Hello folks, as an old user of such tech I am a bit concerned that two things may be confused: Pull up resistors and termination. Often those went alongside, but later on faster bus systems, termination increasingly became the standalone purpose. 1) Pull-up resistors: Will be needed on Open Collector outputs like the 74(LS)38. May also be needed on non OC outputs, depending on the load (inputs per output). On Open Collector outputs you get the advantage of connecting several outputs in parallel, as Adrian said. But essentially, the open collectors in parallel work like a logical Or-Gate without needing extra ICs. 2) Termination: Termination in the strictest sense is used to kill deflections and delayed spikes from open wire ends such as a cable to several disk drives. Imagine such a cable or wire to be a T. A signal coming from a drive goes up to the "fork" and the travels to the left (say, the controller) and to the right (Open cable end). On computers like the Apple 2 you would have such cables hanging out of the back for drives. The signal part that goesto the open wire end gets reflected and goes back towards the controller after hitting the end of the wire and can cause extra spikes or dips on the wire that can throw the signal out of specs (like between 0 and 1), and by doing so cause data integrity problems. A resistor just sends the signal that went to the open end to ground or to +5V. I personally know that better from SCSI chains on my Amiga computers, where every drive had termination that could be enabled or disabled, and the termination resistors IIRC always were connected to ground, and the data lines were NOT Open Collector. Nowadays signals are so extremely high frequency that termination is built into for example DDR2,3,4 and 5 RAMs, being called ODT. On-Die-Termnation, built into the chips themselves. I hope I could help a bit to avoid confusion. Greetings from Germany
Yes, I was also a bit puzzled why such a relatively now speed bus would require terminaton as I understood it in a more HF sense. But of course with lines being controlled by open-collector outputs, it makes sense that they need pull-ups. It's possible that eben in the official TRS-80 terminology they were called terminators, even though they don't serve the same purpose as let's say the 50 ohms "end cap terminators" did on coax thin Ethernet.
@@Colaholiker That would be a good example! I am a bit surprised to see the 150 Ohm value in the video. I am more familiar with 1k-4.7k pull-ups, but the 150 Ohm one could be owed to using the "without letters"-7438. I can however see why Adrian mentioned that issues were more common if the cable length was >10 ft (3m). I suspect this may be where a reflected signal can coincide with the next rising or falling signal flank and add up to something substantial that can confuse such simple old logic. I personally have enver seen such issues myself, since those flat floppy cables in my Amigas and PCs were never longer than maybe 75 cm.
Remember, with reflections, the transfer speed is not as important as the transition time (low to high or high to low.) Sharp transitions are what cause reflections, and with fast data transfer, you have to transition quickly because there’s no time to dawdle! ;-) But even “slow” busses may have fast slew rates, and thus, issues with reflection. Of course, cable capacitance and inductance also affect signal integrity. Termination and pull-up resistors aren’t inherently discrete topics. The world tends to be messy sometimes, and this is one of those. There are many ways to terminate signal lines. One is through pull-up resistors. Another might be to pull down to ground (which wouldn’t work for open collector), or between positive and negative differential lines.
@@amurtigress_mobile365 Right, 150 Ohm is a pretty strong (low resistance) pull-up, but if the logic gates there were still bipolar and not field effect, they sure needed some current and not just a voltage level. However, the entire subject was new to me, as I have never worked with 8" floppies, and with 5.25" and 3.5" this never was an issue. Even though I vaguely remember that our first PC (some Amstrad XT class machine) with an external 5.25" floppy had issues running the intnernal 3.5" floppy when the external drive was disconnected. But I was too young to understand the details back then. To me, the most fascinating part about the 8" floppy drives is that the spindle motor ran all the time and was mains powered. Wow.
Never encountered termination with floppy drives. But did with SCSI. By that time most SCSI devices just had a termination switch, but I do have an external case for SCSI which has the resistor pack.
Yeah me either, only encountered termination on BNC ethernet networks, some kid always stole the terminator which stopped a whole bank of machines working at school. The other place is with DMX for stage lighting, but mostly it's unnecessary for short runs, and we generally run multi universe over artnet now as it's more reliable than one massive chain - one cable fault and you lost the whole rig in that scenario. Now we would just lose the fixtures in that one universe.
Imagine if the ending of the original Terminator were modified slightly: As the limping remains of the cyborg reach out for Sarah at the dead end, she notices a slightly futuristic SCSI terminator, places it on the cyborg and says, "You’re terminated, m…"
I’m really glad to see some of these in-depth videos. I am really getting anxious that some of this practical knowledge is going away, since a lot of the documentation and discussion of this old tech was pre-Google. The gray-beards who knew how it worked aren’t going to be here forever.
Listen, and understand. That terminator is out there. It can’t be bargained with. It can’t be reasoned with. It doesn’t feel pity, or remorse, or fear. And it absolutely will not stop, ever, until your transmission line is pulled up to 5 volts
I've been dealing with floppy drives, cables and controller cards since my first XT, and this is the first time I hear of "termination". they just worked. (I've used 5.25 and 3.5 drives, I've never even seen a 8 drive in reality!). this was fascinating. great video.
New PC drives usually "terminate" with fixed 1kOhm or something like that, which is fine for short cable lenghts and doesn't load the controller too much with the usual up to 2 drives. I have an old full-height drive out of an IBM 5150 right beside me though, which DOES HAVE a removable resistor pack on it.
@@pelgervampireduck Modern SCSI is a completely different animal, but from vintage computing, the basics are thus: It's a daisy-chainable system. There's one controller and 7 other devices (hard drive, zip drive, tape drive, CD-ROM, TWAIN scanner, etc), for a total of 8 devices over a 50-pin cable. You set jumpers to assign the ID number (0-7) of each device. At each end of that long ribbon cable, either the last device has a terminator resistor pack installed (or enabled by jumper) or you have a terminator resistor pack that plugs directly into the cable. Usually the controller will be set as ID-0 or ID-7, but it's completely arbitrary what you set it to. You don't even have to have them sequential, which means that you're free to route the ribbon and mount the drives however is convenient to you, and the ID's each device are set to will determine which order the devices show up in. You could have the controller connected in the middle of the ribbon if you want (you'd have to take the jumper to disable the onboard terminator), and it works just fine. Macintosh computers from the Plus in '86 until just before the i-Mac in '98 came equipped with SCSI controllers (internal 50-pin, and external DB-25), which was extremely convenient for chaining in extra drives or other devices. They were also the primary storage standard used in servers and workstations throughout the 90's, but they weren't too common in the home PC market, since IDE drives were much cheaper.
For those that don't understand the difference between pull-up and termination for transmission line characteristics, this is a very old (but good) video from AT&T which describes wave propagation and termination in a physical way. ua-cam.com/video/DovunOxlY1k/v-deo.html (actual old video starts at 1:30 or so)
the 150 ohm in the spec would be the lowest resistance that was supported by the gate output current sink. higher resistances would work fine, for example 300 ohms or 2 of 300 ohm in parallel.
The UK army was using 8inch disks for archive in their Wavel battlefield communication system. Along with 1mb of mos memory and 2 x 24mb of bubble memory. And a 75bwd teleprinter for input output. Oh those were the days.
Yeah, I've been caught by open collectors not being pulled up before. I was really confused the first time I used a comparator, it just wouldn't work. So I checked the datasheet and noticed the pullup resistor, so I tried adding one to my circuit and it magically started working.
I remember having to do a lot of this when putting an external 8” drive on an ibm AT we got a external ribbon cable with a termination block that was sealed. Nightmare getting it to all work.
I have a Zenith Z-161 which is a XT class luggable from 1984. It has two Mitsubishi M4851 half height double sided double density 360k floppy drives. They have jumpers for connecting the signals to termination resisters on the drive and the last drive must be terminated for the drives to function. So some early PCs did indeed require termination of floppies.
Related topic: Let’s all hold hands and repeat the chant; 90% of all SCSI communications issues are termination related… 90% of all SCSI communications issues are termination related…
Probably you can get away with fuzzy termination because the data rate is so slow. UltraSCSI no way! Very interesting - I love the electronics videos! Keep'em coming!
I=V/R. When that signal is set to a low there would be ~50mA for a 100Ohm pullup resistor or ~33mA for 150Ohm. Times this by however many data lines require terminating, quite a bit of power when you think about it.
When this was defined, most microcomputer chips were still NMOS or even older, even less effective PMOS. In this type of logic, EVERY SINGLE GATE is basically made open-drain with an integrated pullup "resistor". (Actually, it's slightly better than this because it's a FET configured as a constant current source .. that's why old chips often needed +12V .. to bias those CCSs. Later they could drop that requirement by being able to integrate depletion mode FETs there.) So, those things were always pulling current, and quite a lot of it. A simple Z80 computer with NMOS CPU and NMOS memory, accompined by LS-TTL logic, could easily draw a few amps on the 5V line. Nobody cared a bit about those few milliamps there because of this. "Power comes out of the outlet"
The resistors may be on the reciever side to deal with supply domains. The reciever may not like to get voltage on the data-pin while it's without power (or above supply on the data-pins). By having the pullup/termination to the same supply that won't happend. Take the case of unpowered drive connected to powered computer, the signals from the computer won't be high level (thus no powering the drive via data-pins). The open collector outputs is just a transistor and should hopefully have no problems with voltage on the pins without suppy. There is also the case of for example the drive having a 5.5V supply and the computer having 4.5V, no reciever will get a input voltage above it's supply, keeping the system safe. Aka how to not blow up stuff!
Note about "fixing incorrect terminations on your controllers". That could be bad idea, if you are not extactly understand how these things work, especially, if connection is not actually broken.
💪😁🔥🔥🔥Cool video... as always! Btw, how much time do you spend on preparing a typical episode? I'm trying to do sth similar but as for now it's terribly time consuming. Perhaps it's because I'm making my first steps...
Some computers like the IBM 5110 had the floppy drives connected through a very long cable. The computer sat on the desk and the drives were in a separate three foot tall metal box that sat on the floor that you usually put the printer on. I can see how termination would be an issue with a long cable like that.
That's could be not related to "short PC cables", but with different ways of termination. I heard something about "distributed" termination (NB: it's translated word, not original english term).
Fixibg the controllers termination seems like something you'd only need to do if you are using old drives that are built to the shugart spec. Newer drives would probably expect the not to spec termination. Would be interesting to see how the termination differs based om age of drive and the platform
In the early 90's I hooked up a second SCSI drive to my Amiga and couldn't get it to work, nearly sent it back thinking it was defective.. took it into a local shop and the guy laughed, moved the resistor pack from the original drive to the new drive and said good to go. He never explained it beyond "this has to be on the last drive in the chain". I was never a PC guy in those days and went right into Linux on PCs in 99 and had no idea this was also needed on floppies, very interesting! Did MFM/RLL/IDE/PATA/SATA etc drives ever need anyone to "manage" the termination?
I've been a "PC guy" since the 80s and this is the first time I hear of this. I'm in shock. I've used all kinds of drives, cables and controller cards, and they just worked. I didn't know this was a thing.
MFM/RLL: The 34pin cable for both disks on a controller (quite similar to a floppy interface .. pinout is moved all around and instead of a side select there are 4 head select lines and of course no data pins anymore) .. yes .. hard drives had a little resistor pack you had to remove at the one in the middle of the cable if you had two. It's just that nobody remembers, since almost no one could afford more than one of them back then. The 20pin cable per disk was a point-to-point differential signal, so no user-managable termination needed. PATA: Uses series termination on the controller side and quite strictly limited cable lengths. SATA: Differential point-to-point connection, therefore fixed termination resistors in the receivers.
I did not understand termination prior to this video. After watching it, I understand it even less. (This isn't a fault in the video, by the way. I just struggle with this stuff) Also: petition to nickname the TRS-80 Model II resistor pack "Arnold".
Boy, I am puzzled. I worked with hundreds of 5.25" and 3.5" floppy drives, and I don't recall doing termination on them. Bad memory, or lucky they worked? I will have to look at some of my PC's I still have. Great episode.
I feel the same, I didn't know this was a thing. I'm in shock. it's like a "mandela effect", hahahaha. I've done everything you can think of with 5.25 and 3.5 drives, cables, controller cards, and I never heard of this!.
I’ve only recently been aware of it because of working with XT and AT systems. I just installed a YE Data YD-380 360KB 5.25” drive in an XT 286 this afternoon, and had to remove the termination pack since it was going in as the B: drive. But two years ago? Didn’t even know it was a thing with floppy busses.
At 5:40... Isn't that long blue component on the drive a resistor pack? I would have assumed that was doing the termination. Maybe it's for something else.
5:47 "There is certainly no termination resistor block on here" - what about the thin blue stripe next to the 3third pin from the bottom? Looks resistor blocky to me...
Terminating with 150 ohms at 5 volts sounds like a huge waste of power by today's standards. That's 166mW burnt for each line each time the signal is driven high.
Pretty common for some of those old standards. Bear in mind VGA uses 75R, and some drivers actually use a second 75R term to ground (not in-line) at the transmit end. That’s 37.5R! But not at 5V. :-)
if putting 2 drives such as 5 1/4s in a system and both are terminated and controller terminating, it may be too much load for a ttl to sink low enough..
Good explanation! I always wondered how termination on floppydrives work as on a PC you normally don't have to deal with it as termination is built in on most drives. Allow me to ask a (stupid?) question: Seeing that Shugart SA800 from your Tandy, I wonder how those wrapped wires make contact with the pins. They're enamel wires aren't they? In the close up shots they look like they're wrapped with the enamel insulation still on them.. Could be me though ;-)
There's also a portion of the wire exposed that is wrapped around the post that is not obvious without close up macro images. The insulated part at the bottom of the pin is a side product of hand wrapping tools
@@allensmithphotography Trust us... the wire wrapping tool strips the insulation from the wire! Lots of fun fixing your wiring, when it doesn't however!
If you wrap a wire around a pin very tight, the pin edges will cut through the insulation without need to remove it first. It is actually rather reliable connection and can be done fast by mechanised and automated tools, so it was commonly used in 70s-80s for high importance equipment.
if you just said "Shugart Drive" I would have bought into the either/or reasoning, But you give a name: Alan Field Shugart (September 27, 1930 - December 12, 2006) I'm having troubles with my sound card since upgrading to Fedora 35, so, just like back in the 80s, I can't confirm "Shoe-gart" as the pronunciation. But there's certainly a proper pronunciation of the surname and in this time of the internet we should take the time and find out what the proper pronunciation is.
Drive geometry of the M2 single sided 8" drive is identical to a 1.2mb 5.25 drive. 8" requires a write precomp signal though, because of the large velocity difference between inner and outer tracks. I am curious to see how he got around that.
As far as I know, the FDC in PCs is actually based on the 765, so there shouldn´t be any poroblems with that. The FDC is simply "hidden away" in a larger IC which also has other functions added to it. For example, there was a time before all the major support functions in a PC were consolidated into the North Bridge and South Bridge, when SuperIO chips were fairly common; these ICs typically provided interfaces for: up to 4 Floppy Drives, up to 2 IDE HDDs, dual serial port, and a parallel port.
@@Lee_Adamson_OCF There is also the !TG43 (Track greater 43) signal. Not all drives used it, but some reduced the write current on the inner tracks with it. Some also did keep track on which track they are internally and created it by themselves.
![Adrian's Digital Basement ][](http://yt3.ggpht.com/ytc/AIdro_k7CrQRe71qN6c0Ua7A1P_I8KkxjqLa70u0zYeCvdvnKw=s48-c-k-c0x00ffffff-no-rj)
Hello folks,
as an old user of such tech I am a bit concerned that two things may be confused: Pull up resistors and termination. Often those went alongside, but later on faster bus systems, termination increasingly became the standalone purpose.
1) Pull-up resistors: Will be needed on Open Collector outputs like the 74(LS)38. May also be needed on non OC outputs, depending on the load (inputs per output). On Open Collector outputs you get the advantage of connecting several outputs in parallel, as Adrian said. But essentially, the open collectors in parallel work like a logical Or-Gate without needing extra ICs.
2) Termination: Termination in the strictest sense is used to kill deflections and delayed spikes from open wire ends such as a cable to several disk drives. Imagine such a cable or wire to be a T. A signal coming from a drive goes up to the "fork" and the travels to the left (say, the controller) and to the right (Open cable end). On computers like the Apple 2 you would have such cables hanging out of the back for drives.
The signal part that goesto the open wire end gets reflected and goes back towards the controller after hitting the end of the wire and can cause extra spikes or dips on the wire that can throw the signal out of specs (like between 0 and 1), and by doing so cause data integrity problems. A resistor just sends the signal that went to the open end to ground or to +5V.
I personally know that better from SCSI chains on my Amiga computers, where every drive had termination that could be enabled or disabled, and the termination resistors IIRC always were connected to ground, and the data lines were NOT Open Collector.
Nowadays signals are so extremely high frequency that termination is built into for example DDR2,3,4 and 5 RAMs, being called ODT. On-Die-Termnation, built into the chips themselves.
I hope I could help a bit to avoid confusion.
Greetings from Germany
Das.
Yes, I was also a bit puzzled why such a relatively now speed bus would require terminaton as I understood it in a more HF sense. But of course with lines being controlled by open-collector outputs, it makes sense that they need pull-ups. It's possible that eben in the official TRS-80 terminology they were called terminators, even though they don't serve the same purpose as let's say the 50 ohms "end cap terminators" did on coax thin Ethernet.
@@Colaholiker That would be a good example! I am a bit surprised to see the 150 Ohm value in the video. I am more familiar with 1k-4.7k pull-ups, but the 150 Ohm one could be owed to using the "without letters"-7438. I can however see why Adrian mentioned that issues were more common if the cable length was >10 ft (3m). I suspect this may be where a reflected signal can coincide with the next rising or falling signal flank and add up to something substantial that can confuse such simple old logic. I personally have enver seen such issues myself, since those flat floppy cables in my Amigas and PCs were never longer than maybe 75 cm.
Remember, with reflections, the transfer speed is not as important as the transition time (low to high or high to low.)
Sharp transitions are what cause reflections, and with fast data transfer, you have to transition quickly because there’s no time to dawdle! ;-) But even “slow” busses may have fast slew rates, and thus, issues with reflection. Of course, cable capacitance and inductance also affect signal integrity.
Termination and pull-up resistors aren’t inherently discrete topics. The world tends to be messy sometimes, and this is one of those. There are many ways to terminate signal lines. One is through pull-up resistors. Another might be to pull down to ground (which wouldn’t work for open collector), or between positive and negative differential lines.
@@amurtigress_mobile365 Right, 150 Ohm is a pretty strong (low resistance) pull-up, but if the logic gates there were still bipolar and not field effect, they sure needed some current and not just a voltage level.
However, the entire subject was new to me, as I have never worked with 8" floppies, and with 5.25" and 3.5" this never was an issue. Even though I vaguely remember that our first PC (some Amstrad XT class machine) with an external 5.25" floppy had issues running the intnernal 3.5" floppy when the external drive was disconnected. But I was too young to understand the details back then.
To me, the most fascinating part about the 8" floppy drives is that the spindle motor ran all the time and was mains powered. Wow.
Never encountered termination with floppy drives. But did with SCSI. By that time most SCSI devices just had a termination switch, but I do have an external case for SCSI which has the resistor pack.
I used a "termination dongle" of sorts, a fat pack that went on the end of the SCSI cable.
Yeah me either, only encountered termination on BNC ethernet networks, some kid always stole the terminator which stopped a whole bank of machines working at school. The other place is with DMX for stage lighting, but mostly it's unnecessary for short runs, and we generally run multi universe over artnet now as it's more reliable than one massive chain - one cable fault and you lost the whole rig in that scenario. Now we would just lose the fixtures in that one universe.
Imagine if the ending of the original Terminator were modified slightly:
As the limping remains of the cyborg reach out for Sarah at the dead end, she notices a slightly futuristic SCSI terminator, places it on the cyborg and says, "You’re terminated, m…"
I’m really glad to see some of these in-depth videos. I am really getting anxious that some of this practical knowledge is going away, since a lot of the documentation and discussion of this old tech was pre-Google. The gray-beards who knew how it worked aren’t going to be here forever.
Listen, and understand. That terminator is out there. It can’t be bargained with. It can’t be reasoned with. It doesn’t feel pity, or remorse, or fear. And it absolutely will not stop, ever, until your transmission line is pulled up to 5 volts
I've been dealing with floppy drives, cables and controller cards since my first XT, and this is the first time I hear of "termination". they just worked. (I've used 5.25 and 3.5 drives, I've never even seen a 8 drive in reality!).
this was fascinating.
great video.
Anybody familiar with SCSI knows about termination blocks.
New PC drives usually "terminate" with fixed 1kOhm or something like that, which is fine for short cable lenghts and doesn't load the controller too much with the usual up to 2 drives.
I have an old full-height drive out of an IBM 5150 right beside me though, which DOES HAVE a removable resistor pack on it.
@@BlackEpyon
I never had to install or replace a SCSI drive. I know they exist, but I don't have experience with those.
@@pelgervampireduck Modern SCSI is a completely different animal, but from vintage computing, the basics are thus:
It's a daisy-chainable system. There's one controller and 7 other devices (hard drive, zip drive, tape drive, CD-ROM, TWAIN scanner, etc), for a total of 8 devices over a 50-pin cable. You set jumpers to assign the ID number (0-7) of each device. At each end of that long ribbon cable, either the last device has a terminator resistor pack installed (or enabled by jumper) or you have a terminator resistor pack that plugs directly into the cable. Usually the controller will be set as ID-0 or ID-7, but it's completely arbitrary what you set it to. You don't even have to have them sequential, which means that you're free to route the ribbon and mount the drives however is convenient to you, and the ID's each device are set to will determine which order the devices show up in. You could have the controller connected in the middle of the ribbon if you want (you'd have to take the jumper to disable the onboard terminator), and it works just fine.
Macintosh computers from the Plus in '86 until just before the i-Mac in '98 came equipped with SCSI controllers (internal 50-pin, and external DB-25), which was extremely convenient for chaining in extra drives or other devices. They were also the primary storage standard used in servers and workstations throughout the 90's, but they weren't too common in the home PC market, since IDE drives were much cheaper.
Loved this and the companion video. Can't have enough disk drive operation info, especially across systems!
For those that don't understand the difference between pull-up and termination for transmission line characteristics, this is a very old (but good) video from AT&T which describes wave propagation and termination in a physical way. ua-cam.com/video/DovunOxlY1k/v-deo.html (actual old video starts at 1:30 or so)
the 150 ohm in the spec would be the lowest resistance that was supported by the gate output current sink. higher resistances would work fine, for example 300 ohms or 2 of 300 ohm in parallel.
150Ohms is also quite near to the characteristic impedance of the flat cable, which is usually said to be ~133 Ohms.
Ahhh cleverly simple solutions of yesteryear.
The Model II certainly is a nice machine, right through to the drive.
The UK army was using 8inch disks for archive in their Wavel battlefield communication system. Along with 1mb of mos memory and 2 x 24mb of bubble memory. And a 75bwd teleprinter for input output. Oh those were the days.
You missed out on a golden opportunity to use an Arnold Schwarzenegger impression to do a video about terminating floppy drives. :D
Yeah, I've been caught by open collectors not being pulled up before. I was really confused the first time I used a comparator, it just wouldn't work. So I checked the datasheet and noticed the pullup resistor, so I tried adding one to my circuit and it magically started working.
I remember having to do a lot of this when putting an external 8” drive on an ibm AT we got a external ribbon cable with a termination block that was sealed. Nightmare getting it to all work.
That made pretty much perfect sense IMO.
This is a great lesson in scale first of all.
Fascinating and enlightening as always. I didn't understand termination until now.
I have a Zenith Z-161 which is a XT class luggable from 1984. It has two Mitsubishi M4851 half height double sided double density 360k floppy drives. They have jumpers for connecting the signals to termination resisters on the drive and the last drive must be terminated for the drives to function. So some early PCs did indeed require termination of floppies.
Thanks for this, I got a model II without the terminator pack, so this is very helpful.
Related topic: Let’s all hold hands and repeat the chant; 90% of all SCSI communications issues are termination related… 90% of all SCSI communications issues are termination related…
Thank you for this, never understood termination as it is older then me
Unfortunately, this video doesn't explain termination, either. Check Amurtigress' comment.
Probably you can get away with fuzzy termination because the data rate is so slow. UltraSCSI no way! Very interesting - I love the electronics videos! Keep'em coming!
A great explanation!
It reminds me the n64 jumper pack. If your expansion pack was not installed you needed that jumper to use the console.
I=V/R. When that signal is set to a low there would be ~50mA for a 100Ohm pullup resistor or ~33mA for 150Ohm. Times this by however many data lines require terminating, quite a bit of power when you think about it.
When this was defined, most microcomputer chips were still NMOS or even older, even less effective PMOS. In this type of logic, EVERY SINGLE GATE is basically made open-drain with an integrated pullup "resistor". (Actually, it's slightly better than this because it's a FET configured as a constant current source .. that's why old chips often needed +12V .. to bias those CCSs. Later they could drop that requirement by being able to integrate depletion mode FETs there.)
So, those things were always pulling current, and quite a lot of it. A simple Z80 computer with NMOS CPU and NMOS memory, accompined by LS-TTL logic, could easily draw a few amps on the 5V line. Nobody cared a bit about those few milliamps there because of this.
"Power comes out of the outlet"
The resistors may be on the reciever side to deal with supply domains.
The reciever may not like to get voltage on the data-pin while it's without power (or above supply on the data-pins). By having the pullup/termination to the same supply that won't happend.
Take the case of unpowered drive connected to powered computer, the signals from the computer won't be high level (thus no powering the drive via data-pins).
The open collector outputs is just a transistor and should hopefully have no problems with voltage on the pins without suppy.
There is also the case of for example the drive having a 5.5V supply and the computer having 4.5V, no reciever will get a input voltage above it's supply, keeping the system safe.
Aka how to not blow up stuff!
fasinating. I never knew that about floppy drives.
Note about "fixing incorrect terminations on your controllers".
That could be bad idea, if you are not extactly understand how these things work, especially, if connection is not actually broken.
Here it is the floppy drive terminator theme : ua-cam.com/video/R_2fVhArvk8/v-deo.html
The thumbs down are from people who wish they had an "8 inch floppy."
I have three.
..... But none of them "work". :(
Conflating termination resistors with pull-up resistors.
it's confirmed, Adrian has an 8 inch disk
Multiple of them. And quite a few slits to stick them in by now.
💪😁🔥🔥🔥Cool video... as always! Btw, how much time do you spend on preparing a typical episode? I'm trying to do sth similar but as for now it's terribly time consuming. Perhaps it's because I'm making my first steps...
ooooO~ Id wondered about this for years! Thank you. :)
Some computers like the IBM 5110 had the floppy drives connected through a very long cable. The computer sat on the desk and the drives were in a separate three foot tall metal box that sat on the floor that you usually put the printer on. I can see how termination would be an issue with a long cable like that.
That's could be not related to "short PC cables", but with different ways of termination. I heard something about "distributed" termination (NB: it's translated word, not original english term).
The only thing missing from the size comparison at the beginning was a banana.
Whose banana, though?
Fixibg the controllers termination seems like something you'd only need to do if you are using old drives that are built to the shugart spec. Newer drives would probably expect the not to spec termination. Would be interesting to see how the termination differs based om age of drive and the platform
Hard to believe this thing is predecessor of T-1000.
In the early 90's I hooked up a second SCSI drive to my Amiga and couldn't get it to work, nearly sent it back thinking it was defective.. took it into a local shop and the guy laughed, moved the resistor pack from the original drive to the new drive and said good to go. He never explained it beyond "this has to be on the last drive in the chain". I was never a PC guy in those days and went right into Linux on PCs in 99 and had no idea this was also needed on floppies, very interesting! Did MFM/RLL/IDE/PATA/SATA etc drives ever need anyone to "manage" the termination?
I've been a "PC guy" since the 80s and this is the first time I hear of this. I'm in shock.
I've used all kinds of drives, cables and controller cards, and they just worked. I didn't know this was a thing.
MFM/RLL: The 34pin cable for both disks on a controller (quite similar to a floppy interface .. pinout is moved all around and instead of a side select there are 4 head select lines and of course no data pins anymore) .. yes .. hard drives had a little resistor pack you had to remove at the one in the middle of the cable if you had two. It's just that nobody remembers, since almost no one could afford more than one of them back then. The 20pin cable per disk was a point-to-point differential signal, so no user-managable termination needed.
PATA: Uses series termination on the controller side and quite strictly limited cable lengths.
SATA: Differential point-to-point connection, therefore fixed termination resistors in the receivers.
I did not understand termination prior to this video. After watching it, I understand it even less.
(This isn't a fault in the video, by the way. I just struggle with this stuff)
Also: petition to nickname the TRS-80 Model II resistor pack "Arnold".
A too long video? Does not compute. It's like too much money in the bank, or too tasty food. 🤷♂
That is weird as I had an AMSTRAD PC1512 with the UPD765A and I didn't notice any termination resistor pack and this machine used 5.25" discs.
Boy, I am puzzled. I worked with hundreds of 5.25" and 3.5" floppy drives, and I don't recall doing termination on them. Bad memory, or lucky they worked? I will have to look at some of my PC's I still have. Great episode.
I feel the same, I didn't know this was a thing. I'm in shock. it's like a "mandela effect", hahahaha.
I've done everything you can think of with 5.25 and 3.5 drives, cables, controller cards, and I never heard of this!.
I’ve only recently been aware of it because of working with XT and AT systems. I just installed a YE Data YD-380 360KB 5.25” drive in an XT 286 this afternoon, and had to remove the termination pack since it was going in as the B: drive.
But two years ago? Didn’t even know it was a thing with floppy busses.
At 5:40... Isn't that long blue component on the drive a resistor pack? I would have assumed that was doing the termination. Maybe it's for something else.
i used to really like those huge floppies when i was a kid... i thought they were really important
So, an open-collector output is basically a toggle switch? open/closed?
5:47 "There is certainly no termination resistor block on here" - what about the thin blue stripe next to the 3third pin from the bottom? Looks resistor blocky to me...
Terminating with 150 ohms at 5 volts sounds like a huge waste of power by today's standards. That's 166mW burnt for each line each time the signal is driven high.
Pretty common for some of those old standards. Bear in mind VGA uses 75R, and some drivers actually use a second 75R term to ground (not in-line) at the transmit end. That’s 37.5R! But not at 5V. :-)
I was expecting to see video of a T1000 executing a Connor drive. ;)
4:26 Analog voodoo 😄
if putting 2 drives such as 5 1/4s in a system and both are terminated and controller terminating, it may be too much load for a ttl to sink low enough..
Good explanation! I always wondered how termination on floppydrives work as on a PC you normally don't have to deal with it as termination is built in on most drives.
Allow me to ask a (stupid?) question: Seeing that Shugart SA800 from your Tandy, I wonder how those wrapped wires make contact with the pins. They're enamel wires aren't they? In the close up shots they look like they're wrapped with the enamel insulation still on them.. Could be me though ;-)
There's also a portion of the wire exposed that is wrapped around the post that is not obvious without close up macro images. The insulated part at the bottom of the pin is a side product of hand wrapping tools
@@allensmithphotography Trust us... the wire wrapping tool strips the insulation from the wire!
Lots of fun fixing your wiring, when it doesn't however!
@@tekvax01 not all wire wrap tools do that. I know mine doesn't.
If you wrap a wire around a pin very tight, the pin edges will cut through the insulation without need to remove it first. It is actually rather reliable connection and can be done fast by mechanised and automated tools, so it was commonly used in 70s-80s for high importance equipment.
Tomatoes, Tomato''s on the Alan Shugart pronunciation. I've always heard it pronounced more of a Shoe gart going way back to the 80's.
if you just said "Shugart Drive" I would have bought into the either/or reasoning, But you give a name: Alan Field Shugart (September 27, 1930 - December 12, 2006) I'm having troubles with my sound card since upgrading to Fedora 35, so, just like back in the 80s, I can't confirm "Shoe-gart" as the pronunciation. But there's certainly a proper pronunciation of the surname and in this time of the internet we should take the time and find out what the proper pronunciation is.
Don't you need a PC with a UPD765A FDC controller chip in it to make that 8" Floppy Drive work.
Adrian's video on the main channel explains how he got it working.
Drive geometry of the M2 single sided 8" drive is identical to a 1.2mb 5.25 drive. 8" requires a write precomp signal though, because of the large velocity difference between inner and outer tracks. I am curious to see how he got around that.
As far as I know, the FDC in PCs is actually based on the 765, so there shouldn´t be any poroblems with that. The FDC is simply "hidden away" in a larger IC which also has other functions added to it. For example, there was a time before all the major support functions in a PC were consolidated into the North Bridge and South Bridge, when SuperIO chips were fairly common; these ICs typically provided interfaces for: up to 4 Floppy Drives, up to 2 IDE HDDs, dual serial port, and a parallel port.
@@Lee_Adamson_OCF There is also the !TG43 (Track greater 43) signal. Not all drives used it, but some reduced the write current on the inner tracks with it. Some also did keep track on which track they are internally and created it by themselves.
Dude! No backtrack link back to the main video series????!!!!!! 🤦🏻♂️🤦🏻♂️🤦🏻♂️
2
Did you just say "clear F'in volts or zero" at 7:22? I hope you didn't do this without marking your channel Not For Kids.