• Please review our updated Terms and Rules here

8272 FDC to 8" drive connections

Mike_Z

Mike_Z

Veteran Member
Joined
Dec 1, 2013
Messages
1,713
Location
Near Milwaukee Wisconsin
Well..... winter is here in Wisconsin and at full strength. So I'm out of the garage and back to the bench to work on my 8080 machine. I'm working on connecting an 8" drive to it for the purpose of using CP/M 2.2 on this machine. A long time ago I started this project but my family and real job got in the way, now that I'm retired I have resurrected the project. Recently I've been converting all my OLD OrCAD drawings to KiCAD, excellent program. This has afforded my an even closed inspection of my connections. I think that maybe I have had a wiring error for a long time. Actually I'm unsure of how to connect a couple.

I have a Siemens FDD 100-8 drive that has the following outputs
Pin 46 RAW DATA --- this has all the clock and data pulses
Pin 48 SEP DATA --- this is just the data pulses (I'm using single density only)
Pin 50 SEP CLK --- this is just the clock pulses

On the 8272 (through 7414 buffers) I have
Pin 22 RDDATA --- Currently connected to FDD Pin 46 RAW DATA all clock and data pulses, I think this is correct
Pin 23 DWIN --- Currently connected to FDD 48 Data pulses only, I'm thinking this should be Clock pulses only

I have read the INTEL data sheets and some application documents, but they have left my brain in a fog. How should these couple of wires be terminated. Thanks for the help. Mike
 
You may find the data sheet for the NEC 765 more informative--Intel did a cross-licensing deal with NEC on the 765/8272.

Pin 23 should be separated clock--and curiously, the polarity doesn't matter. I think you may want to connect pin 22 to the SEP DATA, but it could be that the chip won't care. Outputs from the drive should be terminated by a 150 ohm pullup to +5--and are active-low. So you may need to invert the read DATA.

You may want to consider using a little 8 pin (WD9216) data separator between your drive and FDC and use the raw data output from the drive. The benefit is that you'll at least be able to read FM and MFM floppies for very little extra cost--it's pretty much a drop-in solution. The data separation is probably better as well--I think the Siemens drives just use a one-shot data separator.
 
Chuck, I have the 150 ohm pull ups on both inputs and both lines have a 7414 inverter on them. I looked at the 765 data sheet and it states, as does the Intel 8272 that pin 23 RD DATA contains both clock and data pulses. Which seems contrary to the label. Pin 22 is labeled 'Data Window' wouldn't you think that would be between clock pulses? The cell where the data pulse is? I've got to read a little closer. Thanks Mike
 
You are correct, the Siemens just uses a mono to seperate data and it only works for single density. I looked at the WD9216. The data sheet shows two devices WD9216-00 and WD9216-01. The 01 is for both 5 1/4 and 8 inch, but still, I'm unsure of the connection to the 8272. What is 8272 RD DATA, the datasheet states, "Read Data: Read data from FDD, containing clock and data bits"; 8272 DW, "Data Windows, Generated by PLL, and used to sample data from FDD". So, DW must be seperate clock or data, but for some reason I can't seem to get it through to my person neurons. Mike
 
If I go to the IBM PC schematics, pins 22 and 23 are labeled FD CLK and FD DATA. On an old clone floppy controller I have, using a 9216 data separator, the SEPCLK 9216 output is connected directly to pin 22. I haven't checked, but I think the data goes through an inverter from the SEPDATA on the 9216.
 
OK, well either way what I have isn't either. I've been looking around in the internet and there is a lot of stuff out there, but it all seems to be just copies of the data sheet. No one seems to explain stuff. Thinking about it, it makes sense that the 8272 inputs should be sperated data and clock. I think that your last post is correct. (whereas the first one seems to be backwards, 22 and 23 are mixed). So far I only have been working on the drawings and haven't dragged out the hardware. Hopefully after Thanksgiving, I'll get to firing up this stuff and trying things out. I'm sure that there will be more questions. I appreciate your help. Mike
 
Well... I've had another set back. With the 8272 chip installed on my FDC board, the console UART stopped working. I found that the computer was still running, but I could not poll the console UART. After trying a bunch of stuff, I found that with the 8272 INPUT pin #2 disconnected and pulled high, every thin worked. So, I figure either 1. the 8272 chip is bad, (don't have a replacement yet) or the fan out of the 8228 for the I/OR and maybe the I/O W is toooooo much. Looking at the 8228 data sheet I'm not sure what the fan out is, it says the IO current is between 15 and 90 ma, but that is short circuited. The voltage would not be correct. As for what is connected to the I/OR and I/OW is 1 timer 8253, 2 Parallel ports 8255, 2 serial ports 8251, 1 FDC 8272, 1 DMA 8237, 1 memory chip 43256, 1 PIC 8259. I also have 16 6116 memory chips and two front panel lights that are connected to I/OR and I/OR, but they are buffered. Should I add a couple of buffers to these lines? Mike
 
You could simply have an error in your address selection circuitry, Mike.

Running all that stuff from an 8228, I certainly would buffer them. I assume that your address lines are also buffered.
 
All the chip selects are correct. I know that because with the 8272 out of the circuit, all the PI's work. I have 74ls367's buffering the address bus. I've been using 74LS04's as buffers, but I need two in series to do it. Since I have to add a chip should I use something else? Another 367? Mike
 
OK, I'll look into those. Say, by the way, I was looking for a Phase Lock Loop you talked about WD9216. I can't find any of those. Some guy in China has some, but I'm not going that way. I've looked on ebay, Mouser and Jameco. Any suggestions? Mike
 
Well, the IC was fairly widely duplicated and so clones are common.

FDC9216 is one
UM8326 is another

I don't know if that helps or not--but if you get stuck, drop me a line--I think I still have most of a tube's worth.
 
Rats! I added a couple of 74LS367 (because I had them) to the I/O lines to the port board. This removed the load of the timer, the two parallel ports and the two serial ports. It didn't help. Apparently my 8272 chip must be bad. I'll have to find a replacement. Mike
 
You can get FDC9266 or FDC9268 chips. They are basically 8272 with integrated FDC9229 data separator. FDC9266/8 and 8272 even have similar pinout. (You can also get an FDC9229 or FDC92C39 data separator and connect it to your 8272, but last time I've checked they were more expensive and difficult to get than FDC9266 or FDC9268).

Check N8VEM Disk I/O V3 for FDC to floppy interfacing. It supports both 34-pin and 50-pin interfaces.
 
If you want to go to higher integration and don't mind working with a PLCC, I've got a tube of WD37C65B chips that will probably never get used.

Personally, though, I'd probably use a WD279x chip--it simply cooperates better with systems that don't use interrupts or DMA.
Or a selected WD1773 that has been tested on high-density.
 
Last edited:
Thanks for the info, but I'm set up for the 8272/765. I have lots of documentation etal. So, I'll stick with the 8272. I figure while I'm waiting for the replacements. I start to work on some test programs for the FDC. Thanks, Mike.
 
This morning the replacement 8272 came in the mail. With high hopes I plugged it in and tried it out. The new chip worked just as the old one did. RATS! So I figured that I had a wiring error. I looked over all my wiring again. Could not find a problem. I looked at the symptoms. My front panel lights told me that when the monitor program was polling the console uart, there was a data bus contention. When the 8272 was out of the circuit everything worked, so I figured that the 8272 was also outputting something on the data bus at the same time. I checked the reset, I/OR, I/OW, CS and A0 connections no problems. I tested the chip select and it was activated when the correct address was issued. The Reset had the correct sense. The I/O lines were working properly. At one time I lifted the I/OR line to the 8272 and then everything seemed to work, but that turned out to be a false lead. More inspecting, a traced the circuit over and over. Tested the lines again, tried the old chip, nothing. Then it occurred to me what else could be making the 8272 want to send something. The DACK line. After all day chasing ghosts I found that I had programmed the DMA DACK line to have the wrong sense. I had it set for an active high and the 8272 was looking for an active low. So, during a non DMA situation the 8272 thought that a DMA cycle was being granted. All I had to do was flip bit 7 of the command register on the DMA. Problem solved. Geeeeeezzzzzz! Anyway now I can start to do some test programs. Seems I can make more mistakes than ever, connect a computer into the mix and it goes up ten fold. Mike
 
Started to write some code for the 8272. My first questions are does the 8272 issue an interrupt for a SPECIFY and SENSE DRIVE STATUS commands. I suspect that it does not for the specify, because there is no results phase. The SDS has a results phase, but it is only ST3, and maybe doesn't need an interrupt. Thanks Mike
 
You must log in or register to reply here.
Back
Top