Thread
Daystar IIsi/SE/30 Adapter, $150 on Ebay!!?
Oh yeah, stupid question JDW:
Do you still have the same floppy drive problems with a DayStar adaptor (either IIsi or SE/30) as with the TwinSpark?
Also, I've got a IIfx ROM SIMM now... I should try and see if my (external) floppy drive works with the Turbo040, IIsi adaptor, and IIfx ROM SIMM.
When we worked on this previously, my lack of a non-stock ROM prevented me from attempting to reproduce your issue...
Do you still have the same floppy drive problems with a DayStar adaptor (either IIsi or SE/30) as with the TwinSpark?
Also, I've got a IIfx ROM SIMM now... I should try and see if my (external) floppy drive works with the Turbo040, IIsi adaptor, and IIfx ROM SIMM.
When we worked on this previously, my lack of a non-stock ROM prevented me from attempting to reproduce your issue...
I would love to see a crisp photo (both sides), at 1024 pixels, of that "nicely elevated" Daystar adapter, trag!
A kindred spirit. That is why I mentioned earlier in this thread that I sold a IIsi adapter for US$35 (after informing the buyer that he would need to cut up his metal chassis), which included international EMS express mail shipping to the buyer in the US. I felt no need to reap tremendous profit off a fellow classic Mac enthusiast.
Nevertheless, both you and I have admitted in this thread that Manabu Sakai has clearly spent a considerable sum on R&D for that TS Adapter; hence his desire to get a decent ROI wherever possible. I therefore will refrain from condemning the man for his $59 price tag on the ROM upgrade. But I will admit, it is "not cheap."
Full version 040.
I would love to borrow someone's IIci ROM to give it a compatibility test. But in the meantime, I await Equill's test results.
It's difficult for me to see it as a "ROM incompatibility" alone because again, the DiiMO has no problems at all, regardless of ROM. Could be the 040, I guess. But it could also be the TS Adapter. Since Equill will be testing without the TS Adapter, he can offer further insight into this matter.
As to borrowing your IIsi adapter, I would first love to see some photos of what you have!
Be that as it may, have a look at my Flickr photo. You will see the VLSI chip (ASIC version) on my card. The B&W photo taken of the Daystar User Manual shows the older, non-ASIC version of the card. I once owned that older version too. Both the ASIC and non-ASIC versions of the card exhibited the 1.4MB floppy incompatibility when used with non-stock ROMs and the TS Adapter. So the problem doesn't appear to be related to the ASIC or absence thereof.
Tyler, I tested that a very long time ago (early 2005? late 2004?). I just don't remember about the result with the IIsi adapter. I can only remember that it works with the Turbo 040 in the SE/30, except for the fact you must cut up your metal chassis to make it fit.
For this reason, I look forward to Equill's test results, as he will be testing with a Daystar IIsi adapter, I believe. If he then puts in a IIfx or IIsi ROM and has no problems with reading/writing 1.4MB floppies, then clearly, it is the TS Adapter that is at fault. But if he too has the same problem, then it is not the TS Adapter but rather a ROM incompatibility which, apparently, is not fixed even in the latest 4.11 ROM. (In which case we need a Daystar engineer to write us an even newer ROM fix. [
] ]'>
Jeff, forgive me. At times, I can be about as dense a neutron star!
Thank you again for the wonderful capacitor offer. And I would encourage my fellow SE/30 owners on this site to take advantage of your offer. It's a great way to replace all the caps on the logic board, with top quality replacements!
Jeff, since I read that you supplied Gamba with some IIci ROM chips (and code) back in 2003, would you happen to have the means to make a IIci ROM? I am skeptical if it would work, but I love to test!
JDW: Last night I tested the floppy drive with this configuration:
SE/30 (80MB RAM)
IIsi adaptor
Turbo040 (later ASIC version) (4.11 ROM IIRC but I can TattleTech it if needs be) (40MHz 68040 @ 48MHz overclock)
Asante MacCon si
IIfx ROM SIMM
7.5.5
External 1.44MB floppy (upgraded external 800k drive with auto-inject SuperDrive mechanism)
Same results as you: proper operation with 800k floppies, inability to read 1.44MB ones (shows 'disk is unreadable' dialog box and formatting fails almost immediately)
Having reproduced your results, I think the Turbo040 and the IIfx ROM SIMM are incompatible in the SE/30. The only better test would be a Turbo040 attached to the proper DayStar SE/30 adaptor, but I don't imagine that there'd be any difference, since they're the same circuit, differently shaped.
SE/30 (80MB RAM)
IIsi adaptor
Turbo040 (later ASIC version) (4.11 ROM IIRC but I can TattleTech it if needs be) (40MHz 68040 @ 48MHz overclock)
Asante MacCon si
IIfx ROM SIMM
7.5.5
External 1.44MB floppy (upgraded external 800k drive with auto-inject SuperDrive mechanism)
Same results as you: proper operation with 800k floppies, inability to read 1.44MB ones (shows 'disk is unreadable' dialog box and formatting fails almost immediately)
Having reproduced your results, I think the Turbo040 and the IIfx ROM SIMM are incompatible in the SE/30. The only better test would be a Turbo040 attached to the proper DayStar SE/30 adaptor, but I don't imagine that there'd be any difference, since they're the same circuit, differently shaped.
Nothing to forgive, but I thought you might not have made the connection when you encouraged me to post a FS message in Trading Post and I already had one there under my trag monicker.
Gamba's IIci ROM worked, but it may not solve the problem that you are seeing. I think we may need to actually decode the PAL on the IIsi/SE/30 adapter and figure out what it is doing to solve this problem. Hmmmm. Thinking along those lines, has this problem been shown with the IIsi ROM in the SE/30? The IIsi works with that adapter (presumably), so why wouldn't the SE/30, with a IIsi ROM.
I still have the ability to make the Flash chips with IIci ROM code on board. That is what I supplied to Gamba. I chose the chips and sent Gamba the specs. He laid out and fabricated the circuit board, with a little bit of advice from me (but not much) and I sent him the programmed chips, which he attached to the board. Shortly after that he disappeared. :-(
I don't have the ability to easily make the ROM circuit board, and none of the existing modules use a variety of chip which could be easily desoldered and replaced.
I do have a board layout designed for the ROM module which will be compatible with the incredibly common PLCC32 packaged flash chips. But turning that into a board would either require many more hours of fabrication than I will realistically be investing any time in the next 6 months at least, or somewhere in the neighborhood of $400 - $600 to pay someone like 4pcb.com to fabricate the design.
I also have the firmware code for the various Turbo040 versions extracted around here somewhere. Probably on the hard drive of that old Windows '95 machine that drives my chip programmer...
Anyone here the least bit skilled at reading Mac machine code or doing a disassembly? Come on, don't be shy. Raise your hands.
In this case, what I would hunt for is code at the beginning of the Turbo040 firmware which checks what the host machine identity is. The host machine identity is going to be taken from the ROM. So when a Turbo040 is in an SE/30, but the SE/30 is using a IIfx or IIsi ROM, then the Turbo040 is going to behave as if it is in a IIsi or a IIfx--if it has code to check the host machine identity. Apparently, this is not a problem except for floppy drive timing. Maybe. It's a decent hypothesis. If such a check is done, we could create a version of the Turbo040 firmware such that it uses the SE/30 results even when it thinks it is in a IIsi or IIfx.
Hmmm. The Turbo040 cannot go in a IIfx. There's no adapter for it. So what set of parameters is it using if it finds itself in a IIfx? I would guess that it is defaulting to IIci parameters, which would imply that using a IIci ROM won't get us any better results than the IIfx ROM does.
Another option would be to find the location of the machine identifier in the ROM code and change it to SE/30's code. As long as one is burning the ROM code onto Flash anyway, it would be easy enough to do a small edit before doing the burn. That is, assuming that the machine identifier is not located in many places.
I was thinking (out loud, too soon) in terms of TattleTech, too, but the DayStar CP QuadControl shows the card's ROM version without even being prompted (other than by opening the CP).
Your neutron star is on the rise again, J.
My DayStar Turbo040 is in my IIci, as posted above, and it's the MicroMac Carrera040 that I shall use in the first SE/30 to be accelerated, not least because of its altogether more compact 030 PDS adapter. I have a IIfx ROM (bought from catmistake, some time ago), but it may not see use in the other SE/30 for a while unless another Turbo040 comes to live with me. That could be soonish, or not. After all, the first one was six years in the waiting.
de
My impression was that the reason the IIfx and IIsi ROMs worked in the SE/30 was that they were "universal" ROMs with drivers for every device in every (modular) Mac ever made up to the date of their manufacture.
With that in mind, the natural assumption would be that the toolbox routine that returns the gestalt ID doesn't just pull it out of the ROM as a constant, but instead queries the hardware and determines the model of the Mac based upon what it finds using some heuristic. The fact that you have to modify the gbly resource in your System file after a ROM swap neither confirms nor denies this: one one hand, the fact that you have to change anything at all means that the ROM swap changes _something_ but on the other hand, the changes you have to make are the same whether you're using a IIsi or IIfx ROM.
So, it doesn't seem like putting a IIfx ROM into your SE/30 makes it identify as a IIfx, nor does using the IIsi ROM identify it as a IIsi, yet putting a non-stock ROM does make it identify as something *different*... can anybody run TattleTech on an SE/30 with a non-stock ROM and tell us what model TattleTech thinks it is? If not, I'll try to do it when I get to work.
If the debian initrd images work, I'll probably try the floppy drive under Linux, where there's supposedly a working SWIM driver. If that works, it at least proves that this problem is fixable in software...
With that in mind, the natural assumption would be that the toolbox routine that returns the gestalt ID doesn't just pull it out of the ROM as a constant, but instead queries the hardware and determines the model of the Mac based upon what it finds using some heuristic. The fact that you have to modify the gbly resource in your System file after a ROM swap neither confirms nor denies this: one one hand, the fact that you have to change anything at all means that the ROM swap changes _something_ but on the other hand, the changes you have to make are the same whether you're using a IIsi or IIfx ROM.
So, it doesn't seem like putting a IIfx ROM into your SE/30 makes it identify as a IIfx, nor does using the IIsi ROM identify it as a IIsi, yet putting a non-stock ROM does make it identify as something *different*... can anybody run TattleTech on an SE/30 with a non-stock ROM and tell us what model TattleTech thinks it is? If not, I'll try to do it when I get to work.
If the debian initrd images work, I'll probably try the floppy drive under Linux, where there's supposedly a working SWIM driver. If that works, it at least proves that this problem is fixable in software...
Yes, same problem when using the IIsi ROM. But again, I am doing the test with a TS Adapter. I am not able to test the IIsi ROM with the IIsi adapter because I sold the IIsi adapter a couple years ago.
So Tyler's test confirms the problem! Yeah! Not only confirms it, but also confirms it on EXTERNAL drives too -- I only confirmed it on the INTERNAL floppy drive. So Tyler's test in combination with mine shows that neither the IIsi nor the IIfx ROM is compatible (IIfx ROM having been tested with the IIsi adapter in Tyler's SE/30, and both ROMs having been tested with the TS Adapter in my SE/30).
Tyler also mentioned his RAM, OS and Ethernet configuration. But I tested with many different RAM sizes, different RAM chips (sometimes filling all slots, sometimes only one bank, etc.), OS's from 6.0.8 through 8.1, with and without the MacCon attached. The same 1.4MB read/write problem occurs regardless of RAM, OS, and Ethernet configuration.
It also occurs regardless of the INITs, including the Daystar QuadControl CP used with the Turbo 040. Yes, I tested with it and without. And when I had the CP running, I changed various cache settings -- every combination I could think of. Nothing worked to solve the 1.4MB floppy problem EXCEPT putting the stock SE/30 ROM back into the SE/30.
I actually wrote extensively about all my various tests on this site, prior to the hard drive crash a year ago. All those hours of typing in my detailed reports vanished when that crash occurred. Here are the only older threads still in existence on this subject: Thread#1 & Thread#2.
Prior to the crash, the threads containing my posts were found at the following URLs:
http://www.68kmla.org/viewtopic.php?p=34873#34873
http://www.68kmla.org/viewtopic.php?p=48778#48778
http://www.68kmla.org/viewtopic.php?p=53786#53786
http://www.68kmla.org/viewtopic.php?p=53921#53921
But even Google doesn't have that info caches anymore.
I've discussed this in the past a bit in this AppleFritter thread. But that thread mainly talks about on screen artifacts and vertical bars. It does have some photos of interest for you though.
I've always wondered why he fell off the face of the planet, but perhaps he literally is no longer with us on this earth?
One last note...
Back in December 2005, Tyler and I exchanged some emails about the .SONY driver and a ROM patch Tyler put together for me. Looking through our old emails, I see that we played with this in a resource editor for a number of days, but ultimately the patch didn't work. :'(
So perhaps if anything needs to be patched, it would appear to be the ROM code of the Turbo 040, which for whatever reason is incompatible with IIfx and IIsi ROMs when used in the SE/30.
Just to clarify, my SE/30 with Turbo040 and IIfx ROM SIMM reports Gestalt ID 9, which is the normal SE/30 Gestalt.
My Turbo040 has 4.01 ROM. I'm having terrible stability problems running 7.5.5 with IIfx ROM. (I don't care because I'm reverting to stock ROM right now, but wanted to mention it.)
My Turbo040 has 4.01 ROM. I'm having terrible stability problems running 7.5.5 with IIfx ROM. (I don't care because I'm reverting to stock ROM right now, but wanted to mention it.)
I found the IIsi and IIfx ROMs to be the same in terms of compatibility when used in conjunction with the Daystar Turbo 040 40MHz card. I didn't notice much of a difference when using the stock SE/30 ROM either. Most of the incompatibilities I saw had to do with the cache in the 040 -- some software just won't run at all or will crash, even if you disable the 040 cache in QuadControl.
The only difference I've seen between the IIfx and IIsi ROMs is that there are no horizontal lines at cold boot with the IIsi ROM. But you get the lines with the IIfx ROM.
The only difference I've seen between the IIfx and IIsi ROMs is that there are no horizontal lines at cold boot with the IIsi ROM. But you get the lines with the IIfx ROM.
I was getting hangs during INIT loading with the IIfx SIMM; they'd go away if I'd take out half my control panels then put them back in over the course of a few reboots. Eventually I'd get it running on full extensions, IIfx SIMM, and Turbo040, but it took a lot of reboots to get there.
I'm guessing that the 4.11 ROM is more compatible with the ROM-swapped SE/30 than 4.01.
I don't think I've ever had any software that's crashed on the Turbo040 except for the VM kernels for MachTen (and those are noted in the Turbo040 readme as being incompatible).
Oh, well, I learned that the gestalt doesn't change with the ROM swap, which is what I wanted to know. :-D
With the IIfx ROM SIMM I don't get any funny types of patterns at cold boot, horizontal lines or otherwise. Just a longer time before the screen clears and the happy Mac shows up.
Also, since I had my SE/30 open to change the ROM SIMM, I took a few photos to document my setup. You can pull them up at http://www.fenestrated.net/~macman/se30_turbo040/ and see what the 'frame slice' modification actually looks like!
You can also see the modifications I made to my Turbo040 when overclocking it to 48MHz. The crystal oscillator has been socketed and the 20MHz unit replaced with a 24MHz one. I added a blower to the factory heatsink, which gets rid of the crashing problem. I affixed the heatsink to the 68040 using the old overclocker's 'super glue and arctic silver' method, removing the frag tape DayStar had used. I also put some little heatsinks on the ASIC and cache SRAM just for fun; they probably don't help much but they can't hurt, right?
I'm guessing that the 4.11 ROM is more compatible with the ROM-swapped SE/30 than 4.01.
I don't think I've ever had any software that's crashed on the Turbo040 except for the VM kernels for MachTen (and those are noted in the Turbo040 readme as being incompatible).
Oh, well, I learned that the gestalt doesn't change with the ROM swap, which is what I wanted to know. :-D
With the IIfx ROM SIMM I don't get any funny types of patterns at cold boot, horizontal lines or otherwise. Just a longer time before the screen clears and the happy Mac shows up.
Also, since I had my SE/30 open to change the ROM SIMM, I took a few photos to document my setup. You can pull them up at http://www.fenestrated.net/~macman/se30_turbo040/ and see what the 'frame slice' modification actually looks like!
You can also see the modifications I made to my Turbo040 when overclocking it to 48MHz. The crystal oscillator has been socketed and the 20MHz unit replaced with a 24MHz one. I added a blower to the factory heatsink, which gets rid of the crashing problem. I affixed the heatsink to the 68040 using the old overclocker's 'super glue and arctic silver' method, removing the frag tape DayStar had used. I also put some little heatsinks on the ASIC and cache SRAM just for fun; they probably don't help much but they can't hurt, right?
Hmmm. Does the floppy drive work properly with the IIfx or IIsi ROM and *no* Daystar upgrade? How about PowerCache vs. Turbo040?
JDW, I remember that earlier thread. But it wasn't a topic in which I had a great deal of interest at the time. I think I was focused on my IIfx RAM project then.
JDW, I remember that earlier thread. But it wasn't a topic in which I had a great deal of interest at the time. I think I was focused on my IIfx RAM project then.
Nice pics. It is fun to see other folks' rigs.
Can you elaborate on how you added the blower to the heat sink? Or did you remove the factory heat sink and then reattach the whole assembly including the blower?You can also see the modifications I made to my Turbo040 when overclocking it to 48MHz. The crystal oscillator has been socketed and the 20MHz unit replaced with a 24MHz one. I added a blower to the factory heatsink, which gets rid of the crashing problem. I affixed the heatsink to the 68040 using the old overclocker's 'super glue and arctic silver' method, removing the frag tape DayStar had used. I also put some little heatsinks on the ASIC and cache SRAM just for fun; they probably don't help much but they can't hurt, right?
Nice overclocking job. Any time I get a Turbo040, after I test it, the first thing I do is add an oscillator socket. They are just too overclockable not to. Especially if you get something like a 25 MHz model.
In short, the blower is just glued onto the side of the heatsink; I glued it after the heatsink was glued to the 68040, but there's no reason it couldn't have been done the other way around.
This setup has evolved over time; here's the story of the cooling solutions my Turbo040 has used...
Originally I mounted a normal fan on the end of the card stack to blow over the stock heatsink. This was highly inefficient because most of the fan was blocked; eventually that fan died. The SE/30 ran most of the time but would occasionally hang.
Next, I removed the stock heatsink from the 68040 and mounted a GeForce 4xx GPU HSF assembly with frag tape. This was also loud and quite inefficient but it was from a really cheap graphics card; the "heatsink" was crimped together in places and had really poor thermal conductivity. Sometimes the SE/30 would crash when it was hot outside. Also, the height of the HSF was slightly more than the stock heatsink, causing even more severe fitment problems in the card stack.
To improve upon this I put the stock heatsink back on and then carefully glued the blower to the side of the stock heatsink. The blower is the CPU fan from a dead white iBook, by the way. This worked OK but I noticed that the chip itself was *way* hotter than the heatsink. That was when I decided to try the "super glue in the corners, arctic silver everywhere else" trick which seems to have worked. When you touch the chip and heatsink, they feel like about the same temperature.
I wish there were enough room to just stick a Pentium III or IV heatsink on it and just call it done, but alas that's just not what it's like being inside an SE/30.
This setup has evolved over time; here's the story of the cooling solutions my Turbo040 has used...
Originally I mounted a normal fan on the end of the card stack to blow over the stock heatsink. This was highly inefficient because most of the fan was blocked; eventually that fan died. The SE/30 ran most of the time but would occasionally hang.
Next, I removed the stock heatsink from the 68040 and mounted a GeForce 4xx GPU HSF assembly with frag tape. This was also loud and quite inefficient but it was from a really cheap graphics card; the "heatsink" was crimped together in places and had really poor thermal conductivity. Sometimes the SE/30 would crash when it was hot outside. Also, the height of the HSF was slightly more than the stock heatsink, causing even more severe fitment problems in the card stack.
To improve upon this I put the stock heatsink back on and then carefully glued the blower to the side of the stock heatsink. The blower is the CPU fan from a dead white iBook, by the way. This worked OK but I noticed that the chip itself was *way* hotter than the heatsink. That was when I decided to try the "super glue in the corners, arctic silver everywhere else" trick which seems to have worked. When you touch the chip and heatsink, they feel like about the same temperature.
I wish there were enough room to just stick a Pentium III or IV heatsink on it and just call it done, but alas that's just not what it's like being inside an SE/30.
Since you mention multiple reboots, I just don't know that it is a software incompatibility in your case. I would be very curious to see if you get that same result with a 40MHz or lower clock speed. Because again, my stock 40MHz card doesn't exhibit those hands during boot, perhaps because I have not overclocked the card.
So we either need to underclock Tyler's card to test this. Or we need Equill and a few others to try a high speed (40MHz preferably) Turbo 040 to see their results in the SE/30.
The 1.4MB floppies read/write PROPERLY when the Daystar Turbo 040 is removed, even though a IIfx or IIsi ROM is left in the SE/30. However, since I am using the TS Adapter, I must also mention that removal of the Turbo 040 also requires removal of the TS Adapter. You cannot simply use the TS adapter's top slot to insert a MacCon Ethernet card. You must use it for an accelerator. In other words, the middle slot is mandatory on the TS Adapter, while the top slot is optional.
So simply saying that "it works fine" without the Turbo 040 does not in any way indicate if the problem is the Daystar card or if the problem is the TS Adapter -- since both are being removed.
PowerCache? I wouldn't know. I tried one 50MHz Daystar 030 card which also had 50MHz 68881 co-processor. But as you can see in my Sad Mac photos, I only had crashes with it. I ultimately had to return it to the seller. I think the reason it didn't work in my SE/30 is because it was not the UNIVERSAL version of the 030 card.
What I can say is that the DiiMO 50MHz 030 (with co-processor) works fine with any ROM. No issues with the floppy drive. However, I do not use the TS Adapter along with the DiiMO because the DiiMO doesn't require an adapter like Daystar cards do. But again, this does not necessarily say the TS Adapter is at fault. It still could very well be a Daystar ROM incompatibility with the floppy drive system of the SE/30.