Intel 82077AA functional replacement? (2023 Update)

I'm willing to bet you are going to need to roll your own in an FPGA, with analog circuitry to handle the signals from the heads. I don't know exactly what the FDC does, do you? I guess if you have an interface spec, and especially source code for the driver, you can reverse engineer from that.

You might try finding computer hobbyist groups on Facebook, Linkedin and the like. I'm willing to bet someone has done this before to keep some gear running.

I don't know too much either, but i know it's slow. I bet you can do it with high end micro, rather than FPGA.

There are no analog signals coming from the floppy drives as we know them. Back in the 80-s I did two controllers, 765 based. It is all TTL to/from the floppy.

Talking to the FDC is very different from talking to FDD.

Just need some circuit to read/write MFM signal, and controlling the stepper motor.

Crystal clock, PLL and magnetic reader/writer.

With the 765 I did that with a few TTL parts back then, one had to do the clock recovery (for either FM or MFM) The interface to the FDD was standard, still is if anybody makes them... Back in the early 90-s I had a SCSI floppy, my newer machines had no FDC. It was really well made, I could talk it into reading the floppies from my older machines which I used with 256 bytes/sector; all the gaps etc. thing was programmable over SCSI. Cost something like $300, I still sort of mourn it... (killed it with a shitty scsi cable I had made. I have no floppies to read, moved all I wanted to images as files on my disks (some while I still had that scsi FDD) but I still mourn it when I think of it :-).

There are plenty of 82077AA devices on eBay if you only need few of those. I would even bet they are not fakes as it is not one of high-demand expensive devices.

So the floppy drive itself has electronics to recover the data from the head signal?

If that's so, you probably can interface to a floppy with an MCU. The harder part might be the interface to the other side, but if you can specify that yourself, you should be good to go with something simple like SPI with a custom protocol.

MFM at reasonable data rates is easily done in an MCU.

Not exactly the data, just to make a TTL bitstream out of what the R/W head reads - and to record in the opposite direction. How that stream is divided into sectors, gaps between them etc. is the job of a disk controller, in my case it was a upd765.

Possibly but it won't be an easy ride. IIRC the bit stream is 1 MHz clocked, was it 2 at high density (don't remember). Then it is coming from a mechanically rotating media plus noise so clock recovery is not the easiest thing to do - but doable, I think what I did was to reset a counter or something to stay in sync, was simple anyway and worked rock stable. Don says it has been all done with a Z80, which must have been quite an achievement if it could do standard MFM floppies. My achievement was to make a 1 MHz 6809 read the data from the

765 at highest speed it could come without using a DMA (prior to that I had done it with one, the 6844) and servicing the interrupts in the gaps between the sectors, was anything but trivial to do. (I think the highest speed data would come from a 5" HD floppy, they rotated somewhat faster at the highest density). >

Perhaps I was not clear. The FDC IC controls the data transfer from the disk to the MCU. The disk drive proper contains the analog electronics to position the heads (most often a stepper motor), and pull out the data. The floppy disk controller does not contain any analog processing components and head positioning software/hardware, all that is done on the drive itself. Yes, a high end micro could probably do the job, but the FDC is a 'cleaner' approach...and if I had to go a different route, I'd use an FPGA instead of the micro. I do have the data sheets for many of the FDC chips of that era, including the 82077AA. While I've done it before for DEC RK07 drives, I don't want to reinvent the wheel on a FPGA if I don't have to.

Hah! This guy has put in the effort and made it, did not know someone would be into that sort of thing. Like I said not an easy ride, but he has done it all, bravo.

you must not have seen many modern timers

One additional point. I firmly believe that controlling devices at the basic level should be the job of hardware. It is a logical separation of concerns as well as reducing cpu utilization to other jobs the processor can do instead of bit banging a disk drive (or what ever). Maintaining sequencing and timing of signals is efficiently done by hardware, leaving the CPU to handle other tasks. Yea, could throw a processor at it as the controller but that would muddy up the architecture (hardware and sw) of the system I am working with.

Those were the days when memory was limited. But for today's system, if you even want to use floppy, would be to load everything into memory while booting. While booting, your CPU would not be doing any other task.

On a sunny day (Thu, 29 Dec 2022 10:04:23 -0800 (PST)) it happened Three Jeeps snipped-for-privacy@gmail.com wrote in snipped-for-privacy@googlegroups.com:

As somebody else suggested: ebay for the chip. That said I wrote a driver for the 8072 and designed some hardware PLL with a 4046

scroll down to floppy disk controller. That was for 8 inch Kaypro II type floppies...

All that said, I still have an old PC with a floppy drive and one day long ago bought an USB floppy drive. That may be the fasted way,,, But if you want to prove yourself ...

These days with 64 GB USB sticks and micro SD cards ... recording HD movies on those... Amazing.

On a sunny day (Fri, 30 Dec 2022 03:57:52 GMT) it happened Jan Panteltje snipped-for-privacy@yahoo.com wrote in <tolr5r$j4s8$ snipped-for-privacy@dont-email.me:

Oops. 5 1/4 inch and 8272 ..

Yes, that is what I mean. I didn't realize there was any electronics in the floppy drive itself. I've never paid close attention to the floppy interface. I was always more interested in the hard drive interfaces. They clearly have lots of on board electronics.

A Z80 is pretty slow compared to today's MCUs. That would be an interesting design to see.

I worked on a tape drive interface which was much slower than this, and done in logic, because we are talking the days of 8080s where a minimum design was more like a half dozen chips. I used to know details about all the serial formats, but they are lost in the haze. I know this was a simple one, like Manchester and only took a few chips of TTL. I believe timing was done by some fixed timing element, so a similar design would probably not be flexible enough for a floppy.

I'm willing to bet an ARM CMx could be made to do the job. They clock pretty fast these days.

Talking about old technology, I remember meeting some guys who worked at NASA in Florida. They said that even today, it would not be possible to design the networked data distribution they had, using off the shelf computing and networking. They provided a few numbers and I would not argue with them. The Apollo system was all TTL and had pretty amazing stats. That was maybe 20 years ago, so maybe now with 10 or 100 Gbps networking... I don't know. I guess we'll never know. It's not like that will ever be built again.