PC floppy adaptor for Amiga

[prowler]

Originally Posted by orange

maybe you should try that caps dumping tool if they are so important or keep trying with x-copy.

Thanks for those suggestions, orange!

[TomCrazy]

Originally Posted by prowler

@Stedy,

I replaced the internal floppy drive of my Commodore A1200 with my Mitsumi PC floppy drive via one of your adapters this evening with the intention of having a disk imaging session, but the floppy drive would not read any disks inserted in it!

The drive was being polled, disks inserted were being detected and the drive was spinning up. DF0:, PC0: and MAC0 icons appeared on the desktop each time, but the disk was unreadable in each case.

Running TSGui resulted in the error message "Cylinder No 0 - No Sector Header", i.e. no formatting.

I will try to explain this. Note however that following information is gathered after a lot of testing on my Commodore A1200. Since I do not own an Escom Amiga, I can not guarantee that these things apply to Escom Amigas as well, but I guess most of it should be the same.


Since you are using TSGui, I assume that you are working in an Workbench environment while having your disk imaging session?


The short version:

The problems you experience has to do with a kind of design flaw in Stedy's adaptor. The problem lies in the circuit that is generating the ready-signal. The ready-signal from Stedy's circuit is not a genuine ready-signal (the same ready-signal that can be found on original Amiga disk drives), thus causing problems in a Workbench environment.
I will explain the technical details more deeply at the end of this post.

Originally Posted by Stedy

@Prowler
Have you got the nRDY jumper set correctly? The escom A1200 uses the jumper wire, the commodore A1200 does not need it.

The symptoms you describe can be caused by lack of nRDY signal.

Actually, the symptoms Prowler is describing can not at all be caused by the lack of a ready-signal. Workbench requires no ready-signal at all in order for the disk drive to function properly. Workbench will in fact work very well even without the presence of a ready-signal.
("Without the presence of a ready-signal" = The ready-signal being at a logic high at all times, which is the default state and the state it will stay in as long as nothing is connected to pin 34.)

What Workbench does NOT like is a faulty ready-signal, which will result in lots of strange error messages when accessing the disk. As usual, writing is more sensitive than reading.

Originally Posted by prowler

@Stedy,
The problem I had with my adapters was solved by switching the NRDY jumper to the 1-2 position (Use Clip), even though I am using them in my Commodore machine without the clip attached. With the jumper in the 2-3 position (Normal), the PC floppy drives will not read any disks!

Yes, this is of course only natural and not strange at all.
When you are using your Commodore Amiga, pin 34 is still connected to the floppy-controller on the motherboard (since it is not an Escom Amiga). So, when you are setting Stedy's adaptor to "use clip"-mode, the ready-signal is not being fed to pin 34 anymore. In this mode, the ready-signal is actually being disconnected from the Amiga, hence everything works fine in Workbench since Workbench doesn't even require a ready-signal at all in order to work.
However, when you pull the switch into "feed into pin 34"-mode, the Amiga is being fed with a ready-signal that is not genuine, causing the Amiga to access the disk drive in an incorrect way.
The result: A lot of strange behavior such as read/write errors and "Object already exist"-messages although you are copying files to an empty disk, among other things.


The long (and more technical) version:

In order to understand more exactly why Prowler is getting these kind of error messages, and why it works with the ready-signal removed from the system, it is necessary to get a bit more technical.

First of all, the ready-signal is of course a signal that will give the Amiga information about when the drive is ready for access. This is a signal coming from the drive, so the direction is [Drive] -> [Amiga].

The ready-signal will be set to a logic high (meaning "NOT ready") when one or several of the following conditions occurs:
* The motor is not spinning
* The 'drive select'-signal is set to a logic high (which means the drive is not selected)
* The step motor is moving and not resting at one of the 79 tracks on the disk

Most disk drives refuse to spin up the motor when there is no disk in the drive, even if they receive a 'motor on'-signal. So "The motor is not spinning" can in a way also mean "No disk in the drive". XCopy is utilizing this feature by monitoring the ready-signal to determine if there is a disk in the drive instead of monitoring the 'disk change'-signal, but that's another story.

Let's focus on the last point I mentioned; "The step motor is moving".
For the drive to be able to read or write to a track on the disk, the head must be resting at one of the tracks. Naturally, if the drive head would be somewhere between two tracks, the drive would not be able to do much of anything since the head is not fixed on a certain track. Doing a read or a write at this point will in most cases result in a read/write error.

So how do the Amiga know when the drive has finished stepping to the next (or previous) track and that the head is resting steadily on the track it has just moved to? Here is where the ready-signal comes in!
The diskdrive will set the ready-signal to a logic high (=not ready) when the head is in movement, to prevent the Amiga from sending a new read- or write request too early, which would result in a read/write error. In other words, the Amiga will wait until the drive has set the ready-signal to low again (=ready) before it starts to issue its next command to the drive. The next command might be another step signal to make the head move even futher, or it might be something like turning the motor off or... READ OR WRITE.
If the ready-signal indicates too early that the drive is ready for its next command, when it in fact is still busy doing a step (moving a step of the head motor), well... The Amiga will send a read/write-request too early. And since the step motor has not been given the time to reach the next track yet (and is still between two tracks) then BAAM... read/write error!
I can not explain exactly why Commodore designed the Amiga so that it works perfectly fine even without the presence of a ready-signal in a DOS-environment (like Workbench). But it seems that when the floppy controller is accessing DOS-disks, it doesn't rely on the ready-signal. Instead it just uses a default waiting time after each step of the motor head. This waiting time is sufficient in order to let the head of the drive settle at the new track before a read or write is issued.
However, if there is a ready-signal present, even in a DOS-environment the Amiga floppy controller will go like "What, the drive has already finished moving its head? Oh well, no point in waiting any longer, I might just as well issue my next command then." But if the drive wasn't really ready, but just sending a false ready-signal to the Amiga and the Amiga tries to write to the track, then... "Volume blablabla has a read/write error on block <something>".
What would you expect, Ms Amiga? Of course you got read/write errors, the diskdrive just lied to you and told you it was ready to read and write when it in fact was still busy moving the head.
It is possible that the floppy controllers of the Escom Amigas work a bit differently, thus eliminating these problems even if a faulty ready-signal is present. I don't know about that and I can not test it, since I only have Commodore Amigas. (An A500 and an A1200)

I had a look at the schematics for this adaptor that Stedy posted on his site. And I can surely say that if he has built his adaptor with the same design that he put into the schematics he posted, then the ready-signal from that adaptor is going to be a quite strange ready-signal, not at all similar to that of an original Amiga diskdrive.
This ready-signal is going to be set to low (=ready) as soon as the drive is selected (drive select is set to a logic low). This being with the only exception of the index signal, which is mostly at a logic high state all the time except for the small index-pulses coming from the drive (pulses consisting of very short logic zero intervals). These pulses are what gives information about the rotation speed of the drive and has nothing to do with the ready-signal at all! So I can really not understand why Stedy has incoperated the index-signal into the logic in this way. This way, the index-signal will only mess up the ready-signal even more, making it being even further from an authentic ready-signal that can be found in an original Amiga diskdrive.

Anyway, since Stedy's adaptor sets the ready-signal to "ready" as soon as the drive is selected, the Amiga will receive information about the drive being ready too early after doing a head motor step. Therefore, if the next command (after a step) is to read or (even worse) write to the disk, the result will in many cases be a read/write error.

All of you who wish to test this on their Commodore Amiga can do that very easily. Here is how:
Open up a PC diskdrive. Modify it in order to work with Workbench by rerouting "select drive" from pin 12 to pin 10 on the floppy connector on the drive, and by rerouting "disk change" from pin 34 to pin 2. (It is important that pin 34 is completely set free and is floating. In other words, just connecting a write between pin 34 and pin 2 will not work since pin 34 will then still be fed with the "disk change"-signal. Freeing it might involve cutting a path in the PCB of the diskdrive in cases where there are no jumper pads available. Do it at your own risk. And by the way, you also have to make sure that pin 2 is disconnected from the density select line of the PCB of course.)
After this, make sure that pin 34 is actually set free from pin 2 by doing a simple continuacy test with a multimeter between these two pins.
Next, take a wire and connect pin 10 (select drive) with pin 34 (disk ready). If you are using more than one diskdrive in your Amiga, you'd be better off using a diode (for example 1N4148) and connect the cathode to pin 10 and the anode to pin 34 (of the diskdrive connector), in order to prevent the ready-signal of the second drive to activate this modified drive by mistake.

Now, every time the disk is selected, it will be set to "ready" and keep that state until the drive is not selected anymore.
The result: All trackloaded games (or maybe 99.9%) will mostly run and work perfectly (as long as the game itself doesn't have a compability issue, i.e an old A500 game being run on an A1200), but in Workbench... Usally it works fine when just reading disks, but writing will most of the time result in a read/write error. Like I mentioned above, this is because the drive is falsly indicating that it is being "ready" too much, even while the head motor is stepping.
In fact, doing this mod directly to the diskdrive which I just described will end up with almost the exact same result you'll get from using Stedy's adaptor because according to his schematics, that's how he has designed it and that's basically what his adaptor is doing. He has also used the index-signal in his logic circuit, but like I mentioned, this index-signal has got nothing at all to do with the ready-signal and should therefore not be implemented in this way at all. It would be better to just leave it out of the circuit. The index will just be pulsing zeroes at an even interval as the drive motor is spinning and these pulses will show up, with regular intervals, as pulses of logic high (=not ready) in the ready-signal. However, these pulses have no relation whatsoever with the drive being ready or not. So in this sense, it is just messing up the ready-signal.
It is also not possible to generate an authentic ready-signal by just using two NAND gates and some pull-up resistors. A bit more complex logic has to be incoperated in order to make the generated ready-signal to function as a true ready-signal that doesn't cause any problems.

I have actually designed my own adaptor with the same purpose; Being able to use PC diskdrives on my Amiga. It is completely my own design and not a reproduction of Stedy's or anyone else's schematics. It involves using a logic circuit which is a bit more complex. It generates a true, authentic ready-signal just like the one of an original Amiga diskdrive. I have monitored the ready-signal of an original working A500 diskdrive and engineered my circuit to generate an identical ready-signal to that of the A500 drive. This includes sending information to the Amiga about when the head is busy moving, like I was talking about earlier in this post.
But I will not tell you more about that here, because this thread is about Stedy's adaptor and not about mine. If any of you guys are interested, I can post pictures and write some more info about it in a seperate thread. I just have to open up my A1200 and dig it out first. The prototype is currently inside my A1200 doing a good job of generating the ready-signal for a PC drive that doesn't have a ready-signal output on the controller-chip. And it works great with basically all trackloaded programs I tried (which are A1200 compatible), Xcopy, and Workbench without read/write errors.
Anyway, more about that in another thread.

Cheers!

[Paul_s]

Tom, you have lived upto your surname with that post Welcome to EAB, your stay will be long, there are no emergency exits and all your money belongs to amigakit.com

[prowler]

Thanks for your explanation, TomCrazy; all the more so if you have joined up just to post it (let's hope not! ).

I'm more interested in this than you might think, and I imagine Stedy will be interested too. As yet, I've read only the short version. I shall read the long version later offline to try to understand the implications fully.

[orange]

iirc, I used something similar; drive_select->rdy via diode, like in one of aminet solutions. anyway it worked for me, don't remember having any problems.

[TomCrazy]

Paul_s:
I hope that Amigakit will not take all my money, but only the amount I am ordering Amiga parts for.


Prowler:
I didn't join this forum only to write that explanation, but I guess you could say that it was part of the reason. But as an old Amiga nerd and hardware modding enthusiast, I couldn't help but to find this place quite interesting. So I decided to join up.

I hope Stedy won't feel offended for what I wrote about the ready-signal on his adaptor by the way. I just wanted to share some information that maybe could prove useful to both Stedy and everyone who is using his adaptor.


Orange:
I can say for a fact that using the solution which involves connecting pin 34 and pin 10 with a diode will not make up for the ready-signal that can be found on original Amiga diskdrives. For just loading trackloaded games and programs, this solution is suffient enough to make the games run. But it's still not a good solution in the long run.

Issue 1:
When trackloading, the stepmotor will be put in overdrive while it's moving, because the ready-signal indicates that it's ready all the time. You could say that this will shorten the seek time of the diskdrive when trackloading (the head is moving faster than normal during 'trackloaded based movement'). The Amiga will be okay since there are no moving parts there, but I don't know about the diskdrive. Common sense tells me that putting the step motor in hyperdrive is not something that makes the diskdrive last longer, anyway.

Issue 2:
XCopy (and similar programs). XCopy checks for the ready-signal in order to determine if there is a disk in the drive or not every time you press "start" or "checkdisk". With a diode connected between pin 34 and pin 10, the Amiga will take it as the drive is ready when you press "start" in XCopy, even if no disk is in the drive. If there is no disk in the drive and you press start, the diskdrive led will light up and XCopy will actually try to copy the non-existing disk in the drive, resulting in a 2 for every track (showing "22222222222222222...." where there is normally "000000000000..." when you copy a disk).

Issue 3:
Workbench will have a lot of read/write errors, especially when writing to a disk. You could try to low-level format a disk and then copy a file to the disk that is at least 500kB (the bigger the better). If you get read/write errors, it's because of the ready-signal. If you don't, then I guess it must be either because of the diskdrive having a really fast step motor and can therefore handle this (although I've never seen a diskdrive like that, all the ones I tested got read/write errors with the diode hack), or because you are using an Escom Amiga which could be developed to be more PC diskdrive-friendly (but again, I have no way of testing this).

Using a diskdrive that really has a ready-signal or using a circuit that's generating a ready-signal identical to that of an original Amiga drive will solve these problems.

When you tested this diode-hack, was that on an Escom or Commodore? Just curious if the Escom Amigas are better at handling PC diskdrives or not. I have a hard time believing so, but you never know.
Also, did you try to use XCopy? Did you try to click "start" even when there was no disk in the drive? If everything works perfectly, the led should light up for just a second, then XCopy will report "No disk in DF0: !". If it doesn't work, then XCopy tries to start copying even though the drive is empty and you will probably see the screen being filled with a lot of 222222.
And did you test to low-level format several disks in Workbench and then copying files to them big enough to fill at least 60% of the disk?


To everyone who has an Escom Amiga:
Just out of curiousity, can you start trackloaded games on a out-of-the-box Escom Amiga, even if the 'internal pin 34 <-> external connector pin 1'-hack has not been performed?
I'm just wondering because if I'm correct, the diskdrive being shipped with an Escom A1200 is really just a slightly modified PC-diskdrive with the lack of a ready-signal (and even if there was a ready-signal, it would not reach the controller anyway if the hack on the motherboard has not been implemented yet).
I bought a new diskdrive for my Commodore A1200 about a year ago... only to receive a slightly modified PC diskdrive in the mail which completely lacked the ready-signal on the controller-chip! The only positive thing was that it fitted perfectly inside my A1200 and that the drive seemed quite Amiga-friendly overall in a DOS-environment.
In fact, that is how my project of building a ready-generating circuit started, which later developed into an external adaptor since the components couldn't fit inside the diskdrive. I'll make a thread about it later.

[rkauer]

Awesome explanation!

How about just tricking the signal inserting a wait time of 8ms on each ready cycle? Even a dumb "tick" generator can do the trick or?

[cosmicfrog]

welcome tom, looking forward to more posts from yourself

plz do start a thread all about your floppy adapter, especially if you got some nice pictures, not essential but very welcome if you have some, some ppl round here need their weekly Hardware p0wn quota hehehehheh

[orange]

I just tested it with 600Kb lha archive in OS4 and AmigaDOS, no problems verifying the archive.
can't test x-copy on this amiga.

I think you're exaggerating; perhaps modern drives don't have rdy because they don't need it..

[Stedy]

@Tomcrazy

I need to re-read your long explanation of the logical chain on the ready signal at a later date.

With regard to the existing design, I am happy with it. It was thoroughly tested by myself and two others prior to release. X-Copy, D-Copy, track loader demos, Workbench copy/format/delete and various copy protected games were tested successfully. The MK1 design broke with X-Copy, a change to the logic was made to make the MK2 design that you see on the website. I have before and after logic analyser capture to prove this. The MK2 design is the version that has sold 50 units with no major complaints so far.

I have been working my way through the Amiga Hardware RKRM whilst working on my updated MK3 design, which uses a CPLD. This has different ready logic, primarily due to the support for HD disks.

@thread

Copying this from my other thread here:

http://eab.abime.net/showthread.php?t=53385&page=2

Floppy drive adapter
--------------------

The updated floppy drive adaptor is working in the double density mode and will operate as DF0-DF3. Currently working on HD support and a better method of drive selection for DF1-DF3. This has been a fiddly one to implement, there are state machines and a reasonable chunk of logic now used in the design, one of the advantages of a CPLD.

The retail price for this will be in the £15-20 range. It may be necessary to get this assembled as the fine pitch CPLD is not easy to hand solder. Once the prototypes have been thoroughly tested, the decisions can then be made.

Finally, happy new year to you all.

Ian

[TomCrazy]

Happy New Year! I hope that 2011 will be a great year for everyone with lots of cool hardware hacking.

Originally Posted by rkauer

Awesome explanation!

How about just tricking the signal inserting a wait time of 8ms on each ready cycle? Even a dumb "tick" generator can do the trick or?

I am not that familiar with tick generators since I have had no reason to work with them up until now. I did a quick search on the net, but mostly found the term "tick generator" being used in computer programming. If you are refering to a kind of pulse generator set to a certain fixed frequency, it is unlikely to work since the "not ready" must be issued when the step motor is moving.
Also, the ready-signal is not set to 'not ready' only when the step motor is moving. It's set to 'not ready' at other occations as well. Consequently, it is not appropriate to design a circuit that adds a delay to every 'not ready'-cycle. Instead, the wait time should be triggered somehow when the step motor starts moving. I think there might be several solutions for how to do this. I will get more deeply into that later.

Originally Posted by orange

I just tested it with 600Kb lha archive in OS4 and AmigaDOS, no problems verifying the archive.
can't test x-copy on this amiga.

I think you're exaggerating; perhaps modern drives don't have rdy because they don't need it..

Exaggerating? Hehe, then I guess I just happened to get read/write errors every time I tried various diode-hacked diskdrives on my A1200, and I guess that my scope just happened to give me incorrect readings every time I was observing the logics while the diskdrive was in operation.
But seriously, I can ensure you that what I have written in my last posts is 100% true when it comes to my A1200. But of course, there are different Amigas out there, there are different floppy diskdrives and there are also different operating systems out there. Just because my system has a big problem with the diode hack doesn't mean that other systems will have the same problem. If it works fine for you and you don't even get any write errors, then that's great! You don't need to take any futher action, then. But I'm sure there must be other people like me out there, whose system is very sensitive to a non-authentic ready-signal. My long explanation was mostly directed to Prowler since he experienced very similar symptoms to the ones I experienced with a bad ready-signal. So Prowler's Commodore Amiga is probably just as sensitive as mine in this case.
It is up to each and everyone what they will do with the information in my posts, but I can guarantee that this is at least how my A1200 behaves. But you are welcome to scold my Amiga anytime for it's behavior if you want. I can't help that it's picky when it comes to the ready-signal. And I just took the behavior of my A1200 and put it into words here on this forum. I wouldn't really call that exaggerating.

I have also noticed that most modern PC diskdrives don't have the ready-signal on their controller chip. As far as I know, the PC floppy controllers have never used the ready-signal at all. (At least I have never seen or even heard of a PC that uses the shugart interface.) And since the Amiga is not such a new machine anymore, I guess the manufacturers of diskdrives choose to design their diskdrives without any ready-signal if they can save a few cents in production cost for every drive by doing so. They probably consider systems that make use of the shugart interface as "dead" or "too old for the general public to use such systems anymore". They probably care a lot more about money than they care about Amiga enthusiasts like us when manufacturing their diskdrives.
To me it seems like it's getting harder and harder to even find PC diskdrives without a ready-signal in computer stores today. The ordinary PC user has probably given up the idea of even using floppy disks by now since there are faster and more convenient alternatives like USB-sticks and memory cards.

Originally Posted by Stedy

@Tomcrazy
The MK1 design broke with X-Copy, a change to the logic was made to make the MK2 design that you see on the website. I have before and after logic analyser capture to prove this. The MK2 design is the version that has sold 50 units with no major complaints so far.

I have been working my way through the Amiga Hardware RKRM whilst working on my updated MK3 design, which uses a CPLD. This has different ready logic, primarily due to the support for HD disks.

You don't need to prove anything. If you say it works fine then I take your word for it.

I saw screenshots earlier in this thread from your logic analyzer capturing program. I am curious how you managed to capture so many channels at the same time. Do you have a sound card for your computer with a lot of line-inputs?

So if I understand things correctly, the schematics for the MK2 design are the schematics that can be found on your website? When you say that it has sold 50 units with no problems so far, I'm getting kind of curious if it would work on my machine too. I think I will order one soon and test it on my picky A1200. Can I order the PCB + components unassembled in order to make it a bit cheaper, in comparison to ordering a fully assembled unit?

I haven't been doing much studying on how to utilize the HD feature of PC diskdrives on an Amiga. But I guess it would require more advanced logic compared to just make a PC drive operate properly as a DD drive on an Amiga. Using CPLDs is probably a good choice in such a case. You have given up the idea of using PAL chips, right? I think I saw something about that earlier in this thread if I'm not mistaken.
Btw, do you know if it's also possible to implement things like timing logics and flip-flops in CPLDs? If so, maybe I should start using CPLDs myself since it will greatly reduce the numbers of components being used. I just need to find a programmer first.

[Stedy]

It has been some time since my last update. Work is progressing well.

I have tested the Double Density (DD) support and I am happy with operation as DF0:-DF3 with the new logic design. This works with X-Copy, D-Copy, various techno demos and numerous games, hours of testing

The next build of firmware I have just finished simulating is for the high density mode. To identify the drive as HD there is a sequence I have to respond to:

1. Drive MTRXD- low: Turn on the disk drive motor.
2. Drive SELxB- low: Activate drive select x, where x is the number
the selected drive.
3. Drive SELxB- high: Deactivate drive select x..
4. Drive MTRXD- high: Turn off disk drive motor.
5. Drive SELxB- low: Activate drive select x.
6. Drive SELxB- high: Deactivate drive select x.
7. Drive SELxB- low: Activate drive select x.
8. Read and save state of RDY .
9. Drive SELxB- high: Deactivate drive select x.

repeat steps 7 through 9, 31 more times for a total of 32 iterations,

A double density drive responds with $00000000 an HD drive with an HD disk responds with $AAAAAAAA, so my logic must detect the disk type and respond accordingly. This has been implemented using a state machine.

The half speed trick is something that I will be experimenting with this
weekend.

If the HD experiment works Ok and after sufficient testing, an order for 100 PCBs will be made. I will be getting them assembled by a sub-contractor. Will confirm the final pricing later. Do not expect PCBs for a while yet.

Any updates will be in this thread first, no need to email me asking for information.

Ian

[prowler]

Originally Posted by prowler

I've blown the 74LS38 chip on one of my adaptors by inadvertently shorting it to the Amiga's RF shield whilst removing it from a floppy drive with the (external) power still applied.

I've just bought a pair of replacement 74LS38 chips on eBay for £3.58 including p&p.

[Eclipse]

Originally Posted by prowler

I've just bought a pair of replacement 74LS38 chips on eBay for £3.58 including p&p.

Sorry, I forgot I offered some

[prowler]

Originally Posted by Eclipse

Sorry, I forgot I offered some

Actually, so did I.

No harm done, though, as I have been looking for a reliable mail-order electronic components supplier for a while now. This one on eBay turned out to be J.H Components Limited, Lichfield, Staffs, and they're great people to do business with.

[Supamax]

Originally Posted by prowler

turned out to be J.H Components Limited, Lichfield, Staffs, and they're great people to do business with.

Have you an eBay link to them?

[prowler]

Sure. J.H Components Limited on eBay.

[Eclipse]

I used them to buy my capacitors for my A500 some time a go. They are pretty good.

[cpiac64]

the 74ls38 you can find on _www.farnell.com or _www.rs-online.com or _www.distrelec.it for 1€ exclude p&p

p.s.
on rs are 5 pz soic no dil (4€ exclude p&p)

[Stedy]

Hi,

I have some 8x74LS38 DIL devices in stock if anyone needs some. Message me if interested.

You can buy them from ESR (www.esr.co.uk) for £0.22 each but P&P is £2.50

Will post an update on the updated design once I have completed the current tests.

Ian