SE/30 with Micron Xceed Grayscale & Socketed Daystar Acceler

[JDW]

The holy grail of SE/30 computing. What compact Mac user among us hasn't lusted after this elusive upgrade combo? I myself certainly have for years. Well, today I am pleased to announce that thanks to an extremely considerate tip by my friend Gary_W in this forum, I am now a proud owner of this truly "rare" setup and I wish to share some details about it.

First some historical tidbits. Although I had previously used C-64's and Radio Shack computing devices in the early 80's, my first introduction to a "real" home computer was the Macintosh 128k and ImageWriter I that my father purchased for our family in early 1984. Being only 13 at the time and a techno-geek, I had no need for computer classes to teach me all the ins-and-outs of this revolutionary new machine. Of course, I supplemented my hands-on knowledge with tech news from various sources, including Macworld, MacUser, Byte, and MacWeek magazines. Unfortunately, I never got onboard the DTP bandwagon because I couldn't afford a Laserwriter. But I always loved wowing school teachers and friends with reports and printed graphics on my ImageWriter I (and later ImageWriter II) that were impossible to reproduce on any other kind of home computer at the time. Then in 1989, a tidal wave of new, jaw dropping machines came out, starting with the SE/30 and concluding with the IIci and Mac Portable a few months afterward. That was the year I began working as a service tech at the now defunct Micro Age Computers in Fresno, CA, and allowed me some serious hands-on. Suffice it to say I was green with techno-lust while working there. It was only "window shopping" for me though because even the black-and-white SE/30 cost in excess of $5,000 with hard drive at the time. Ownership of those glorious 1989 machines was merely a dream for me.

Skip forward to 2004. Now living in Japan, I found myself nostalgic for the glory days of computing back in the '80s. I opened an EB*Y account and made my first purchase from Sun Remarketing. (I had known about Sun for many years, seeing their ads in Macworld and MacUser magazines in the early 1990's.) It wasn't long after that I had acquired my first SE/30, and for less than 6% of original cost in 1989. In the months that followed I studied up on the SE/30 and spent way to much money on enhancements for it. Specifically:

• • DayStar Turbo 040 40MHz PDS card
  • The outlandishly expensive TS Adapter from Manabu Sakae at ARTMIX in Japan that allows you to install multiple PDS cards at once.
  • DiiMO 50MHz 030 PDS accelerator (because it's more software compatible than the Turbo 040, albeit slightly slower)
  • MacCon PDS Ethernet card (connects great with my G4 Cube and gets my SE/30 on the internet)
  • 128MB RAM (16MB SIMMs)
  • 4.5GB SCSI IBM DGHS hard drive mechanism originally used in servers back in the day
  • External Apple HD20 SCSI hard drive enclosure
  • Apple IIgs ADB keyboard
  • Apple Pro powered speakers
  • PowerKey Classic (purchased new from the manufacturer)
  • Replacement logic boards
  • Replacement 1.4MB internal floppy drive
  • External 800k floppy drive (Apple)
  • Floppy drive cleaning kit
  • Recapping kits from our very own Trag
  • Silenx fan to replace the mind-numbingly loud stock Elina
  • Secondary PSU to allow me to install lots of upgrades and the SCSI drive internally without overtaxing the primary PSU
  • Another fan to cool the Secondary PSU inside the SE/30
  • All topped by some RetroBrighting, 3% H202 courtesy of our very own Lumpydog
  

Despite having built a very tricked-out SE/30 system, I still longed for more. The upgrade to lust after was the Micron Xceed + Grayscale setup that I had seen twice on EB*Y from about 2005 through 2008. But alas, they were priced into the stratosphere ($700 or so).

It was after this that our very own Gary_W acquired the coveted Grayscale setup for his SE/30, and I while reading of his purchase in these forums, I was happy for him while green with envy at the same time. I wrote to Gary privately to ask if he had seen any other sales of such devices. Being the kind soul Gary is, he promised to keep an eye out for me and let me know if he ever saw another grayscale setup for sale. And you know, those were not just idle words. Gary kept his promise.

The first recommendation I received didn't work out, but the second did. I was skeptical at first because Gary's second recommendation was a Craig's List sale for the San Francisco area, and I didn't know anyone in that area. But I spoke to the seller and it just so happened that he was driving down South for vacation, and he kindly agreed to deliver it to one of my family members in Fresno. Knowing that I would take a vacation in June this year to visit family at home, I made a deal with the seller. And even though I paid the seller a bit extra to drop it off in Fresno, the total sale didn't come anywhere close to the outrageous $700 I'd see on E*BAY for the Xceed grayscale setups alone. With the socketted Daystar boards being even more rare than the grayscale Xceeds (I myself not having spotted a single one on E*AY through the years), I would expect their value to be at least as much as the Xceeds, if not more.

Prior to purchasing from the seller in SF, I knew only two things about what he was selling: (1) This setup couldn't be booted due to SimasiMac (which is no surprise as virtually every SE/30 these days is now in need of recapping), and (2) his Mac definitely had the Micron Xceed PDS video card and CRT yoke board that would drive grayscale on the internal monitor. However, what I wasn't told was the extra bonus that I only discovered upon opening up the Mac in Fresno last month -- the elusive Daystar 50MHz 030 accelerator that attaches directly to the logic board via the CPU socket. That is a goldmine find because it means one can then add an Ethernet card to the mix, making the ultimate SE/30 experience.

I have since returned to Japan, and I am having the machine shipped to me now. As such, I cannot report on anything further until I have recapped the logic board and got it booting.

For now, I would like to share the Micron Xceed photos I took while in Fresno last month:

https://picasaweb.google.com/jameswages/SE30MicronXceedGrayscaleVideoWithDaystarSnapOnAccelerator

(I put these on Picasa instead of my normal Flickr account since Picasa allows me to upload larger photos than Flickr without being forced into a paid PRO account. So be sure to click the magnifying glass icon to zoom in on the photos.)

I will provide more photos and screen captures of the grayscale once I get the Mac up and running, which may be another month or so in the future.

I post this not to gloat, but to share an important find. I also seek to share this information to show that perseverance, patience and friendships here on the 68kMLA pay off. If you have been seeking something rare, don't give up. It's still out there somewhere. And even if you cannot find it yourself, perhaps a kind soul can lend a hand.

Once again, special thanks to Gary_W for being a true friend. You promised me you would keep an eye out for an SE/30 grayscale kit, and that you did. Bless you, Gary! And thanks to the owners of the 68kMLA for giving vintage Mac enthusiasts a wonderful venue to share useful information and enhance their computing experience!

 

[Mk.558]

Does that Xceed make it possible to have a VGA output with that dongle?

I haven't seen any video output cards around on "EB*Y" in awhile. I thought it would be cool to use one in a presentation, particularly if it had a 90* adapter so I can keep the Ethernet card installed (show off online at the same time).

That Exceed card looks really simple. I wonder how easy it would be to reverse engineer one with a sacrificial SE/30 (in case it encounters spontaneous combustion) and some copper trace blank PCB?

 

[trag]

Very cool, JDW. What a find! Thank you for the awesome photos.

The VRAM on your board is fascinating. Perhaps it used to be common, but I've never seen anything like it. I looked up the datasheet for it.

It is 256K X 4 DRAMs but internally, they have a 512 X 4 SRAM component as well. There's some kind of command to make the DRAM transfer data to the SRAM and vice versa. And the SRAM can be output serially (four bits at a time, but on the ticks of a clock input).

Very useful for video applications.

I wonder if that's organized as 256K X 64 or if the Micron board treats it as two banks of 256K X 32 for a 512K X 32 memory.

 

[Bunsen]

the extra bonus that I only discovered upon opening up the Mac in Fresno last month -- the elusive Daystar 50MHz 030 accelerator

That's ...

(wait for it)

... Xceedingly cool!

 

[Gorgonops]

The VRAM on your board is fascinating. Perhaps it used to be common, but I've never seen anything like it. I looked up the datasheet for it...

I found the data sheet myself (I love looking at them and pretending I understand any of it) and I believe it's "yer standard of-that-vintage in expensive video cards" dual-ported VRAM. It's that SRAM output cache which operates asynchronously with the main read-write data bus that lets you use it in an (almost) non-contending manner during host accesses.

I'm impressed that card has 2MB of VRAM. The one manual I saw for a Micron Xceed implied it had at most 512k. (based on its resolution/color depth limitations) This must be the deluxe "MacroColor HR-30" mentioned on the Gamba page which claims to support 16M colors up to at least 640x870.

(Coincidentally, the data sheet mentions that the RAM on this card can be treated similarly to a previous 64kx4 part. Populate the board with that and that's 512k.)

Obviously I don't have any idea how the card arranges the RAM, but I don't see the point of having a 64 bit wide bus. (neither the DAC nor the host can take advantage of that and there's no high-level video accelerator CPU in play either.) My guess would be it's interleaved as two 32bit even/odd word banks.

 

[trag]

I don't see the point of having a 64 bit wide bus. (neither the DAC nor the host can take advantage of that and there's no high-level video accelerator CPU in play either.)

It's a question of eliminating bottlenecks. If you need 32 bit width for your DAC (24 bit color and it's just easier to store 32 bits) and you need, let's say 70 Mwords/s of bandwidth, and your memory can only run at 35 MHz (35 Mwords/s) then you can overcome the speed limitation by making the memory twice as wide.

Now you're getting 2 X 35 Mwords/s, which is the bandwidth you need. Of course, as you noted, the DAC can't accept it that way. So you add some buffers.

Sixty-four bits leaves the memory and enters eight 8-bit wide register chips. Your board control chip activates the Output Enable pin on half of them and 32 bits go to the DAC. Then the control chip activates the Output Enable pin on the other half of them, and the other 32 bits go to the DAC. This is really just a method of building a registered mux.

The tricky part is getting the controller chip to do all that stuff at the right times and that is what the Xceed Maverick chip does (or can do, according to the description of it).

Of course, the register chips must be higher speed than the memory, but that's never been a problem.

 

[Gorgonops]

I don't see the point of having a 64 bit wide bus. (neither the DAC nor the host can take advantage of that and there's no high-level video accelerator CPU in play either.)

It's a question of eliminating bottlenecks. If you need 32 bit width for your DAC (24 bit color and it's just easier to store 32 bits) and you need, let's say 70 Mwords/s of bandwidth, and your memory can only run at 35 MHz (35 Mwords/s) then you can overcome the speed limitation by making the memory twice as wide...

But... why not interleave instead? If you read every other word from a separate bank then you achieve the same bandwidth with the same overall duty cycle on the memory. What you described here:

Sixty-four bits leaves the memory and enters eight 8-bit wide register chips. Your board control chip activates the Output Enable pin on half of them and 32 bits go to the DAC. Then the control chip activates the Output Enable pin on the other half of them, and the other 32 bits go to the DAC. This is really just a method of building a registered mux.

Works just as well if you're interleaving. Read 32 bits straight off one bank into the DAC or, if necessary, preload one of two rotating 32 bit buffers which feeds the DAC, then do the same with the other 32 bit bank. It's the same thing with another beard and mustache, with the only difference being with the 64 bit single-bank your "50% off-duty cycle" is the same for the whole framebuffer, while in the interleaved version the "off-DAC-duty" cycles alternate. The complexity of either architecture is probably pretty similar, although in the 64 bit wide version you always need a buffer for feeding the DAC while if the read performance of the memory is fast enough you can do without it in the interleaved version. (You're also going to need a circuit to shift every other written word into the "correct" half of the "wide" bank while writes just need an even-odd address chip select in the narrow interleaved. Of course, you may need a more complicated arrangement of control and address lines for interleaving, so... I dunno, I have *no idea* which would be the more conservative approach. If you have documentation saying they used a 64 bit wide bus maybe they used a 64 bit wide bus. I apparently missed the link for it.)

The PDF says the serial read access time of this RAM is 25ns. Assuming I did the math correctly worst-case a 2MB framebuffer at refresh rates typical of old Mac monitors will need about 35Mwords/sec, and that would in theory just barely work with 25ns RAM and a 32 bit wide bus. (Or in other words, it implies that you could read a 32 bit wide bank at sufficient speed to not cause a problem, so you don't *need* 64 bit wide and a buffer for the fill rates this card needs. Besides, since the *bulk* performance of the RAM is much lower than 25ns you probably wouldn't be able to do 70Mwords/sec with it no matter how you arranged it.)

Of course, these days you could just use 10ns or faster SRAM and forget all the complications. A similar amount of pure fast SRAM would cost far less than this exotic dual-port DRAM/SRAM hybrid cost back in the day. But if you *really* wanted to use dual-port RAM like this there might be parts like it still on the shelf. A Google turned up data sheets for some 2MB by 16bit parts with a similar sounding architecture. (Data sheet was dated... 1998, or so?)

 

[Gary_W]

Thanks for the kind words James. However I must say if I knew that the machine had the socketed Daystar accelerator, the outcome would have been different. That's MY holy grail and I can't believe it slipped through my fingers without me knowing it. So close, but yet so far! What the heck, the fun is in the search. I will keep looking.

I know you will do a great job documenting the Xceed cards and I look forward to reading your write-ups and viewing the photos.

FYI - For those interested I scanned the manuals for the Xceed cards and made them available here:

Xceed color 30 manual: http://www.box.net/shared/3p7e0rlm1x

grayscale 30 manual: http://www.box.net/shared/y8bd40ic7g

Gary

 

[tt]

Congrats JDW! I know you have been on a long trail of seeking.

I have also been on this search, it all started around 2001 when I came upon an SE/30 at a Goodwill for $10 and found Gamba's SE/30 corner when it was still being updated. If anyone has a DayStar board they would like to part with, PM me... :b&w:

 

[TylerEss]

I've gotta say congrats, JDW. That's a delightful config you've finally managed to put together. The chase can go on for years, but it's great fun when (if) you finally succeed!

 

[tt]

JDW, I was looking at your excellent photos and I was wondering, do you have one of the back side of the DayStar PowerCache board? It would be cool to see what it looks like on the other side.

 

[JDW]

...do you have one of the back side of the DayStar PowerCache board? It would be cool to see what it looks like on the other side.

I shot and uploaded several photos for you today:

https://picasaweb.google.com/103365672326265854011/SE30MicronXceedGrayscaleVideoWithDaystarSnapOnAccelerator#

The photos also show the massive amount of capacitor leakage this logic board experienced, most likely due to the higher than normal current draw required for the Daystar PowerCache and the PDS video card.

While taking the photos, I noticed that the brass part of my SE/30's flyback transformer had broken, as shown here:

https://picasaweb.google.com/103365672326265854011/SE30MicronXceedGrayscaleVideoWithDaystarSnapOnAccelerator#5641720139650991042

I'm not sure if that metal piece is just to hold the flyback in place to ensure the PCB pins don't break during transport, of it is required electrically (I've not yet fully restored this SE/30 so I can test). Your thoughts on this broken part would be appreciated.

To zoom in on any of my Picasa photos, just look immediately above the photo and you will see: Full Screen, Share, Prints, Actions, and then a little Magnifying Glass icon with a "+" beside it. Click on that magnifier to zoom in.

 

[trag]

Nice photos, James. Thank you. It looks like the upgrade has four cache RAM chips, two tag RAM (stores the index for the cache), five octal latches, four octal bus transceivers, six GAL20V8 and three GAL16V8 PLDs.

The latter GAL chips would be the difficult part to reproduce, unless Daystar didn't lock the chips...

I wonder why the thing has five latches. Typically, latches are used to hold the address from the bus, but there are only 32 bits in the address, so the card should only need four of them. Perhaps the fifth one latches some of the control signals?

Now to find an image of a PowerCache '030 card for the IIci and see if it has the same components...

I'd look at mine, but it's in the attic.

 

[tt]

Sweet board. The PowerCache looks really clean. I like that back-lit shot of the motherboard. It looks like the capacitor goo is more evenly spread, maybe the heat mobilized it more so it is not so concentrated to eat away metal. Does it boot?

It would be nice to have an SE/30 board repository for us to upload our photos. Maybe we could do a Picasa group or something like that. I looked at the Nubus Mafia on fritter, but the upload mechanism wasn't working.

Trag, the DayStar Universal PowerCache has similar chips/counts on it (I just checked mine).

 

[JDW]

Trag, I'm still trying to figure out what this Xceed video board is:

http://home.earthlink.net/~gamba2/microngray.html

But the Part Numbers on that Gamba page all start with XCD. But if that web page is accurate, it would indicate that the card I have can only be on of the following models:

• • Color 30
  • Color 30HR
  • MacroColor 30
  • MacroColor 30HR

You can see in my video board photo here that it is PGD-900091:

https://picasaweb.google.com/103365672326265854011/SE30MicronXceedGrayscaleVideoWithDaystarSnapOnAccelerator#5620905452158378386

I looked up the FCC ID which shows the ID was granted on December 31, 1991.

tt, it won't boot but it does show classic signs of Simasimac. If I have time today, I will start the recapping process for the logic board. Once that board is restored, I am hopeful everything else will JustWork™.

I personally like Flickr a lot better than Picasa for storing photos. I have a lot more flexibility when writing text descriptions under the photos too. But I don't want to host ALL my photos online, nor do I have more than a couple hundred total I want to host, as such I refuse to pay for Flickr's PRO service. The only reason I used Picasa to host these photos is because Picasa gives will host a bigger photo free of charge. Stupid Flickr dumbs down the photos to at most 1024px (longest side) unless you pay. I can host photos bigger than 1024px on FaceBook now! I just don't like to use FaceBook for hosting my vintage computing stuff.

 

[trag]

Trag, I'm still trying to figure out what this Xceed video board is:http://home.earthlink.net/~gamba2/microngray.html

But the Part Numbers on that Gamba page all start with XCD. But if that web page is accurate, it would indicate that the card I have can only be on of the following models:

• • Color 30
  • Color 30HR
  • MacroColor 30
  • MacroColor 30HR

You can see in my video board photo here that it is PGD-900091:

https://picasaweb.google.com/103365672326265854011/SE30MicronXceedGrayscaleVideoWithDaystarSnapOnAccelerator#5620905452158378386

I looked up the FCC ID which shows the ID was granted on December 31, 1991.

Yeah, I'm not going to be much help figuring out the model number, as I don't know any more about them, than what's on Gamba's page. It might be educational to look at ads in old MacUser and MacWorld magazines.

I can figure out the amount of video RAM, for what that's worth. And you can do some rough calculations of how much video RAM each card must have from the resolutions supported listed on Gamba's page. That could help you match things up, but if you have a model of card that Gamba never heard of, then it's not going to correspond with the only list we have to work from. Hence my suggestion of old magazine ads or articles.

From the components on your board, it closely matches the description in the file named 221706.pdf which I got from somewhere. It looks like a scan of notes on the theory of operation of the Micron video cards. It also mentions NuBus cards, as their Maverick chip could be used on either NuBus or PDS cards. That's not all that informative though, because one can clearly see the Maverick chip on your board. It's the chip labeled "Micron" near the bottom right of your photo.

The chip labeled BT473 is the RAMDAC. That's the chip which takes in video pixel data and outputs analog signals a monitor can understand. In order to drive an external monitor and the internal grayscale simultaneously, they would have needed to add a second one of those.

Notice the date code on the DAC. It's the 50th week of 92. So your board was probably assembled in '93.

The chip in the far left lower corner (ICS1494) is the clock generator which generates the different frequencies needed as one changes screen resolutions.

The Maxim chip is just an electronic switch. I assume it is used when switching between the two possible outputs.

The two PALCE16V8 at the bottom of the screen are probably what 221706.pdf calls the Pixel Mixer PAL. The Pixel Mixer receives video data in 8 bit chunks and formats it appropriately for lower color (gray) depths. I'm not sure what the other two PLDs are for.

 

[JDW]

I added 4 more pics to the collection today:

https://picasaweb.google.com/103365672326265854011/SE30MicronXceedGrayscaleVideoWithDaystarSnapOnAccelerator#

One of the 4 shows the recapped motherboard, using Trag's replacement cap kit. Sadly, it still shows Simasimac (horizontal stripes and a sad mac chime) at cold boot. There was a lot of leaked fluid on this board, so it could be that a trace had been eaten through. Even so, I could not physically see such when I carefully recapped this board. I know the problem is not RAM or ROM, because I tested two sets of known-good RAM and ROMs. I also tested the board with a stock Motorola CPU instead of the Daystar accelerator. And I even tested in two different SE/30's which have different PSUs. So the board is bad somewhere. Perhaps some of the leaked fluid got under the PDS connector. That is very hard to clean, and I cannot easily remove it. But I know a lot of fluid was leaked there because the metal frame of the SE/30 was physically corroded in that area about 3cm.

Thankfully, I had another socketed SE/30 logic board and I was able to get the Daystar accelerator and Grayscale kit up and running. I first tested it on my known-good SE/30 rig. It worked at first and then I got lockups. I did all manner of swap-outs to debug the problem until it I realized I should check the voltage. It was 4.85v and 12.7v at the external floppy drive connector. These voltages are not new. I had such readings in the past. But for whatever reason, 4.85v is insufficient to keep this socketed accelerator and Grayscale rig from locking up. So I put everything back in the original SE/30 and it worked fine. On that box, I read 4.95v and 12.3v at the external floppy connector. Interestingly, on the 4.85v box, the PSU is a SONY. And the 4.95v box has an Apple branded PSU (Astec?) with lots of air vent holes all around it. In the past, I was not able to get my SONY PSU to kick out any higher voltage on the 5v line. So either the SONY PSUs don't age well, or this other brand of PSU (Astec?) just puts out more voltage.

I must say grayscale is quite nice. It's not as glorious as color, of course. But it's really nice. It's like it should have always been that way from the very beginning. I can switch between B&W, 4 grays, 16 grays and 256 grays.

I'm still not sure what video card I have. The SE/30 rig I bought did come with Micron manuals and an XCEED floppy, which sheds some light on it. The paper manual says "Xceed MacroColor 30, 24-bit Video Card for the Apple Macintosh SE/30." But the floppy disk says "Xceed Technology, MacroColor 30HR, Video Utility Disk 3.5". The paper manual gives the following info:

Video Specifications

                  13" & 14"       Apple 12"

Resolution: 640 x 480     512 x 384

Refresh:      67Hz            60Hz

Horiz. Scan: 35kHz         24.5KHz

Bandwidth: 30.24MHz     24.5MHz

Pixel Depth:  24-bits       24-bits

Video Memory: 2MB

Output Type: RGB

Synch Type: TTL Level Synch

Bus Width: 32-bits

Connector: DB-15

The only consistent problem I've seen is that there is a vertical stripe of 1px thickness, dotted, that is occasionally visible on the screen. It seems to appear more often when I have mostly dark grays or black on the screen. And it's almost impossible to see when the screen is white or has light grays. One other confusing thing about this vertical stripe is that it always appears in the same position, and it even appears briefly and goes away during cold boot. When the screen is set to 256 grays, the vertical line appears toward the left of the CRT. When I change to 16 grays, the vertical line shifts to the right about 1.5 inches. And at 4 grays it shifts to the right another 1.5 inches or so.

I swapped the motherboard, accelerator, video card, grayscale card, and wire harness between two different SE/30s to see if the PSU, CRT and Analog board had anything to do with it. Nope. Exactly the same vertical stripe on both boxes. I also tested without the Daystar accelerator. Still the same stripe. So I know it must either be in the PDS video card or the grayscale card. I've adjusted the screen width and height via the SE/30's analog board POTs, but that doesn't make the vertical stripe go away, nor does adjusting Sharpness. The grayscale card is rather simple electronically, so I don't know if that card is the culprit. But if the PDS video card, what could it be?

UPDATE: Yet Another Problem. Over the past hour I was running Norton SpeedDisk on the SE/30. I had a DMM attached to the external floppy connector's 5v line. It was reading 4.95v and running fine. But I just checked a moment ago and found the system locked up. The voltage reading had not changed. It was still 4.95v. The 12v line was fine too. I restarted the computer and tried to boot off the floppy, but it locked up half way through. This is exactly the same problem as I had when the boards were in my other SE/30 rig. I thought it was the low voltage, but now I am not so sure. The machine does get warm as time goes on, but it was in a cool room with the back side of the case removed. Plus the internal fan was going. It's really strange. I'm not sure what to make of it. Your thoughts would be appreciated.

 

[trag]

I added 4 more pics to the collection today:https://picasaweb.google.com/103365672326265854011/SE30MicronXceedGrayscaleVideoWithDaystarSnapOnAccelerator#

One of the 4 shows the recapped motherboard, using Trag's replacement cap kit.

Just looked at the new pictures. Ouch. Sorry about the fit problem. I never even thought about the CPU socket upgrade when I chose the new capacitors. I just thought, "There's room under the frame. This works."

I'm surprised that the 220uF is enough bigger to be a problem though. The size isn't that different, but I guess Daystar designed for the most clearance they could get. If you have the 50V 470uF and it was under the upgrade, that would be the larger problem....

Sadly, it still shows Simasimac (horizontal stripes and a sad mac chime) at cold boot. There was a lot of leaked fluid on this board, so it could be that a trace had been eaten through. ... Perhaps some of the leaked fluid got under the PDS connector. That is very hard to clean, and I cannot easily remove it.

Do you have, or can you get a spray can of flux cleaner, or contact cleaner spray? Something like this:

http://www.all-spec.com/products/Chemtronics/Cleaners_and_Aerosols|Electronics_Chemicals|CLN-04/ES1697.html

With a spray can and the little tube it comes with, you should be able to squirt solvent in the hard to reach spots. I use a similar product, the name of which escapes me at the moment. I thought it was FluxOff, but mine isn't by Chemtronics and it comes in a blue can, and it's cheaper, but it leaves a white residue that bothers me. Everything I've used by Chemtronics has been good quality, so the above stuff is probably better than what I use (hopefully, doesn't leave a residue). Only eight cans to go, and then I can switch to the Chemtronics....

You could also try the dishwasher method. A soak in hot water will probably get the hidden spots.

But....it's also possible that the goo ate a via. Vias are the holes in the circuit board which are coated in copper and connect different layers of the circuit board together. In most places, the traces are protected by the solder mask, and damage to an exposed pin/pad should be fairly visible. But corrosion in a via can be stealthy. The copper can be corroded and chemically changed, yet, it still appears that the via is normal because it's filled with roughly copper colored stuff.

If other things (such as more thorough cleaning) fail, you might try gently poking in the vias with a pointy metal implement and see if the metal in any of the vias is softer than it should be.

This is what killed my IIci back in the mid '90s. Capacitor goo pooled in a via and ate the copper away, and I had to run wire from one side of the board to teh other to reconnect the two traces.

The only consistent problem I've seen is that there is a vertical stripe of 1px thickness, dotted, that is occasionally visible on the screen.

Unfortunately, this is a VRAM problem. If it was an address line or data line (wire/circuit board) problem, the line would appear in more than one place, I think. When you shift the color depth, the bad locations in video memory are now mapping a different location on the screen.

Hmmm.. What resolution is this happening at? 512 X 384? And it's vertical, so it's every 512 pixels. Or it could be every 1024 pixels if it is a dotted line -- every other vertical pixel in a line. That could be an address line problem. If the A9 (A0 being lowest) address line had a problem... No, but that should cause everything larger than that address to be wrong. Hmrph. Gotta think about that. It'd be easier if we just new exactly how the map the video image into the VRAM. Are they using two banks of 1 MB? Or are they using all sixteen chips as one big 64 bit wide memory bank.

The fact that it goes away at a cold boot, might indicate a heat problem. If you have some cold spray (or can get it) similar to this:

http://www.all-spec.com/products/1672-10S.html

you could get the machine going. Wait until the line appears. Then spray each memory chip in turn and see if the line disappears. Of course, this is really only useful if you can source a replacement memory chip, and if it bothers you enough to be worth the effort.

However, the fact that it takes a while to appear at boot time could just mean that the video card starts out using its VRAM one way at boot time and switches to a different organization later. I saw that on an ixMicro something-or-other. The one after the Twin Turbo.

UPDATE: Yet Another Problem. Over the past hour I was running Norton SpeedDisk on the SE/30. ... I restarted the computer and tried to boot off the floppy, but it locked up half way through. This is exactly the same problem as I had when the boards were in my other SE/30 rig. I thought it was the low voltage, but now I am not so sure.

Was this in a machine where the fatter 220uF capacitor causes the Daystar upgrade to stand off a bit? It could be that the connection is good when it is cool, and that the pins aren't making good connection when it heats up.

Oh, that reminds me, on your Simasimac logic board, there are some socket chips. Have you tried removing those and reseating them? The SIMM sockets look nice and clean, so I don't think bad contacts are a problem there, but one of those socketed PALs might have worked its way loose.

 

[JDW]

Do you have, or can you get a spray can of flux cleaner, or contact cleaner spray?

Not "flux cleaner" in name, no. But I have long used 99.5% dehydrated Ethanol to clean my electronics:

http://web.archive.org/web/20080131132706/http://chie.kenko.com/chie/product/item/itm_7731096072.html

I apply it with a spay bottle and then gently scrub the board with an old toothbrush. I just can't get the toothbrush under the PDS connector. Even so, your right about it possibly being an via. That will take more time to detect since there are many. But when I have the time, I will give the board another look-over. I want to get that board up and working because it's my only other board that has a socket for the CPU.

(And yes, I was thinking about those socketed ICs too. They are in there really good though so I didn't want to apply too much pressure to get them out. But since the board is basically non-functional, I'll take the risk, yank them, then clean beneath them and reseat them.)

Unfortunately, this is a VRAM problem.... What resolution is this happening at? 512 X 384? The fact that it goes away at a cold boot, might indicate a heat problem.

It happens on the native 512x384 resolution of the SE/30's internal CRT.

I don't think it is a heat problem. I say this because the problem appears at the same time, under the same conditions. At cold boot, the machine is cold! And it flashes on and then off at the same point during the boot process. And that remains true even though the screen background gray level is consistent. It's almost like it comes on and then the screen refreshes and it goes away.

And then after the machine is booted, the vertical dotted line appears and disappears in accordance with the background gray level. Or so it "seems" in my testing.

Was this in a machine where the fatter 220uF capacitor causes the Daystar upgrade to stand off a bit? It could be that the connection is good when it is cool, and that the pins aren't making good connection when it heats up.

I thought about that. But if you look at my photo, you will see that the CPU pins are in there pretty deep, even for the section that isn't seated 100%. I brought home a physically smaller (lower voltage rated) radial cap. So I could use that to swap out the fat axial. I just haven't tried that yet. But again, I am very skeptical if the CPU pins are the problem. I am wondering about the voltage.

After I posted last night, I yanked the motherboard and put in another recapped logic board that has a soldered CPU (no acceleration) and 32MB of memory onboard (all RAM slots filled). I was able to run that board for more than 3.5 hours last night, and it did not lockup once! However, when I put the DMM on it, I measured 4.98v~5.00v and 12.4v at the external floppy connector. So you see, the voltage is getting higher because there is less demand on the PSU. So it could be that the Daystar board refuses to work right if the voltage goes below 4.98v for extended periods?

I haven't every opened this style of PSU before (the non-SONY kind with numerous air vent holes throughout it) to see if a higher voltage setting would solve the crashing problem on the socketed board with the socketed Daystar accelerator.