On 2010-06-23, Jon Kirwan <jonk@infinitefactors.org> wrote:> On Wed, 23 Jun 2010 21:04:55 +0000 (UTC), Grant Edwards>>>>Funny, all the discussions about baudrates and errors. I'd use an ISA COM >>>>card and simply replace the crystal... done it, works like a charm. Only 5 >>>>minutes work. >>> >>> Hehe. But this means you actually _have_ something with an >>> ISA bus on it!! These days... well. >> >>Then use a PCI card. >> >>> On your point, yes. A crystal change on any of the usual >>> spate of old ISA boards would easily solve the problem. >>> Forgotten lore. >> >>It works for PCI cards as well. > > I like ISA and simpler software. > > Although I understand reflection wave principles, clock line > skew and serpentine clock lines,And swapping out the buad-rate clock oscillator on a PCI serial card requires no understanding of any of that.> and the like, I very much appreciate being able to use simple logic, > wire-wrapping techniques, and custom circuit design with the ISA bus. > It is a low-tech bus that can be reached by hobbyists.We're not talking about building a card from scratch. We're talking about taking an off-the-shelf serial card and swapping out the xtal oscillator.> PCI, and not merely because of the hardware but also because of other > aspects (plug and play), out of reach of most hobbyist tools and > skills.Swapping in a 2.00MHz oscillator for a 1.843MHz oscillator on a PCI card requires _exactly_the_same_ skills as doing it on an ISA card.> And I also understand the desire to get rid of the south bridge, > chipset side-band channels to support ISA DMA over a bus that simply > cannot and does not support ISA DMA timing requirements, interrupt > mapping, and so on. It's pretty obvious this was a huge source of > continuing chipset bugs and needed testing regimes, as well.I have absolutely no idea what any of that has to do with this thread. -- Grant Edwards grant.b.edwards Yow! Mary Tyler Moore's at SEVENTH HUSBAND is wearing gmail.com my DACRON TANK TOP in a cheap hotel in HONOLULU!
Monitor RS232 comms with millisecond resolution
Started by ●June 20, 2010
Reply by ●June 23, 20102010-06-23
Reply by ●June 23, 20102010-06-23
On Wed, 23 Jun 2010 21:58:21 +0000 (UTC), Grant Edwards <invalid@invalid.invalid> wrote:>On 2010-06-23, Jon Kirwan <jonk@infinitefactors.org> wrote: >> On Wed, 23 Jun 2010 21:04:55 +0000 (UTC), Grant Edwards > >>>>>Funny, all the discussions about baudrates and errors. I'd use an ISA COM >>>>>card and simply replace the crystal... done it, works like a charm. Only 5 >>>>>minutes work. >>>> >>>> Hehe. But this means you actually _have_ something with an >>>> ISA bus on it!! These days... well. >>> >>>Then use a PCI card. >>> >>>> On your point, yes. A crystal change on any of the usual >>>> spate of old ISA boards would easily solve the problem. >>>> Forgotten lore. >>> >>>It works for PCI cards as well. >> >> I like ISA and simpler software. >> >> Although I understand reflection wave principles, clock line >> skew and serpentine clock lines, > >And swapping out the buad-rate clock oscillator on a PCI serial card >requires no understanding of any of that.Agreed. I just took Meindert's comment regarding an ISA board and ran with it. :P>> and the like, I very much appreciate being able to use simple logic, >> wire-wrapping techniques, and custom circuit design with the ISA bus. >> It is a low-tech bus that can be reached by hobbyists. > >We're not talking about building a card from scratch. We're talking >about taking an off-the-shelf serial card and swapping out the xtal >oscillator.We could be, but I expanded Meindert's comments in a different direction for the pure pleasure of doing so. Threads are sometimes like that.>> PCI, and not merely because of the hardware but also because of other >> aspects (plug and play), out of reach of most hobbyist tools and >> skills. > >Swapping in a 2.00MHz oscillator for a 1.843MHz oscillator on a PCI >card requires _exactly_the_same_ skills as doing it on an ISA card.Not so. I'm looking right now at two such cards, one ISA and one PCI. The ISA board has a large, socketed crystal module. The PCI a tiny, SMT unit. The skills required for modifying one is much different (and the tool tips required, too.) It _may_ be the case, but not necessarily so as these two boards easily illustrate to me.>> And I also understand the desire to get rid of the south bridge, >> chipset side-band channels to support ISA DMA over a bus that simply >> cannot and does not support ISA DMA timing requirements, interrupt >> mapping, and so on. It's pretty obvious this was a huge source of >> continuing chipset bugs and needed testing regimes, as well. > >I have absolutely no idea what any of that has to do with this thread.Then you are lost to what I've been saying. Which is fine. That happens, as well. Jon
Reply by ●June 23, 20102010-06-23
On Jun 23, 4:50=A0pm, Jon Kirwan <j...@infinitefactors.org> wrote:> On Wed, 23 Jun 2010 21:04:55 +0000 (UTC), Grant Edwards > > > > > > <inva...@invalid.invalid> wrote: > >On 2010-06-23, Jon Kirwan <j...@infinitefactors.org> wrote: > >> On Wed, 23 Jun 2010 14:13:14 +0200, "Meindert Sprang" > >><m...@NOJUNKcustomORSPAMware.nl> wrote: > > >>>"Paul Keinanen" <keina...@sci.fi> wrote in message > >>>news:9pe3269vevlnos3fn3j1ne6dv7tnfgm293@4ax.com... > >>>> At 62.5 kbit/s the bit time is 16 us, thus +/-8 us error from the > >>>> nominal sampling point from the middle of the bit period would be > >>>> allowed. > > >>>Funny, all the discussions about baudrates and errors. I'd use an ISA =COM> >>>card and simply replace the crystal... done it, works like a charm. On=ly 5> >>>minutes work. > > >> Hehe. =A0But this means you actually _have_ something with an > >> ISA bus on it!! =A0These days... well. > > >Then use a PCI card. > > >> On your point, yes. =A0A crystal change on any of the usual > >> spate of old ISA boards would easily solve the problem. > >> Forgotten lore. > > >It works for PCI cards as well. > > I like ISA and simpler software. > > Although I understand reflection wave principles, clock line > skew and serpentine clock lines, and the like, I very much > appreciate being able to use simple logic, wire-wrapping > techniques, and custom circuit design with the ISA bus. =A0It > is a low-tech bus that can be reached by hobbyists. =A0PCI, and > not merely because of the hardware but also because of other > aspects (plug and play), out of reach of most hobbyist tools > and skills. > > And I also understand the desire to get rid of the south > bridge, chipset side-band channels to support ISA DMA over a > bus that simply cannot and does not support ISA DMA timing > requirements, interrupt mapping, and so on. =A0It's pretty > obvious this was a huge source of continuing chipset bugs and > needed testing regimes, as well. > > But I like ISA.Just slap something like a PLX 9052 on your PCI board. The front side handles the PCI, the back side is ISA.
Reply by ●June 23, 20102010-06-23
On Wed, 23 Jun 2010 16:28:56 -0700 (PDT), "robertwessel2@yahoo.com" <robertwessel2@yahoo.com> wrote:>On Jun 23, 4:50�pm, Jon Kirwan <j...@infinitefactors.org> wrote: >> On Wed, 23 Jun 2010 21:04:55 +0000 (UTC), Grant Edwards >> >> <inva...@invalid.invalid> wrote: >> >On 2010-06-23, Jon Kirwan <j...@infinitefactors.org> wrote: >> >> On Wed, 23 Jun 2010 14:13:14 +0200, "Meindert Sprang" >> >><m...@NOJUNKcustomORSPAMware.nl> wrote: >> >> >>>"Paul Keinanen" <keina...@sci.fi> wrote in message >> >>>news:9pe3269vevlnos3fn3j1ne6dv7tnfgm293@4ax.com... >> >>>> At 62.5 kbit/s the bit time is 16 us, thus +/-8 us error from the >> >>>> nominal sampling point from the middle of the bit period would be >> >>>> allowed. >> >> >>>Funny, all the discussions about baudrates and errors. I'd use an ISA COM >> >>>card and simply replace the crystal... done it, works like a charm. Only 5 >> >>>minutes work. >> >> >> Hehe. �But this means you actually _have_ something with an >> >> ISA bus on it!! �These days... well. >> >> >Then use a PCI card. >> >> >> On your point, yes. �A crystal change on any of the usual >> >> spate of old ISA boards would easily solve the problem. >> >> Forgotten lore. >> >> >It works for PCI cards as well. >> >> I like ISA and simpler software. >> >> Although I understand reflection wave principles, clock line >> skew and serpentine clock lines, and the like, I very much >> appreciate being able to use simple logic, wire-wrapping >> techniques, and custom circuit design with the ISA bus. �It >> is a low-tech bus that can be reached by hobbyists. �PCI, and >> not merely because of the hardware but also because of other >> aspects (plug and play), out of reach of most hobbyist tools >> and skills. >> >> And I also understand the desire to get rid of the south >> bridge, chipset side-band channels to support ISA DMA over a >> bus that simply cannot and does not support ISA DMA timing >> requirements, interrupt mapping, and so on. �It's pretty >> obvious this was a huge source of continuing chipset bugs and >> needed testing regimes, as well. >> >> But I like ISA. > >Just slap something like a PLX 9052 on your PCI board. The front side >handles the PCI, the back side is ISA.I still can't just wire-wrap the PCI bus side of it. 1.5" +/- 0.1" clock line, serpentined, 2ns clock skew at 33MHz and 1ns at 66, etc. I suppose pre-built boards with chips down and the ISA back-side might do it. But this, again, is getting way over my hobbyist toolset. If there is a problem with the PCI bus, I simply don't have the tools to diagnose. They are kind of expensive, too. Also, no plug-and-play, required PCI transaction support and registers to deal with, etc. ISA is __very__ easy. Add PCI, even if only on a corner of it, and complexity goes ^^^ up. I'm glad to keep the older machines around for hobby playing. They work VERY well, even to this day. And Win98SE licensing fits my model and boots _very_ quickly, well. Jon
Reply by ●June 24, 20102010-06-24
On 23/06/2010 23:50, Jon Kirwan wrote:> On Wed, 23 Jun 2010 21:04:55 +0000 (UTC), Grant Edwards > <invalid@invalid.invalid> wrote: > >> On 2010-06-23, Jon Kirwan<jonk@infinitefactors.org> wrote: >>> On Wed, 23 Jun 2010 14:13:14 +0200, "Meindert Sprang" >>> <ms@NOJUNKcustomORSPAMware.nl> wrote: >>> >>>> "Paul Keinanen"<keinanen@sci.fi> wrote in message >>>> news:9pe3269vevlnos3fn3j1ne6dv7tnfgm293@4ax.com... >>>>> At 62.5 kbit/s the bit time is 16 us, thus +/-8 us error from the >>>>> nominal sampling point from the middle of the bit period would be >>>>> allowed. >>>> >>>> Funny, all the discussions about baudrates and errors. I'd use an ISA COM >>>> card and simply replace the crystal... done it, works like a charm. Only 5 >>>> minutes work. >>> >>> Hehe. But this means you actually _have_ something with an >>> ISA bus on it!! These days... well. >> >> Then use a PCI card. >> >>> On your point, yes. A crystal change on any of the usual >>> spate of old ISA boards would easily solve the problem. >>> Forgotten lore. >> >> It works for PCI cards as well. > > I like ISA and simpler software. > > Although I understand reflection wave principles, clock line > skew and serpentine clock lines, and the like, I very much > appreciate being able to use simple logic, wire-wrapping > techniques, and custom circuit design with the ISA bus. It > is a low-tech bus that can be reached by hobbyists. PCI, and > not merely because of the hardware but also because of other > aspects (plug and play), out of reach of most hobbyist tools > and skills. >ISA has also always been out of the reach of the hobbyist. While there is no doubt that it is simpler to design an ISA card than a PCI card, in the days of ISA it was hard to make such a card. Information about the bus wasn't as easily available (no Google), and even if you knew how to do it, the design of a card was not insignificant. But the biggest hurdle for a hobbyist would be testing - you need an expendable spare computer to test your card, because of the high risk of frying the whole machine. These days you can get a cheap PCI bus computer for very little, and second-hand ones for practically nothing. When ISA was the main bus, a spare computer was a big investment.
Reply by ●June 24, 20102010-06-24
"Jon Kirwan" <jonk@infinitefactors.org> wrote in message news:8715265o00dgu3msi6bra5d51n3sj2npvs@4ax.com...> Not so. I'm looking right now at two such cards, one ISA and > one PCI. The ISA board has a large, socketed crystal module. > The PCI a tiny, SMT unit. The skills required for modifying > one is much differentOnce you've done it a couple of times, you wonder why you ever made such a fuss about it :-) Bu then again, I am an embedded software and hardware guy.... Meindert
Reply by ●June 24, 20102010-06-24
"David Brown" <david@westcontrol.removethisbit.com> wrote in message news:4c23020b$0$4113$8404b019@news.wineasy.se...> ISA has also always been out of the reach of the hobbyist. While there > is no doubt that it is simpler to design an ISA card than a PCI card, in > the days of ISA it was hard to make such a card. Information about the > bus wasn't as easily available (no Google), and even if you knew how to > do it, the design of a card was not insignificant.I designed a few cards back in the 90's. Information was not hard to come by, Intel provided the spec, IIRC, and as long as you double-triple-quadruple checked the power connections, the only serious thing that could happen is that your computer locked up when the card was inserted. And even that actually never happened to me. The hardest thing for me was to find a decent mechanical specification to make it fit into every computer case.. Meindert
Reply by ●June 24, 20102010-06-24
On Thu, 24 Jun 2010 09:10:13 +0200, "Meindert Sprang" <ms@NOJUNKcustomORSPAMware.nl> wrote:>"Jon Kirwan" <jonk@infinitefactors.org> wrote in message >news:8715265o00dgu3msi6bra5d51n3sj2npvs@4ax.com... >> Not so. I'm looking right now at two such cards, one ISA and >> one PCI. The ISA board has a large, socketed crystal module. >> The PCI a tiny, SMT unit. The skills required for modifying >> one is much different > >Once you've done it a couple of times, you wonder why you ever made such a >fuss about it :-):) I have put some effort into it, before.>Bu then again, I am an embedded software and hardware guy....Well, there is that. I am only a hobbyist on the hardware side. Jon
Reply by ●June 24, 20102010-06-24
On Thu, 24 Jun 2010 08:57:40 +0200, David Brown <david@westcontrol.removethisbit.com> wrote:>On 23/06/2010 23:50, Jon Kirwan wrote: >> On Wed, 23 Jun 2010 21:04:55 +0000 (UTC), Grant Edwards >> <invalid@invalid.invalid> wrote: >> >>> On 2010-06-23, Jon Kirwan<jonk@infinitefactors.org> wrote: >>>> On Wed, 23 Jun 2010 14:13:14 +0200, "Meindert Sprang" >>>> <ms@NOJUNKcustomORSPAMware.nl> wrote: >>>> >>>>> "Paul Keinanen"<keinanen@sci.fi> wrote in message >>>>> news:9pe3269vevlnos3fn3j1ne6dv7tnfgm293@4ax.com... >>>>>> At 62.5 kbit/s the bit time is 16 us, thus +/-8 us error from the >>>>>> nominal sampling point from the middle of the bit period would be >>>>>> allowed. >>>>> >>>>> Funny, all the discussions about baudrates and errors. I'd use an ISA COM >>>>> card and simply replace the crystal... done it, works like a charm. Only 5 >>>>> minutes work. >>>> >>>> Hehe. But this means you actually _have_ something with an >>>> ISA bus on it!! These days... well. >>> >>> Then use a PCI card. >>> >>>> On your point, yes. A crystal change on any of the usual >>>> spate of old ISA boards would easily solve the problem. >>>> Forgotten lore. >>> >>> It works for PCI cards as well. >> >> I like ISA and simpler software. >> >> Although I understand reflection wave principles, clock line >> skew and serpentine clock lines, and the like, I very much >> appreciate being able to use simple logic, wire-wrapping >> techniques, and custom circuit design with the ISA bus. It >> is a low-tech bus that can be reached by hobbyists. PCI, and >> not merely because of the hardware but also because of other >> aspects (plug and play), out of reach of most hobbyist tools >> and skills. >> > >ISA has also always been out of the reach of the hobbyist.Are you just in a contrary mood, today? No other explanation is possible. Out of reach??? Hardly. But I suppose it depends upon what is, is. :)>While there >is no doubt that it is simpler to design an ISA card than a PCI card, in >the days of ISA it was hard to make such a card.IBM, itself (and I know this because I still have them here), sold very cheap proto boards for the ISA bus as far back as I can recall. Very beautiful boards, too. I believe I paid close to $30 for each. 1984/1985.>Information about the >bus wasn't as easily available (no Google),Hogwash. I still have my technical reference multi-volume binder set from IBM with __complete__ documentation. Very complete. Complete BIOS listings, with comments, included, and schematics as well. I used these routinely for hobby playing. They were available in 1983 and IBM sent regular updates as new boards came out. For example, volume 3 includes a section I used called "IBM Personal Computer General Purpose Interface Bus Technical Reference," that is dated August 15th, 1984. Part number 6138155. (It wasn't called ISA, then, of course.) Subsequent to this, there were some books out (latish, but for late-comers quite good all the same) that appeared. As a supplement book to those who didn't have access to IBM's manuals (a shame), Edward Solari's books were a great help. His AT book first came out circa 1989/1990, memory serving. And then of course MindShare's multi-editions (my shelf copy here is the 3rd edition) on the ISA (and other buses.) Personally, I used the rather complete IBM information early on and liked Solari's contribution. MindShare was fine, but came too late.>and even if you knew how to >do it, the design of a card was not insignificant.Trivial, for trivial projects. I think so, anyway. Plus, IBM provided, on their proto boards, a complete, laid out 7400 series style decode section. So I guess maybe I just had it too easy.>But the biggest >hurdle for a hobbyist would be testing - you need an expendable spare >computer to test your card, because of the high risk of frying the whole >machine. These days you can get a cheap PCI bus computer for very >little, and second-hand ones for practically nothing. When ISA was the >main bus, a spare computer was a big investment.Since I do NOT consider myself to be anything other than a hobbyist in electronics -- barely that -- and since I already know just how darned easy this was for me when I knew far, far less than I do now about it, I don't know at all where you are coming from. Or, at least, I simply didn't find all the struggles you suggest. And by comparison with PCI??? No discussion. I used tools developed for PCI at Intel. I know how much they paid for them. And I know what including PCI on the Pentium did to "mom and pop" motherboard manufacturers, too. (Intel staff told me this was an internal goal, in fact.) I won't even debate it. Not worth the trouble. Jon
Reply by ●June 24, 20102010-06-24
On Thu, 24 Jun 2010 09:16:27 +0200, "Meindert Sprang" <ms@NOJUNKcustomORSPAMware.nl> wrote:>"David Brown" <david@westcontrol.removethisbit.com> wrote in message >news:4c23020b$0$4113$8404b019@news.wineasy.se... >> ISA has also always been out of the reach of the hobbyist. While there >> is no doubt that it is simpler to design an ISA card than a PCI card, in >> the days of ISA it was hard to make such a card. Information about the >> bus wasn't as easily available (no Google), and even if you knew how to >> do it, the design of a card was not insignificant. > >I designed a few cards back in the 90's. Information was not hard to come >by, Intel provided the spec, IIRC, and as long as you >double-triple-quadruple checked the power connections, the only serious >thing that could happen is that your computer locked up when the card was >inserted. And even that actually never happened to me. The hardest thing for >me was to find a decent mechanical specification to make it fit into every >computer case..IBM provided __complete__ documentation. I always felt like I had enough. And that goes back to late 1983 and early 1984. So long before 1990, when Solari's book on the AT arrived. I did terrible things, risky things, and didn't have any trouble to speak about. Just the usual. I avoided the 'mechanicals' issue by simply using the excellent and relatively cheap (in fact, they were less expensive than 3rd-party sources [Jameco, later] which were much inferior and more expensive, to boot) IBM prototype boards. These were perfectly built, bus interface already laid out, and slick as heck to use. No troubles. And that was 1985/1986. And by that time EPROMs that were clearly fast enough for the bus were rather readily available to a meager hack like me. (BIOS extension.) Jon
