On Sat, 15 Jun 2013, Dave McGuire wrote: Ahhh. Okay. I had a feeling I might've been wrong there. I couldn't recall if the USB spec mentioned DMA or not...guess either it did or controllers added it anyway -- -- Cory Smelosky http://gewt.net Personal stuff http://gimme-sympathy.org Projects

On Sat, 16 Jun 2013, Dave McGuire wrote: Hmmmmm. True. That would depend on case-specific methods and shouldn't be in a spec. -- -- Cory Smelosky http://gewt.net Personal stuff http://gimme-sympathy.org Projects

On 06/15/2013 01:04 PM, Tony Duell wrote: That's funny--I've never heard any claims by any of the promulgators of the above-named interfaces about being able to transfer data at 4.0/3.2 Gbit/sec, but for USB. --Chuck

On 06/16/2013 01:19 PM, Tony Duell wrote: Well, that's the burst rate anyway. But with online storage even for the home computer user running into the multi-terabyte range, I can see an application for a very high-speed interface. Most of the complexity is in the host, not the peripheral. Indeed, even without the FTDI-style interface chips, it's still possible to inexpensively create a device-level interface to USB using nothing more than an AVR ATTiny MCU--see the VUSB project list for minimalist interfaces. Indeed, my only gripe is with the registration requirements and costs involved with dealing with the USB-IF people. Some Chinese outfits have gotten around that one by simply picking VID and PIDs out of thin air. But at its most basic, USB is not terribly complex and is a good match for today's devices. --Chuck

On 06/17/2013 11:41 AM, Tony Duell wrote: What, if anything does MSI mean any more? I can remember a time when the 74181 was considered to be LSI. And what does IEEE 388 have to do with this discussion? According to my publication catalog, it's "Standard for Transformers and Inductors in Electronic Power Conversion Equipment " I suppose one could also argue that the same task could be accomplished with a bucket of 2N2222s as well. What's the point? A functional USB interface can be done in an 8 pin microcontroller that probably costs an order of magnitude less than the sockets used in your 20 MSI package implementation. 232C and USB are both bit serial, but USB has higher speeds, friendly voltage levels, power from a host included and well-defined protocol. --Chuck

I'm not sure what's supposed to be unfriendly about RS-232C. The major unfriendly part there, it seems to me, is the space in the middle, -3V to +3V IIRC, for which the logical signal state is undefined. Though I suppose it's a question of "friendly to what/whom".... The part of RS-232C I _really like_ is the impedance spec, the part that says that any pins or combinations of pins may be shorted to one another or to any voltage source within, I think, the valid RS-232C voltage range (-15V to +15V or some such - I forget the exact limits), with the shorts maintained indefinitely, without damage. I also like the splitting of protective ground and signal ground. You'd almost think it had actually been designed by competent electronics engineers or somethin'.... /~\ The ASCII Mouse \ / Ribbon Campaign X Against HTML mouse at rodents-montreal.org / \ Email! 7D C8 61 52 5D E7 2D 39 4E F1 31 3E E8 B3 27 4B

True! Especially when "misused" for that for which it was not intended. RS232 had a documented fatality - a homicide directly attributed to frustration of DTE (microcomputer) to DTE (printer) interfacing. I was luckier, I succeeded in getting ink onto paper with my first serial printer (DTC-300 Hytype 1) connected to TRS80 Model 1 after about 5 weeks. But, the only way that I ever got text into a Merganthaler was with a box of solenoids sitting on its keyboard, and that was slower than my publisher's wife's typing. -- Grumpy Ol' Fred cisin at xenosoft.com

On Tue, 18 Jun 2013, Tony Duell wrote: But, you probably knew what you were doing. OTOH, I had a guy working for me who started with NO hardware experience. He had taken two of my beginning programming classes, and decided that THAT was what he wanted to do with his life, so he begged me to let him work unpaid. After a probationary period, I made him Vice President. On his first attempt, he reached into one of my bins of cables, and tried the first one that had the right gender on each end. It turned out to have exactly the right lines crossed and worked. You had schematics??!? My first attempt was with the mimeographed "preliminary" docs for the Radio Shack Serial interface add-on card. and no Google. -- Grumpy Ol' Fred cisin at xenosoft.com

On 18/06/2013 20:13, Tony Duell wrote: My first ever serial printer took more like 3 days (maybe half a dozen days). I have excuses, though: my Exidy Sorcerer had two programmable baud rates, 300 and 1200, and my Creed 7B did neither; my Sorcerer was RS232 and the Creed was current-loop; and the Sorcerer spoke ASCII and the Creed spoke Murray. I still got it printing listings fairly quickly. -- Pete Peter Turnbull Network Manager University of York

On 19/06/2013 21:12, Tony Duell wrote: 50 baud, yes. So one part of the interfacing was making a little daughterboard to divide by 6 and feed that to the UART when serial (but not cassette) was selected. That part had been taken care of by the previous owner. It had an 80VDC supply. I just had to do a little extra. Not sure about Murray/Baudot, to be honest. But several character codes were non-spacing and think the shifts were. Anyway, I modified the print drum a little to make it easier to get certain things by using overstrikes. And then wrote a driver that translated the Sorcerer's 8-bit codes into 5-bit plus shifts and overstrikes where appropriate. -- Pete Peter Turnbull Network Manager University of York

On 06/17/2013 07:52 PM, Mouse wrote: Well, the problem is that to go from mark to space, you need to swing AT LEAST 6 volts to be defined. Given that most logic is either 3V or 5V, that's a bit of a problem. OTOH, USB is differential, not single-ended. Had you specified RS-422 or RS-486 as a comparison, I would admit that both were competitive with USB. Every citation of the TIA/EIA/RS-232 spec talks about the range of voltages used for signaling, but I can't find a citation of "voltage into what LOAD" defined. Is 2 ohms too small; is 2 megohms too big? How much current at the specified voltage range is to be supplied by the driver? RS?EIA?TIA-232 is a single-ended protocol, which makes it less tolerant to various signal issues, such as ground loops, induced noise, etc. I remember that the old MC1488 driver were power-hungry devices; about half a watt per package if memory serves. The 1489 receiver was nearly as bad in terms of power. '232 was a good interface for the 1960s and 70s, but it's not a great choice today. --Chuck

On 06/17/2013 09:50 PM, Mouse wrote: Well, it goes to my first point about +/-3-+/-15 volts into what? Resistance, capacitance, what are limits? Where in the spec does it say? Us 10 pF too little and 10 uF too much? So the old devices had to be practical, meaning lots of current and voltage into a loosely-specified load. I suspect that a MAX232 probably wouldn't have enough oomph to drive some of the old 1960s gear. As far as pin count--differential is only 2 pins; on single-ended signals, you still need a return line for the signal line. --Chuck

Charge pumps... Step up transformers... Alas very few machines had RS422 ports (the Mac serial port wasn't truely RS422 IIRC). But covnerting between RS232 and RS422 is a pretty simple job. I've never seen a single-chip converter, was there one? It's the sort of thing that Maxim might make, but I've not come across it. Did you mean RS486? RS485 is the multi-drop RS422, I think. I've not looked, but is it in the appropriate CCITT spec (V.24?) ? That may well be availble on-line. -tony

_You_ mentioend RS486 I think. I'd never come across it. I have come across RS485 which is the multidrop version of RS422 (several devices on one bus), I wondered if that's what you meant. Thanks, I will take a look -tony

Since when has cost had anything to do with complexity? And if that's an 8 pin microcontroller, then if you use 2 pins for power, and 2 pins for the USB signals, you only have 4 pins left to tlak to wahtever. Which probably implies a serial interface on that side too. So more circuitry is needed to link that to whatever yoy are working with. It appears that you measure complexity by the number of devices you solder to the PCB, or perhpas the total number of soldered joints. I do not. I consider things like the number of active devices (or gates, or whatever) inside the ICs, the number of states of state machines, etc. let me ask you a simple quesiton. You have to flash an LED at about 1Hz -- the exact frequency is not important. Nor is the duty cycle (within reason). You also have a suitable supply line (so no need to consider 3 terminal regualtors, etc) Now, off the top of my head there are 3 obvious solutions (there are many more too, of course ) : A discrete component multivibrator -- 2 transsitors, 4 resistors, 2 capacitors, LED A 555 astable --555 Timer IC, 3 resistors, 1 capacitor (2 if you are a purist and want to decouple 'control voltage), LED An 8 pin microcotnroller, 1 resistor, LED Which of those 2 you consider to be the simplest? Actually, the 'friendly votlage levels' could be a minus-point in that it is likely a USB interface conenctor is connected directly to some large and complex IC, while an RS232 port will almost certainly invlove a simple buffer IC or two. So that if you accidentally connect too high a voltage to the cable, in the case of USB you have a lot of work to do, in the case of RS232 it's one cheap IC that will take a few minutes to replace (even if soldered in). And the USB protocol is significantly more complex than RS232 and that to me is a major disadvantage _where the extra capabiliities of USB are not needed__. As an interface to large-ish remveable memory devices it seems fine. As am interface to a fairly low-speed ADC system it seems ridiculous. -tony

On 6/18/2013 2:30 PM, Sean Conner wrote: A NE-2 bulb , resistor and cap is simpler and test-able* assuming you have the power supply free already. Ben. * Does hardware testing and programing time count as well.

So, put a LED in series with the NE-2. :-) (Hmm, been too long - is the NE-2 on-state current high enough to light a LED?) I was trying to design something analogous with a LED...maybe a zener and a SCR could be combined for similar effect? V+ ---- /\/\/\/ ----+---+------------+ | | - | | | | /| | | | | +---|< |--. --- +-----| |------+ | \| \ V | | | - ------- | ___ | | / \ | | +--- /\/\/\/ -----|-|<|-|---+ | | \___/ V- --+---+ Provided the two resistors combined keep the steady-state current below the SCR's dropout current, that ought to work. Add three transistors so that current through the LED shuts off V+ to the charging circuit and it should work even without that. Showing the transistors as boxes because of the limitations of text graphics, and using an NPN as the inverter: --------- V+ --+----+--------| E C |---- /\/\/\/ --+ | | | PNP | | | | | B | | | | --------- +-------+ | | | | | | | | +-- /\/\/\/ --+ _____| +-------| |---------+ | | | ^ | | | | | | /_\ | | | +--- /\/\/\/ --+ Zener | | | | | +----. --- SCR | | | \ V | | | ------- LED | | | | ___ | +---|-------- /\/\/\/ ----------+ | / \ | | | +---(-|>|-)-----+ | --------- | \___/ | | | C | | | | | NPN B |-----+-- /\/\/\/ --+ | | | E | | | | | --------- | | | | | | --------- | | +-- /\/\/\/ --+ | C | | | | | NPN | | | V- ------+------------------+---| E B |---+--- /\/\/\/ --+ | | --------- | | +---- /\/\/\/ ----+ | +------------------------------------+ The only way I can see that failing is if the amplification of the switching transistors is so low that SCR holding current isn't enough to end up switching off the PNP. By this point, though, you might as well use the two-Q two-C four-R astable circuit, possibly with another Q as a switch for the LED. The major advantage of this is that the charge accumulated in the charging capacitor is the power used to light the LED - though you need to use FETs instead of BJTs for the switching for that. But build the astable with high Rs, low Cs, and FETs instead of BJTs, and you have very low parasitic power draw there too. Oh well, it was an interesting gedanken circuit. :) /~\ The ASCII Mouse \ / Ribbon Campaign X Against HTML mouse at rodents-montreal.org / \ Email! 7D C8 61 52 5D E7 2D 39 4E F1 31 3E E8 B3 27 4B

On 6/19/2013 1:10 PM, Tony Duell wrote: Now that I think about don't they make flashing leds. Ben.

On 06/18/2013 11:16 AM, Tony Duell wrote: I count 3 choices. None of the above. A UJT, capacitor and resistor with LED would be simpler than your 2-transistor gizmo. A DIAC (or other suitable 4-layer diode) would be even slightly more simple. Perhaps a nitonol wire comprising a thermal switch might even be simpler. Now, let's talk about what's most flexible (i.e. bang for the buck). The 8 pin MCU can generate *precise* pulses with a *precise* duty cycle. Heck, you might even be able to get rid of the resistor if you employed some PWM on the output, such that average maximum ratings on the LED weren't exceeded. EDN (ISTR) had a column on this subject last year concerning replacing a 555 with 8 bit MCUs. It was a lot of "it depends", as I recall--and no clear winner. Interesting--I've seen more EIA 232 line receivers zapped by ESD than USB devices, but that could be because 232 has been around longer. But who lives in a world where a single person writes both sides (host and client) software for a given USB application? My point is that as a general purpose-common-as-dirt interface, USB is hard to beat these days, whether you're trying to push 50 bits or 50 megabits per second down the line. 20 years ago, the picture was not the same. --Chuck

On 06/18/2013 03:44 PM, Dave Wade wrote: Well, now I think you're just trying to get Tony's goat... :) But MCU vs. FPGA is an ongoing argument for many applications with no clear winner. I know that the 1488/1489 combo was popular, but the lead engineer on the comms project at the time that I was involved insisted that the 75150/75154 combination was much superior. I don't recall what his reasoning was, but we didn't run into any popped 7515x parts that I can remember. --Chuck

On 06/18/2013 07:03 PM, Fred Cisin wrote: Well, the 75150 datasheet says that it can withstand (indefinitely) applied short-circuit low-impedance voltages of +/- 25V. The 154, similarly is rated for input voltages of =/-25V. So, I guess the answer is probably "yes". --Chuck

On 19/06/2013 21:45, Tony Duell wrote: I saw an IBM Quietwriter destroyed by having RS-232 fronm an old XT stuffed in its centronics port. The printer port failed in an XT so some one put a gender changed in and plugged it into the serial port which on an XT is on a standard 25-pin socket.... Dave G4UGM

On Wed, 19 Jun 2013, Tony Duell wrote: thanks I wonder what else our lab staff was doing that kept blowing 1488/1489 chips?

On 06/19/2013 06:53 PM, Fred Cisin wrote: The most common is really bad static. I've also seen people pulling RS232 plugs with either piece live, if ground is the pin that breaks first then you may 24V at the wrong place. The other is safety ground that has potentials from one outlet to another. Most common reason for that is neutral and ground swapped or a peice of gear using ground instead of neutral for the return. I've seen more than a few outlets in NYC and other places where from the ground pin on one outlet to an outlet ground pin on a different breaker had 25V AC! I even had one that had 119Vac (ground and line swapped!). Be very careful what you call ground. Allison

On 06/19/2013 06:52 PM, Fred Cisin wrote: I acquired an industrial PC some time back with a fairly high-end ATX PSU in it (can't remember the brand, but it wasn't a consumer piece of junk). As I touched the back of the unit, I could feel a "tingle" in my fingers. A check with a DMM showed about 50VAC between the chassis and neutral. At the AC socket, the reading was 0V or very close to it. The AC cord checked out fine. I haven't opened the PSU yet to find out what the source of the mystery is, but I suspect that the wire going from the AC socket to chassis doesn't. What I'm feeling is probably from the AC mains filter in the PSU. So it does happen. --Chuck

On 06/22/2013 12:54 PM, Tony Duell wrote: Generally, I don't worry too much about 3-wire equipment. I am particularly careful with old 2-wire stuff, such as lamps with metal bodies. There's it's not capacitor leakage that'll do you in, but the accidental shorting of the body to the hot side of the line. It's happened to me with brand-new lamps right from the store. I don't understand why 2-wire lamp cord is still legal in these days of grounded outlets being almost universal. I guess I should pop open that PSU today and see what's going on. --Chuck

----- Original Message ----- From: "Tony Duell" <ard at p850ug1.demon.co.uk> To: <cctalk at classiccmp.org> Sent: Sunday, June 23, 2013 8:39 PM Subject: Re: USB (Re: Next completely OT topic:) And if you can run the workshop or work area through an RCD, maybe with a sepperate circuit for things that dont like the power droppingg out , like RK05's. You may also get the occasional niusance trip, like when you power up a large rack, I need to turn on the 2 psu's of my 11/50 one at a time, but this is a small price to pay for the extra safety. BUT this is no substitute for probber earthing. Dave

How do you end up with 6V between the earth pins of 2 sockets on the same bench? If I could measure any voltage at all, I would want to know the reason why. -tony

On Thu, 20 Jun 2013, Tony Duell wrote: To be fair, the worst offenders of "plug stuff anywhere that the plug fits" were "Professors of Computer Science" The Microsoft "Bus Mouse" (1st original green-eyes) had an interface card with a DE-9 female. There were some who refused to believe that it was not appropriate to plug it and the monitor (CGA/MDA/EGA) in randomly and swap around until it worked or could never again work. -- Grumpy Ol' Fred cisin at xenosoft.com

On 6/19/2013 2:21 PM, Tony Duell wrote: Moo.. My beef ... USB cables force you to have everything on the desktop. Sometimes you need to put someting over there rather than right by the computer.

That was my own reaction on seeing that, too - needing an entire CPU, albeit a simple one, is being cited as evidence of how simple it is?! Depending on what it's to be connected to, I could come pretty close to an async serial interface using discrete transistors. I'd need a lot of them, probably nearly a hundred by the time it's all done, but it's not out of reach. USB probably _is_ out of reach. Dunno 488, never done much with that. But 232, yeah. Indeed, crank the baud rate down to ca. 15 baud and a little practice lets you read off character values by eye. Even at higher rates I've sometimes been able to tell useful things by watching blinkenlights. /~\ The ASCII Mouse \ / Ribbon Campaign X Against HTML mouse at rodents-montreal.org / \ Email! 7D C8 61 52 5D E7 2D 39 4E F1 31 3E E8 B3 27 4B

:-) Actually, I think a maechancial RS232 interface is impossible too. Electromechanical, sure. You could switch a polar relay fro man RS32 signal and use the contacts of that to operate the receive magnet in your asynchronous receiver (e.g. ASR33 selector unit) But I doubt you could make a receiver magnet that could be driven directly from an RS232 line and which would have enough force to driectly operate a similar selector mcchanism. -tony

On Mon, Jun 17, 2013 at 07:41:48PM +0100, Tony Duell wrote: [...] So what? You're evidently building embedded systems with hard real time requirements. That is not the target market for USB.

On 6/15/2013 8:48 PM, Cory Smelosky wrote: USB has RS232 adapters, now you can have the best of both worlds. 2 GHZ download for your 110 baud TTY. Ben. PS. I have the DE1/DE0 FPGA boards here, So I hope I can get a nice hardware version of some classic equipment to play with rather than just Windows.

Err, no. That seems to be a USB slave device only. Surely a computer should be cpable of acting as the host, communciating with peripherlas (in this case USB peripehrals). Why would I want ot make a Spectrum a periphaeal of some other machines? And I'll bet there is a lot more logic on that interface board than is inside the Spectrum, which seems a little ridiculous... -tony

[tony] And? It is...in the right context. (Which doesn't occur much around you, of course.) Or am I being insufficiently inflammatory to Uphold this Tradition? /~\ The ASCII Mouse \ / Ribbon Campaign X Against HTML mouse at rodents-montreal.org / \ Email! 7D C8 61 52 5D E7 2D 39 4E F1 31 3E E8 B3 27 4B

I remember those days; we used to have the likes of Joe Rigdon (wherever you are), Megan Gentry (don't know if she still lurks here), and some others I can't remember anymore. We had a somewhat different cast of characters, and of course it was a different hobby back then as well (but I won't go there). This hobby gave me a peaceful, easy feeling in those days (to borrow a line from the Eagles). Folks were always eager to help, always willing to lend a hand (and maybe a tape image or two), and we'd have a good laugh while we were at it. We loved our machines, and we loved to expend the effort just to make them 'go' and talk about how we did it. That's what our hobby (and this mailing list) was really all about. I'm not saying that the above isn't true anymore: I'm simply saying that the mood and character of our hobby has changed over the years, and that there are folks who used to have alot of interesting stuff to say, are now silent. Flame suit on. N0body ____________________________________________________________ FREE 3D EARTH SCREENSAVER - Watch the Earth right on your desktop! Check it out at http://www.inbox.com/earth

Thank you, sir. You ventured where I dared not. Playing with the hardware/software was the joyous part; our machines begged to be tinkered with. It is my fear, that with mostly PC-type hardware available to beginners (usually kids w/ little or no money) these days, they will miss out on the joy that was 'hacking'. In fact, it may spur little interest in computer history at all. Perhaps I'm wrong; but from here, it's not looking so good.... N0body ____________________________________________________________ FREE 3D EARTH SCREENSAVER - Watch the Earth right on your desktop! Check it out at http://www.inbox.com/earth

You should then be pleased to know that I added a parallel port to my ETH Lilith, which might very well be the only functional one in existance. Also an IDE interface is in the works. I'll keep on hacking this machine, despite its rarity. Jos