• Please review our updated Terms and Rules here

ISA USB adapter

N

NobodyIsHere

Veteran Member
Joined
Dec 21, 2006
Messages
2,410
Hi! Would there be any interest in an 8 bit ISA USB adapter? I saw a recent thread on CCTALK mailing list on the subject and wondered if there were sufficient interest to do VCF group project. Working with Chuck and Hargle, there have been a couple recent VCF homebrew projects that turned out well (XT-IDE, AT2XTKBD)

Here is the thread on CCTALK:

http://classiccmp.org/pipermail/cctalk/2009-December/279840.html

Of particular interest is this comment

http://classiccmp.org/pipermail/cctalk/2009-December/279929.html

The SL811HS is available in 28 pin PLCC which would aid the DIP PTH approach. However, it may be worthwhile to try an SMT board and see how that goes since the parts appear more readily available in 48 TQFP. Not that I am thrilled with an SMT part but I wouldn't want that to be the only reason to squelch and otherwise good idea.

Thanks and have a nice day!

Andrew Lynch
 
I am under the impression that the software for a USB device is much more difficult than what we've done with XTIDE. What kind of drivers would we have to start with?
 
Doesn't a USB controller depend on the CPU to function?

Its an interesting proposition to add USB to an 8 bit machine, but what do you envision it being used for? Are we talking mostly HID devices? or Mass storage?
 
Drivers for the various devices might be a problem on 8-bit platforms, unless you wanted to restrict yourself to generic classes (e.g. mouse, disk, etc.).

I know it's surface mount, but you may want to look at the FTDI Vinculum host controller. Because it has a 32-bit RISC core, it has a lot of generic device support built-in. This is as opposed to, say, Windows support with its layered (and large) driver structure.

Strictly speaking, there's nothing magic about a UHCI; you can even make an AVR uC service as a host controller (although not with full-speed functionality, for obvious reasons).
 
It would be great but indeed I always expected this to be a total different difficulty level (not that I have any useful knowledge about it..)
Using USB sticks as mass storage and bypassing the 360K floppy problem when transferring files would be great.
When WLAN USB sticks would work... that would also be nice....
 
Could we not build it with its own bios that has built in support for HID devices and mass storage devices, but also with a pass through mode (for custom drivers and other devices)?

It could even be built to translate USB mass storage devices to either floppy or Hard drives.

Is this conceivable?
 
Hi! I don't know the specifics of the software effort needed to implement USB on an 8 bit ISA bus nor how useful it would be. My intent is to gather some information and see if there is any interest. The datasheet for the SL811HS actually mentions the chip can provide USB for

"The SL811HS is an Embedded USB Host/Slave Controller capable of communicate with either full-speed or low-speed USB
peripherals. The SL811HS can interface to devices such as microprocessors, microcontrollers, DSPs, or directly to a variety of
buses such as ISA, PCMCIA, and others. The SL811HS USB Host Controller conforms to USB Specification 1.1."

http://www.datasheetarchive.com/pdf-datasheets/Datasheets-316/606532.pdf

Since it mentions ISA specifically, I am thinking it must be possible and may be practical to do this. It may be there is software already available that could be adapted. I just don't know enough about the subject and thought it would be an interesting group discussion and potential project similar to XT-IDE, AT2XTKBD, etc. I'd be willing to participate in any group project if one comes about.

Thanks and have a nice day!

Andrew Lynch

PS, at least notionally, this project would be just a "scoop out the guts" of the XT-IDE board and use this chip instead. Good re-use of the XT-IDE design and get some more mileage out of the considerable investment the VCF community has placed in it already.
 
If you wanted to support the entire range of USB devices, including, for example, hubs, on an 8-bit real-mode platform, it could be a monster of a software project (and one I wouldn't want to do). I remember how long it took Linux to get acceptable USB support--and how buggy it initially was. And Linux still doesn't have support for many devices (e.g. my USB DSL modem).

Still, if one was determined, one could probably borrow a lot from Linux, but I suspect that the memory overhead would be more than most people would like.

I suggested the Vinculum because it comes with built-in-support for a number of devices and a simple way to update the firmware. While not exhaustive, it would support generic HID, printer, flash, music and even a slave-to-PC hookup. Protocol is simple ASCII strings.
 
For what it is worth, the RetroClinic DataCentre for the Acorn BBC Micro has some USB functionality. The range of devices useable on the Beeb depends on drivers and application software. It probably has nothing to do with an (8-bit?) ISA interface, except as an example of what has already been done and how it is used.
 
carlsson said:
For what it is worth, the RetroClinic DataCentre for the Acorn BBC Micro has some USB functionality. The range of devices useable on the Beeb depends on drivers and application software. It probably has nothing to do with an (8-bit?) ISA interface, except as an example of what has already been done and how it is used.
Click to expand...

The DataCentre also uses the FTDI Vinculum, and though as Chuck(G) mentioned it is a surfice mount chip, FTDI do do a prototyping version that is mounted on a DIP carrier http://www.vinculum.com/prd_vdip1.html I have a couple of these and one day will get round to plating with them.....

Phill.
 
How about CPU load and bus bandwidth?

How about CPU load and bus bandwidth?

Just wondering... How CPU-intensive would it be to drive USB peripherals over the ISA bus, particularly on vintage systems? And wouldn't USB devices (especially those that can use 10mbit/s bandwidth) easily saturate the available ISA bus bandwidth? I guess it would be okay for HID devices (keyboards, mice, game pads, etc), but for anything else???

Don't take me wrong, I still think it's awesome to retrofit USB interfaces on vintage machines, but I'm just wondering... :)
 
tantilizeing idea!

tantilizeing idea!

....and one worth of development.

As digger has pointed out, USB is very demanding (from what i understand, anyhow) on the host controller. Running another processor to handle to overhead, tho i understnad its possible,
has got to be a nightmare......But it could work, and it would be one I'd glady support as much as possible.

Linux support (as chuck has mentioned) was extremely buggy to start off with...but now (excepting a few oddball devices, like the usb dsl modem) is now very stable and would prove to be a good starting point.

What i would like to know is how one could impliment more then say, half a dozen devices
without running out of address space. Say, for example, one has a mouse, a keyboard, a usb flash drive, and, just because, a serial to usb converter. Could the usb board driver re-direct all
access to/from these devices through a single I/O address and IRQ? I know, on modern machines, IRQ's are all but forgotten and the host controller only needs one, if that to talk to alot of things attached to it.

I dont know how it could all work out....but count me in!
 
This is one of the reasons that I suggested a Vinculum. All the PC has to do is toss and catch the data--if you buffer it, the bus speed is irrelevant to USB operation. If you have a very slow bus, the time between transactions increases, but USB protocol proceeds at the rated speed.
 
I'd love to have an ISA card that I can attach an optical mouse to :)

I have absolutely no experience with creating drivers though
 
Since i am really clueless i can afford to throw clueless fantasies in the air, but ever since i heard about that the pc piezo pc speakers could generate sounds directly from digital signals from motherboard?

I've been dreaming a bit of a usb AD/DA without quantisation basicly using none quantisised audio streaming to and from HD. I guess it would be a 1-bit AD to record your old vinyl, and then be able to use the an XT to stream it back over a 1-bit DA. So basicly depending on port speed and CPU speed you could chose any sample rate you wanted in 1-bit format over this unit.

Of course no compression modes just plain bitstreaming from your 500 GB hooked up with the XT control card streaming over your XT USB card. Freedos support FAT32? is 500 GB partitions possible?

Superb high definition audio delivered at 4 or 8 Mhz to your HIFI system.

I have to read up a little on Nykvists sampling theorem before i start ;D but i did actually wrote a program that could stretch pitched samples so the donald duck effect dissapeared, and you could use one over all octaves.

I think i did not care about quantization i just averaged the pitched resample by fill in the blanks with nearby bitcopies using statistics to match the length of the original sample, strangely enough it worked.

It was a bit of a bitwig i guess :D

JT
 
Oooops at 4 mhz samplerate that is only 34 our of music on 500 GB?
Ok let put the sample rate at 1 mhz, that would be 136 hour music, but pristine music ;D

Maybe the ISA ide conctrolcard support 1 TB?
272 hours of music at 1 mhz samplerate, well if you have medium sized collection.........

JT
 
http://www.cs.tut.fi/sgn/arg/rosti/1-bit/

Now when storage at 1 TB is around 100 euro i think people interested in HIFI would like their recordings in prestine digital format.
Would be nice with an open source format supporting free samplerates at 1-bit.

Been reading about flash ADC, and wilkinson ADC, succesive approximation ADC i admit i do not understand it at all is this examples of ciruits design or available ADC chips, i guess it is design?

I guess you can buy an 1-bit AD circuit of sigma delta design, is there much analog conversion of the line signal todo before you feed it to the AD circuit?

And then buffering and synching with the ISA buss?

I guess what i really wonder is what level of skill it would take to implement a sigma delta AD converter on an ISA card, just so you could feed the signal over the buss to the HD?

This is what i dream about doing, although i have no real idea about what a sigma delta AD really is, but i understand that 1-bit sampling is not that computation exhaustive as quantized sampling since there there is no dithering.

Many here seem to be more or less techno gurus an open source project for none quantized audio (1-bit?) with free sample rates would be nice.

JT
 
Many modern CD players use 1-bit DAC of SDM type. The SACD format i beleive was built around the SDM, but what is needed is an opensource format with free samplerate. I think the 8-bit ISA buss is fast enough to give a very descent sound, can USB 1.1 run at the speed of the original ISA buss would love to see an 8-bit ISA card with a 1-bit SDM ADC/DAC streaming to and from disc.

Or maybe the idea build a USB interface for the 8-bit ISA buss be would able to support a modern USB soundcard playing PCM WAV or is there more then just streaming that the processor must handle, computation overload?

JT
 
Chuck(G) said:
This is one of the reasons that I suggested a Vinculum. All the PC has to do is toss and catch the data--if you buffer it, the bus speed is irrelevant to USB operation. If you have a very slow bus, the time between transactions increases, but USB protocol proceeds at the rated speed.
Click to expand...

Hi Chuck! Yes, the Vinculum would work fine for those who can assemble SMT devices. The chip is only available in LQFP-48 so that makes amateur assembly problematic. I am sure a lot of us could assemble it but for broad acceptance that would be an issue.

There are PICs and AVRs in 40 pin DIP that can also handle USB. We'd need about 12 spare IO pins (D0-7, A0, CS*, IORD*, IOWR*) give or take a few plus uC software to interface to the ISA bus.

The board would be pretty simple as it would be a modification of the XT-IDE but the software particularly the uC side would be difficult.

Another alternative would be to use one of those Vinculum DIP adapters like these http://www.vinculum.com/prd_vdip1.html

I have no idea of the cost but am confident it will make most people choke. I think that's where the 40 pin DIP AVR/PICs have an advantage.

My plate is pretty full right now with work and am also backed up on home brew projects. I have one N8VEM ECB board open and two S-100 boards in the works at the moment. This project will have to wait until someone picks it up or things free up elsewhere.

Thanks and have a nice day!

Andrew Lynch
 
You must log in or register to reply here.
Back
Top