Well I doubt you will find that. Very few people could do it and the market being negligible, chances are it will not be around in the foreseeable future.
But it should be doable.
There are MCUs with enough RAM (>640k), which are fast enough to emulate the 386 to an equivalent of well above 100 MHz (this should be OK?); the monitor output will be harder to do, the floppy disk is as challenging. I only vaguely remember those MCUs with that much RAM so I don't know whether they have the interfaces which can be tweaked to do monitor and floppy, they were some Freescale DSP based things, easy to look up. If you would settle for a CPU+RAM+flash (3 or 4 chips), I know it can be done :-).
See also my earlier post. There are no x86 On-Chip-Flash microcontrollers - probably the eZ80 is closest, and Zilog do have an OS for that. Once you have an OS, you tend to need more flash, which is why these modules exist. So, to answer your question above:
Q:" What can be done with them, that cannot be done with a single chip microcontroller? " A: You can fit more code/Data into them, and in some cases, that's a vital feature. A Micro that is suddenly too small, is no use at all :)
So, which to use is decided by your code and data set sizes.
Example: The Zilog ZDots (they call it Single Board computer) has
256KF+16KR on chip, plus 1MByte of FLASH, 512K of SRAM, and Ethernet, at under 0.5W
Our customers often don't have the time or resources to purchase, install and learn a toolchain for a microcontroller.
They can develop their application on their PC in Borland C or Quickbasic and upload the 16-bit executable to our board and run it.
Since our boards have a read/write DOS filesystem and a serial console, our users can log data, upload revised software, and run various utilities. Booting time has been optimised and is typically .5 to
1.5 seconds.
No. There are no single-chip microcontrollers that I know of that can do that.
Well, the Wikipedia definition of microcontroller states that it *usually* integrates read-only or read-write memory. No x86 chip that I'm aware of does that.
O.K. Take one of the first old PCs with a floppydisk-drive, a keyboard and a Hercules Monitor and look at it as a black box. Take an identical PC and remove all the interiority but leave the floppydisk-drive and the connectors to the monitor and the keyboard. Take a singlechip-microcontroller. Connect every pin of the floppydisk-drive, the keyboard and the monitor to a pin of the microcontroller. Write a program for the microcontroller, which makes the second blackbox identical with the first blackbox. Now you can put a (MS-)DOS installation diskette in both and both will install it. Now you can put another diskette with an original MS-DOS program from the old times and start it. Both blackboxes will execute the programm the same way. The microcontoller has become a (MS-)DOS PC.
There is only one thing left. The pins of the microcontroller will not be able to drive i.e. the monitor-pins directly, because the pins of the microcontroller only drive a few milliamperes. Thus it is allowed, to adapt the (5V/3,3V or similar) pins of the microcontroller to the voltage and the current of the floppydisk-drive, the monitor and the keyboard by accordant transistors. But it is not allowed to give up the one-pin-peripherie to one-pin-microcontroller concept. The "brainpower" for driving/receive every pin of the periphery has to come from the microontroller and not from any external chip. I hope, the job definition is clear now.
And the question is: Are nowadays singlechip microcontrollers powerful enough, to manage this problem.
Are you talking about an MCU (say an AVR32 UC3A, or ARM7, etc) emulating an x86? Or using an actual x86?
This is a completely different question. As PC's started off using 8088s (or 8086s, or whatever), then the answer to this question is definitely "yes".
--
Regards,
Richard.
+ http://www.FreeRTOS.org
13 official architecture ports, 1000 downloads per week.
+ http://www.SafeRTOS.com
Certified by TÜV as meeting the requirements for safety related systems.
That is an answer just the way I expected. The 100 MHz is a helpful graduation. The question is simply theoretic. As I wrote in my OP this would be a useless application. I do not expect a real product. But I could imagine, that it could be a challenging project, to say: Yes, it is true. The few squaremillimeters of nowadays microcontrollers are enough to build a 1980s PC.
Its more than that really - Didi is saying that there are micro that are powerful enough to *emulate* an x86, so are more than powerful enough to provide a native application that was functionally similar to a PC.
--
Regards,
Richard.
+ http://www.FreeRTOS.org
13 official architecture ports, 1000 downloads per week.
+ http://www.SafeRTOS.com
Certified by TÜV as meeting the requirements for safety related systems.
I am aware this "little" difference. The "functionallity" i.e. of the Hercules Monitor is possible on a simple text or graphical display. I think, its controller will laugh at the 1980s PC.
Short answser : No Longer answer : There is no problem with Core-CPU power, and even 'most' peripherals, but modern silicon does not 'match' with your job definition.
Storage: It's been a long time since I've seen anyone target a Floppy drive, but users ARE active with USB Flashdrives, and also Compact FLASH cards. There are USB host microcontrollers, but they are not x86 cored.
Display: There are no uC that can drive VGA directly, but some have varying levels of LCD support, but those devices are Microprocessors - they expect large memory pools, and do NOT expect all that on one chip.
We actually have a commercial OEM product, our Vga-232, that is designed to add a Flicker free Colour VGA to any small uC - but it used RS232 to make remote annunciator readouts practical.
If you target Compile-and-run, then you need a VGA etc BIOS interface that clones the memory map/BIOS of the older PCs - and again, you step outside the realm of single chip microcontrollers.
There is no commercial/volume merit in what you describe, tho there could be some educational benefit. So big Embedded X86 players, like VIA, will not be interested.
Zilog is one company who _could_ do this (via CP/M or a variant), but I think they see more volume in Ethernet support, and web page rather than duplicating a ZX81.
Having said all the above, there is renewed focus by many X86 players, at the 'sub 1W market', for highly portable devices. Indications are Intel has some devices coming that might get close to your "job definition", but they will not focus on DOS - so someone would need to hack that. Stacked-die chips are used in these highly portable devices, and they morph the line between Microprocessors, and Microcontrollers. See also SST, for some stacked memory solutions.
Thank you for your detailed answers. :-) I took this (hard) example (replace the interior of a MSDOS PC by a microontroller) to make clear what I was thinking about. It is clear to me, that flash and displays are the better alternative to the floppy and monitor.
By the way: I have to attend to some old industry robotors and PLCs, which are still in use and do it well. And these have to be interfaced by MSDOS and RS232. For these applications I have to use very old ebay notebooks at 33Mhz. It is not possible to run the communication on a modern notebook, neighter with an emulator nor freedos. Further some of them have to run with a date before 2000. ;-) Otherwise they do not compile correctly, although they have no time and date features.
I had a look at the Freescale parts I suggested and it appears that we are not that far yet - single chip MSDOS PC emulation.
The monster DSPs with enough RAM will likely do it all easily, but they come with no flash.
There are a few PPC MCUs which come close, but don't have enough RAM and interfaces.
The PSCs (Programmable Serial Interfaces) on the MPC5200 and on other parts are definitely usable for emulating a floppy disk controller and a not too demanding monitor output, but the 5200 needs flash and RAM (this is the part I know how to do it with). A small CPLD will be needed to make all it takes for the monitor and FD interfaces.
Another few years and some MCUs with >512k RAM and flash + the tweakable serial ports with FIFOs are bound to come out, but I guess we don't have any yet available. However, I have not looked around to exhaust all possibilities so I may well be missing some.
Please don't strip attributions (the "joe wrote:" lines) for material you quote.
PCs started long before IBM entered the market. The earliest were
8008 based, and there were not many of them. The next, and popular, level was the 8080, later replaced by the Z80. Both ran CP/M as the DOS (and other OSs). They also can be easily replaced by todays microcontrollers, although the timing complexities of the floppy drive may require separate floppy control circuitry, as may the video output.
I built the equivalent of 8008 pcs about 1971 or so, and of 8080 pcs about 1974.
--
Chuck F (cbfalconer at maineline dot net)
Available for consulting/temporary embedded and systems.
It is not imported into English (probably because few English-speakers can spell it!), and should therefore be written /gedankenexperiment/ (in italics), but it is a well-known technical term. It was made famous by Albert Einstein - most of the "experiments" for relativity and quantum mechanics were "gedankenexperiments", or "thought experiments", although the term was used before that.
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here.
All logos and trade names are the property of their respective owners.