Announcing the 'IWMless' IWM substitution - call for beta testers

Hi fans -

While the various propaganda pictures posted by 'Khaibitgfx'  in posts #47, #51, #55 above look nice, they are a bit off topic for this 'beta test thread', as they seem to imply 'victory' for the 'IWMless', which would only be true if all the beta tests were completed (they are not !) and yielded results which allow to start production (a milestone not yet reached).

So let's better stay honest (no propaganda pictures anymore !) and stick to the facts:

'IWMless' ON APPLE IIc PLATFORM

- as of today, only (!) four (!) Apple IIc have been repaired by replacing a dead original IWM with an 'IWMless' beta test unit. This count of four includes my own Apple IIc, so only three beta testers so far. The feedback from beta testers is in this thread:

https://www.applefritter.com/content/results-reports-iwmless-beta-testers

- one further beta tester is still struggling to make his Apple IIc work after the replacement, but this seems to be no 'IWMless' problem so far, it looks as if this Apple IIc has more bad ICs other than the dead original IWM.

'IWMless' ON APPLE IIe / 'Liron' CARD PLATFORM

- two examples of the 'IWMless TEST CARD' have been populated so far and they are now being tested to see if the 'IWMless' could work in a 'Liron' style card. I have one of these two, and 'softwarejanitor' has another one. The third empty PCB is on its way to 'VIBR' in France, so he will eventually have a test platform to make his floppy emu work with it.

- 'softwarejanitor' had some success with running TOTAL REPLAY, see his post #58 above, but with my own 'IWMless TEST CARD' which is identical to 'softwarejanitor's card I saw some freezes and crashes of TOTAL REPLAY.

CURRENT STATUS SUMMARY

While the repaired Apple IIc so far look good, the tests with the 'IWMless TEST CARD' already have reveiled some issues which question the functionality of an 'IWMless' in such a card. And because the 'IWMless TEST CARD' is a 100% identical circuit and firmware copy of the original 'Liron' card, except that it has two 20-pin headers instead of the DB-19 connector to attach floppy disk drives or floppy disk emulators, this also implies that the 'IWMless' in its current form probably can't fix original 'Liron' cards having a dead original IWM. I don't deem this to be catastrophic, because fixing 'Liron' cards was not an initial mission objective for the 'IWMless' - all I wanted is a viable solution to fix Apple IIc having a dead IWM. The potential opportunity to fix 'Liron' cards appeared later, and requires further investigations to see if this is possible at all.

WHO IS THE CULPRIT ?

The big question, as always, is "who is the culprit". My take on this is that the whole 'SmartPort' complex is a really rotten engineering rathole full of wonky and unreliable hardware, firmware and software. My IWM reverse engineering work has exposed nasty bugs in the original IWM and my decision to fix them may have caused some incompatibilities, who knows - this typically happens  when software was 'hacked' to dodge a hardware bug and to make everything work, everybody is happy, until later, the hardware bug is found and fixed, and then the hack doesn't work anymore as expected and instead of making the system work breaks the system. Alas, this is how bug fixes can break things - I have seen enough engineering documents explaining hardware flaws in ICs which must not be touched to keep everything out there in the field working.

(No good because now you have to design that same bug into all the ICs of the same family forever and ever. TELETEXT is one of the known examples for that - the first generation of these ICs made by Philips had bugs and unexpected behaviours and because the TV studio side had developed and deployed hacks to dodge these bugs, all the knockoff TELETEXT ICs made by other companies had to have the very same bugs, too. An engineering nightmare !)

Note that the Apple IIc (the primary target) is a very closed machine with not much variability. Plug in the ROM 4 upgrade and you are good to go and use 'SmartPort'. There are NO other variables in this game, as long as it's an Apple IIc. 'SmartPort' won't work with ROM FF (the first firmware version for the Apple IIc) and ROM 4 is believed to be the latest and greatest firmware version where 'SmartPort' works well enough to be useful. Consequently, I was able to run TOTAL REPLAY hosted on a BMOW Floppy Emu on my 'IWMless' equipped Apple IIc with ROM 4 upgrade 24/7 and no trouble.

But anything 'Liron'-ish on the Apple IIe or Apple II+ is wonky, with or without the 'IWMless', mind you - I always try both the original IWM and the 'IWMless'. Every possible combination of slot number, drive number, CPU type, TOTAL REPLAY version was tried, and some either refuse to start at all, or only run briefly, or run for a few hours, before it hangs or crashes.

Now, tell me who is the culprit ? - I could not tell, as it also happens with the original IWM.

ISSUES WITH WONKY 'Liron' BANK SELECT

More disturbing, I found out that the 2k ROM bank select mechanism of the 'IWMless TEST CARD' does not work flawlessly on my Taiwanese PC-48 Apple II clone. The activation of the 2k ROM page works, but then , when the 2k ROM is accessed, it will turn off at some point (at the same address, and not only when accessing $CFFF). This also depends on the type of the CPU, NMOS or CMOS. I'm still investigating what is wrong there. The very same 'IWMless TEST CARD' works fine in the Apple IIe, as far as the 2k ROM bank select mechanism is concerned. And the bank select circuit is the same as in the original 'Liron' card. Maybe it's a timing problem in this particular PC-48 Apple II clone.

A CAN OF WORMS ?

So you can see that the whole 'SmartPort' complex seems to be  a can of worms, aka "engineering rathole". I can explain some of the problems and know remedies for them, but so far some of the weird behaviour can't be explained (in a "root cause" sense).

DID 'SmartPort' EVER WORK RELIABLY ?

I have a hunch that the whole 'SmartPort' design may be botched and never did work all too well, and this might be the reason why Apple may have decided to not support it any further and to not sell a 'SmartPort' based HDD for the Apple II family. The equivalent to this is Atari's "Parallel Port" of their Atari XL and XE family which, despite of having fancy specs and built in firmware-in-ROM support for "Parallel Port Devices", it never really worked and so they never made and sold any such devices - the only thing that worked was the Atari 1064 memory expansion module for the 600XL, and this is not a true "Parallel Port Device" as specified by their documentation full of untested functions that never worked as intended.

WHAT CAN WE DO ?

It boils down to more testing and analysis work on the 'IWMless TEST CARD' to find out why the hangs and freezes do occur. Once root cause is found, the software/hardware/firmware/'IWMless' RTL may be patched to make it work. This might mean that there may be two versions of the 'IWMless', one which works in the Apple IIc, and one which works in the 'Liron' card. Who knows.

THE 'softSP' TESTS

I've also looked into softSP operation (I finally got one softSP specimen) with the 'IWMless' and it almost works if the 'IWMless' is put into 'compatibility mode' where it emulates a DISK II controller. Alas, TOTAL REPLAY also freezes / crashes after running for a while. Then I repeated the 'softSP' experiment with a real DISK II card, and it froze / crashed after running a while, too. So who is the culprit there ? --- Maybe my Apple IIe is sick.

ONION MODEL ?

It looks like we have the dreaded "onion model" of system debugging at hand: peel away layer after layer of bug ridden code and wonky hardware and "weep" in the process. Until the budgeted money runs out and management dumps the whole project into the trash can. At least this is how it's done in the industry. As hobbyists, we never will run out of budget, and neither will we miss deadlines, because there ain't any, but we might run out of patience. This is because we don't know if it ever will work robustly enough to be acceptable. We don't even know if it ever did work in the first place, back in the 1980s ! --- maybe some people who did use 'SmartPort' back then, perhaps with one of the very rare third party HDDs, could tell us more about whether it worked or not - at least in the 1980s, the hardware platforms were new and not full of 40+ year old electronic components who could contribute to the observed issues.

Comments invited !

- Uncle Bernie

After removing all slot cards except the 'IWMless TEST CARD' and also removing the 'AIIE 80COL/64k MEMORY EXPANSION' card from my Apple IIe, TOTAL REPLAY was able to run for many hours glitch free, no crashes and no hangs, until it reached the 'SERPENTINE' demo where it is expected to hang, which of course it did. 'SERPENTINE' does not hang when running on the Apple IIc equipped with the 'IWMless'. More on this issue with 'SERPENTINE' here:

https://www.applefritter.com/content/asking-you-do-small-experiment-needs-liron-card-or-clone

... the screen photo in post #15 of this thread shows how 'SERPENTINE' freezes / hangs when TOTAL REPLAY is running on a Apple IIe.

So what does this recent experiment prove ?

It proves that the 'IWMless TEST CARD' works fine in a 64k RAM equipped Apple IIe. But it does not allow a conclusion whether the hangs / freezes did come from certain 128k games within TOTAL REPLAY, or just from a bad 'AIIE 80COL/64k MEMORY EXPANSION' card.

To be able to find out, I started to repair a known dead original 'AIIE 80COL/64k MEMORY EXPANSION' card, Apple part #820-0067-B.  At the moment, all the suspect 'uT' DRAMs have been desoldered, and I also desoldered the poor quality socket for the 74LS245:

AIIe_64k_in_repair.JPG

I did not attempt to diagnose or repair the other card I used before, which is a clone of the original Apple card. This leaves the clone card as is, and gives me the opportunity to swap it back. The idea is that if I can get trouble free operation of 'SmartPort' (no crashes / hangs) when using the repaired card, then swapping it against the previous card should bring the crashes / hangs back. If so, this would prove that the previous card was the culprit for the hangs / freezes, and not some issue with the 'IWMless TEST CARD'.

But if both cards fail, then it could be an issue with the 128k game demos or some yet unknown fault in my PAL Apple IIe. It recently got a MMU donated by 'softwarejanitor' to bring it back to life. Most earlier tests with 'Liron'-ish slot cards equipped with the 'IWMless' a few months ago were done with a NTSC Apple IIe sent to me by 'softwarejanitor' for repairs. I repaired it and then used it as a test platform for the 'IWMless', and the 'Liron'-ish card was based on 'bradleyb's 'Liron' clone which he had lent to me under the condition I'd populate it and send it back to him including an 'IWMless'. I only did very brief tests with the MMU from the NTSC Apple IIe transferred into my own PAL Apple IIe, just to see if the 'Serpentine' freeze happens in both machines (it did, and this proves that the 'freeze' is not caused by a difference between NTSC and PAL machines).

So I'm now dealing with a different Apple IIe test platforms than with the one I had a few months ago. This of course means that all the problems I see now might come from defects within the new test platform.

But all of this is not a surprise. We are dealing with > 40 year old hardware here and so all bets are off. Instead of chasing bugs in the 'IWMless' I might see and chase faults in the old hardware. My Taiwanese PC-48 Apple clone is known to be wonky and I saw a previously known good DISK II card fail in its slot #6 when trying out the 'softSP' in slot #5. It would load TOTAL REPLAY just fine and then after a few demos it would hang or crash. Wiggling the DISK II card did help, at least for a while,  hinting at a possible contact problem. Oh the joys of working with 40+ years old hardware !

So there are a lot of suspected defects to sort out. If I can prove the original IWM in the original 'Liron' card is bad, I will desolder this IWM and put a socket in, and then I have a real 'Liron' card to test the 'IWMless' with.

Once I have enough proof that the 'IWMless' works in a 'Liron'-ish slot card for the Apple IIe, then I might launch another beta test series aimed at repairing original 'Liron' cards.

The beta test series for repairing Apple IIc having a dead original IWM will soon enter the 2nd round, I have already built seven more 'IWMless' units, but as I ran short of those ultra low profile sockets, there will be a delay until I have found viable substitute sockets. They are on Digikey but before I have obtained a sample I can't say if it fits.

- Uncle Bernie

In post #61, 'DigitalNZ' wrote:

" I'd love to be a beta tester if you're still looking for volunteers !"

Uncle Bernie answers:

Indeed, more 'IWMless' beta testers are needed !  The first round of beta testers did not yield much insight into the compatibility of 'IWMless' with copyprotected software titles. All feedback I got so far is that "it works !"

Fine. A few Apple IIc saved and brought back to life !

It also was found out that a few of the beta testers had one or more bad DRAM in their Apple IIc - and this must be fixed before attempting to replace the IWM with an 'IWMless'. (I have a whole conjecture how bad DRAMs may lead to a path of destruction of the original IWM, as a bad DRAM can also lead to the dreaded "Check Disk Drive" error, and then while tinkering with the disk drive, users may blow up the IWM).

For this purpose I've published the binary of a DRAM test program here:

https://www.applefritter.com/content/dram-test-eprom-all-apple-ii-except-iigs-maybe

... and I recommend it to any prospective 'IWMless' beta tester wanting to repair an Apple IIc. All you need is a 27128 or 27256 pin compatible EPROM and an EPROM burner. Follow the instructions in the above link. This way you can find out whether your Apple IIc needs the 'IWMless' at all. Repair the DRAM before ordering an 'IWMless' - because the original IWM may be intact, and the culprit for the "Check Disk Drive" error message may have been the DRAM alone.

I'm currently finishing up the BOM I need to build some 'YAAKEN' keyboard encoders and once I have that, I'll order components along with new ultra low profile sockets. Once I have them and found them fit (literally, as in: mechanical fit), then I will make another call for 'IWMless' beta testers here in this thread. Alas, the price will increase as these new ultra low profile sockets are very expensive and USPS has increased the minimum postage for parcels again (seems to become a bad habit, it's called "inflation", also seen in grocery bills ... just yesterday I paid $73 at Whole Foods just for two tiny filet mignon steaks (1.3 lb meat), one "Bonhomme Brie" cheese box imported from France, one tub of vanilla icecream and two bags of organic flour, as I bake all my bread myself since they dropped their marvelous organic bread and only offer the same toxic crap bread you can find everywhere for half the money --- imagine if you have to feed a family at these food prices !)

Stay tuned ! (Oh, and buy durable goods you need to replace soon now ... prices may increase dramatically, except for real estate, where I expect the prices to collapse, but of course, predicting the future is difficult and uncertain)

- Uncle Bernie

In post #65, 'tiktok4321' wrote:

" It appears that the IWM on my IIc+ is failing.  Is there a potential solution for that ? "

Uncle Bernie answers:

Sorry, but the 'IWMless' was deliberatly specified to not support the FAST mode of the IWM. Which is needed only for the 3.5" floppy disk drives in the Apple IIgs, the Apple IIc+ and the Macs. The Apple IIc runs only at 1 MHz CPU clock which is too slow to handle FAST mode, and it is never used in the IIc nor in the 'Liron' card for the other 1 Mhz Apple II members, so I could leave FAST mode out., allowing to fit the design into a smaller and cheaper CPLD.

Another reason for this is that for the IIc+ and the IIgs there is not a large enough marketplace / demand for an 'IWM' replacement and I'm not interested in any of the Macs, as I never had one. Development of the 'IWMless' was driven by my desire to a) repair my own Apple IIc, and, later, b) to allow my 'Replica 2e' project to also emulate the Apple IIc with SmartPort etc. ('Replica 2e' will never emulate the Apple IIgs, as the IIgs deviates far too much from the basic Apple II architecture and never was successfully cloned, so if the Taiwanese cloners could not do it back in the day when there was money in it, I better don't waste my own time on trying that feat).

HOW TO REPLACE An IWM IN THE APPLE IIc+

If your Apple IIc+ really needs a new 'IWM' you can always use an Apple IIc as an organ donor: take the original IWM out of the IIc to fix your IIc+, and then use an 'IWMless' the make the donor whole again - it even will be better as with the original IWM, as 'IWMless' has the DISK II compatibility mode.

HOW TO DIAGNOSE MOTOR TIMER PROBLEMS

I think it's unlikely that your drive motor keeping spinning problem comes from the IWM. You can check the logic level on the /ENBL1 pin of the IWM (pin #20) to see if it ever tries to turn the motor off (going from low to high) but fails to reach a good 'high' level so the disk drive does not know what to do. A poor 'high' level due to a weak high side driver transistor in the IWM could be helped with a pull up resistor of 1k (or around that value) to +5V (IWM pin #26). You could also enter the monitor program to give motor on and motor off commands manually. If the internal motor timer of the IWM was faulty, it would not attempt to rise the /ENBL1 pin from low to high.

Again, I think it is unlikely that the IWM develops such an internal fault.

THE MOST LIKELY REASON WHY 'IWM's GET DAMAGED

The most common way how the IWM 'dies' has to do with overvoltages applied to its SENSE and RDDATA inputs which may happen when people tinker with the floppy disk drive after getting a "Check Disk Drive" error, which may have been caused by a bad DRAM. All the three Apple IIc with a dead IWM I had acquired over the last two decades also had a bad DRAM, which might have set the chain of events in motion that then did cause damage to the IWM.

In any case, manual motor on / motor off commands and measuring the actual logic levels (a cheap multimeter is enough) will allow you a diagnosis of the problem.

- Uncle Bernie

P.S.: Motor on command: read location COE9,  Motor off command: read  COE8. Allow ~1 sec of motor off delay time.

In post #67, 'microbmen' wrote:

 "I have a iic motherboard with a failing IWM.  It behaves odd, working with some smartport devices, but not all.  Seems like it's a power draw issue.  Too much pull and the chip fails."

Uncle Bernie comments:

Unless the IWM gets really hot to the touch it does not draw too much power. I'd recommend you to check your IIc's 5V supply first, and also disconnect the internal floppy disk drive and plug in a Floppy Emu configured to boot DOS 3.3   .  .  . this can tell you if the problem is in the floppy disk drive.

Other than the DRAM problem (for which the IIc is notorious) there is another systematic problem in the IIc and this is a race condition between the FCLK (pin #24) and Q3 (pin #25) clocks on the IWM. This is a "hazard" present in all Apple II but generally does not manifest in any slot card that uses only one clock. The IWM does use both clocks and hence has the "hazard". Apple knew about the problem and added a series resistor on their "Liron" card to delay Q3 by a few 10 ns, hereby defusing the race condition. I did not have enough Apple IIc on my lab bench to know if they ever tried the same remedy on the IIc. The only IIc I have left over is my "lab rat" for 'IWMless' work, and on this motherboard the FCLK and Q3 go directly from the TMG PAL to the IWM. Which is why the 'IWMless' has a RC delay and a Schmitt Trigger on its PCB to delay Q3 in a more controlled manner than a series resistor alone could do (which uses the parasitic capacitance of the IWM's Q3 input pin and has a poorly defined logic threshold).

This said, here is a proposed way to tackle the problem:

- check your DRAMs first ... the exact failure mode of these uT4264-15 is still unknown but it is tricky, as not all DRAM tests can catch it. Some people with a lot of skill and patience replace all DRAMs, but this may be excessive. I have repaired the DRAM in maybe half a dozen Apple IIc, and it always has been one or two bad DRAM ICs and all were  the lower right corner of the PCB (positions 16-19). But this anecdotal evidence based on only half a dozen cases does not mean it's a general rule.

- once the DRAM has been brought to fully functional condition, and you still have an issue with IWM related functions, try a Floppy Emu in lieu of the internal disk drive. If the problem goes away, the culprit is the disk drive.

- If the internal disk drive was not the culprit (it can be defective and interfere with SmartPort operations) try to add a delay in the IWM's Q3 clock. This involves cutting a trace and adding a resistor over the cut, try values from 220 Ohm to 470 Ohm.

I would recommend to anyone suspecting problems with the IWM in an Apple IIc to go through the above procedure first. Desoldering the IWM is tricky and the 40+ years old PCBs are not in good condition anymore, in most cases I saw some "bubbling" of the solder when soldering in the ultra low profile sockets and this is a bad sign hinting at a problem with the PCB. So you would want to avoid swapping any ICs in the Apple IIc unless absolutely necessary. This is why I don't change all the DRAMs at once ... doing so has a 16 x higher risk of damaging the PCB than just changing one DRAM. Sure, a skilled operator could also fix damaged PCBs by adding flight wires but it's tricky to find out exactly where the defect is (i.e. hairline crack, or a plated through hole gone bad - the solution for the latter is to resolder the IC pins on both sides of the PCB).

The bottom line is that only if the IWM has been confirmed to be bad, replace it.

I've still three 'IWMless' beta test kits left, send me a PM if you want one, but sorry, I can't send them to the EU or any other country with insane customs paperwork requirements or "RoHS" bans, etc. The "EU" is probably the worst offender, see my post #9 of this thread:

https://www.applefritter.com/content/smartport-disk-emulators-and-daisy-chaining

to find out why I can't send them to the EU. It's really bad, and I don't think it will get any better. The times where you could buy collectible electronics from all over the world is definitely over ... they are now being treated as "toxic waste" due to lead based solder and won't be allowed to enter the EU (they get a red sticker with the text "not conforming to the way the EU works" or such, and go back to the sender, if you are lucky, otherwise they just disappear). Oh, and here in the USA we can't order stuff from abroad either, because of unpredictable tariffs and, worse, insane and surprise "handling fees" the carriers slap onto parcels, these are independent of the declared value, so your few components bought in China worth $2  might be hit with additional $16 "handling fees" extorted from you by the greedy carrier ... looks like a criminal racket and a scam.

- Uncle Bernie

In post #69, 'superbenk' wrote:

" The drive does the head banging thing at power on but then just spins the disk constantly without seeking ...  Pin 1 (PH0) is high while Pin 2 (PH2), Pin 27 (PH4?), Pin 28 (PH3?) are all stuck low.  I do NOT get the "Check disk" message at all (with or without a disk in the drive) ... I would love to be a beta tester of IWMless if they are still available. "

Uncle Bernie answers:

These symptoms are not necessarily caused by a dead IWM. Can you run the built-in memory test ? It is  quite weak but still good enough to detect most trouble with the DRAMs. It will also give you a sign of life for the CPU being able to execute code, you can see this from the screen pattern changing all the time (some short pauses are normal, when the memory test works elsewhere). If this test runs and runs, the basic machine is most likely OK and trying to change the IWM against an 'IWMless' may be worth an attempt. But success is not guaranteed. Bad Apple IIc can suffer from many ailments. I once had half a dozen IIc and ended up with two that work (with 'IWMless', all of them had bad IWMs).

At the moment I have two 'IWMless' beta test kits left over, while the beta testers slowly churn through the Apple II software base (3500+ titles). So far it looks good, but progress is slow, and I did not get progress reports for a while. But this is understandable, the volunteers also have a bread job and a family life, so they can't work all the time on the testing.

If you want a beta test kit, just send me a PM. I can send it easily to the lower 48, or any other country other than the EU, Norway and Switzerland.

- Uncle Bernie

In post #72, 'superbenk' wrote:

" I think UncleBernie's RAM tester ROM is 16k - I'm assuming I can just copy it 4 times & not need to mess with the solder bridges on W1/2?  Of that works I'll give it a try.   "

Uncle Bernie comments:

For my DRAM memory test, you can always use larger EPROMs as long as they fit into the socket and have the industry standard pinout all 27XXX EPROMs do have. For FLASH devices you need to be more careful about the few pins which are different because they control write/erase functions. And for EPROMs, check that VCC (+5V) is present on all the "other" pins (other than pin #28) where VCC might be required. Just consult the datasheets and find out the differences between the usual IC in the target socket and the EPROM you want to use. Then program that EPROM with the DRAM test image repeating as often as needed to fully fill it - this prevents any issues with banks selects etc. After successful programming, bend the conflicting pins by 45 degrees so they don't contact the socket and flight wire them to the VCC or GND as needed to get the regular READ function for the EPROM or FLASH memory IC.

I would not attempt to replace an original IWM unless the DRAM of the machine has been confirmed to work for the lower 48k - this is all you need to boot DOS 3.3 and once you can boot DOS, you can run more sophisticated DRAM tests which test all the banks in all the various models of Apple II. My DRAM test in ROM intentionally tests only these lower 48k - it is meant to be used for near dead Apple II and it can even telegraph the bad bit# out even when the DRAM is so bad that no useful text screen can be obtained. But I forgot how that telegraph trick works - would need to find the source code, which is difficult. In the Apple-1, the same DRAM test algorithm fits into 256 bytes but does not have the "telegraph" feature as it assumes the terminal section of the Apple-1 works.

- Uncle Bernie