31:53 <[g2]> that'd be a problem :) Jun 10 10:31:57 I believe we are working with a harvard architecture. Jun 10 10:32:05 I and D are separate address spaces. Jun 10 10:32:05 <[g2]> nod. Jun 10 10:32:45 We *happen* to merge them after the caches, but that is a choice made by the system designers. Or, perhaps the chip designer makes the merge. Jun 10 10:33:19 It would be clever, though, to use icache routines for JTAG functions. Jun 10 10:33:45 We can 1) initialize sdram, 2) perform high-speed programming to flash, 3) ??? Jun 10 10:34:46 <[g2]> I'd think it'd be 1) initialize sdram, 2) high-speed load sdram and program, 3) run Jun 10 10:35:00 prpplague: I hate to admit this, but I think I remember encountering the same thing that you did with the LCD controller bit. Jun 10 10:35:23 [g2]: I just mean that we can perform flash programming from icache without needing sdram. Jun 10 10:35:36 It wouldn't even be that difficult to automatically generate the code. Jun 10 10:35:37 <[g2]> beewoolie, and where does the data come from ? Jun 10 10:35:46 All load immediates. Jun 10 10:36:23 beewoolie: hehe, yea, well it was a nasty one imho Jun 10 10:36:25 <[g2]> that'd be a cute little program :) Jun 10 10:36:34 I'll have to review my flash programming code. IIRC, the block programming mode requires that a stream of 32 bytes be written to the flash in one go. Jun 10 10:36:56 We don't care how many instructions it takes and the performance boost is tremendous. Jun 10 10:37:25 <[g2]> well the icache is only 32K Jun 10 10:37:50 [g2]: We don't have to program the whole file in one pass. Jun 10 10:38:01 We just have to do 32 bytes at a time to get the boost. Jun 10 10:38:15 <[g2]> well I'd think we'd just boot APEX Jun 10 10:38:25 <[g2]> then it could load via serial Jun 10 10:38:41 usually flash is far far slower than sending data into icache. Jun 10 10:39:03 Sure. That'll work too. But, this is a general solution that doesn't require any knowledge of the platform. Jun 10 10:39:14 Except for the address of flash. Jun 10 10:39:35 <[g2]> well the JTAG stuff is pretty platform specific :) Jun 10 10:41:13 <[g2]> I'm also not sure how we'd handle the timing of the flash write completing Jun 10 10:43:10 <[g2]> wrt allowing more code to be loaded into the icache Jun 10 10:43:53 Interesting question. I know that this is what the BDI does. Jun 10 10:44:29 There must be some sort of signaling available. Jun 10 10:46:08 <[g2]> we could use a GPIO for signalling I'm sure as a hand-shake Jun 10 10:50:41 I don't know if JTAG can read the GPIO lines without interfering with the operation of the CPU. Jun 10 10:51:36 <[g2]> actually, it'd work with one of the hw control lines on seral Jun 10 10:56:48 Again, I'm suggesting that there is a way to do this without knowledge of the platform. BDI doesn't care which ARM9 or XSCALE core is being used. Therefore, it must be possible to signal in a way that doesn't use peripherals. Jun 10 11:04:00 <[g2]> I'd guess that the BDI knows exactly the platform Jun 10 11:04:18 <[g2]> because it read the ID from the chips on the JTAG chain Jun 10 11:04:37 <[g2]> and I think you are correct, there is probably some way of signalling Jun 10 11:04:50 <[g2]> bbiab Jun 10 11:33:43 beewoolie: ping Jun 10 11:46:47 ep1220: ping..ding Jun 10 11:47:10 do you have time to discuss Flas Prog over JTAG ? Jun 10 11:47:15 ep1220: do Jun 10 11:47:20 ok Jun 10 11:47:34 BTW, I'm excited about the progress. Jun 10 11:47:34 JTAG has a way to transfer data Jun 10 11:47:42 Me too :-) Jun 10 11:47:51 transfer + handshake Jun 10 11:48:05 Are you talking about DCC? Jun 10 11:48:36 Do not know DCC (?). there is a TxReg and a RX Reg plus Flags Jun 10 11:48:56 It is not standard JTAG Jun 10 11:49:06 Part of the xscale core? Jun 10 11:49:18 Yes, part of the JTAG debug controller Jun 10 11:49:34 Is this in the ixp users guide? Jun 10 11:49:48 in the core developer manual Jun 10 11:50:05 Hmm. Let me see what I've got. I might need to get that doc. Jun 10 11:51:12 Intel Doc Number is 27347302 Jun 10 11:51:18 I;ve got the IXP42x Product Line ... Developer's manual March 2005 Jun 10 11:51:30 Ah, no different doc... Jun 10 11:51:40 this one does not have a section on SW-debug Jun 10 11:51:48 the other one is dated Jan 2004 Jun 10 11:52:20 it is general XScale not 42x sepcific Jun 10 11:52:31 * beewoolie is download it Jun 10 11:52:56 baseline is: You load flasher stub into IC Jun 10 11:53:04 then feed it data thru JTAG Jun 10 11:53:17 Nice. That's better than I'd expected. Jun 10 11:53:41 TxReg is 32bits Jun 10 11:53:47 Where in the doc is this? Jun 10 11:54:01 ...9.14? Jun 10 11:54:02 Chapter 9 Software Debug Jun 10 11:54:20 9.14 is for cache Jun 10 11:54:36 9.11 describes JTAG side of Tx/Rx Jun 10 11:54:50 9.8-9.11 Cscale side Jun 10 11:55:19 Xscale side Jun 10 11:55:22 OK. Looks like I'll need to read up on this. Jun 10 11:55:27 What is the question? Jun 10 11:55:48 i have no question. Just wanted to point to You these registers Jun 10 11:56:27 OK. Looks like a straightforward way to handle this. Jun 10 11:56:37 Yes. there is a FTDI2232 issue which You should know if You plan to write code Jun 10 11:56:55 it is slow when You change transfer direction. Jun 10 11:57:05 e.g You write a Word. Jun 10 11:57:12 then read a handshake Jun 10 11:57:20 this takes 1millisecond Jun 10 11:57:26 Oh, That's bad Jun 10 11:57:29 So You get 4KB per sec max. Jun 10 11:57:39 still: 256K in approx. 1 minute Jun 10 11:57:51 gs said soething of 30minutes with parallel .. Jun 10 11:57:56 IMHO, that makes this part really unattractive. Jun 10 11:58:03 Now you can speed it up. Jun 10 11:58:19 try to transfer e.g. 4 words Jun 10 11:58:34 Let me honest here. It isn't very attractive to spend lots of development time with a setup that doesn't meet our design goals. Jun 10 11:58:56 what is the goal ? Jun 10 11:59:29 We want to be able to achieve commercial-grade performance with this device. Jun 10 11:59:39 While cheap is important, it also needs to be effective. Jun 10 12:00:00 It ought to be feasible to write the whole of flash in a production environment. Jun 10 12:00:19 in what time ? Jun 10 12:00:38 Well, 30 minutes is kinda slow. Jun 10 12:00:53 I said 1 minute - slowest .. Jun 10 12:01:08 if You do 4 words in direct sequence You are at 20seconds Jun 10 12:01:09 30 minutes for 8MiB. Jun 10 12:01:29 I suppose I need to look at this a little closer. Jun 10 12:01:43 you have a larger FLASH in mind .. Jun 10 12:01:59 Sure. I'm talking production. Jun 10 12:02:03 one alternative. You can download to SDRAM Jun 10 12:02:16 This can be done at 5Mbit per second Jun 10 12:02:18 Also, should be be able to write to more than one device at a time? Jun 10 12:02:51 with this scheme the Xscale writes to the devices. Jun 10 12:02:59 s/be be/we be/ Jun 10 12:03:11 so i think no problem (?) Jun 10 12:04:11 I suppose it isn't much of a problem since the device is going to be relatively cheap and USB ports are plentiful. Jun 10 12:04:32 I'll take a look at the guide. Jun 10 12:04:40 You many several devices on several boards. Jun 10 12:04:43 BTW, what would it take for me to get setup with the same HW that yo uhave? Jun 10 12:05:07 ep1220: In English? Jun 10 12:05:57 the fastest way: You buy the eval module (35$), plus have to add a voltage regulator and a JTAG header Jun 10 12:06:07 What You mean "In english ?" Jun 10 12:06:21 ep1220: you wrote: You many several devices on several boards Jun 10 12:07:06 Did you buy the eval board from the chip manufacturer? Jun 10 12:07:16 at this time i understood: Yu want to program several boards in parallel Jun 10 12:07:45 That would be the primary application. It may come down to being able to bring-up several boards at once. Jun 10 12:08:00 I bougth from FTDI because they are in Europe. Jun 10 12:08:01 it is also desirable for this procedure to be automate-able. Jun 10 12:08:21 k. Jun 10 12:09:06 the module is made by dlpdesign. Yu also can buy in the US http://www.dlpdesign.com/usb/2232m.shtml Jun 10 12:10:23 So, I need this board, a power source (voltage regulator) and an adapter for the JTAG header. Jun 10 12:10:43 yes. plus have to wire up these parts Jun 10 12:11:03 And a USB cable Jun 10 12:11:06 beewoolie: in the states is easier to go through digikey Jun 10 12:11:15 prpplague: Don't you like mouser? Jun 10 12:11:38 they are ok, but i prefer digikey's search engine Jun 10 12:11:53 prpplague: ditto. Mouser tends to be cheaper. Jun 10 12:12:22 beewoolie: With this setup You can do 3.3V targets only Jun 10 12:13:04 That's all I have. We've been planning to build a board capable of 5v, too. Jun 10 12:14:00 then You need to add a 5V capable driver. Jun 10 12:14:29 The plan has been to use a CPLD to handle that logic. Seem reasonable? Jun 10 12:15:01 I think costly. afik ka6sox has dropped this. Jun 10 12:15:12 Interesting. I wasn't aware of that. Jun 10 12:15:20 last time he was talking open collector. Jun 10 12:15:48 My thinking has been to develop a set of software that is capable of working with multiple emulators. Jun 10 12:16:06 Openwince is OK. It doesn't meet my expectations for ease-of-use. Jun 10 12:16:16 You can say this loud Jun 10 12:16:24 ;-) Jun 10 12:16:25 I shortly looked at source. Jun 10 12:16:43 seems not easy to add a device like the FTDI. Jun 10 12:17:02 their JTAG probe abstraction is at a fairly low level Jun 10 12:17:06 Did you use the FTDI libraries? Jun 10 12:17:21 I started with FTDI driver and JTAG lib. Jun 10 12:17:25 but it has a bug Jun 10 12:17:31 is unusable Jun 10 12:17:33 Say it isn't so! Jun 10 12:17:46 so now i wrote my own lib. Jun 10 12:18:00 as Jtaglib was windows only. I continued there. Jun 10 12:18:29 there is a linux driver, but not sure how well supported (seems binary only) Jun 10 12:18:45 in theory this driver should be compatible with the Win driver Jun 10 12:18:47 Honestly, I'm a couple of weeks away from getting into this. I'll need to buy the HW and assemble it. I want to make this a real FOSS tool for developers. Jun 10 12:19:18 There are some plans to make a PCB here. Jun 10 12:19:21 I appreciate all that you've done to get us motivated to make progress. Jun 10 12:19:28 nP Jun 10 12:19:47 personally i am also a bit disappointed by the performance Jun 10 12:19:50 When do you think a PCB is feasible? Jun 10 12:20:01 Do we have any other alternatives? Jun 10 12:20:23 I plan to look into FPGA + USB (Spartan 3) Jun 10 12:20:34 Will it be affordable? Jun 10 12:20:39 <$100 Jun 10 12:20:45 <$100USD Jun 10 12:21:04 FTDI I think 60-70$. Jun 10 12:21:18 if one can sell a few hundred, 50-60 Jun 10 12:21:27 the FPGA around 150$ Jun 10 12:21:39 but that would rivalan Abatron or Majic Jun 10 12:22:05 A lot depends on which voltage You target Jun 10 12:22:20 For a little more cash, that's really attractive to me. Ultimately, the cost isn't that important. The reall bugger with BDI is that they want $$$ for supporting other cores. Jun 10 12:22:46 I know. 2500EUR base + 800EUR per core Jun 10 12:24:02 We've been going around this spartan solution for some time. Is the hard part doing all of the FPGA logic to do the the JTAG emulation? Jun 10 12:24:36 if You do JTAG only, no. Jun 10 12:25:04 If You want fast JTAG debug you need more intelligence in there. Jun 10 12:25:33 the 8051 which does the USB is not fast enough Jun 10 12:25:46 Would it be feasible to start with a simple JTAG interface and do the coding on the host. Then, adding optimizations to handle faster debug? Jun 10 12:26:25 the reason the FTDI is slow is the USB bus. Jun 10 12:26:32 Sounds like we need an ethernet interface Jun 10 12:26:35 as they say in texas, shit or get off the pot Jun 10 12:26:39 ya gotta start somewhere Jun 10 12:26:57 I looked into Ethernet Jun 10 12:26:57 prpplague: It is also important to chart a good course. Jun 10 12:27:14 but that is a lot more expensive then the 8051 USB Jun 10 12:27:16 I don't have a problem starting with the FTDI part for the time being. I can get some code working. Jun 10 12:27:31 also my thinking. Jun 10 12:27:32 beewoolie: agreed, just think you should do some basic tests with the ftdi part first Jun 10 12:27:45 prpplague: on it. Jun 10 12:27:56 is a cheap way to try algos Jun 10 12:27:58 beewoolie: hehe Jun 10 12:28:09 ep1220: Seems like we should be able to get a cheap ethernet interface chip somewhere. Jun 10 12:28:40 I'll make an order in the next few days. BTW, what sort of regulator did you use? Jun 10 12:29:31 You need a 3,3V Low-Drop-out reg. I used an LT1085 - because it was in stock :-) Jun 10 12:29:40 is relative expensive Jun 10 12:29:46 100mA suffice Jun 10 12:29:46 k. thx Jun 10 12:30:19 Vin can be as low as 4.3V on USB Jun 10 12:30:52 Can we pull power from the USB cable and regulate it down? Jun 10 12:31:04 Exactly what I did. Jun 10 12:31:09 Excellent. Jun 10 12:34:31 one could make a very simple PCB taking the 2232M, the regulator and JTAG header in a short time. Jun 10 12:34:47 not sure if it is worth it Jun 10 12:35:09 I like to start with what we know works. Jun 10 12:37:30 prpplague: case in point. This voltage regulator is $5 on digikey, $1.30 on mouser. Jun 10 12:38:12 beewoolie: they are available in different cases, with different current ratings Jun 10 12:40:51 I know. Still, the mouser prices are significantly lower on average. Jun 10 12:41:05 yes, Mousers: sells You the one from ST version, not from LT Jun 10 12:41:26 donot know if there is an important difference. Jun 10 12:42:47 Do I care? Jun 10 12:43:45 I figure I'll get several and buy ones with a little larger output capacity. Jun 10 12:44:01 depends. If the drop-out voltage is higher, You might Jun 10 12:44:39 beewoolie: You are a HW engineer ? Jun 10 12:44:41 You can get around the USB latency, if you have a processor on the other end that can do complex tasks Jun 10 12:44:41 Help me out here. What does the drop-out voltage indicate? Jun 10 12:45:08 so you don't do single shift operations, but like, a batch memory read or write Jun 10 12:45:10 The min difference between input and output voltage. If it gets smaller, the regulator will power off Jun 10 12:45:14 ep1220: I have a EE degree, but I don't do this stuff on a regular basis. Jun 10 12:45:35 So, the idea is to have a relatively low drop-out voltage. Jun 10 12:45:43 yes. Jun 10 12:45:55 In our case USB gives You 4.3V worst case. Jun 10 12:46:12 Ah, so a 1.1 drop-out voltage is too high. Jun 10 12:46:30 If You build 1000: yes. Jun 10 12:46:45 For one most likely it will work. Jun 10 12:47:01 BTW: You will see 4.3 only if You use the device behind a bus-powered hub Jun 10 12:47:24 if YOu connect directly to the PC you have close to 5V most of the time. Jun 10 12:47:36 Ah. That's what I figured. I usually only connect direcly since USB ports are manifold these days. Jun 10 12:48:08 (except maybe You have another device which draws a lot of power, like a bus-powered TV grabber) Jun 10 12:51:09 Looks like they stock only one. It's got 3A output and a drop-out voltage of 1.3V Jun 10 12:51:23 For Your protoype I am sure You get away with this 1.3V (which is guaranteed worst-case) Jun 10 12:51:35 Indeed. Jun 10 12:51:43 It is 1.3 at 3A if we look at the same datasheet Jun 10 12:51:54 Right. Jun 10 12:53:07 You take the TO220 package ? Jun 10 12:53:28 That's what they have. Seems easy to work with. Jun 10 12:53:34 nod Jun 10 12:54:54 requires a capacitor on in and out. Jun 10 12:55:16 I think I've got those on hand. Jun 10 12:55:56 I forgot: You also need a MOSFET to get an open-drain ouput for RESET Jun 10 12:56:04 I used a BS107 Jun 10 12:59:28 Hmm. OK. Jun 10 13:00:31 my schematic is on paper for now. I will post it tomorrow. So maybe wait a day before odering Jun 10 13:00:57 e.g i use the 2nd channel in the module as a 3.3V RS232 to USB converter. Jun 10 13:01:05 k. Jun 10 13:01:21 That's what I had expected. Jun 10 13:01:40 will you use it with the NSLU2? Jun 10 13:01:47 That's the idea. Jun 10 13:02:11 in this case You also must add a TRST and RESET to the JTAG connector Jun 10 13:02:11 My nslu2 is presently wired for a digilent pinout. Jun 10 13:02:33 that does not suffice. You need the other 2 as well, else no load to mini-IC Jun 10 13:02:36 We shouldn't need those, IIRC. The JTAG bus can be managed without them. Jun 10 13:02:47 Oh. Interesting. Jun 10 13:03:00 np. Jun 10 13:03:37 means removing one resitor, adding 2 wires and 2 more connector pins Jun 10 13:03:54 s/resitor/resistor Jun 10 13:04:50 for mini-IC download You need to control RESET (the RESET button on the slug does not assert reset ..) Jun 10 13:04:59 I have a second slug. I'll make the mods to it. Jun 10 13:05:38 good. i updated the WIKI/JtagConnector section Jun 10 13:05:47 thx Jun 10 13:05:53 to tell were to connect these 2. Jun 10 13:06:15 together with the schematic i will post photos of my setup Jun 10 13:07:56 Good. This part search thing is really tedious. There are so many variables. I find the right part...and it's in the wrong package. Jun 10 13:09:36 yea. eg. there are also different BS107. the important thing here is Gate Threshold Voltage, should be lower than 2.5V Jun 10 13:10:30 basically i chose the parts based on what i had here, there are many equivalent ones Jun 10 13:11:07 As long as I you make your BOM available, I should be able to get equivalent parts. Jun 10 13:11:40 I will add key parameters. Jun 10 13:12:40 however, e.g. i used a LT1085 with adjustable ouput - so i have to Resistors which you would not need with the 3.3V fixed version Jun 10 13:13:14 s/to/two/ Jun 10 14:41:18 * prpplague heads home for the weekend Jun 10 16:13:21 Anyone interested in taking a job in Shanghai? **** ENDING LOGGING AT Fri Jun 10 23:59:57 2005