On Apr 22, 2012, at 10:47 AM, Alexandre Souza - Listas wrote: Someone pointed this out a few months ago: http://a4000t.com/store/index.php?main_page=product_info&cPath=66_73&am… I believe someone else mentioned they'd been trying it out (might have been Glen Slick?) and that there were varying degrees of success with anything but the CF card part on most classic OSes, since each port exists as a separate LUN on the same SCSI ID. I could be spuriously recalling a conversation that didn't happen, though. I'm still interested in trying one out, but I'd be interested in making a multi-format flash memory device SCSI reader specifically for booting older machines. At some point, when I acquire some more round tuits. - Dave

On Sun, Apr 22, 2012 at 8:25 AM, David Riley <fraveydank at gmail.com> wrote: On those devices the PCCard slot is LUN 0 and the CF slot is LUN 1. I didn't have any luck using the device with the built in SCSI controllers on either a VaxStation 3100 M76 or an InfoServer 1000, regardless of which LUN I tried to use. The device LUNs would show up in >>>SHOW DEVICES but errors would occur when trying to use them. The device seemed to work fine with the limited testing I did with a CMD CQD-220. The configuration menu for the CMD CQD-220 lets you specify both the SCSI target ID and the target LUN. I was able to use a CF card in the LUN 0 PCCard slot with a CF-PCCard adapter and in the native LUN 1 CF slot. Booted both RT-11 and XXDP from the device with the CQD-220. -Glen

On Apr 22, 2012 3:31 PM, "Alexandre Souza - Listas" <pu1bzz.listas at gmail.com> wrote: You can change the SCSI target ID with jumpers but not the LUN to card slot mapping. I have no information about any methods to tweak the firmware in this device.

? On Sun, Apr 22, 2012 at 5:06 PM, ben <bfranchuk at jetnet.ab.ca> wrote:

On 4/22/2012 2:31 PM, Glen Slick wrote: I picked up one of these as well, and it seems to work pretty well with systems that don't care about the distinction between "fixed" and "removable" media -- the cards in the slots are reported as being removable which VMS doesn't seem to like much for using as a system drive (though NetBSD seems to be perfectly happy). One other caveat -- if you're using CF cards, make sure your they support the CF interface, and not just IDE; I have 3 Hitachi 4GB microdrives (tiny hard drives in a CF form-factor) and it turns out this particular model only has an IDE interface. The CF slot in these devices don't support the IDE interface and so they will not be recognized (which is a bummer...) Josh

There's a major DISADVANTAGE of a lot of solid-state media, including CF cards and SD cards. Because the erase block size inside the card is large, they have to do blocking and deblocking to that size. When you write a sector, the erase block containing that sector is read into a buffer in the controller. (The controller is part of the memory card, not part of the computer; the computer has a host adapter.) The sector data gets buffered, but not immediately written to the flash, because you don't want to wear it out with unnecessary erase cycles. When you write more sectors in the vicinity, they go into the buffer. Eventually the block does get erased and the data rewritten. Now suppose that you're running a multitasking OS with n files open for write. If you nave at least n of these buffers in the controller, things work smoothly. If the controller has fewer than n, it has to do the erase/write cycles more often. If n is much smaller than the number of files you are writing, the performance becomes very bad, and the wear to the flash goes up substantially. For CF and SD cards, the number of buffers in the controller is very small. Some have as few as two; the most I've seen is six. These cards are not meant for replacement of disks that have a lot of parallel activity. That's a serious concern when using them to replace SCSI or IDE drives. The intended usage mode for these cards is with MP3 players, which don't write much at all, cameras that generally only write one file at a time, and thus need only two buffers (data and FAT), and for manually transferring files, which also is usually only done one file at a time. At one point I saw a web page where various memory cards were benchmarked, and the number and size of the controllers' internal buffers were deduced, but I didn't save a link to it and am too lazy to find it again right now. SSDs with an IDE or SATA interface are intended for use as general-purpose disks, so (some? many? most?) of them don't have this problem, although there is a lot of variation between vendors. Off-brand SSDs, and even some name-brand ones, have been known to have this and other problems, but the name-brand vendors generally have gotten their act together. Eric

On 23 Apr 2012 at 22:11, Liam Proven wrote: Well, there's a third option--Microdrives. There was (and may still be) a guy on eBay selling 4GB Seagate Microdrives NOS for less than $5 shipped. I use one on my mail server and it's been running for 4+ years 24/7. I used it mostly to keep heat and power down vs. a normal 3.5" IDE drive. No write fatigue, but like any mechanical drive, it will eventually wear out from just sitting there and spinning. But so far, so good. --Chuck

On 24 apr. 2012, Jules Richardson wrote: Loved those things back then. From a distance though, considering the prices they went for when they were hot :) re, Sander

On 24 Apr 2012 at 9:00, Jules Richardson wrote: Reminds me of the Datasonix Pereos that attempted to get 1GB on a Sony NT microcassette. I wonder if any backups have survived? https://en.wikipedia.org/wiki/NT_%28cassette%29 http://homepage3.nifty.com/kenk/imx/ndc-04.jpg --Chuck

W O W!!! This is something I never heard of :oO

On 24 April 2012 15:58, Chuck Guzis <cclist at sydex.com> wrote: Oh my yes, those things! I remember those. I think we reviewed one in PC Pro. My colleague Dick Pountain - an experienced "elder statesman" of IT - thought it sounded great and bought one. I do not believe he stayed delighted for all that long, but at first, he thought it was great. He could carry around a gig in his pocket, which in 1995 or so was pretty impressive - a big modern hard disk then was around a gig. The best one, which most of our review machines had, was the Quantum Fireball 1200. -- Liam Proven ? Profile: http://lproven.livejournal.com/profile Email: lproven at cix.co.uk ? GMail/G+/Twitter/Flickr/Facebook: lproven MSN: lproven at hotmail.com ? Skype/AIM/Yahoo/LinkedIn: liamproven Tel: +44 20-8685-0498 ? Cell: +44 7939-087884

When I first read that, I thoguh you were talking about an endless loop tape drive that Sinclar used on the Spectrum and QL machines.. I wouldn't recomend those for anything... -tony

On Tue, 24 Apr 2012, Chuck Guzis wrote: Depending on the file structure, you could RECORD infinite amounts in that continuous loop. But you had no chance of reading back anything that wasn't on that last pass through the loop. And with many, such as the Exatron Stringy Floppy, you had little chance of reading ANYTHING back, even if you stayed within their file structure.

On Tue, 24 Apr 2012, Chuck Guzis wrote: The Vogons did.

Chuck Guzis wrote: Probably due to the inconvenience of needing 6.3Vac for the Vff supply.

Why does Vff have to be AC? Surely it's the rms vlaue that really matters, so 6.3V DC would be OK too [1]. And IIRC the unregulated output of the mains adapter was on the expanison connector of the Speccy, so you could use that with a dropping resistor. [1] Vff = 6.3V nearly always implies an indrectly-heated cathode, so the polarity and whether it's AC or DC make very little difference. -tony

Of course. I have a reprint of it here. I also know what Vff stands for [1], and relaise the significance of 6.3V RMS. Given that, I don't see why DC souldn't be used... [1] I have used some serious device that quote Vff as a parameter. I've also used some where having an AC supply is important [2]. I don't think that would apply to a WOM though. [3] VF displays. If the filament is heated from DC, oen end of the display is brighter than the other. -tony

On 4/27/2012 2:58 PM, Chuck Guzis wrote: But then you have to reverse the blades of the 6' fan to keep the dust bunnies* blowing on to the chips. Ben. The bunny bucket chips is the B version. Still in development as bit or two randomly *HOP* out every year or so. Lose bits are a serious problem with 10^13 bits traveling every second or so per computer.

On 28 Apr 2012 at 19:34, Tony Duell wrote: Well, note that there's a requirement for 115VAC anyway. I guess the perfect way is to couple a small alternator to the 6 foot fan motor needed for cooling. That way, if the fan fails, the filament supply will automatically be discontinued, allowing for graceful degradation. --Chuck

On 29 Apr 2012 at 20:42, Tony Duell wrote: Take a look at the block diagram near the "Chip Destruct/" pin. Note that the 115vac is connected to the output of the inverter that feeds into the 25102. --Chuck

Yes, but AFAIK it doesn't specify if this supply has to be floating, centre-tapped to ground, or whatever. That implies to me an isolated filament heating a separate cathode. So I can't see why AC is essential. -tony

On 2012 Apr 27, at 11:39 AM, Tony Duell wrote: The datasheet shows only a single connection for Vff, not a floating connection. AC is required by the WOM for two reasons: - Vff is used in a critical area of the chip. DC used here would create a unidirectional field-differential which promotes asymmetric migration of ions or impurities in the cystalline structure of the chip. This critical region holds a few atoms in a particular quantum state where - in accordance with the principles of quantum computation - they are able to sink bits with a probability of 1, thus allowing an infinite number of bits to be written to the device (like a digital black hole). - The AC is also used as a clock to periodically empty the WOM bit- bucket caches. This is the WOM equivalent to RAM refresh; without the clock, the bit-buckets would fill up. The bit-bucket caches are a performance improvement, they sit between the write interface and the quantum sink, every now and then the bit-buckets are emptied into the quantum sink. The quantum sink was top-secret military technology when developed over 30 years ago. The military didn't really want the WOM device released but Signetics wanted to try marketing it, so a compromise was reached: the drain shown on the chip in the datasheet. This was actually a misdirection, so people wouldn't ask how the device was able to sink an infinite number of bits. The military need not have feared, there wasn't much of a commercial market at the time. People thought the WOM was a joke, or at best a pointless device. Recently however, WOMs have found a significant application in digital audio systems. New design and research has resulted in digital reflection-suppressor (DRS) technology (patent applied for) - an alternative to the green-pen solution for suppressing unwanted reflections in a CD. With DRS technology, the reflections (photons) are allowed to bounce around in the CD, eventually being detected and making it out into the (electronic) bit stream intermixed with the desired audio bits. Sophisticated proprietary digital filter algorithms identify the unwanted reflections in the bit stream, separate them from the true audio bits, and send them to the WOM where they disappear. Obviously, if Vff is not AC the WOM will eventually not be able to sink the reflections, leading to degraded audio quality. NDAs prohibit me from disclosing full details of the quantum sink, but quantum computation is now public enough that this much can be disclosed without concern of legal consequences. Finally, you can benefit at home from this taxpayer-funded military technology - it's about time. Standalone DRS products will be available soon, or contact to discuss incorporating DRS technology into your production audio systems. We're currently working on new filter algorithms to send noise bits into the WOM. If you thought digital audio was quiet before - just wait.

On 2012 Apr 28, at 11:57 AM, Tony Duell wrote: A not unusual I/E curve. It's not incompatible. On 2012 Apr 28, at 11:34 AM, Tony Duell wrote: Which I pointed out, and negates your presumptions of an isolated filament and DC compatibility. The datasheet specifies AC. (Why do I bother)

On Sun, 22 Apr 2012 23:54:43 -0700 Eric Smith <eric at brouhaha.com> wrote: Modern NAND-Flash chips have a usual erase block size of 512 bytes. This fits the 512 byte blocks of disks. So no (de)blocking. The write buffering is mainly there to mitigate the effects of several consecutive writs to a single block. (E.g. directory structures.) There is an other problem: MLC. Multi Level Cells. In the old days a Flash cell stored a single bit. There was a single comparator with a single threshold to decide 0 or 1. Current SD cards are MLC. They compare the voltage in a single flash cell to e.g. three thresholds and thus can store two bits per cell. Newer devices can store three bits per cell. This means the thresholds get closer to each other. Unfortunately reading the contents of a cell draws a few electrons of its charge. So the voltage of the cell drops a tiny bit with every read. Given the narrow voltage windows of a triple bit cell its only a question of time when read cycles drop the voltage below the next threshold. - Reading becomes destructive! I addition these MLC devices have a very small count of write cycles. Down to 1000 writes per erase block. Thats OK for a MP3 player or digicam, but in no way this is a reliable storage medium. SD cards are cheap consumer junk. Yes, at work we have been bitten by this problem. Servely. :-( -- \end{Jochen} \ref{http://www.unixag-kl.fh-kl.de/~jkunz/}

Jochen Kunz wrote: Further to my previous comments about this, a quick survey of recent NAND flash parts does not seem to support your statement at all. It appears that for "modern" parts, even the small-block devices now typically have an erase block size of 16KB or more. The "page size" can be as small as 512 bytes, but you can't erase those independently. Back in 2005 Micron published a technical note on small-block vs. large-block NAND that discusses some of the technical tradeoffs, though it focuses more on write performance than cost: http://www.micron.com/~/media/Documents/Products/Technical%20Note/NAND%20Fl… While the device densities have continued exponential growth in the seven years since that was published, as far as I can tell the engineering tradeoffs are still quite similar. There are NAND flash chips (or System-in-Package devices) that contain both the controller and the NAND flash; these have an external interface that uses 512-byte sectors, but they still have to deal with larger erase blocks internally, just like packaged flash cards and USB flash drives do. Eric

On Wed, 25 Apr 2012 01:52:01 -0700 Eric Smith <eric at brouhaha.com> wrote: "draws a few electrons" is simplified. Its about what a colleague at work explained to me. He cares about the SD cards we build into our products. We where bitten by this problem when we evaluated a new type of high capacity MicroSD card. He can't be more speciffic due to NDAs, probably the same you signed too... Still: (Most) SD cards are _not_ a reliable, long term storage medium. IIRC that colleague talked about special "industrial grade" SD cards. They are guaranteed to be robust and reliable as old fashined NOR Flash. But they are expensive and have less capacity. -- \end{Jochen} \ref{http://www.unixag-kl.fh-kl.de/~jkunz/}

Um, Dreamcast VMUs are more than just SD cards; they've got an 8-bit CPU and a small 1bpp display in there too. And the SH's serial port is brought out to a connector on the back panel.... Unless you mean something other than the Sega gaming console? /~\ 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

It's not clear from what I've been able to find that the adapters between the serial port and SD cards are passive (as opposed to having active circuitry in them); I haven't found any description of what's in them beyond "it Just Works with this software". Do you have reason to think the adapter is a passive device? /~\ 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

Sounds like a pretty convincing reason. :) Ah! Excellent. Thank you very much; I think that page answers my remaining questions here. /~\ 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

AFAIK, this serial port is bitbanged...but I don't have that much info. If you wish, I have the schematics of the mainboard :)

Here you have :o) http://www.4shared.com/dir/13490883/b476a0ff/DREAMCAST_VA0.html

On Apr 24, 2012, at 12:03 PM, Mouse wrote: The short answer is that the easy way to interface with SD/MMC cards is as an SPI device. This schematic uses Tx as MOSI, Rx as MISO, RTS as #CS and CTS as SCK. The card most certainly doesn't drive the clock; it behaves as an SPI slave. - Dave

On Sun, 22 Apr 2012 11:59:53 -0500 Jules Richardson <jules.richardson99 at gmail.com> wrote: I had a similar idea a while back: A thing that emulates a SCSI disk on one side and a iSCSI initiator on the other side. That way I can connect it to a iSCSI target on a modernish PeeCee and have the "SCSI disk" of the vintage machine as an image file on the PeeCee. Or skip the iSCSI stuff alltogether: Reprogramm a NCR / Symbios 53Csomething to act as a SCSI target, emulating a SCSI disk that is backed by an image file on the host machine. -- \end{Jochen} \ref{http://www.unixag-kl.fh-kl.de/~jkunz/}

On Sun, 22 Apr 2012 17:48:37 -0400 (EDT) Mouse <mouse at rodents-montreal.org> wrote: Yes. Thats a typical problem. The 53C??? might be a good candidate. AFAIK it contains a RISCy processor and is quite far programmable. Once I got a AlphaServer with PCI DSSI cards. The DSSI was implemented with 53C525... It would be especially interresting if the SCSI chip of the Sun ELC is capable of implementing a target. The ELC board isn't that large, fits in an external SCSI enclosure, can be driven with a 5 V / 2 A PSU, is a very nice piece of hardware, boots NetBSD nicely over Ethernet and would be able to do that iSCSI trickery. I have several ELC boards around from machines with dead CRTs. In the days of NetBSD 1.5 or 1.6 I built a CDR copy applience out of one... -- \end{Jochen} \ref{http://www.unixag-kl.fh-kl.de/~jkunz/}

That I would have liked to have seen. :) I've got something like a dozen of them, extracted from SLC/ELC machines - the experience behind the "headless SLC/ELC" page I wrote for sunhelp.org. I want to, someday, do something with them.... Target-mode SCSI would be a cool thing. /~\ 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 Sun, 22 Apr 2012, David Riley wrote: Indeed, they can. I once used some dual-53C96 TURBOchannel boards to form a network between a VAXstation 4000/60 and a VAXstation 4000/96. Did it to investigate what it took to write a network device driver for VMS. Had DECnet and TCP/IP running over it, but the plug got pulled before I could look at forming a cluster. The point of using the dual-53C96 board was so that one chip on each board could act as a target while the other acted as an initiator. -- roger ivie rivie at ridgenet.net

Alexandre, You sent me an email on November 10 saying "Hey, nice to see you are on this list too" in Portuguese, remember? :-) If there are any more of us here, I would like to ask if anybody has technical details about machines like the Cobra 530 minicomputers. I have learned much about all kinds of old machines, specially through the documentation available in BitSavers. But the only local stuff I have found was a non technical thesis about the "Patinho Feio" which cites some technical texts about that and the G-10 machines. All other information I have is from historical books and newspaper articles. While it is interesting to read that the Cobra 530 was more advanced than the other 16 bit minis of the time, I would rather judge for myself. After all, they say more or less the same thing about Cobra's SOX (Unix clone) and my experience with it was that it wasn't very good at all. -- Jecel p.s.: a week ago a new science museum was opened in my town (S?o Carlos, SP). Though it isn't that large, there is space that could be used to have a computer museum and there are several people interested in making that happen.

On Sun, Apr 22, 2012 at 5:52 PM, Alexandre Souza - Listas <pu1bzz.listas at gmail.com> wrote: Perhaps you can use one of these: http://aminet.net/docs/hard/hddmem.readme http://de3.aminet.net/docs/hard/hddmem_exp_a50.readme or this: http://www.students.tut.fi/~leinone3/ide/ide68k.html HTH -- Regards, Torfinn Ingolfsen

The problem is to buy internationally. Brazil is passing thru a moment that mostly everything that comes by mail is taxed And I can build a board at home :) --- Enviado do meu Motorola PT550 Meu site: http://www.tabalabs.com.br ----- Original Message ----- From: "Joachim Thiemann" <joachim.thiemann at gmail.com> To: "General Discussion: On-Topic and Off-Topic Posts" <cctalk at classiccmp.org> Cc: "General Discussion: On-Topic Posts Only" <cctech at classiccmp.org> Sent: Tuesday, April 24, 2012 5:51 AM Subject: Re: Amiga IDE Non-autoboot interface

On 22 Apr 2012 at 12:01, David Riley wrote: I meant that for aging hands and eyes, they're physically too small to manipulate easily--and far too easy to drop on a carpet and be gone forever--until discovered by a vacuum cleaner--too late. --Chuck

On 04/22/2012 03:22 PM, Fred Cisin wrote: Huh? Not at all. It is the mainstay of pro-level digital camera work. It has just disappeared from consumer-shit-grade stuff because it is "huge". Heh. (still aching from moving an RK07 last night, and that's on the "light" side of my stuff!) -Dave -- Dave McGuire, AK4HZ New Kensington, PA

Last time I moved an RK07, I took it apart. The thing is very modular, you cvna take off the covers, disk pack holder, filter, blower, PSU assembly and cardcage very quickly. THen take off the mains input bracket (it hides one of he screws, don't ask how I found that out) and 3 more screw then retain the main chassis. That is fairly heavy but I carried it with no problems. Did the same with an RA60.... -tony

[RK07] Ah, mine had been taken off the stand (I did get the stand too, it;s now back on it). And, yes, there were steps, tight corners, etc to negotiate. It was a lot less hassle ot take it to bits (it took only a few minutes) than to have to manhandle it. And I learnt how the drive went together which was a help later on. -tony

On Apr 22, 2012, at 3:32 PM, Dave McGuire wrote: It's also very heavily used in the embedded computing industry; lots of PC-104 (or PCI/PCIe versions of same), Mini-ITX, etc. small form- factor motherboards available with CompactFlash slots for their main hard drive. That whole thing where it's a legitimate ATA disk right out of the box is its main advantage there. It's being slowly replaced by CFast, which as I understand it has the advantage of running as a SATA drive, but beyond that, I'm not really sure what the difference is. I haven't really looked, but I don't think CompactFlash is going anywhere anytime soon. - Dave

In CF you get transfer speeds waaayyyy higher than with a SD card. So it will be the mainstay for digital cameras (which do not evolve to XQD) for some time Interesting is the price http://www.google.com/products/catalog?q=xqd+card&oe=utf-8&client=f… - $170 for the 128 GB SD gem http://www.google.com/products/catalog?q=xqd+card&oe=utf-8&client=f… - $695 for the 128GB CF...