As various questions about CoCo floppy drive issues keep surfacing on the CoCo mailing list/newsgroup, comp.sys.tandy, etc., I decided to begin archiving information on that topic in one place. This is the beginning of that archive.
For now, this page will be divided into three sections: one dealing with CoCo floppy controllers, the second with CoCo floppy drives, and the last containing information pertaining to the use of non-Tandy drives with the CoCo. Note that this page is in its earliest stages, and is nowhere near complete at this time. The primary goal of this page, at least initially, is to answer some of the common questions about interfacing various drives (particularly those from the IBM-compatible world) to our beloved CoCo.
Tandy offered floppy drives for the CoCo in various packages. Each of these included a controller cartridge which plugs into the computer’s system bus, as well as an external case with a floppy drive mounted in it. The controller cartridge (or ‘pak’) contains one of several Western Digital floppy controller chips, the Disk BASIC ROM chip, and the necessary interface logic. The drive unit contains a linear power supply and a 5.25″ single-sided (in most cases), double-density, half-height (in most cases) floppy drive. A captive (in most cases) ribbon cable exits the drive unit, terminated in a 34-pin card-edge connector which attaches to the controller pak. The drive unit also has an AC power cord. Even though the drives and controllers were sold as packages, they can be interchanged almost universally.
CoCo Floppy Controllers
This first model of CoCo floppy controller was packaged in a long cartridge, and was introduced for use with the original CoCo. It uses a WD1793 controller chip, which requires +12V DC. For this reason, a multi-pak interface (or other source of +12V) is required to utilize this controller with a CoCo 2 or CoCo 3. This controller also uses an analog data separator circuit, infamous for its inherent unreliability. However, this controller is the only one capable of being modified (though with considerable difficulty) to allow use with high-density 3.5″ floppy drives.
All subsequent controllers use the WD1773 controller chip, do not use the unreliable analog data separator, do not require +12V (so they will work with any CoCo model), and are capable of double-density operation only.
This model was introduced for use with the CoCo 2 (as it lacked the original CoCo’s +12V power supply), and is also mounted in a long cartridge.
This model is also mounted in a long cartridge. It appears to be identical to the 26-3029, with the exception that it carries the Tandy name rather than the TRS-80 name.
One of the more ‘modern’ controllers (mounted in a shorter cartridge than the older ones), this is also the only Radio Shack model ever to feature gold-plated card-edge connectors. Due to the much higher reliability of gold contacts over tinned PC fingers, this is my personal favorite among all of the Radio Shack controllers. This model was introduced in the 1986 Radio Shack catalog.
This is the last Tandy controller. It reverted back to mere solder-tinned edge connectors, for whatever reason (cost?!). Unlike every previous model, this one uses a 28-pin ROM rather than a 24-pin ROM for Disk BASIC. This model was introduced in the 1988 Radio Shack catalog.
All Radio Shack (and some third-party) CoCo floppy controllers use a pair of 7416 open-collector buffer chips to communicate with the floppy drive(s). These chips can fail and cause various strange symptoms. They may be replaced with 7406 chips, which are more rugged (the 7416 is rated at 15V, while the 7406 is rated at 30V). Socketing the replacement chips makes future replacement much easier, should it become necessary.
CoCo Floppy Drives
26-3022 ‘Color Computer Mini Disk’
These original CoCo floppy drives were gray in color, matching the exterior of the original CoCo (and other early TRS-80 machines and peripherals). They use a 5.25″ full-height TEC (Tokyo Electric Corp.) drive mechanism, infamous (like the mating controller) for reliability problems. The drive unit stands vertically, making it taller than it is wide. It was reportedly capable of accessing only 37 tracks or so physically, prompting Tandy’s incorporation of the 35-track limit into Disk BASIC. These (and all subsequent) drives were double-density and, like all but the very latest model (FD-502), were single-sided. This resulted in a capacity of approximately 160 kilobytes per disk.
26-3029 ‘Color Computer Mini Disk’
This drive model had a white case, but was still a full-height, vertically-mounted assembly. These first two styles have 34-pin card-edge connectors on their rear panels rather than captive ribbon cables. The addition of extra drives required using additional case/power supply/drive units.
FD-500 (26-3129), FD-501 (26-3131)
All subsequent drives were also white in color, but were mounted horizontally and used half-height drive mechanisms. They shipped with a single drive (Drive 0) in the lower ‘bay’, and a blanking plate over the upper opening. These drives could be upgraded to a dual-drive system by installing the second drive kit in the existing case. Note that the kit included, and all dual-drive installations in these cases require, a 12-volt cooling fan. The units run acceptably cool with a single drive, but not with two drives. The 4 cm diameter, 1 cm thick brushless fans (as used on many CPU heatsink assemblies) work well in these cases.
This drive unit is like the FD-501, with two notable exceptions. This is the only Tandy model ever to incorporate true double-sided drives, and its drive power connectors are like those of typical 3.5″ floppy drives rather than those of normal 5.25″ drives. The smaller power connectors are also reportedly wired differently than typical 3.5″ drive connectors. As noted elsewhere, either a patched Disk BASIC or an alternative operating system is required to gain access to the second side of disks. A cooling fan must be installed for safe operation if a second drive is added.
CoCo Use With Non-CoCo Drives
Yes, drives other than those from Tandy may be used with the CoCo. In particular, the common 360K PC drives are CoCo-compatible. To put it into the proper perspective, IBM chose for the PC the same pre-existing floppy standard that Radio Shack chose for the CoCo. Actually, IBM deviated from the standard more than Radio Shack did!
The standard specifies the possibility of up to four drives connected to one controller. All decent floppy drives have four drive selection jumpers, usually numbered 0 through 3 (occasionally 1 through 4). While they were used properly in some early PCs, the standard was still not utilized to its full potential (by allowing only two drives). The CoCo supports all four possible drives. All ‘modern’ PCs use a twisted floppy interface cable, with all drives permanently set to ID 1 (or 2; the second of the four possible choices). The same selection method may be used with the CoCo, but will result in the same limitation of only permitting two drives. The usual selection method Tandy used was different. They jumpered all drive select lines on all drives, and removed all but the desired ‘teeth’ from the edge connectors on the ribbon cable. This is commonly referred to as the ‘dental extraction’ method of drive selection. PC and CoCo floppy drives alike may be connected to their controllers in the proper way, with a straight-through ribbon cable feeding all edge connectors, and the proper use of drive select jumpers on the drives themselves.
Another area in which the PC ‘world’ deviated from the standard is the issue of drive cable termination. The standard specifies that a terminating resistor array shall be installed at the end of the interface cable farthest from the controller, and nowhere else. As with SCSI interface cables, 10base2 networking, and all other RF applications, proper termination prevents reflections and standing waves in transmission lines. CoCo drives typically are (and should be) terminated in the proper manner. The Wintel crowd, on the other hand, decided to eliminate that little additional bit of required knowledge (along with drive selection elimination by twisting the interface cable). What they do is terminate ALL drives; that way, one cannot forget to install the terminator at the proper place. Of course, with only two possible drives in the first place, it does not cause as much of a problem as four terminators would.
Another issue to be aware of is that all drives in a CoCo system must spin up when told to do so by the ‘motor on’ line, and not by their drive select lines. PC drives may be jumpered wrong for CoCo operation. Problems may arise when copying information from one drive to the other, as the CoCo assumes that all drives are spinning while any drive is accessed. No delay is inserted between reading from one drive and writing to another. The name of the jumper which controls this behavior depends on the specific drive.
5.25″ 80-track 720K (‘quad-density’) and 3.5″ 720K drives are also CoCo-compatible, within the preceeding constraints. They are electrically identical to each other. Disk BASIC will only access the first 35 tracks on the first side of any disk, however, so they will still only hold about 160K of information each. Various patches exist to Disk BASIC to allow access to the extra space, and the OS-9 operating system can make use of it as well. Note that 720K 5.25″ drives were virtually unknown on the PC platform (except in the Tandy 2000); they were fairly popular in the CP/M and OS-9 communities.
5.25″ 1.2 meg drives are virtually useless on the CoCo. They are reportedly more like the old 8″ drives than either other 5.25″ or 3.5″ drives. They also spin at a different rate (360 RPM rather than 300). Some may reportedly be jumpered to spin at 300 RPM and function as either 360K or 720K drives, but they would still not be reliable for use with true 360K media (just as they are not in PCs).
3.5″ 1.44 meg drives can sometimes be jumpered to function as 720K drives; one should always use true 720K media when doing so. The previously-mentioned hack for the original 26-3022 controller may make these drives useful (under patched Disk BASIC or OS-9, of course) as true 1.44 meg drives; I have not attempted to implement that hack. UPDATE: supposedly, just the act of using true 720K media (without the high-density hole) will make a high-density 3.5″ drive function as a double-density drive. I intend to confirm this.
Something else should be noted about 1.44 meg 3.5″ floppy drives. I encountered such a drive which appeared to be nonfunctional. Further testing revealed that it worked with 720K disks, but not 1.44 meg disks. I checked the density hole sensor switch, and found it to be good. I then traced the circuit from that switch and found that it connected to one conductor in a small mylar ribbon cable connecting the two halves of the drive’s logic circuitry. It turned out that the ribbon cable was not making solid contact. Disconnecting it from the small connector on the one PC board and reseating it fixed the drive. I have since repaired other bad drives in the same manner.