**** BEGIN LOGGING AT Fri Jul 08 23:59:56 2005 Jul 09 02:59:09 ep1220: What about the EZNIOSUSB? is that one compatible with 3.3V ? Jul 09 03:00:00 It should be. Jul 09 03:00:20 I seem to remember inspecting that schematic a few month ago Jul 09 03:01:26 I see no level changer circuitry, but there is a 3.3V regulator Jul 09 03:01:44 Does the 2232 need 3.3 V externally regulated? Jul 09 03:05:48 pin 3,42 VCC +4.35 - 5.25V Vcc for device core Jul 09 03:06:07 there is a convenience 3v3 out too Jul 09 03:06:36 Oh, I think my penance is up, you probably dont reconize this nick. Jul 09 03:13:22 Hm, looks like the chip can be configured to do either 3.3 or 5 V for I/O Jul 09 03:17:57 on 2 ports, theres a vccio for both a and b Jul 09 03:26:50 Jmunakra: Yes, EZNIOUSB is 3.3V Jul 09 03:27:44 and so should be the DLP-2232M Jul 09 03:28:11 the DLP-2232M does not have a voltage regualarot on board. Jul 09 03:28:19 You have to add it externally Jul 09 03:28:30 So you have an external 3.3V regulator to support your chip Jul 09 03:28:36 Yes. Jul 09 03:28:42 OK, got it Jul 09 03:29:17 is there other boards that might be suitable for development? Jul 09 03:29:55 the DLP looks nice. Jul 09 03:29:57 With 2232M, I am not aware of any other Jul 09 03:30:20 the DLP is the cheapest method. Jul 09 03:30:27 DLP is nice, but You need to add more parts Jul 09 03:30:44 Theres a dual serial port adapter with a 2232 in it, but its $80. Jul 09 03:30:57 JMunakra: where did You see the DLPM at 19$ Jul 09 03:30:58 ep1220, not too much extra stuff though right? Jul 09 03:31:30 at the FTDI site when you click "Buy online" Jul 09 03:31:54 dyoung: do you have the link to that serial adapter Jul 09 03:32:12 JMunakra; this are UKP not USD Jul 09 03:32:23 Oh :) Jul 09 03:32:29 Is approx 35USD Jul 09 03:32:48 the dlp: http://www.mouser.com/index.cfm?handler=displayproduct&lstdispproductid=560644&e_categoryid=284&e_pcodeid=62602 Jul 09 03:33:25 dyoung: what You need to add depends on what You want to do. Jul 09 03:34:07 If You only want to do FLASH programming thru the JTAG scan-chain all U need is a 3.3V regulator (plus capacitors) Jul 09 03:34:11 Hmm, the serial adapter is $40 now; I guess I was imagining things. Jul 09 03:34:32 ... or pilfer the 3.3V from the DUT.. Jul 09 03:34:41 http://www.usbgear.com/computer_cable_details.cfm?sku=XC-232-C&cats=199&catid=626%2C106%2C199%2C465 Jul 09 03:34:53 The 3.3C on-chip can only be laoded with a few mA Jul 09 03:35:01 Not enough to drive VCCIO Jul 09 03:35:18 oh, I meant the real DUT. Jul 09 03:35:23 like a nslu2 for instance... Jul 09 03:35:31 Taking 3.3V from DUT has following problem: Jul 09 03:35:48 the 2232C needs VCCinternal and VCCIO from same source Jul 09 03:36:10 e.g it does not like having internal power from USB and VCCIO unpowered Jul 09 03:36:45 dyoung, how do you know it has the 2232C in it? Jul 09 03:37:07 I think thats the one that prpplague looked at. Jul 09 03:37:31 I msyelf am not positive it has it. Jul 09 03:37:47 I saw one at the store that looks similar that does though. Jul 09 03:39:26 Still, You have to open these devices and add JTAG cabling, disable/remove the RS232 level converter, .. Jul 09 03:40:04 I guess you do not come away cheaper than with the DLP2232M Jul 09 03:40:05 change the EPROM Jul 09 03:40:15 THis is easy. SW only Jul 09 03:40:28 ep1220, exactly. Thats why I like the dlp ! Jul 09 03:41:09 One could think of making a small PCB to take the DLPM + other parts Jul 09 03:41:27 thts actually the current plan... Jul 09 03:41:41 The ideal thing for us would be a commercial product with a connector where only a matching cable needs to be made. Jul 09 03:41:59 JMunakra: this does not exist. Jul 09 03:42:20 Though I heard there are JTAG emulators using this chip (500$+) Jul 09 03:42:44 We could use their software :) Jul 09 03:43:08 the openjtag board that the dlp2232 plugs into should be cheap. Jul 09 03:43:48 I routed what I have and asked for a quote. Jul 09 03:43:52 If we had a JTAG software that supports more than just the slug, this thing could quickly become big Jul 09 03:44:12 Setup fee 50EUR, plus 7EUR per board Jul 09 03:45:10 (in Q10) Jul 09 03:45:27 JMunakra: Writing this SW is a lot of work Jul 09 03:45:54 ep1220, I have a working ARM debugger Jul 09 03:46:19 and there is a MIPS module, but I haven't tested it Jul 09 03:46:42 JMunakra: JtagPack ? Jul 09 03:46:49 jtagpack.sourceforge.net Jul 09 03:46:56 I looked at it: Jul 09 03:47:06 It ought to be easy enough to have "personality modules" to make it work for lots of targets Jul 09 03:47:09 Most files are copyright by a Japanese guy Jul 09 03:47:19 go and make a driver for 2232C for it and try, it shouldn't take more than a day now that you have the code Jul 09 03:47:37 Copyrighted, but they're GPL Jul 09 03:47:45 You saw: http://www.nahitech.com/jtag-en/license.html Jul 09 03:47:57 There he says it is not GPL Jul 09 03:48:30 Oh yeah, he switched license. The files I got were released under GPL. Jul 09 03:49:00 JMunakra: Now if he claims someone else has added the GPL lines ? Not that I believe you did this !!! Jul 09 03:49:36 I don't believe that's a big threat. Jul 09 03:49:53 I have his original packages, and copies of his old web sites. Jul 09 03:50:12 And even if I didn't... Jul 09 03:50:40 If he were to do anything about it, we could still dump his code and use the IP we produced so far. Jul 09 03:51:59 Anyway, I don't think he's doing a debugger right now. Jul 09 03:52:27 oh, yeah, he is :) Jul 09 03:54:16 dyoung: jtagpack is very modular. You need <500 lines of code to add a new hardware (bit shift level) Jul 09 03:54:43 JMunakra: the DLP-2232C is not good at bitshift level. Jul 09 03:54:49 There it is very slow. Jul 09 03:54:55 what else do you do with it? Jul 09 03:55:56 MPSSE... Jul 09 03:56:03 With bitShift level I mean. Your API is like setTMS(), SetTDO(), toggleTCK() Jul 09 03:56:15 You refer to the same thing ? Jul 09 03:57:01 For the 2232 your API is writeIR(), writeDR(), ... Jul 09 03:57:40 it knows about the TAP state machine? Jul 09 03:57:44 that's great! Jul 09 03:57:54 No it does not. Jul 09 03:58:20 If you say writeIR(), the stae machine has to change. Jul 09 03:58:28 how is that done? Jul 09 03:59:01 Yes, but this is done in SW. Yu send a combination of commands and bitstrings over USB Jul 09 03:59:38 Anyway, the jtagpack API is basically TAPSetState(), shiftInOut(). Jul 09 04:00:27 so that should go together easyly. Jul 09 04:02:10 That abstraction layer fits better Jul 09 04:02:36 night all Jul 09 04:02:52 nighty-night Jul 09 04:04:25 JMunakra: First i must finish the Linux Port of my library, anyway Jul 09 04:04:47 What does your lib do? Jul 09 04:05:27 implement commands like: WriteIR(), ReadDR(), WriteReadDR(), .. Jul 09 04:05:57 Then I have a few XSCALE specific higher level routines Jul 09 04:06:14 Like: Read Xscale DBGTX register, .. Jul 09 04:06:52 jtagpack utilizes a bitmask to indicate which bits need to be read Jul 09 04:07:38 the xscale functions would go into a higher abstraction level in jtagpack, over the core, while the bitshift is below the core Jul 09 04:08:31 I will take another look at it. Jul 09 04:08:46 the core (libjtag) abstracts away the physical chain (multiple chained chips etc.) Jul 09 04:09:32 I looked at it once, and saw it does a much better job at abstraction then OpenWince Jul 09 04:10:07 Still I prefer to have a core independent of jtagpack. Jul 09 04:10:31 Are you going to release your source? Jul 09 04:10:45 And maybe have another layer to fit it to jtagpack (so it can be separated easily if needed) Jul 09 04:10:50 Yes Jul 09 04:11:14 that should be easy Jul 09 04:11:39 I also have a faster JTAG HW in planning Jul 09 04:11:52 and want to make sure that one drops in easily :-) Jul 09 04:12:09 do you use the open source libftdi? Jul 09 04:12:30 I have it, and use it as a reference. Jul 09 04:12:40 I started with the FTDI API on Windows. Jul 09 04:12:42 libusb? Jul 09 04:12:53 is your code portable? Jul 09 04:13:15 And now I am implementing the (neede) functions from their library for Linus over libusb Jul 09 04:13:21 s/linus/Linux Jul 09 04:13:50 I want Win + Linux Jul 09 04:14:05 I want Mac OS X Jul 09 04:14:15 and Linux Jul 09 04:14:27 You have gcc on Mac OS X, right ? Jul 09 04:14:33 yeah Jul 09 04:14:40 and libusb Jul 09 04:14:48 This is all You need- Jul 09 04:15:05 jtagpack is also autotoolized. Jul 09 04:15:38 I have no experience with the autoconfig and co on Linux. Jul 09 04:15:51 I will leave this part to other people Jul 09 04:17:11 but until other people do have 2232C based compatible HW, it will not be of much use anyway Jul 09 04:18:28 What target do You use for You work on jtagpack ? Jul 09 04:18:38 ARM7TDMI Jul 09 04:18:50 evaluation board ? Jul 09 04:19:19 no, a board we made for a product at my company Jul 09 04:20:05 You are a HW engineer ? Jul 09 04:20:31 hm, usually the target powers the JTAG hardware, but do you say, in the FTDI case the JTAG hardware needs to power the target? Jul 09 04:20:35 no, SW Jul 09 04:21:16 For most commercial devices the target only powers the Level-shifters of the JTAG hardware Jul 09 04:21:35 With the FTDI there is no power connection between target and probe Jul 09 04:21:41 Yeah, but since we don't have level shifters Jul 09 04:21:45 (except for GND) Jul 09 04:22:20 OK, so the target doesn't need to be powered by the probe. Jul 09 04:22:54 Yes. however this leaves some issues when the probe is powered and the target not (or vice-cversa) Jul 09 04:23:09 ep1220: Taking 3.3V from DUT has following problem: Jul 09 04:23:09 ep1220: the 2232C needs VCCinternal and VCCIO from same source Jul 09 04:23:28 yes Jul 09 04:24:25 Oh, but the IO pins don't need to be supplied from VCCIO. I get it. Jul 09 04:25:02 so when the target is now powered? Jul 09 04:25:11 I don't see the problem Jul 09 04:25:46 If only one device is powered, the other is backdriven -> this can damage the IO drivers Jul 09 04:27:13 Hm, shouldn't an input be high-z? also when the device is not powered? Jul 09 04:28:42 You know how a diode works ? Jul 09 04:28:55 yeah :) Jul 09 04:29:18 Asume You have a diode backwards biased from the IO-pin to the VCC pin Jul 09 04:29:49 Normally this one does not have current flowing thru it Jul 09 04:30:06 as VCC is larger than the level at the IO pin. Jul 09 04:30:14 or the same Jul 09 04:30:26 Oh, I see Jul 09 04:30:37 Now the device is powered off, so VCC is at GND. Jul 09 04:30:40 and the diode is there for what? Jul 09 04:31:04 to protect the IO? Jul 09 04:31:12 e.g. Jul 09 04:31:27 but it can just be a "byproduct" of the design. Jul 09 04:31:35 E.g. with TTL Jul 09 04:31:40 Is that a general problem in circuit design? Jul 09 04:31:46 Yes. Jul 09 04:31:59 OK Jul 09 04:32:12 You have to design output stages to be immune to back-driving Jul 09 04:32:18 so we would be better off if we provided a level changer Jul 09 04:32:47 Yes, but one which is designed with this in mind. Jul 09 04:32:50 but I'm sure you designed that in as an option already :) Jul 09 04:34:00 This is primarily a matter of cost and what You expect. Jul 09 04:34:20 E.g. if i debug a NSLU (70EUR value) i might no care much. Jul 09 04:34:46 If this were my only prototype (2 weeks to replace it) I'd think differently. Jul 09 04:35:22 So far there was no conclusion here what features an openjtag HW should support. Jul 09 04:35:33 What performance is expected, ... Jul 09 04:35:44 At which price Jul 09 04:36:28 It would be good to have the option, so it can compete with a commercial product Jul 09 04:37:03 Of course, hackers will want to save the money, but we should have a broad adoption in mind. Jul 09 04:38:24 If You make it an option it adds to cost. Jul 09 04:38:41 i.e. more than when it is "mandatory" Jul 09 04:39:42 I mean in the sense of a stuffing option Jul 09 04:40:03 the board will be a tad bigger, but... Jul 09 04:40:56 but then You make 2 boards = more work, 2x setup cost ... Jul 09 04:41:10 why 2 boards? Jul 09 04:41:29 You das "will be a tad bigger" Jul 09 04:42:01 Actually the shifters are not that expensive. Jul 09 04:42:18 but hard to get in small numbers Jul 09 04:43:57 alright Jul 09 06:20:12 <[g2]> ep1220, morning Jul 09 06:20:38 morning g2 Jul 09 06:21:01 <[g2]> you had a nice chat with JMunakra yesterday :) Jul 09 06:22:27 You mean 4 hours ago ? Jul 09 06:26:43 <[g2]> yes :) Jul 09 06:27:36 was nice Jul 09 06:32:57 <[g2]> I'm gonna be making boards soon Jul 09 06:33:01 <[g2]> CPU boards Jul 09 06:33:09 XScale ? Jul 09 06:33:13 <[g2]> :) Jul 09 06:33:17 <[g2]> IXP422 Jul 09 06:33:24 :-) Jul 09 06:33:43 that one has encryption unit, right ? Jul 09 06:33:44 <[g2]> Ok... I'm not *making* the boards, I'm having an OEM make them for me Jul 09 06:33:52 <[g2]> yes Jul 09 06:34:03 which interfaces ? Jul 09 06:34:17 <[g2]> interfaces ? Jul 09 06:34:26 <[g2]> you mean what's on the boards ? Jul 09 06:34:31 yes Jul 09 06:35:08 <[g2]> 2 ethernet, serial, JTAG, miniPCI, CF, 64MB memory, + extras Jul 09 06:35:40 <[g2]> metal case Jul 09 06:35:59 sounds good. Jul 09 06:36:12 <[g2]> we'll see Jul 09 06:36:24 <[g2]> I'm excited and nervous about it Jul 09 06:36:34 I can feel with You. Jul 09 06:36:50 How many do You have to buy ? Jul 09 06:36:55 <[g2]> 200 Jul 09 06:38:30 Plan to sell them as development kits ? Jul 09 06:38:48 <[g2]> yes and LAMP/e-mail/other stuff Jul 09 06:39:06 Does it have harddisk IF onboard ? Jul 09 06:39:29 <[g2]> no. the CF + 2 USB 2.0 ports Jul 09 06:39:43 external PSU I asume. Jul 09 06:39:49 <[g2]> nod Jul 09 06:40:24 <[g2]> Target price is $249 US for all but the CF which will probably be sized but what ppl want on it Jul 09 06:40:44 <[g2]> that includes case and ps Jul 09 06:42:04 Does it have a boot FLASH ? or boot from CF ? Jul 09 06:43:04 <[g2]> tiny boot flash to load from CF Jul 09 06:43:35 You already have setup Your webshop ? Jul 09 06:43:43 <[g2]> from a cost perspective I think thats better and more useful than a fixed 8MB NOR flash Jul 09 06:44:09 <[g2]> I've got Apache 2.0 and PHP5 running on my slug :) Jul 09 06:44:15 I do not have Flash prices in my head. Jul 09 06:44:15 <[g2]> with ssl Jul 09 06:44:38 <[g2]> I'll need to setup some pages and full specifications Jul 09 06:45:21 <[g2]> I'll be launching in the next week or two Jul 09 06:45:55 Would be great if the PSU delivers enough power for a USB powered HD. Jul 09 06:47:06 <[g2]> :) Jul 09 06:47:44 The CPU is at 266MHz ? Jul 09 06:47:44 <[g2]> I think high-powered usb devices are .5A @ 5V right ? Jul 09 06:47:48 <[g2]> nod. Jul 09 06:48:05 <[g2]> we couild do 533, but again cost Jul 09 06:48:19 Yes. 500mA (is also what USB housings for 2.5 drives are rated) Jul 09 06:49:09 <[g2]> I think after having some devices out in the market (with some success) future builds could support that faster processor Jul 09 06:49:36 nod Jul 09 06:50:01 <[g2]> I don't know whether it'll be well received or a big yawn Jul 09 06:50:31 Hard to say. Jul 09 06:50:43 <[g2]> sure is Jul 09 06:51:49 I was thinking about a similar device for the SOHO market. Jul 09 06:52:04 there You might well be sucessfull. Jul 09 06:52:27 <[g2]> that's the market for the web server / e-mail server I was thinking of Jul 09 06:53:35 Do U have it FCC certified ? Jul 09 06:54:00 <[g2]> it'll have full compliance Jul 09 06:54:04 <[g2]> FCC, CE, etc... Jul 09 06:54:22 the OEM does the certification ? Jul 09 06:54:50 <[g2]> it costs Jul 09 06:55:01 I know :-( Jul 09 06:55:14 Don't know about FCC but you can self certify for CE Jul 09 06:55:36 Just woe betide you if you say it is compliant and then somebody finds out it isn't... Jul 09 06:55:53 but You might have to show measurements .. Jul 09 06:56:15 You're supposed to have measurements but you don't have to file them anywhere first Jul 09 06:56:21 true Jul 09 06:56:46 so if you're really on a budget you can just hope it's compliant and only get the measurements done if there are challenges. living life really on the edge then Jul 09 06:57:17 <[g2]> It's not *that* expensive Jul 09 06:57:19 I know a couple of people that have done that. They've designed it to be compliant to the best of their knowledge and just self certified it Jul 09 06:57:39 For one thing it's not but you're supposed to do it for every revision and model Jul 09 06:57:49 that can add up quickly Jul 09 06:58:03 or maybe it's just I know a lot of real cheapskate people :) Jul 09 06:58:36 I guess, You put it besides Your TV set - and if it does not harm receiption Your customers likely will not complain :-) Jul 09 06:58:36 <[g2]> well certainly for a spin you can just get some lab time and do your own measurements Jul 09 06:58:47 yep Jul 09 06:59:09 <[g2]> but having the first design checked out makes a bunch of sense Jul 09 06:59:14 Renting the gear/lab and doing it yourself is the cheapest method Jul 09 06:59:42 <[g2]> except for the "doing it yourself" that's what's being done Jul 09 06:59:48 yeah Jul 09 07:00:49 Assuming you follow good practices and it's in a case, it's unlikely to fail anyway Jul 09 07:00:58 <[g2]> nod. Jul 09 07:04:47 Getting a little distribution made up that does VPN or acts like a AP should help the popularity of it Jul 09 07:06:33 <[g2]> Tiersten, VPN is a possibility, I'd convert the RV082 with firmware first Jul 09 07:06:57 <[g2]> unless logging was required in which case it's a toss-up Jul 09 07:07:19 <[g2]> A transparent bridge / logger will be an option though Jul 09 07:07:45 <[g2]> I think the RV082 has the 533Mhz 425 and 16MB flash Jul 09 07:08:03 <[g2]> same basic price point $250-260 US Jul 09 07:08:19 <[g2]> Plus 8 extra ethernets Jul 09 07:10:04 Just having something people can load and then play I mean Jul 09 07:10:08 <[g2]> I chrooted into statically built nslu2 recovery partition Jul 09 07:10:25 <[g2]> that's the whole point Jul 09 07:10:45 <[g2]> I'm hoping I don't totally lose my shirt/pants/etc.... Jul 09 07:10:55 <[g2]> socks Jul 09 07:11:18 <[g2]> I think it makes total sense, but time will tell Jul 09 07:11:59 The CF, it in IDE mode? Jul 09 07:12:14 <[g2]> think so Jul 09 07:16:21 [g2]: will the case stand vertically? Jul 09 07:16:35 <[g2]> for you it will Jul 09 07:16:39 heh Jul 09 08:36:47 hi g2 Jul 09 08:39:08 <[g2]> hey jacques Jul 09 08:39:11 <[g2]> doh! Jul 09 08:39:13 <[g2]> hey JMunakra **** ENDING LOGGING AT Sat Jul 09 23:59:56 2005