PDA

View Full Version : 8" Floppy Drive Termination Question



mkstabd
December 4th, 2009, 03:27 PM
Is there such a thing as an external floppy cable terminator? My S-100 system can run with 2 or 3 8" floppies at a time. 2 of the drives are in the dual enclosure with one terminated. But when I add the 3rd external drive to the chain, I currently have to pull the terminators off the 2nd drive, which means popping the case open. I'd prefer to just keep the end of the cable terminated and add/remove the 3rd drive as needed. I'd then remove all the drive terminator chips and let the cable handle it. Can anyone point me to a device or schematic?

Thanks,

Ty

Chuck(G)
December 4th, 2009, 04:00 PM
Is there such a thing as an external floppy cable terminator? My S-100 system can run with 2 or 3 8" floppies at a time. 2 of the drives are in the dual enclosure with one terminated. But when I add the 3rd external drive to the chain, I currently have to pull the terminators off the 2nd drive, which means popping the case open. I'd prefer to just keep the end of the cable terminated and add/remove the 3rd drive as needed. I'd then remove all the drive terminator chips and let the cable handle it. Can anyone point me to a device or schematic?

The problem is that the floppy termination is 150 ohms to +5. Since there's no source of +5 on the floppy signal cable, you'll also need a source of +5 to provide termination.

But given the power source, you can easily build your own on a bit of perfboard. Take a 50 pin header, ground all of the odd-numbered pins, and connect 150 ohm resistors between all of the signals from (not to) the host and +5.

In other words, the host drives all of its signal output lines with open-collector drivers.

mkstabd
December 4th, 2009, 06:12 PM
The problem is that the floppy termination is 150 ohms to +5. Since there's no source of +5 on the floppy signal cable, you'll also need a source of +5 to provide termination.

But given the power source, you can easily build your own on a bit of perfboard. Take a 50 pin header, ground all of the odd-numbered pins, and connect 150 ohm resistors between all of the signals from (not to) the host and +5.

In other words, the host drives all of its signal output lines with open-collector drivers.

Thanks, Chuck. Does the attached schematic properly capture what you described?

Chuck(G)
December 4th, 2009, 06:20 PM
Thanks, Chuck. Does the attached schematic properly capture what you described?

Looks good to me!

mkstabd
December 8th, 2009, 03:46 PM
Okay, so here's my idea: Taking a cue from my Sun SCSI devices, I want to be able to flip a switch between enabling and disabling termination inside my dual floppy enclosure. If the switch is set to "enable", then the last drive inside the dual enclosure is terminated and any external drives chained after the enclosure are electrically isolated. If the switch is set to "disable", then the terminator circuit is bypassed and any external drives chained after the enclosure are reconnected to the floppy cable as they would be with no termination circuit in place.

I was proposing it could look like the attached updated schematic. I'd install a DPST toggle switch to the back of the enclosure for the enable/disable selector and tap +5V from the dual enclosure's power supply to power the circuit. Does this seem reasonable?

I want to make an inline power tap extension using the exact same connectors that the floppy drive and power supply use now (I don't want to cut any original wires in this box!) but I'm not sure what the sockets are called or how they're sized. I'm just now learning how to navigate websites like Digi-key. I live next to a Fry's here in Renton, WA, and they don't have these connectors. The ones they have are much smaller for a 6 hole socket.

Thanks for your patience with my basic questions guys, especially to Chuck. I am not an EE, just a mad scientist who's very eager to learn this stuff. This seemed like a reasonable first project. I'm currently reading up on copper etching and, if this circuit works on my breadboard, I'll build it out. BTW, I burned my hand with my soldering iron last night, so I must be learning something!!

paul
December 8th, 2009, 03:57 PM
You still have signal lines connected together via 2 x 150 ohms. Disconnecting the 5 volts doesn't disable that loading.

Can you not make a longer cable and loop it outside before the internal terminated drive?

mkstabd
December 8th, 2009, 04:18 PM
You still have signal lines connected via 2 x 150 ohms. Disconnecting the 5 volts doesn't disable that loading.

Can you not make a longer cable and loop it outside before the internal terminated drive?

DOH! That's right, sheesh. Even though the resistors aren't connected to ground, I guess they're still connected to each other. Thanks! Just for yucks, what would be the most efficient way to disconnect the resistors too?

Yes, I had thought about the longer cable idea. It would be far simpler. I'm still not used to externally run ribbon cables and just thought this might be a fancier upgrade.

Chuck(G)
December 8th, 2009, 05:29 PM
DOH! That's right, sheesh. Even though the resistors aren't connected to ground, I guess they're still connected to each other. Thanks! Just for yucks, what would be the most efficient way to disconnect the resistors too?

Absent an n-pole switch You could go with a CMOS switch or even a TTL tristate buffer to terminate the resistors. Just set it to high-Z when you don't need it.

You might find a more elegant solution in Pericom's lineup.

Another idea is to isolate the terminating resistors from each other with schottky diodes. Then switch the +5.

antiquekid3
December 8th, 2009, 05:51 PM
What about those DB-25 parallel switches that people used to use to select printers? You could definitely use a switch like the one in there for your application, I would presume. The wafer switch has to switch a lot of wires, so it'd be great to connect/disconnect those resistors. Definitely not as elegant as two octal tri-state buffers, though.

Kyle

MikeS
December 8th, 2009, 08:33 PM
In my experience it's not usually a problem if the cable isn't terminated at the end (as long as it's terminated somewhere). Why not try it as is, i.e. only put the terminating resistor pack in the last drive inside the cabinet and just extend the cable to the (unterminated) external drive (as short as possible).

Switching the grounds as per your diagram doesn't look like a solution; besides, the grounds would be connected through the drives' power ground.

Chuck(G)
December 8th, 2009, 09:59 PM
Switching the grounds as per your diagram doesn't look like a solution; besides, the grounds would be connected through the drives' power ground.

No, but a diode in series with each resistor (schottky would be best to keep down the voltage drop) would isolate the pins, so removing the +5 would essentially disable the terminator.

MikeS
December 8th, 2009, 10:08 PM
No, but a diode in series with each resistor (schottky would be best to keep down the voltage drop) would isolate the pins, so removing the +5 would essentially disable the terminator.That was my first thought too, but it does seem needlessly complicated...

mkstabd
December 9th, 2009, 03:12 AM
Thanks to all for the responses.

Here are some updated schematics using tri-state buffers (http://bartosh.us/s-100/schematics/Floppy%20Cable%20Terminator%20Circuit%20with%20Ena ble-Disable-tri%20state%20buffer%20isolation.pdf) and schottky diodes (http://bartosh.us/s-100/schematics/Floppy%20Cable%20Terminator%20Circuit%20with%20Ena ble-Disable-schottky%20diode%20isolation.pdf). I also moved the switch to the +5V side of the power supply and went with a DPDT toggle switch.

From Chuck(G)
Absent an n-pole switch You could go with a CMOS switch or even a TTL tristate buffer to terminate the resistors. Just set it to high-Z when you don't need it.

You might find a more elegant solution in Pericom's lineup.

Another idea is to isolate the terminating resistors from each other with schottky diodes. Then switch the +5.
@Chuck: It seems like all 3 options would work (CMOS switch, tri state buffers, schottky diodes). IMHO, the CMOS analog switches look to be the most reliable, yet the most expensive (aside from the Pericom solution.) The tri-state buffers seem like a great solution. Are there any drawbacks to this approach? According to what I found on the web, the schottky diodes lose only 0.2V or less, so that seems like the most practical solution. Which way would you go?


From antiquekid3
What about those DB-25 parallel switches that people used to use to select printers? You could definitely use a switch like the one in there for your application, I would presume. The wafer switch has to switch a lot of wires, so it'd be great to connect/disconnect those resistors. Definitely not as elegant as two octal tri-state buffers, though.
@antiquekid3: I had initially thought of a wafer switch too, but I was worried about arcing and the fact that it's a mechanical switch. I was hoping to go with as little mechanical as possible, if for no other reason than it's a good science project for me to try to implement it with circuits. I agree with you about the tri-state buffers being an elegant way to go, which is kind of what I'm shooting for. My concern about wafer switches may be unjustified, but I'm just erring on the side of caution to protect my SCP drive controller board.


From MikeS
In my experience it's not usually a problem if the cable isn't terminated at the end (as long as it's terminated somewhere). Why not try it as is, i.e. only put the terminating resistor pack in the last drive inside the cabinet and just extend the cable to the (unterminated) external drive (as short as possible).

Switching the grounds as per your diagram doesn't look like a solution; besides, the grounds would be connected through the drives' power ground.
@MikeS: I hadn't considered terminating any other drive than the last drive in the chain. I'll have to try that out.

Also, I updated the diagram to switch on the +5V side vs the ground side for the "enabled" mode. Please help me understand why sharing the drive ground is a bad thing. Do I need to look at a completely isolated +5V source? I was hoping to use the +5V source already in the enclosure. I thought I would be using the same ground that the chip resistor terminators normally plugged into the drive were using. I'll take another look at the drive schematics.

Dutch
December 9th, 2009, 06:03 AM
For your switch, consider using half of a 74LS00 or a 74LS279 latch for debouncing the contacts.

-Dutch

Chuck(G)
December 9th, 2009, 09:13 AM
I'd take the diodes--cheap and simple. You can even get them in SIP and DIP packages (e.g. SN74F1056).

But Mike is also probably right--unless the cable going to the external drive is longer than about a meter, you could be okay just terminating the middle drive, given the data rates involved.

MikeS
December 9th, 2009, 02:48 PM
Thanks to all for the responses.

@MikeS: I hadn't considered terminating any other drive than the last drive in the chain. I'll have to try that out.Should work.


Also, I updated the diagram to switch on the +5V side vs the ground side for the "enabled" mode. Please help me understand why sharing the drive ground is a bad thing. Do I need to look at a completely isolated +5V source? I was hoping to use the +5V source already in the enclosure. I thought I would be using the same ground that the chip resistor terminators normally plugged into the drive were using. I'll take another look at the drive schematics.
Sharing the ground (and +5V) is a good thing; I'm just pointing out that switching the ground as you're still doing is a bad thing, although in this case the switch probably won't do anything anyway except possibly introduce noise since the grounds are still connected together through the power supply.

Also, the way you have the buffers wired defeats their purpose; you'd connect all the inputs to +5V and connect the switch to the enable inputs (also connected together); I don't think key bounce would be a problem ;-)

But I still suspect that none of this is really necessary.

BTW, what are you using to draw those diagrams? I'm always looking out for new tools.

mkstabd
December 10th, 2009, 02:29 AM
Sharing the ground (and +5V) is a good thing; I'm just pointing out that switching the ground as you're still doing is a bad thing, although in this case the switch probably won't do anything anyway except possibly introduce noise since the grounds are still connected together through the power supply.

You've made me realize something: I was only thinking of operating the switch with everything powered off; it hadn't occurred to me to flip the switch when the bus was powered up. If the switch was flipped with power, what would the worst case scenario be? Could the noise you describe fry any chips in my floppy drives or host controller? Or would the effects be more like corrupting data transfers if any were in progress at the time?


Also, the way you have the buffers wired defeats their purpose; you'd connect all the inputs to +5V and connect the switch to the enable inputs (also connected together); I don't think key bounce would be a problem ;-)

LOL @ key bounce :-)

The tri-state buffer schematic was my attempt at capturing what I thought Chuck was describing, which was to isolate the resistors from each other when that side of the circuit was powered down (switch in disable/pass thru mode.) If there was no +5V flowing to the buffers, then all the paths between the resistors would be blocked. Conversely, if there was +5V to the buffers, then they'd activate, allowing the resistors to rejoin the circuit.

It's kind of like an anti-fix in that their primary purpose was meant to support no signal flow, rather than allowing signal flow. In this case, the buffers achieve the same effects as using the schottky diodes, but with no voltage drop. Perhaps not the most efficient use of the buffers, but I still think it's a way cool idea!


BTW, what are you using to draw those diagrams? I'm always looking out for new tools.

I threw those together with Visio. :blush: Can you recommend a good freeware modeling tool?

Ty

Dutch
December 10th, 2009, 06:00 AM
LOL @ key bounce :-)





Yeah, that was dumb. I misread the previous post.

Duh!