it gets really tedious if things go pear shaped and one is running out of ideas what's wrong. and even the community, too. Props for your patience and not losing motivation!
I built a 1581 replica using the same PCB you have in this video (from PCBWay designed by Thomas Christoph in Canada). Mine was exhibiting similar issues, random working and then not. What I found after staring at the PCB for seemingly hours was that I installed one of the RP1/RP3 resistor networks the wrong way around. Pin 1 on each are towards the outside edge of the PCB. I originally had both in the same orientation.
Hi. I just finished 1581 today. I struggled with this project for two weeks because it didn't work, but now it works 100%. One binding finding. An inverter must not be used to power this device, but a classic transformer and rectifier. The converter causes a lot of communication errors both on the 1581 board and on the serial line. The power supply during operation must not drop below 5 Volts.
I like that you show that things go wrong sometimes. Almost all RUclipsrs don't. I had plans to build a 1581 myself, but now I'm hesitant to do so. Super thanks for this video. Has saved me both time and money.
I will always leave in the mistakes, even if it makes me look like a total idiot (plus, it might help someone to not make the same mistake), and also if I break something or rip off a trace etc... I don't strive for "perfection", just what most people can achieve - you get what I did, warts & all. Mostly there is no possibility for a "take 2". I had enough footage of the drive working that I COULD have edited together in such a way that it looked like the drive just worked and always worked. But that would be misleading, and I'd never do that. I'm still annoyed that things only seem to let out the magic smoke AFTER I turn the camera off though, lol
I had to use the adapter board for the floppy you pictured - and all LS and 7406/7407 ICs (no tolerance for any variations in chips) on the DIYChris board. It gets along fine with two other 1571 drives on my C128 with a Panasonic JU261 drive in it. The light does stay on; that's a way the floppy light is wired - just unsolder the LED - you're supposed to reply on the Commodore lights. Stay with in tolerance spec value components as well. It's a good kit and works well, and the first time I built it the issues were all mine (bad 8520, bad WD IC - I got a chip tester after that.)
Sorry things didn't work out from the start - that is really frustrating. I hope collectively you are able to diagnose what's wrong and get all of your 1581s working. I've ordered a replacement drive for my unit, but I believe my particular issue was self-inflicted (reversed the 12 and 5v lines going to the floppy drive, whoops!!). Will keep you posted when I know more!
Please try to increase the supply voltage. Use a bench power supply to see what your voltages are. Increase in steps of 100mv until it works or until it is almost dangerous. I'm almost sure this is a voltage issue.
Really stupid question to ask, but I noticed that when you power on the plus-4 first and then power up the 1581 second you always get errors (as if the serial bus is not being initialised correctly). But when you power up the 1581 first and the power up the plus-4 you seem to get it to work - which shows that the 1581 is powering up and running on its DOS and when you turn on the plus-4 that will re-initialise the serial bus thus putting the plus-4 and 1581 in a good state - as doing a directory after that power up sequence seems to show a directory correctly. But when you eject the disk and put in a new disk does the 1581 re-initialise the serial bus, as something must happen to tell the 1581 that the disk as been swapped. It might be a good idea to put a scope on the serial bus between the plus-4 and 1581 and see what happens when you eject a disk and put in a new one.
That's an interesting observation, and worth investigating further. Nothing happens on the serial bus, since communication is always instigated by the bus master (ie the computer), not the slave (ie 1581). The problem is that sometimes the 1581 knows the disk has changed and sometimes it doesn't. This is all internal to the 1581.
I'm curious about your ground routing, also what is the noise polution like, as in are there certain days that are much noiser? If atmospheric noise is not a problem then I would suspect connectors and cables may have intermitten issues.
Man! When I remember these cursed projects in 80's electronic magazines. No circuit diagram was correct, no pcb layout was correct, Homebrew pcb's, holes drilled by hand. Not enough money to buy ic sockets ... Brings back nemories.😁
Thank you for your in-depth testing of this kit, Tim! I built a similar kit a little over a year ago and could not get it to reliably work with my C16. I do not have a C64 to test with, and I'm not sure if that is a factor; after all, the I/O DOS is the same in both machines. I initially had several issues with bad I/Cs (verified with a BackBit Chip Tester): bad 8520, bad 65C02 (and later a mis-labeled 6502A). The USB cables and adaptors I used to power the device also introduced too much of a voltage drop leading to instability in the power source. However after rectifying those issues, I am currently in the same state that you presented in your video: the drive will not reliably retrieve data. Periodically, a directory listing or file retrieval is truncated and the drive hangs requiring a reset. I have tested with the stock and JiffyDos ROMs with the same results. I have not tested the disk changing issues you experienced, but given that our builds share the same issues, I'm certain it will occur on my unit as well. The only components on the PCB that I was dubious about when purchasing were the ferrite beads. I bought ones that best matched the diameter and length of the original specs but seeing as the BOMs for this build state something like "just get them from a spare C64", I'm uncertain that I installed the correct values. Thank you again for documenting the faults. Hopefully someone who has gone through this build and has also corrected these problems can share their remediation procedure!
This is quite interesting, particularly because I have an original 1581 which exhibits several of the same symptoms. It has all its original ICs as far as I know and the original Chinon drive as well. I have been meaning to investigate it closer at some point, so I hope you get some clues to what would be causing this.
You mentioned that Captain Commodore got a Go-Tek to work well. That would suggest that the power consumption of the physical drive is a lot higher on the reproduction boards or that the voltage regulation is out. it would definitely be worth scoping the voltages and power draw to compare the physical versus SD-Card based drives. Is there a newer, more efficient, pin-identical voltage regulator that doesn't require so much cooling? In my limited experience of mid-90s PC components, a single component running too hot would often bugger communications between more sensitive items on the board.
I'm not in a position to test it right now, but Captain Commodore's figures are 570mA with the drive, and 290mA with the Gotek. My power supply (CD32) is rated at 2.2A on the 5V line. Neither drive uses 12V. The official 1581 power supply I think was 5V 1A, 12V 0.5A
Hello Tim. Very good video. I believe you didn't do anything wrong. My suspicion is that the problem lies in many revisions of the PC drive. Over the years, manufacturers have continually improved the floppy drive electronics. Your Samsung SFD-321B seems to be a REV.T6. I use a REV.T1 and it works well. The floppy drive adapter isn't bad, but you can do without it. The older revisions of the drives can be easily modified so that it is not needed. There are many instructions on the internet (Amiga modification).
So I'm using a TEAC Amiga drive (taken from my A500) and a modified 321B (unknown revision since it was missing the top plate). Both still exhibit the problem. The two PC drives are Revs T5 and T6. Retrobits (see stickied link) is using a rev M2, but on a Toms01 board. Revision being a factor is an interesting idea, but it's a very small sample size at the moment, and without knowing the differences... Since the adapter is performing the same function as the conversion, what makes the conversion "better"? What does it do that the adapter doesn't?
I also have a Toms01 board. I first tested it with the adapter and an ALPS DF354H168F drive. That did not work. With a DF354H090F it worked, including changing the diskette. I found parts of a damaged original case. The adapter doesn't fit in there very well because of the shielding plate. I know from pazuzu72 that the adapter is not necessary. The conversion is easier with Samsung drives. You don't have to remove any of the controller's legs or cut any traces. The fact that older drives work better is just a suspicion that I can't prove. My adapter was designed by androSID & toms01. He does exactly the same thing. The difference is that an SN74HCT02 is installed on it. @@TimsRetroCorner
@@black_forest_76 Because I don't have a shielding plate with my 3d printed case, the adapter not fitting is something I hadn't considered. It's a very good reason for doing the conversion instead! 👍 Anecdotally at least, it's looking like Toms01 is the most reliable of all the replica boards. (my adapter is also AndroSID & Toms01). Also anecdotally, converted drives work better than adapted, which suggests the conversion does something that the adapter does not. But I don't know what that is yet.
Hi Tim, I have a 1581 replica myself. The drive is always selected as the drive select signal is directly connected to GND. In the original 1581 you won't recognise this as the drive itself used there has no LEDs. The drive you use is a Samsung SFD-321, right? I also use this drive for my replica. There are different versions, later ones are cost reduced to the max. One of the two drives I own even is hardwired to HD and does not have the HD detection switch. When I remember correctly I used my second drive as reference and boched the PCB to have it permanently set to DD. Using the drive with the wrong settings (presets for the heads like magnetic field strength, filter setting of the head amplifier etc.) can lead to read/write errors. And there's more: In my drives some of the switches had contact problems so that a closed switch had a resistance of several kOhms. If that's the case with the disk presence switch or the write protect switch you can have all sorts of funny issues like disk changes not detected at all or permanently (I don't know by heart which state the switches have with an inserted disk or with no write protection). So it's definitely worth to check a) if your drive has an HD detection switch and even if it has to think about hardwiring it to DD and b) to check if the switched are working properly. And a final point: These drives use 5V only as you noted correctly. In general the power supply scheme of the 1581, the Amiga and most likely the C64 is quite poor in the way that the 5V rail is directly powered from an external supply over long cables while the currents are quite high. When the drive is stepping or starting the motor this can lead to voltage dips. I noticed that you use quite thin and long cables to you bench power supply. I can imagine that this leads to malfunctions while a Gotek drive that mainly needs to power a 3.3V microcontroller is a much lower load than a real drive. So you could/should check the power rail with an Oszilloscope directly on the PCB. P. S. As I have no spare custom chips I bought a ready built PCB as nowadays chances are quite high to get fake or defective parts.
Thanks. I hadn't been aware there were different versions of the SFD (someone else mentioned they spotted mine is a Rev 8) I'll have to look into what the differences are and how they might impact it working.... I did also notice that there are other PC drive adapter boards (on ebay etc) that have extra smoothing caps on the supply lines. Maybe that is beneficial.
I've had all of the same problems and I'm still dead in the water on this build. I have tried three different floppy drives, two are different PC drives, one is an Amiga 500 drive. I have also tried multiple different IC's, different ROM's, different SRAM, different CIA, different floppy controller 1770 vs 1772. All a failure. I just get a drive that spins forever and can't read anything. It also locks up my Commodore Plus/4 and causes it to hang and I have to power it off to free the computer. I've also tried 74LS vs 74LSHCT logic and that didn't work. I can't see how anyone would have managed to build this and get it working and stay working. If they did, then I'd sure like to know how because for me, it's been nothing but endless frustration as well as a lot of money sunk into this project.
The resistor packs there are 3 of them. Make sure you got the right values in the correct positions 1 is a different value. Kit works perfectly with either 6502 or 65c02. I think you have on in the wrong position. Or you used 3 of the same value.
Nope. 2K7 on the CPU pullups, and 1K on the bus pullups. Exactly as per. Triple checked. The second board even used discrete 1K resistors instead of the packs.
For a Gotek, you will need to buy and flash the HxC Firmware to the device as for the free "Flashfloppy" does not support both d64 or d81 files.... Well, it is actually free for a period of time, but normally it costs you 10 Euros...
I can now report that after following the instructions provided by Retrobits, I now appear to be having a reliable working drive. I removed the PC adapter board completely and did the modifications on my Samsung SFD-321B. The only other odd thing I encountered is "device not found". Even though the instructions I could find online for a 1581 said both switches in the rear need to be UP for device #8, I ended up only able to get it to work with both switches DOWN. I'm not sure where the discrepancy here is.
I think it depends on the actual switches you have - some are down for 'on', others are up. Glad you have yous working. Can I ask which PCB you are using, and also what is the make/revision of the drive mechanism? (it'll be on the label)
Man that Sucks you are having a problem I built 2 of them and they just worked. the only problem I had was with HD Disks and the drive density switch I had to jump the switch so it thinks all disks are DD
There should be no problem using a 6526A - it's only different in terms of the TOD clock, and that's not an issue for the 1581. In fact, you may even get away with a straight 6526 (I did), but that's not guaranteed to run at the higher speed.
No they don't. The twist is there in PC drives because they default to 'drive 1' so need the signals swapped, but Amiga drives are all 'drive 0' so don't need the twist.
have you tried testing continuity of the /RDY and /disk_chng signals? ideally from the back of the connector on the drive to the top of the pins on the 8520/6526
You said you changed out several chips, also the 1770 with a 1772, you know these have different default step rates. You can run them but they work better with modified firmware.
Weird that it's so quirky. I have been accumulaiting the kit to build some of these. I'll have to try it and see what happens.Thanks for a great video about it, now I'll go in eyes wide open.
You don't have to close the line ever time you run one of those disk commands. Why do you think you have to keep closing after opening? Why not just leave it open and then continue to use "print#15..."?
@@TimsRetroCorner: But why would you have even generated that habit in the first place, instead of closing after a sequence (or never, since I can't think of a time when we really would need to close; I'd love for someone to make a video that demonstrates any necessity to close those at all)?
I'm not an expert but maybe something slightly off with some timing somewhere as suspected. This old tech was designed with old chip timing difficulties in mind. A few milliseconds difference between old and new chips makes a difference.
How does this board differ from a salvaged 1571? I believe the 1571 and 1581 use the same circuit, only a different floppy mechanism. David Murray (The 8-bit Guy) built a 1581 using a 1571 board and an Amiga floppy and it worked without a problem.
A 1571 is very different internally. As with other 5.25" Commodore drives, all the drive electronics are on the main PCB - it directly drives the motors, has all the head amplifiers and signal encoding/decoding etc. The 1581 just talks to a floppy controller IC (WD1770) and lets the drive handle the signals, motors etc internally. You can't plug a 3.5" Amiga floppy into a 1571, at least not without major surgery and a ROM rewrite. I couldn't find a 1581 build video on the 8-bit Guy's channel, maybe you have a link?
@@TimsRetroCorner I remember a video on his channel building a replica but it may have been removed. It's possible he used a NOS case and a 1581 board from a dead drive. As for the 1571: "The drive uses the MOS 6502 CPU, WD1770 or WD1772 floppy controller, 2x MOS Technology 6522 I/O controllers and 1x MOS Technology 6526." So in the very least you can use the chips off a dead 1571 to populate the board, as it uses the same controllers but some more additional logic to talk to the drive heads directly. The only difference would be that the analog board on the drive was pulled and integrated into the 1571 board instead of leaving it on the drive. An MFM interface is very simple and low-level. If you look at this video there's a top view of the 1571 board: ruclips.net/video/hi_3T00MctI/видео.html It looks like the basic circuit matches that of the 1581 but the connector where the floppy would attach is now covered with some potted terminating resistors and probably traces that go off to the left-hand side of the board, which contains the electronics that would usually be on the drive.
@@Stoney3K You are correct - I've even had one apart on this channel! My fault for misremembering (the 1772 I had was a spare for the 1571, I should have remembered that at least!!). Getting old, ho hum!
Almost at the end now. I feel your frustration. I got myself 5 drives only to get one working. The original one in my 386 pc just spontaneously refused to respond. (General failure reading drive A). I know, not C64, but I get it. I have been blessed with an original 1581 but plan to build one or two (just because I can, it’s a hobby ;) ). In relation to that; what filament did you use for the “beige” one? I can’t seem to find a good supply PLA in retro-white. Looking forward to a followup in which hopefully you have found some answers.
The CMD FD-2000 has the same issue of not recognizing when a new disk is inserted. CMD stated that when swapping disks, a reset command must be sent, similar to the "I" command. So perhaps that's an issue with all 3.5 inch drives? (From CMD's Commodore World Issue 24 Pg 16: "If you don't do this, the drive often ends up using incorrect information stored in variables") However it seems to be the least of your problems; fingers crossed you solve this!
"incorrect information stored in variables" is an understatement! Basically the BAM and directory will be from a completely different floppy, so if you write to the drive you'll corrupt the disk. I did try the "I" command, sometimes it worked, sometimes it didn't. Especially it didn't if the drive was throwing 74 Not Ready errors.
Have you tried it with the floppy drive on its back to get it away from the board ? If something is marginal on the board, like say a missing or badly soldered pull up resistor, the EMI from the drive mech itself could explain the seemingly random behaviour.
Yep, tried it all ways up, down, sideways. Happens on multiple boards and at least 4 builders. Even more going by the comments. So it seems to be something endemic, not just a bad joint.
@@TimsRetroCorner endemic to the recreation? you mentioned replacing 74HC with HCT parts. The original C= design used only 74LS apart from the 7406/7407 line drivers. How much of your (and others?) board has HCT rather than LS parts? For q proper design they are largely interchangeable and most people dont think twice. This all falls apart rapidly if there are missing pullups on an input somewhere.. A recreation of a design could easily miss a pullup resistor and with 74LS would probably just work fine as 74LS floating inputs tend to read high. 74HC and HCT dont.
@@Penfold42 The HCT part is on the small PC Amiga Interface converter board, and it's specified that way in the BOM. Amiga drives don't use that board, so it's ruled out as an issue for them at least. (I put an HC in it originally as that was all I had handy. I later swapped it for an HCT and also tried an LS for completeness) The drive itself is a mix of LS and straight 74 (eg 7406) as per the original. The two boards that I built differ in that one uses bussed resistor pullups and the other uses discrete resistors (I wanted to rule that out as a factor)
Love the video. I'm sure you've thought of this, but could there be something funny going on with the Plus4? I wonder if you tried it in another Plus4 or a 64 do you get the same issues?
Well, at least the 1581 replica enhanced board (courtesy PCB Way) sure has some layout flaws. The Replica Design is (as every board design, which is currently available) derived from the original Commodore 1581 board layout. According to these schematics, the ceramic capacitor marked "C5" on the board is only connected to ground on one side, but not to 5 V on the other, just as it should be. Seems like the Via of one side of the capacitor is not connected to the +5V on the 8520.This Via leads to nowhere , so the capacitor does not do anything.This cannot be right. As this capacitor acts as a filter cap, it may be a source for the drive not working reliably. Don´t know how that looks on other board designs. I guess there should be such a capacitor on other designs as well. Other flaws are minor, but might be irritating such as there are two "U4" markings printed on the board, the design needs some corrections nonetheless...
I just built the 1581 Enhanced V2 from PCBWay. While I had e few initial issues I did not have the same problems as you. I have a panasonic PC drive that has been modified, and it works fine. The disk swap is also detected. However, I tried a Samsung PC drive (which I modified my self) and it worked fine to, but on that one the disk swap is not detected. Not sure why. Video coming soon@@TimsRetroCorner
These will work occasionally, but more often they just return an error. And there's no rhyme or reason that I can see. It isn't the drives. It isn't the ICs. It isn't the cables. It isn't the board (one of them is a DIY Chris board). There has to be something marginal in the way they interact, is what I figure. But I don't know what.
@@TimsRetroCorner my non-educated guess would be that it's something with the board design. I have no clue about it but the passive components may have some wrong values that mess with signal quality. Too many caps can pull up resistance, too little little resistance can fail to pull up a signal to count as "high". It being intermittent kind of fits that, with some signal being marginal. Probably requires some Adrian Black level Oscilloscope artistry to figure out where it goes wrong. With schematics.
These boards are all supposedly reverse engineered from the OG original, and were shown working by their original designers. Now some of these boards have 'enhancements" I guess they could be a factor. Also, we have 50Hz mains here, could that be a factor with some 50Hz ripple getting through on the power lines? vs 60Hz ripple in the US? Maybe some signals got rerouted resulting in different length traces that cause timing issues? There are still lots of possibilities
Early on I got mixed up between J1 and JP1, and placed pin jumpers when I should have placed solder blobs. That did get rectified (seems daft to me to mix blobs and pins - why not just have pin jumpers on the board? It would make things much simpler!)
That's a dodgy cap in the Plus/4 modulator (you can see it in several earlier videos too). I have replacement modulator boards on order (should give me SVIDEO), and they will be another project. The flickering has nothing to do with the drive.
would have been nice if they had incorporated that adaptor in to the main board as 95% of the people building them are using PC drivers.. all so keep an eye on that 7805 mine got MAD hot like burring your fingers, ended up replacing it with a 5v switching reg that had the same formfactor.
It's nice having the option to switch between drive types. I did change the wall wart at one point in the video from 12V to 9V in the hopes that it would run cooler. The "burnt finger test" didn't report much difference though (don't have a thermal camera at this point). The switching supply might be the way to go though. I disabled the internal supply in favour of the CD32 external brick, but it would be nice to be able to run both drives at the same time...
Retrobits has posted an update to his own 1581 build. Watch it here: ruclips.net/video/-mKsMX0BG1Y/видео.html
it gets really tedious if things go pear shaped and one is running out of ideas what's wrong. and even the community, too. Props for your patience and not losing motivation!
Finally a color-coordinated 1581 for my Plus 4!
I built a 1581 replica using the same PCB you have in this video (from PCBWay designed by Thomas Christoph in Canada). Mine was exhibiting similar issues, random working and then not. What I found after staring at the PCB for seemingly hours was that I installed one of the RP1/RP3 resistor networks the wrong way around. Pin 1 on each are towards the outside edge of the PCB. I originally had both in the same orientation.
Yep, they're back to back. I also have a board with discrete resistors, which exhibits the same problem.
You have the patience of a saint Tim. Very frustrating but you will crack it in the end. 😊
Hi. I just finished 1581 today. I struggled with this project for two weeks because it didn't work, but now it works 100%. One binding finding. An inverter must not be used to power this device, but a classic transformer and rectifier. The converter causes a lot of communication errors both on the 1581 board and on the serial line. The power supply during operation must not drop below 5 Volts.
I'm just in the process of editing the follow up video, where I get to the bottom of the problems with SFD-321 drives, and work out a fix...
I like that you show that things go wrong sometimes. Almost all RUclipsrs don't. I had plans to build a 1581 myself, but now I'm hesitant to do so. Super thanks for this video. Has saved me both time and money.
I will always leave in the mistakes, even if it makes me look like a total idiot (plus, it might help someone to not make the same mistake), and also if I break something or rip off a trace etc... I don't strive for "perfection", just what most people can achieve - you get what I did, warts & all. Mostly there is no possibility for a "take 2".
I had enough footage of the drive working that I COULD have edited together in such a way that it looked like the drive just worked and always worked. But that would be misleading, and I'd never do that.
I'm still annoyed that things only seem to let out the magic smoke AFTER I turn the camera off though, lol
I had to use the adapter board for the floppy you pictured - and all LS and 7406/7407 ICs (no tolerance for any variations in chips) on the DIYChris board. It gets along fine with two other 1571 drives on my C128 with a Panasonic JU261 drive in it. The light does stay on; that's a way the floppy light is wired - just unsolder the LED - you're supposed to reply on the Commodore lights. Stay with in tolerance spec value components as well. It's a good kit and works well, and the first time I built it the issues were all mine (bad 8520, bad WD IC - I got a chip tester after that.)
This sounds like a signal integrity problem to me. Are the termination resistors enabled in the floppy drive?
Sorry things didn't work out from the start - that is really frustrating. I hope collectively you are able to diagnose what's wrong and get all of your 1581s working. I've ordered a replacement drive for my unit, but I believe my particular issue was self-inflicted (reversed the 12 and 5v lines going to the floppy drive, whoops!!). Will keep you posted when I know more!
This drive drove me nuts, I will get round to scoping it out properly..... Great vid chap
Thanks 👍
It's a real puzzler. I'm rooting for you guys to figure it out.
Please try to increase the supply voltage. Use a bench power supply to see what your voltages are. Increase in steps of 100mv until it works or until it is almost dangerous. I'm almost sure this is a voltage issue.
I gave this a thumbs up immediately even before watching! Great stuff
Nice video Tim, it's always a pleasure to watch you. Thx.
Really stupid question to ask, but I noticed that when you power on the plus-4 first and then power up the 1581 second you always get errors (as if the serial bus is not being initialised correctly). But when you power up the 1581 first and the power up the plus-4 you seem to get it to work - which shows that the 1581 is powering up and running on its DOS and when you turn on the plus-4 that will re-initialise the serial bus thus putting the plus-4 and 1581 in a good state - as doing a directory after that power up sequence seems to show a directory correctly.
But when you eject the disk and put in a new disk does the 1581 re-initialise the serial bus, as something must happen to tell the 1581 that the disk as been swapped. It might be a good idea to put a scope on the serial bus between the plus-4 and 1581 and see what happens when you eject a disk and put in a new one.
That's an interesting observation, and worth investigating further.
Nothing happens on the serial bus, since communication is always instigated by the bus master (ie the computer), not the slave (ie 1581). The problem is that sometimes the 1581 knows the disk has changed and sometimes it doesn't. This is all internal to the 1581.
@@TimsRetroCorner so the disk change line goes direct to the 8520, those are all used ic are they not?
I'm curious about your ground routing, also what is the noise polution like, as in are there certain days that are much noiser?
If atmospheric noise is not a problem then I would suspect connectors and cables may have intermitten issues.
Man! When I remember these cursed projects in 80's electronic magazines. No circuit diagram was correct, no pcb layout was correct, Homebrew pcb's, holes drilled by hand. Not enough money to buy ic sockets ... Brings back nemories.😁
Thank you for your in-depth testing of this kit, Tim! I built a similar kit a little over a year ago and could not get it to reliably work with my C16. I do not have a C64 to test with, and I'm not sure if that is a factor; after all, the I/O DOS is the same in both machines.
I initially had several issues with bad I/Cs (verified with a BackBit Chip Tester): bad 8520, bad 65C02 (and later a mis-labeled 6502A). The USB cables and adaptors I used to power the device also introduced too much of a voltage drop leading to instability in the power source. However after rectifying those issues, I am currently in the same state that you presented in your video: the drive will not reliably retrieve data. Periodically, a directory listing or file retrieval is truncated and the drive hangs requiring a reset. I have tested with the stock and JiffyDos ROMs with the same results. I have not tested the disk changing issues you experienced, but given that our builds share the same issues, I'm certain it will occur on my unit as well.
The only components on the PCB that I was dubious about when purchasing were the ferrite beads. I bought ones that best matched the diameter and length of the original specs but seeing as the BOMs for this build state something like "just get them from a spare C64", I'm uncertain that I installed the correct values.
Thank you again for documenting the faults. Hopefully someone who has gone through this build and has also corrected these problems can share their remediation procedure!
This is quite interesting, particularly because I have an original 1581 which exhibits several of the same symptoms. It has all its original ICs as far as I know and the original Chinon drive as well.
I have been meaning to investigate it closer at some point, so I hope you get some clues to what would be causing this.
You mentioned that Captain Commodore got a Go-Tek to work well. That would suggest that the power consumption of the physical drive is a lot higher on the reproduction boards or that the voltage regulation is out. it would definitely be worth scoping the voltages and power draw to compare the physical versus SD-Card based drives. Is there a newer, more efficient, pin-identical voltage regulator that doesn't require so much cooling? In my limited experience of mid-90s PC components, a single component running too hot would often bugger communications between more sensitive items on the board.
I'm not in a position to test it right now, but Captain Commodore's figures are 570mA with the drive, and 290mA with the Gotek. My power supply (CD32) is rated at 2.2A on the 5V line. Neither drive uses 12V. The official 1581 power supply I think was 5V 1A, 12V 0.5A
Hello Tim. Very good video. I believe you didn't do anything wrong. My suspicion is that the problem lies in many revisions of the PC drive. Over the years, manufacturers have continually improved the floppy drive electronics. Your Samsung SFD-321B seems to be a REV.T6. I use a REV.T1 and it works well. The floppy drive adapter isn't bad, but you can do without it. The older revisions of the drives can be easily modified so that it is not needed. There are many instructions on the internet (Amiga modification).
So I'm using a TEAC Amiga drive (taken from my A500) and a modified 321B (unknown revision since it was missing the top plate). Both still exhibit the problem. The two PC drives are Revs T5 and T6. Retrobits (see stickied link) is using a rev M2, but on a Toms01 board. Revision being a factor is an interesting idea, but it's a very small sample size at the moment, and without knowing the differences...
Since the adapter is performing the same function as the conversion, what makes the conversion "better"? What does it do that the adapter doesn't?
I also have a Toms01 board. I first tested it with the adapter and an ALPS DF354H168F drive. That did not work. With a DF354H090F it worked, including changing the diskette. I found parts of a damaged original case. The adapter doesn't fit in there very well because of the shielding plate. I know from pazuzu72 that the adapter is not necessary. The conversion is easier with Samsung drives. You don't have to remove any of the controller's legs or cut any traces. The fact that older drives work better is just a suspicion that I can't prove. My adapter was designed by androSID & toms01. He does exactly the same thing. The difference is that an SN74HCT02 is installed on it.
@@TimsRetroCorner
@@black_forest_76 Because I don't have a shielding plate with my 3d printed case, the adapter not fitting is something I hadn't considered. It's a very good reason for doing the conversion instead! 👍
Anecdotally at least, it's looking like Toms01 is the most reliable of all the replica boards. (my adapter is also AndroSID & Toms01). Also anecdotally, converted drives work better than adapted, which suggests the conversion does something that the adapter does not. But I don't know what that is yet.
Hi Tim,
I have a 1581 replica myself. The drive is always selected as the drive select signal is directly connected to GND. In the original 1581 you won't recognise this as the drive itself used there has no LEDs.
The drive you use is a Samsung SFD-321, right? I also use this drive for my replica. There are different versions, later ones are cost reduced to the max. One of the two drives I own even is hardwired to HD and does not have the HD detection switch. When I remember correctly I used my second drive as reference and boched the PCB to have it permanently set to DD. Using the drive with the wrong settings (presets for the heads like magnetic field strength, filter setting of the head amplifier etc.) can lead to read/write errors.
And there's more: In my drives some of the switches had contact problems so that a closed switch had a resistance of several kOhms. If that's the case with the disk presence switch or the write protect switch you can have all sorts of funny issues like disk changes not detected at all or permanently (I don't know by heart which state the switches have with an inserted disk or with no write protection). So it's definitely worth to check a) if your drive has an HD detection switch and even if it has to think about hardwiring it to DD and b) to check if the switched are working properly.
And a final point: These drives use 5V only as you noted correctly. In general the power supply scheme of the 1581, the Amiga and most likely the C64 is quite poor in the way that the 5V rail is directly powered from an external supply over long cables while the currents are quite high. When the drive is stepping or starting the motor this can lead to voltage dips. I noticed that you use quite thin and long cables to you bench power supply. I can imagine that this leads to malfunctions while a Gotek drive that mainly needs to power a 3.3V microcontroller is a much lower load than a real drive. So you could/should check the power rail with an Oszilloscope directly on the PCB.
P. S. As I have no spare custom chips I bought a ready built PCB as nowadays chances are quite high to get fake or defective parts.
Thanks. I hadn't been aware there were different versions of the SFD (someone else mentioned they spotted mine is a Rev 8) I'll have to look into what the differences are and how they might impact it working.... I did also notice that there are other PC drive adapter boards (on ebay etc) that have extra smoothing caps on the supply lines. Maybe that is beneficial.
I've had all of the same problems and I'm still dead in the water on this build. I have tried three different floppy drives, two are different PC drives, one is an Amiga 500 drive. I have also tried multiple different IC's, different ROM's, different SRAM, different CIA, different floppy controller 1770 vs 1772. All a failure. I just get a drive that spins forever and can't read anything. It also locks up my Commodore Plus/4 and causes it to hang and I have to power it off to free the computer. I've also tried 74LS vs 74LSHCT logic and that didn't work. I can't see how anyone would have managed to build this and get it working and stay working. If they did, then I'd sure like to know how because for me, it's been nothing but endless frustration as well as a lot of money sunk into this project.
Agreed, it's been hugely frustrating. At this point a genuine 1581 would have been cheaper, even at ebay prices!
hi! which model floppy drive did you use for this? Thanks!
SFD-321B. And a bunch of others in the followup.
The resistor packs there are 3 of them. Make sure you got the right values in the correct positions 1 is a different value. Kit works perfectly with either 6502 or 65c02. I think you have on in the wrong position. Or you used 3 of the same value.
Nope. 2K7 on the CPU pullups, and 1K on the bus pullups. Exactly as per. Triple checked. The second board even used discrete 1K resistors instead of the packs.
You have a power supply problem, check the voltages with an oscilloscope that do not have ripple when the floppy drive works
Check out the follow up video. It's not the power supply, but rather an issue with one particular make & configuration of drive... ;)
I think most people would upgrade to a Gotek anyway, but troubleshooting the problem would interesting, indeed :)
For a Gotek, you will need to buy and flash the HxC Firmware to the device as for the free "Flashfloppy" does not support both d64 or d81 files....
Well, it is actually free for a period of time, but normally it costs you 10 Euros...
I can now report that after following the instructions provided by Retrobits, I now appear to be having a reliable working drive. I removed the PC adapter board completely and did the modifications on my Samsung SFD-321B. The only other odd thing I encountered is "device not found". Even though the instructions I could find online for a 1581 said both switches in the rear need to be UP for device #8, I ended up only able to get it to work with both switches DOWN. I'm not sure where the discrepancy here is.
I think it depends on the actual switches you have - some are down for 'on', others are up.
Glad you have yous working. Can I ask which PCB you are using, and also what is the make/revision of the drive mechanism? (it'll be on the label)
I'm using Thomas Christoph enhanced 1581 Drive PCB. The drive I'm using is a Samsung SFD-321B Rev T5@@TimsRetroCorner
Man that Sucks you are having a problem I built 2 of them and they just worked.
the only problem I had was with HD Disks and the drive density switch I had to jump the switch so it thinks all disks are DD
I chanced the 8520 with a 6526a and my problems where over, i use the 128 fast serial and it worked to, but Maybe i need to test it some more
There should be no problem using a 6526A - it's only different in terms of the TOD clock, and that's not an issue for the 1581. In fact, you may even get away with a straight 6526 (I did), but that's not guaranteed to run at the higher speed.
Don't Amiga drives have a twist in the cable?
No they don't. The twist is there in PC drives because they default to 'drive 1' so need the signals swapped, but Amiga drives are all 'drive 0' so don't need the twist.
have you tried testing continuity of the /RDY and /disk_chng signals?
ideally from the back of the connector on the drive to the top of the pins on the 8520/6526
Yep.
You said you changed out several chips, also the 1770 with a 1772, you know these have different default step rates. You can run them but they work better with modified firmware.
Agreed there's no enhancement from a 1772 with stock firmware. But it'll still work.
Weird that it's so quirky. I have been accumulaiting the kit to build some of these. I'll have to try it and see what happens.Thanks for a great video about it, now I'll go in eyes wide open.
Please let us know how you get on.
Have you tried a different power supply
Yep.
You don't have to close the line ever time you run one of those disk commands. Why do you think you have to keep closing after opening? Why not just leave it open and then continue to use "print#15..."?
Force of habit, I guess.
@@TimsRetroCorner: But why would you have even generated that habit in the first place, instead of closing after a sequence (or never, since I can't think of a time when we really would need to close; I'd love for someone to make a video that demonstrates any necessity to close those at all)?
I'm not an expert but maybe something slightly off with some timing somewhere as suspected. This old tech was designed with old chip timing difficulties in mind. A few milliseconds difference between old and new chips makes a difference.
We're probably talking microseconds or even nanoseconds, but you could be right. Finding it though...
How does this board differ from a salvaged 1571? I believe the 1571 and 1581 use the same circuit, only a different floppy mechanism. David Murray (The 8-bit Guy) built a 1581 using a 1571 board and an Amiga floppy and it worked without a problem.
A 1571 is very different internally. As with other 5.25" Commodore drives, all the drive electronics are on the main PCB - it directly drives the motors, has all the head amplifiers and signal encoding/decoding etc. The 1581 just talks to a floppy controller IC (WD1770) and lets the drive handle the signals, motors etc internally. You can't plug a 3.5" Amiga floppy into a 1571, at least not without major surgery and a ROM rewrite.
I couldn't find a 1581 build video on the 8-bit Guy's channel, maybe you have a link?
@@TimsRetroCorner I remember a video on his channel building a replica but it may have been removed. It's possible he used a NOS case and a 1581 board from a dead drive.
As for the 1571: "The drive uses the MOS 6502 CPU, WD1770 or WD1772 floppy controller, 2x MOS Technology 6522 I/O controllers and 1x MOS Technology 6526."
So in the very least you can use the chips off a dead 1571 to populate the board, as it uses the same controllers but some more additional logic to talk to the drive heads directly.
The only difference would be that the analog board on the drive was pulled and integrated into the 1571 board instead of leaving it on the drive. An MFM interface is very simple and low-level.
If you look at this video there's a top view of the 1571 board: ruclips.net/video/hi_3T00MctI/видео.html
It looks like the basic circuit matches that of the 1581 but the connector where the floppy would attach is now covered with some potted terminating resistors and probably traces that go off to the left-hand side of the board, which contains the electronics that would usually be on the drive.
@@Stoney3K You are correct - I've even had one apart on this channel! My fault for misremembering (the 1772 I had was a spare for the 1571, I should have remembered that at least!!). Getting old, ho hum!
Great sweater and great video!
Glad you like it!
Almost at the end now. I feel your frustration. I got myself 5 drives only to get one working. The original one in my 386 pc just spontaneously refused to respond. (General failure reading drive A). I know, not C64, but I get it. I have been blessed with an original 1581 but plan to build one or two (just because I can, it’s a hobby ;) ). In relation to that; what filament did you use for the “beige” one? I can’t seem to find a good supply PLA in retro-white. Looking forward to a followup in which hopefully you have found some answers.
It's called "Ivory Beige" by Copymaster. I got it from Amazon.
@@TimsRetroCorner thanks :)
The CMD FD-2000 has the same issue of not recognizing when a new disk is inserted. CMD stated that when swapping disks, a reset command must be sent, similar to the "I" command. So perhaps that's an issue with all 3.5 inch drives? (From CMD's Commodore World Issue 24 Pg 16: "If you don't do this, the drive often ends up using incorrect information stored in variables")
However it seems to be the least of your problems; fingers crossed you solve this!
"incorrect information stored in variables" is an understatement! Basically the BAM and directory will be from a completely different floppy, so if you write to the drive you'll corrupt the disk. I did try the "I" command, sometimes it worked, sometimes it didn't. Especially it didn't if the drive was throwing 74 Not Ready errors.
Have you tried it with the floppy drive on its back to get it away from the board ?
If something is marginal on the board, like say a missing or badly soldered pull up resistor, the EMI from the drive mech itself could explain the seemingly random behaviour.
Yep, tried it all ways up, down, sideways. Happens on multiple boards and at least 4 builders. Even more going by the comments. So it seems to be something endemic, not just a bad joint.
@@TimsRetroCorner endemic to the recreation? you mentioned replacing 74HC with HCT parts. The original C= design used only 74LS apart from the 7406/7407 line drivers.
How much of your (and others?) board has HCT rather than LS parts?
For q proper design they are largely interchangeable and most people dont think twice.
This all falls apart rapidly if there are missing pullups on an input somewhere..
A recreation of a design could easily miss a pullup resistor and with 74LS would probably just work fine as 74LS floating inputs tend to read high. 74HC and HCT dont.
@@Penfold42 The HCT part is on the small PC Amiga Interface converter board, and it's specified that way in the BOM. Amiga drives don't use that board, so it's ruled out as an issue for them at least. (I put an HC in it originally as that was all I had handy. I later swapped it for an HCT and also tried an LS for completeness) The drive itself is a mix of LS and straight 74 (eg 7406) as per the original.
The two boards that I built differ in that one uses bussed resistor pullups and the other uses discrete resistors (I wanted to rule that out as a factor)
Nice.
Love the video. I'm sure you've thought of this, but could there be something funny going on with the Plus4? I wonder if you tried it in another Plus4 or a 64 do you get the same issues?
I've tried it with a C64, so I'm pretty confident that the problem isn't there.
Well, at least the 1581 replica enhanced board (courtesy PCB Way) sure has some layout flaws. The Replica Design is (as every board design, which is currently available) derived from the original Commodore 1581 board layout.
According to these schematics, the ceramic capacitor marked "C5" on the board is only connected to ground on one side, but not to 5 V on the other, just as it should be. Seems like the Via of one side of the capacitor is not connected to the +5V on the 8520.This Via leads to nowhere , so the capacitor does not do anything.This cannot be right. As this capacitor acts as a filter cap, it may be a source for the drive not working reliably. Don´t know how that looks on other board designs. I guess there should be such a capacitor on other designs as well. Other flaws are minor, but might be irritating such as there are two "U4" markings printed on the board, the design needs some corrections nonetheless...
Nice video. I am going to build this, so this was good for preparation. Hopefully I you figure it out some time later
Thanks Tommy. There will be an update of course when the problem is resolved, and I look forward to seeing your build on @Arcticretro
Still sourcing the parts I need, but I will start soon@@TimsRetroCorner
@@tommyovesen Which of the various boards are you building?
Enhanced v2 from PCBWay@@TimsRetroCorner
I just built the 1581 Enhanced V2 from PCBWay. While I had e few initial issues I did not have the same problems as you. I have a panasonic PC drive that has been modified, and it works fine. The disk swap is also detected. However, I tried a Samsung PC drive (which I modified my self) and it worked fine to, but on that one the disk swap is not detected. Not sure why. Video coming soon@@TimsRetroCorner
I have built two replicas, with 1581 boards from DIY Chris, both worked at first power on, on both are PC drives with adapters.
These will work occasionally, but more often they just return an error. And there's no rhyme or reason that I can see. It isn't the drives. It isn't the ICs. It isn't the cables. It isn't the board (one of them is a DIY Chris board). There has to be something marginal in the way they interact, is what I figure. But I don't know what.
@@TimsRetroCorner my non-educated guess would be that it's something with the board design. I have no clue about it but the passive components may have some wrong values that mess with signal quality. Too many caps can pull up resistance, too little little resistance can fail to pull up a signal to count as "high". It being intermittent kind of fits that, with some signal being marginal.
Probably requires some Adrian Black level Oscilloscope artistry to figure out where it goes wrong. With schematics.
@@TimsRetroCorner You placed TWO yellow jumpers at the beginning instead of one, are they supposed to be there?
These boards are all supposedly reverse engineered from the OG original, and were shown working by their original designers. Now some of these boards have 'enhancements" I guess they could be a factor. Also, we have 50Hz mains here, could that be a factor with some 50Hz ripple getting through on the power lines? vs 60Hz ripple in the US? Maybe some signals got rerouted resulting in different length traces that cause timing issues? There are still lots of possibilities
Early on I got mixed up between J1 and JP1, and placed pin jumpers when I should have placed solder blobs. That did get rectified (seems daft to me to mix blobs and pins - why not just have pin jumpers on the board? It would make things much simpler!)
59 minutes in roughly u touched the drive and the screen flickered
That's a dodgy cap in the Plus/4 modulator (you can see it in several earlier videos too). I have replacement modulator boards on order (should give me SVIDEO), and they will be another project. The flickering has nothing to do with the drive.
Could be a bad chip socket
Use your scope and measure voltages under load...
Which I have done.
The 6526 will work fine in fast serial mode.
Thanks. On reflection, thew C128 uses a 6526 so yeah it definitely should work.
@@TimsRetroCorner I've built 2 1581s. One from a VIC-20 board and then i repeated the insanity by heavily modifying a 1541-II board into a 1581 :)
@@Penfold42 I've never seen the insides of a 1541-II, does that have a WD controller too?
would have been nice if they had incorporated that adaptor in to the main board as 95% of the people building them are using PC drivers.. all so keep an eye on that 7805 mine got MAD hot like burring your fingers, ended up replacing it with a 5v switching reg that had the same formfactor.
It's nice having the option to switch between drive types. I did change the wall wart at one point in the video from 12V to 9V in the hopes that it would run cooler. The "burnt finger test" didn't report much difference though (don't have a thermal camera at this point). The switching supply might be the way to go though. I disabled the internal supply in favour of the CD32 external brick, but it would be nice to be able to run both drives at the same time...