8

Early microcomputers sometimes supported several peripherals in a 'daisy chained' arrangement, i.e. instead of the computer having several ports, one per peripheral, it had a single port, which each peripheral would pass through to the next one. For example, per Wikipedia, the Vic-20 had "a serial CBM-488 bus (a serial version of the PET's IEEE-488 bus) for daisy chaining disk drives and printers".

But the Vic-20 cassette drive was not daisy chained. It had a different, dedicated port.

Each added port had a nontrivial cost, not just in terms of manufacturing expense but development complexity, an extra thing that could go wrong, and increased difficulty of RF shielding. I would expect ports to be added only when there was compelling reason. Was there a reason it was difficult to daisy chain a cassette drive?

Did any early microcomputer actually have a daisy chained cassette drive?

  • 8
    A cassette drive port is often connected to dedicated circuitry in the computer for decoding the analog waveform from the cassette. It's not a general purpose interface. And there is very little point to having multiple cassette drives connected to the same computer. – user3570736 Jul 26 '18 at 9:49
  • 2
    @user3570736 the Coleco Adam has two 'tape drives' built in. But they're oddball fully-automated devices that are supposed to provide exactly the same random-access experience as a disk drive, just slower, so that might explain why Coleco saw utility in having two of them when usually there wouldn't be. – Tommy Jul 26 '18 at 13:18
  • 1
    @user3570736 Most (if not all) Commodore PETs and CBMs would accept two cassette drives, but this feature disappeared with the VIC-20 and subsequent home computers. It definitely was handy for duplicating files on cassette, especially sequential data files, which were uncommonly used but occasionally useful. – Jim MacKenzie Jul 26 '18 at 14:55
  • 2
    You are aware that Apple seems to enjoy daisy-chaining monitors now? It's not just early microcomputers. – UKMonkey Jul 26 '18 at 15:22
  • 1
    The VIC-20 is very similar to the PET. The cassette interface is the same for both and the tape drives are interchangeable. For disk drives and printers, both systems use a serial bus, the only difference being that the PET uses an IEEE-488 parallel bus, while the VIC-20 uses a serial derivative that looks the same to application programs but uses a simpler hardware connector. The VIC-20 doesn't use daisy-chained tape drives because the PET didn't, it does use daisy-chained disk drives because the PET did as well. This helped software compatibility and made it easier to port PET programs. – Ken Gober Jul 26 '18 at 17:40
7

TL;DR

Except for the high price area of HI-IL and HP-IB the answer is no, as a single interface raises total system cost and is, especially for cassette drives, which need to be cheap, contraindicated.


The long read

Early microcomputers sometimes supported several peripherals in a 'daisy chained' arrangement, i.e. instead of the computer having several ports, one per peripheral, it had a single port, which each peripheral would pass through to the next one.

First of all daisy chained is only a special case for a bus system. The argument of having only one port (connector) also works for IEEE488.

Second, it was not just early ones using such buses. Just take USB. One port and it replaced next to all others by now.

But the Vic-20 cassette drive was not daisy chained. It had a different, dedicated port.

As it got a different interface. The VIC20/C64 serial bus devices where connected using a serial interface, while the cassette drive did need a specific because of its audio nature.

Each added port had a nontrivial cost, not just in terms of manufacturing expense but development complexity, an extra thing that could go wrong, and increased difficulty of RF shielding.

Not really. Separate ports simplified design a lot, as there was no need to have an integrated handling and an abstract protocol. This saved investment in hardware as an interface was directly controlled, well as in software due less complicated drivers.

I would expect ports to be added only when there was compelling reason.

Strictly cost and simplicity. A parallel or serial port, a mouse or joystickport is much cheaper if the main CPU drives these ports directly. Controlling them via a standardized, digital bus means the addition of this interface and a processors to each connected device.

A built-in interface, only needs its controller integrated into the computer and a port connector. When doing the same via a bus like with the VIC-20, the interface-box requires in addition to the controller a local CPU (plus RAM, ROM, etc., if not integrated) to drive the controller and another interface for host communication. This is true for any kind of device connected via the bus. No matter what device to be controlled. A serial or parallel interface got the same cost burden than a serial, a video controller or a floppy drive.

It becomes less of a 'bus tax' if the device does already need a local CPU, like for a printer, or a floppy (when it comes to slow computers). Still, an Apple Disk II had always a lower manufacturing cost (when including the controller) than a Commodore 1540 (*1). But that's a special case. In general, with such a peripheral bus total system cost is much increased.

There are only three reasons why a designer would choose such a more expensive bus over separate interfaces:

  1. Maximum flexibility with minimum configuration hassle - that's why HP introduced the HP-IB in the first place. Beside that their devices were high end priced anyway, it allowed flexible configuration of measurement and computing equipment.

  2. Keeping the base unit price artificial low - which fits the VIC-20

  3. Create a lock in situation - similar here.

Today it has become different as adding a CPU to a peripheral is no big deal, and even the most primitive devices already got one.

Was there a reason it was difficult to daisy chain a cassette drive?

It wasn't difficult, but costly. To make it work with a bus, the cassette drive would have need its own CPU and interface. This would have cranked the cost up to half way to a disk drive and way beyond 150 USD (early 80s). But to sell a cheap home computer (like the VIC) some cheap mass storage was esential to make it useful. So adding a separate connector and driving the cassette with as least additional hardware as possible was mandatory.

Atari had the same issue as they (rightfully) believed that a computer cassette recorder may be priced higher than a similar audio device, but not 3-4 times as much. So the SIO connector is in fact a bastard of a dedicated cassette interface and a serial bus (*2). When operated as cassette port the serial lines are used different. Clock was not driven, but data-out carried a (square wave) audio signal when writing to cassette. For reading the recorder has a set of filters to directly convert the tones into a bitstream (*3) - which the CPU had to synchronize and decode. The whole operation was enabled by the MOTOR signal (*4). During cassette operation no SIO device could be operated interleaved (*5).

Did any early microcomputer actually have a daisy chained cassette drive?

Yes - as usual depending on the value of early.

The 1979 HP41C got in 1982 a HP-IL controler (82160A) enabling the use of a 82161A Tape Drive. Likewise the HP150 (1983) and HP110 (1984) desktop computers featured default HP-IL enabling the use of the same tape deck.

There where also (third party) cassette drives for IEEE 488 (HP-IB) - I remember one operating with a Commodore CBM (*6).

Other than that I can't remember any.

The Atari SIO did only support this in a mechanical sense as there could only be one cassette drive and that could only be operated alternating with SIO devices.

Much like the cassette docking stations for Sharp's handheld computers. Here also a generic interface was fitted with some additional lines for cassette handling.


Now, when stretching this a bit, every computer bus can be used in daisy chain fashion. Just think of the TI99/4 (literally) wide extensions or Sinclair's rucksack devices.


*1 - That Apple translated this advantage into an even higher profit is not a technical story.

*2 - More correct so it started out as a cassette port that turned into a serial bus during development.

*3 - The hardware frequency discriminator war also the reason why there where no real fast speedtape formats for the Atari.

*4 - Well, not really, as regular SIO-Devices (non-cassette that is) did not monitor the motor line. To them nothing happened on the bus as there was no clock signal. While this should have been sufficient as locking, it could lead to a very strange problem with unfinished disk operations when starting a cassette read, resulting in locking up the disk drive until a reset is send out.

*5 - Yes, of course did clever programmers find a way to write low level routines to create some parallel operation, but that was confined to very special usecases.

*6 - That setup was even more special, as it was used to control the nightly transmission of a radio station. The cassette decks (there where in the end several) weren't used for data storage as maybe intended, but to hold prerecorded shows/music and the Commodore did select which drive to operate at the right time. In the beginning it was only whole segments (a little less than 15 minutes each), later on the software got extended to select single songs of the repertoire (the collection of all cassettes in all drives) according to a classification system (they called it 'program colour') and shuffle them in random order while taking into account when a specific song has been played last time (avoiding twice the same within a few hours). But that's a separate story :))

This might be fairly standard since some years, but in ~1981 it was revolutionary.

9

I’m not sure whether this counts as early enough, but the Atari 1010 cassette drive could be daisy-chained; it was produced from 1983 to 1985, replacing the earlier 410 which had to end the SIO chain. The XC11 (1986) could also be daisy-chained, but the XC12 couldn’t. (Although as Raffzahn points out, you can’t use more than one cassette drive on an SIO bus; so the daisy-chaining ability only allows you to connect other devices to the chain via the drive. Strictly speaking, only one “SIO motor control” device can be connected; cassette drives are motor control devices.)

See the Atari 8-bit FAQ for details.

As a side note, unlike the VIC-20 cassette drive, the Atari cassette drives all used the SIO port; there was no dedicated port for cassette drives on Atari computers. The drives decoded the signal internally.

  • 2
    I guess it's a matter of how 'daisy chained' is interpreted, as the Atari cassette drives are not ordinary SIO drives, but use a different signaling and there can only be one. In fact, it's more like the SIO connector carying two different interfaces - SIO and cassette recorder. – Raffzahn Jul 26 '18 at 11:15
  • 2
    I would agree with @Raffzahn while you can daisy chain the recorders it's not like there is any addressing logic on the cassette player. – PeterI Jul 26 '18 at 11:42
  • 1
    Indeed, thanks @Raffzahn, I’ve fixed my answer accordingly. – Stephen Kitt Jul 26 '18 at 12:02
  • 2
    One interesting consequence of Atari's SIO port was that cassette drives and modems played their audio through the Atari's audio port, which was both useful and annoying, depending on the circumstances. :) – Jim MacKenzie Jul 26 '18 at 14:56
  • 2
    @JimMacKenzie assuming they had motor control, it could have allowed the best port of Mel Croucher's Deus Ex Machina. – Tommy Jul 26 '18 at 16:38
9

You are talking about two different things, the Commodore Datasette, which had an idiosyncratic connector (like the Atari SIO as mentioned in another answer here), and ordinary cassette drives as used with other home computers (BBC Micro, Tandy TRS-80, MSX, Dragon, et al.). The latter were ordinary analogue audio equipment. They weren't computer peripherals with digital data and control interfaces.

The connections from a home computer to a cassette player were the audio line out/earphone and line in/microphone connections, carrying monaural analogue audio, and a motor power relay control input line (as also used by hand-held microphones). That one input (actually two lines) was the extent of the control interface, which not every home computer even employed; and the data were analogue.

A BBC Micro cassette lead

There is simply no concept of daisy chaining or pass-through in such a system. There is no bus. This is analogue home audio recording/playback. The computer does not even get to command whether the player is playing or recording. There is no concept of multiple peripherals addressable via commands from a controller of some kind.

Further reading

  • 1
    "Datassette" not "Dataset". :) Stack doesn't deign me worthy of such a trivial spelling change, so I can't edit the post. – Jim MacKenzie Jul 26 '18 at 14:58
  • 2
    "The computer does not even get to command whether the player is playing or recording. " * PRESS CASSETTE RECORD CS1 THEN PRESS ENTER ... of course it does, it just does it via the operator as proxy. :) – Jules Jul 26 '18 at 16:13
  • 1
    When you've pressed play instead of play+record as often as I and others have, you will realize that that is not a command, but merely a suggestion. (-: A command interface would of course give the computer direct control over enabling the erase, playback, and record heads; as indeed it has on true magnetic tape drives. – JdeBP Jul 26 '18 at 16:29
  • 1
    Did the Dragon pass through the second audio channel for voice playback? The pinout suggests "no". – Maury Markowitz Jul 27 '18 at 1:24

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.