Insider knowledge disclaimer-- I worked at Supra. So yeah, removing the part numbers (early on) was just to frustrate competitors trying to either reverse engineer (or simply figure out our costs). Early units had pretty much ALL the part numbers scrubbed by electric-erasers... I also remember a little discussion about the 68HC000 in particular in that the top speed grade from Motorola at the time was 16MHz (obviously we were overclocking) and we didn't want to have to have our (two) Amiga tech support people have to argue about 'out of spec' parts with nerds on the phone. ;-) Engineering also thought it was funny imagining GVP or some other competitor trying to figure out if we had a special deal with Motorola to get ~33MHz 68000's or something... We ran the accelerator in engineering up to 40MHz (it was originally going to be a 36MHz accelerator, but we got cold feet about compatibility/reliability and turned it down). Running slower also gave purchasing some elbow room in the speed grades of the SRAMs (and maybe PLDs?) which helped keep it cheaper and easier to source parts for. In addition to making it smaller, the Intel PLDs also helped with the design obfuscation/protection -- at the time there wasn't any real fear of anyone being able to bypass the read back protection easily. (Back then there wasn't any Chinese outfits with a SEM and a FIB that'd crack programmable logic for a few hundred bucks!) Those parts are N85C090's -- kind of an early CPLD with ~24 macrocells (900 gates, hence the part number). They had a decent amount of IO though (more IO pins than macrocells). The 'clock' lines on those are just that-- a globally routed clock if the logic needed it. One of the PLDs acted as a state machine (so it used the clock pin as intended) and the other was just a bunch of combinatorial glue, so no clock edges needed. The benchmarks were indeed legit, but we would definitely pick the scenarios where we won! (e.g. things that fit in cache were very quick, so compare that to a slower clocker '030 upgrade over the 16-bit memory bus and the Turbo28 could outperform.) The Turbo28 cached and accelerated system ROM (0xF80000-FFFFFF), slow expansion RAM (0xC00000-D8FFFF), and fast expansion RAM (0x200000-9FFFFF). Removing the 'option' jumper punched a hole from 0x600000-0x9FFFFF for hardware that did DMA or otherwise messed with the cache. At boot time the cache is actually disabled-- there were timing loops in the boot ROMs that early acceleration messed with, so the cache only becomes functional (if it's enabled by switch/jumper) once autoconfig starts.
Late reply - you wouldn’t happen to have access to the PLD equations so that substitute parts might be retrofitted. Or a board redesign created with current parts..
@@bruce_just_ I *might* be able to come up with something! I have a couple of early pre-production units (and probably some of the hardcopy documentation) and there's always a chance that we weren't popping the security fuses on internal test hardware at the time. I'll see if my BP Microsystems programmer supports the N85C090 and if so, I'll try reading what I have and see if the pattern looks legit or not. If I can find the old source (I'm pretty sure it was written with PALASM, or possible ABEL in Synario-- definitely an HDL like those at the time) they'd both fit in a single modern CPLD with enough pins and maybe ~32-36 macrocells or so...
@@ClayCowgill wow fingers crossed. Please consider reaching out to Glen/CRG if it pans out, at least for archiving/digitizing the documents or the PLD files to help the broader Amiga community if you’re able to do so!
Sorry for the late reply I've been on holiday but thank you for the detailed information. I have now done a bit more investigation on this card and I am sure that the PLDs are indeed the fault. There is just nothing happening on the CPU, it is trying to do a read but not getting anywhere. Both UDS and LDS are stuck high isolating the databus on the CPU. I assume both these signals along with R/W are routed through the PLDs so I can only conclude that a failure in there is stopping everything in its tracks before it even gets started. I unfortunately don't have any means of attempting to read out the contents of the PLDs, all I've got is the TL866 and that doesn't support them. The Altera EP910 series of PLDs seem to be pin compatible and are still available on the likes of eBay although they are pricy. It's just getting the code for them and figuring out how to program them which is the current stumbling block.
IC8 and IC9 are cache RAMs used to store the Tag address. They are not used for storing any user DATA. That's what the other two SRAMs are for. The logic on the accelerator needs to know which location of main memory is stored in the data SRAMs as there can be many: cachable area / 16kByte in your case. If you have a look at a standard 486 mainboard you will probably see 4 SRAMs for storing data (4 x 8bit = 32bit) and a fifth SRAM. This is used for storing the tag address. But on all PC mainboards I have ever seen these are standard SRAMs. The IDT 71x74 do have two important features: they can be bulk erased and they compare the pattern on I/O[0..7] with the internally stored pattern at the current address on the address pins. But you get a thumbs up from an Atari ST maniac! Why don't you check /BERR and /HALT lines on the CPU as you did with /RESET? Do you have a link for those IDT you ordered on aliexpress?
@@CRG Thank you. There have been two similar accelerators for the ST: HBS640T28 and T36 running at 28MHz and 36MHz respectively with 64KB L1. I got a T28 and it ROCKS. Hope you will get yours running again. I would really check /HALT. It is a bidirectional signal. I assume it is asserted simultaneously with /RESET when you push the reset button. But if it gets asserted later on the CPU ran into serious trouble. A logic-analyser would come in handy to check the addresses that were accessed just before the assertion of HALT.
A handy tip ive found by accident, using a smidge of white oxide thermal compound on the top of the sanded chips then gently rubbing it about & off with a bit of kitchen roll sometimes can reveal remnants of the sanded off text if was laser etched, its not a 100% success rate, but ive found in a lot of cases i can get some info doing this, the fine oxide gets stuck in the remains of the etching bringing it up visible thru the sanding damage.
To come out of reset a 68000 also needs HALT to start low then go high in sync with the reset signal. If it then returns to low again then that is the CPU itself double bus faulting due to its connected address decoding failing. It would be interesting to check the HALT pin on the motherboard to see if the accelerator is disabling the onboard 68000 by keeping it pulled low.
Had this together with the SupraRAM 500RX on my old Amiga 500 back in the early 90s. Was actually a very speedy accelerator with fast RAM attached considering it's a 68000, made Wing Commander extremely playable! Hopefully you can get it working again! Good luck!
This was my favorite accelerator for the 500 because it gave you just enough speed to play those 3-D intensive games smoothly, but it did not break normal games or other games
As far as the clocks of the 68000 are concerned... The standard 68000s went up to 16 Mhz and the 68HC000 CMOS versions up to 20 Mhz. Both are still made I think, by Freescale. It was a sure bet that the HC CMOS variant would easily hit 28 Mhz, but some such boards offered up to 40 MHz settings to let you experiment. I also remember that sometimes those chips needed a little more voltage, like 5.2V, to run a bit faster. Maybe that's where the hole for an external power supply comes into the story. an external PSU could have 5.3V and help overclocking but they never integrated it into the final product. IIRC the CMOS specifications allowed for up to 6V.
Thanks for the info. I think you are correct and the 68k is still in production and yes those HC chips can go a lot faster than their labelled clock. Supposedly the chip on this card is a 16Mhz part and I do have a spare 20Mhz HC here but since the original chip came up ok in the 14Mhz card I thought it best to just reuse that. The voltage though is a very good point and its the one thing I didn't check properly so that will be the starting point of the next video looking at this card.
I've checked the caps and they certainly aren't shorted but I can pull them off easily enough and test properly out of circuit. I might even have replacements in stock so might just swap them out and as you say those contacts, and maybe even the entire edge connector could use a reflow, just to be sure.
In a similar way and just after, i had to use a "Dremel" for one of my projects (And I wasn't sure of the outcome). I watched the door several times, fearing one of the Glen triplets would appear 😅😂
the main reason for sanding the chip identification is not to hide any shenanigans but to prevent reverse engineering by its competitors. by using out of spec parts it would lead to high returns and would increase costs by an uncertain amount and is very stupid practice.
I can understand that for say a rom or a pal or gal but these are just off the shelf ram and a CPU. Of course they are still hiding what the chips are making reverse engineering harder but if someone wanted to do that with this card I think the more difficult part would be getting the equations out of the PAL or PLDs. I think in the case of this card it was more likely to be hiding the fact that the components are being run out of spec, especially the CPU.
I'm with others in these comments, in that you might be on to something with those ram chips. To generate that much heat, "something" is going on in them, and it's likely not something good. Fingers crossed the replacements may return this thing back to life. I would be rather curious to see it in action and put through some bench marking as well. Very cool video as usual, bud. Looking forward to the follow up on this unit. The little skit was good for a laugh. 🍻😆
Hi Glen, not wanting to teach you to suck eggs but try re-flowing the solder joints on the connector for the Zorro slot. It has worked for me many times and now is my usual first point of attack for this old stuff and considering the Supra soldering was bad...
The Amigas power light always starts dim then should go bright as the system comes out of reset and boots. I'm not sure exactly were it changes brightness in the boot sequence but it's early on somewhere. That said though your comment has made me think about the VCC reading of 4.6v I seen on the scope and that might be because of the shorted tag ram and that might be why its not starting... Thanks for the question, cause its made me think about the problem again 👍
There is an I/O signal which controls the LED and the low-pass audio filter. It is not possible for the A500 to turn the LED all the way off under software control, it is either dim (and the low pass filters are off) or bright (and the low pass filters are on). Off means that there is no 5V rail going to the board (i.e. power switched off or unplugged)
Yeah, the industry was rife with manufacturers fiddling their benchmarks back in those days. Wouldn't surprise me if they "accidentally" left a fast RAM expansion in their machine while doing the tests.
Actually, tantalum caps are notorious for blowing up if exposed to uncontrolled inrush current. It is easy for a new engineer to not be aware of this, but they would find out pretty quickly once units started coming back from the field.
Its possible to sand chips due licensing.. So other company`s cant use those chips on their products. (or have to pay a fee). But sanding it, they couldnt know which ICs are used. So no court cases after all.
I have my EATX2000 although I'd need to strip that machine down a fair bit to try it. What I can do though is try it in the A500 side car to A2000 CPU socket.
@@CRG I think it did once or twice, but it just crashed all the time. i gave up with it in the end. A friend of mine had the 500 version like you and yes it made Elite very playable.
I think I considered getting one of these Supra units back in the day, but IIRC, I heard that there was a high failure rate so I went with the ICD AdSpeed instead, which is the 14.32MHz 68000 in the normal HMOS 64-pin DIP package that displaces your original processor. Plus, I thought it would look goofy along with my Trumpcard 500 hard disk on the side (although honestly it's the Trumpcard 500 that looks goofy, not matching the lines of the A500, like literally WTF). A more electronics-savvy friend of mine helped install the AdSpeed (well, switching the CPU, I could handle) along with a toggle switch on the back of the case (that part, he did, also simple but he had the parts and tools) to enable/disable it. It was a pretty slick mod and it sped things up considerably - IDK, maybe 50-80% depending on the task. To be fair, this was a pretty well-expanded A500 with a fully populated 4MB BaseBoard in the trapdoor. You could feel the difference compared to a stock Amiga. It definitely ran Frontier better but back in the day there was no way to quantify it - no Afterburner or FRAPS or Mangohud or what-not. But honestly that game runs best on a 68030 or better.
While it of course would be best to get it repaired - would it be possible to place a pistorm inside the box? I remember were is an adapter for the PiStorm to connect externaly to an A500, but it is flat and uses much mor space on the side. However, a Pistorm could be more compact if using a Pi zero 2 and both the Pi and the Pistorm having no headers assembled, so you coud connect them as a sandwich. Would be realy cool to use this original extension case of the time with more modern hardware inside...
Oh, I forgot it wasn't yours. However, would you mind to publih detailed photos and some measurenents? Maybe someone would design a replica of this case for 3d printing. Back in the days I never was a big fan of theese kind of external extensions. Esp. theese heavy ones with a hard drive inside. One of the reasons I didn't bought an A500 but an A2000 back then. I always doubted the mechanical strenght. But theese case is so smal, I realy like it. And while you could do a pistorm inside the A500(+), there is still a compatibility issue with some older software. As the internal version replaces the original 68k, the external just deactivates it. As the case is smal an a pistorm is light, I think putting it in such a case would be the best solution.
I had a Supra Turbo for my A500 back in the day. However, not that speedy one that can seemingly outperform a 25MHz 68030, oh no no. Mine had the blistering speed of a 10MHz 68010. Seriously, what was the point, it barely did anything. I did later get a hard drive and attached it via the passthrough connector. I remember wishing I had a real accelerator instead in one of the Zorro slots in the hard drive enclosure.
I think to make best use of this card you will need some sort of fast ram although I still don't think it's getting near a 25mhz 030. We all wished for the hardware we could never afford back in the day. My Amiga in the 90s was bare bones throughout that period, it wasn't until the mid 00s when I got it out again that it started getting upgrades.
@@CRGIt's entirely possible that I'm misremembering as I can't find any info on the Supra Turbo other than the 28. I do remember a 68010, maybe it was 14MHz and a separate speed "hack" I tried. I'm getting old. I did have trap door RAM, but if I'm remembering correctly that is slow, not fast RAM. I don't remember at all if my hard drive enclosure came with additional fast RAM (I bought it used).
@@QunMang ICD AdSpeed? That was a 14.3MHz 68000, not a 68010. It did have a bit more circuitry, I think 8K each of cache and cache tag...? It also had a jumper that you could connect a switch to, to enable/disable it but you have to reset the machine for it to change IIRC. Or not. It was over 30 years ago... my memory's fuzzier than my sac nowadays
You should've sysinfo'd the CPU whilst you had it out. I think you'll need to find someone with a working version, so you can dump the pld's and back them up.
Glen, as stupid and painful as this sounds, what about just remove a component, test it, and if it checks out, mark it off a board print out, and then apply fresh solder and reflow it back in, doing the whole board front and back?
I'm not sure I can properly test the likes of the PLDs or the PAL. I might be able to test the SRAMs using the TL866 although I'd need to make some sort of adapter for it.
@@NorthWay_no @CRG As far as the clocks of the 68000 are concerned... The standard 68000s are available up to 16 Mhz and the 68HC000 CMOS versions up to 20 Mhz. Both are still made I think, by Freescale. It was a sure bet that the HC CMOS variant would easily hit 28 Mhz, but some such boards offered up to 40 MHz settings to let you experiment. I also remember that sometimes those chips needed a little more voltage, like 5.2V, to run a bit faster. Maybe that's where the hole for an external power supply comes into the story. an external PSU could have 5.3V and help overclocking but they never integrated it into the final product. IIRC the CMOS specifications allowed for up to 6V.
@amurtigress_mobile365 thanks for the info. Seems they were overclocking regardless of what chip they used and from what I've since been told its the 16mhz version they fitted to these boards. I do have a spare HC 20mhz chip here that I was considering fitting but since the current cpu started ok at 14mhz in that other card I thought it just best to continue to use it. It must have worked at 28mhz at some point.
Those 2 sanded of RAMs are probably the same type as the regular ones near the CPU, probably the tag memory. Would be cool if you were able to read out the programmable logic contents! I always wondered how the state machines in cache controllers worked. Lets suppose one of those tag rams are broken - the controller would most likely try to fetch data into data memory and update tags, but tags wont update (because the broken RAM chip) - tag isn't equal to whatever part of address it should be equal to - cache controller tries to fetch same data in a loop?
no, not Super, this was SUPRA, as in the brand that (in)famously made that 2400bps external modem that everyone (including me, until I got my USRobotics Courier HST) used back in the day, that got hotter than the devil's balls. It had an aluminum billet case that doubled as a heat sink; I wondered what kind of circuitry was being tortured inside of it, because it just got soooo hot... Anyway they got bought by Diamond Multimedia in 1995, which itself was party to mergers and acquisitions... the hardware market was wild in the 1990's.
I was worried when Glen got the Dremel out, but thankfully Glen was there to notice and then Glen came in and finally got Glen to put the Dremel away.
Just as well I caught myself in the act before it ended in disaster 😂
Multi-Diss-ity!
Insider knowledge disclaimer-- I worked at Supra. So yeah, removing the part numbers (early on) was just to frustrate competitors trying to either reverse engineer (or simply figure out our costs). Early units had pretty much ALL the part numbers scrubbed by electric-erasers... I also remember a little discussion about the 68HC000 in particular in that the top speed grade from Motorola at the time was 16MHz (obviously we were overclocking) and we didn't want to have to have our (two) Amiga tech support people have to argue about 'out of spec' parts with nerds on the phone. ;-) Engineering also thought it was funny imagining GVP or some other competitor trying to figure out if we had a special deal with Motorola to get ~33MHz 68000's or something... We ran the accelerator in engineering up to 40MHz (it was originally going to be a 36MHz accelerator, but we got cold feet about compatibility/reliability and turned it down). Running slower also gave purchasing some elbow room in the speed grades of the SRAMs (and maybe PLDs?) which helped keep it cheaper and easier to source parts for.
In addition to making it smaller, the Intel PLDs also helped with the design obfuscation/protection -- at the time there wasn't any real fear of anyone being able to bypass the read back protection easily. (Back then there wasn't any Chinese outfits with a SEM and a FIB that'd crack programmable logic for a few hundred bucks!) Those parts are N85C090's -- kind of an early CPLD with ~24 macrocells (900 gates, hence the part number). They had a decent amount of IO though (more IO pins than macrocells). The 'clock' lines on those are just that-- a globally routed clock if the logic needed it. One of the PLDs acted as a state machine (so it used the clock pin as intended) and the other was just a bunch of combinatorial glue, so no clock edges needed.
The benchmarks were indeed legit, but we would definitely pick the scenarios where we won! (e.g. things that fit in cache were very quick, so compare that to a slower clocker '030 upgrade over the 16-bit memory bus and the Turbo28 could outperform.)
The Turbo28 cached and accelerated system ROM (0xF80000-FFFFFF), slow expansion RAM (0xC00000-D8FFFF), and fast expansion RAM (0x200000-9FFFFF). Removing the 'option' jumper punched a hole from 0x600000-0x9FFFFF for hardware that did DMA or otherwise messed with the cache. At boot time the cache is actually disabled-- there were timing loops in the boot ROMs that early acceleration messed with, so the cache only becomes functional (if it's enabled by switch/jumper) once autoconfig starts.
Late reply - you wouldn’t happen to have access to the PLD equations so that substitute parts might be retrofitted. Or a board redesign created with current parts..
@@bruce_just_ I *might* be able to come up with something! I have a couple of early pre-production units (and probably some of the hardcopy documentation) and there's always a chance that we weren't popping the security fuses on internal test hardware at the time. I'll see if my BP Microsystems programmer supports the N85C090 and if so, I'll try reading what I have and see if the pattern looks legit or not. If I can find the old source (I'm pretty sure it was written with PALASM, or possible ABEL in Synario-- definitely an HDL like those at the time) they'd both fit in a single modern CPLD with enough pins and maybe ~32-36 macrocells or so...
@@ClayCowgill wow fingers crossed. Please consider reaching out to Glen/CRG if it pans out, at least for archiving/digitizing the documents or the PLD files to help the broader Amiga community if you’re able to do so!
Sorry for the late reply I've been on holiday but thank you for the detailed information. I have now done a bit more investigation on this card and I am sure that the PLDs are indeed the fault. There is just nothing happening on the CPU, it is trying to do a read but not getting anywhere. Both UDS and LDS are stuck high isolating the databus on the CPU. I assume both these signals along with R/W are routed through the PLDs so I can only conclude that a failure in there is stopping everything in its tracks before it even gets started.
I unfortunately don't have any means of attempting to read out the contents of the PLDs, all I've got is the TL866 and that doesn't support them. The Altera EP910 series of PLDs seem to be pin compatible and are still available on the likes of eBay although they are pricy. It's just getting the code for them and figuring out how to program them which is the current stumbling block.
You should have let Dremel Glen fix it "properly" 😂
IC8 and IC9 are cache RAMs used to store the Tag address. They are not used for storing any user DATA. That's what the other two SRAMs are for. The logic on the accelerator needs to know which location of main memory is stored in the data SRAMs as there can be many: cachable area / 16kByte in your case.
If you have a look at a standard 486 mainboard you will probably see 4 SRAMs for storing data (4 x 8bit = 32bit) and a fifth SRAM. This is used for storing the tag address. But on all PC mainboards I have ever seen these are standard SRAMs. The IDT 71x74 do have two important features: they can be bulk erased and they compare the pattern on I/O[0..7] with the internally stored pattern at the current address on the address pins.
But you get a thumbs up from an Atari ST maniac! Why don't you check /BERR and /HALT lines on the CPU as you did with /RESET?
Do you have a link for those IDT you ordered on aliexpress?
Thanks for the detailed info. This is the link to the aliexpress listing for those ICs. www.aliexpress.com/item/1005006302608145.html
@@CRG Thank you. There have been two similar accelerators for the ST: HBS640T28 and T36 running at 28MHz and 36MHz respectively with 64KB L1. I got a T28 and it ROCKS. Hope you will get yours running again. I would really check /HALT. It is a bidirectional signal. I assume it is asserted simultaneously with /RESET when you push the reset button. But if it gets asserted later on the CPU ran into serious trouble. A logic-analyser would come in handy to check the addresses that were accessed just before the assertion of HALT.
"he got that dremel oot again?" 🤣
Can't trust him, always up to no good 🤣
I love that bit of triple Glen editing haha
26 years after it was started, still warms my heart to hear about the BBoAH.
It's a fantastic resource for looking up details of expansions.
A handy tip ive found by accident,
using a smidge of white oxide thermal compound on the top of the sanded chips then gently rubbing it about & off with a bit of kitchen roll sometimes can reveal remnants of the sanded off text if was laser etched,
its not a 100% success rate, but ive found in a lot of cases i can get some info doing this, the fine oxide gets stuck in the remains of the etching bringing it up visible thru the sanding damage.
To come out of reset a 68000 also needs HALT to start low then go high in sync with the reset signal. If it then returns to low again then that is the CPU itself double bus faulting due to its connected address decoding failing.
It would be interesting to check the HALT pin on the motherboard to see if the accelerator is disabling the onboard 68000 by keeping it pulled low.
Thanks for the info, I will certainly check this.
@@CRG Also I wonder what hidden DMA is?
Had this together with the SupraRAM 500RX on my old Amiga 500 back in the early 90s. Was actually a very speedy accelerator with fast RAM attached considering it's a 68000, made Wing Commander extremely playable! Hopefully you can get it working again! Good luck!
Yeah I think it really needs fast ram added to make the difference, hopefully I can get this one going then we can put that theory to the test.
When bad joke warning appears and the joke is actually good.
baffiling problem glenn......you have checked everything pretty well..good video...something will come up!
It is indeed but I've got a few ideas of more things to try and you never know, the new tag ram chips might resolve it.
I had one of these when they came out. It was a great add on!
Haha the Dremel joke 13:50. Nicely edited with "your whole family" 😅
Thanks😆
Multi-Glen shots are awesome. That could be your "schtick". I loved that 😂
Good video, thank you! Can't wait for the attempted fix ❤
Fascinating, hope you get it working so we can learn exactly what tricks they were up to, and if it lived up to the hype.
Yeah here hoping I can figure it out. It certainly would be interesting to put it to the test.
This was my favorite accelerator for the 500 because it gave you just enough speed to play those 3-D intensive games smoothly, but it did not break normal games or other games
Hopefully I can get this one working, then I can put it to the test myself.
@@CRG I know it has a history of FAIL but dang when we had it in our store. The performance and compatibility was great. For games.
Cannot wait for the next one. Hilarious take with your two twins 😀. They look nice too !!!
Love the Alexai Sayle-type bit you did there. Cheers!
As far as the clocks of the 68000 are concerned...
The standard 68000s went up to 16 Mhz and the 68HC000 CMOS versions up to 20 Mhz. Both are still made I think, by Freescale. It was a sure bet that the HC CMOS variant would easily hit 28 Mhz, but some such boards offered up to 40 MHz settings to let you experiment.
I also remember that sometimes those chips needed a little more voltage, like 5.2V, to run a bit faster. Maybe that's where the hole for an external power supply comes into the story. an external PSU could have 5.3V and help overclocking but they never integrated it into the final product. IIRC the CMOS specifications allowed for up to 6V.
Thanks for the info. I think you are correct and the 68k is still in production and yes those HC chips can go a lot faster than their labelled clock. Supposedly the chip on this card is a 16Mhz part and I do have a spare 20Mhz HC here but since the original chip came up ok in the 14Mhz card I thought it best to just reuse that.
The voltage though is a very good point and its the one thing I didn't check properly so that will be the starting point of the next video looking at this card.
I'd resolder those contacts, 20:37 probably test the tantalum caps, and replace if necessary.
I've checked the caps and they certainly aren't shorted but I can pull them off easily enough and test properly out of circuit. I might even have replacements in stock so might just swap them out and as you say those contacts, and maybe even the entire edge connector could use a reflow, just to be sure.
In a similar way and just after, i had to use a "Dremel" for one of my projects (And I wasn't sure of the outcome). I watched the door several times, fearing one of the Glen triplets would appear 😅😂
Its a wonder I ever get anything done with that Glen fella hanging over my shoulder 😂
I had one for a time. It performed more or less like an A1200 with FastRAM if i remember correctly.
the main reason for sanding the chip identification is not to hide any shenanigans but to prevent reverse engineering by its competitors. by using out of spec parts it would lead to high returns and would increase costs by an uncertain amount and is very stupid practice.
I can understand that for say a rom or a pal or gal but these are just off the shelf ram and a CPU. Of course they are still hiding what the chips are making reverse engineering harder but if someone wanted to do that with this card I think the more difficult part would be getting the equations out of the PAL or PLDs. I think in the case of this card it was more likely to be hiding the fact that the components are being run out of spec, especially the CPU.
I'm with others in these comments, in that you might be on to something with those ram chips. To generate that much heat, "something" is going on in them, and it's likely not something good.
Fingers crossed the replacements may return this thing back to life. I would be rather curious to see it in action and put through some bench marking as well.
Very cool video as usual, bud. Looking forward to the follow up on this unit. The little skit was good for a laugh. 🍻😆
You've done all you can, and checked everything you can. Frustrating, but a good video!
Done all that I can for now but I've got a few other ideas so not finished with this card just yet.
I always get fond memories when I see an Amiga....
Hi Glen, not wanting to teach you to suck eggs but try re-flowing the solder joints on the connector for the Zorro slot. It has worked for me many times and now is my usual first point of attack for this old stuff and considering the Supra soldering was bad...
Fingers crossed! I think you are right about the RAM chips.
Hopefully so, would be fun to have it working and be able to test its speed.
Why was your power light dim? Was the 5volts loaded down by a shorted tag ram?
The Amigas power light always starts dim then should go bright as the system comes out of reset and boots. I'm not sure exactly were it changes brightness in the boot sequence but it's early on somewhere. That said though your comment has made me think about the VCC reading of 4.6v I seen on the scope and that might be because of the shorted tag ram and that might be why its not starting... Thanks for the question, cause its made me think about the problem again 👍
There is an I/O signal which controls the LED and the low-pass audio filter. It is not possible for the A500 to turn the LED all the way off under software control, it is either dim (and the low pass filters are off) or bright (and the low pass filters are on). Off means that there is no 5V rail going to the board (i.e. power switched off or unplugged)
Good job you have 3rd Glen to call on should Glen 1 get too Dremely😂 that cracked me up
Glen always needs to keep an eye on Glen, no telling what he'll get up to next.
Yeah, the industry was rife with manufacturers fiddling their benchmarks back in those days. Wouldn't surprise me if they "accidentally" left a fast RAM expansion in their machine while doing the tests.
Enjoyed that video. What happens if you run the card without its CPU?
I didn't try it but I'd expect it to do nothing as the accelerator will be disabling the onboard CPU.
Actually, tantalum caps are notorious for blowing up if exposed to uncontrolled inrush current. It is easy for a new engineer to not be aware of this, but they would find out pretty quickly once units started coming back from the field.
The tants certainly aren't shorted but I can pull them from the board and test properly, or maybe just replace them since they'll be off anyway.
I wonder if you just removed those suspected chips if it would work without them?
That's not a bad idea actually, I wonder if it would show any life with the ram chips removed. I can test this in the next video. Thanks for the idea.
I should have tested all the pins around the chips to see if every pins are solid. Not only look at them ...
I may yet do that but to be honest I think it was a long shot that there would be any cracked joints.
Its possible to sand chips due licensing.. So other company`s cant use those chips on their products. (or have to pay a fee). But sanding it, they couldnt know which ICs are used.
So no court cases after all.
Possibly yes but I would think the RAM and CPU to be off the shelf parts so surely anyone would be free to use them on any product?
could it be possible it just doesnt work with a 500+ and needs a standard 500?
i had an amiga 500+ and the Supra28 worked fine with it with a HD addon.
Nice idea but just tried it on my rev 6 A500 and it doesn't work there either. Will show this in the next video.
the chips with the ionfo sanded off might be to proof of that it all original, and the chips haven't been swapped out
I dare say you are correct, there was certainly no evidence of it being worked on in the past.
What does the Machine jumper change? Just a wild thought but maybe try it in a 2000 if you have one.
I have my EATX2000 although I'd need to strip that machine down a fair bit to try it. What I can do though is try it in the A500 side car to A2000 CPU socket.
Very interesting! I have one of these and it exactly as dead as yours. I sent it to y youtuber too, but nothing came of it...
I had one for my 2000 back in the day and mine was like yours. Dead. I think it did allow the machine to work in bypass mode.
Out of interest do you remember if it gave you much of a speed boost? Did you have it paired with any fast ram?
@@CRG I think it did once or twice, but it just crashed all the time. i gave up with it in the end. A friend of mine had the 500 version like you and yes it made Elite very playable.
I think I considered getting one of these Supra units back in the day, but IIRC, I heard that there was a high failure rate so I went with the ICD AdSpeed instead, which is the 14.32MHz 68000 in the normal HMOS 64-pin DIP package that displaces your original processor. Plus, I thought it would look goofy along with my Trumpcard 500 hard disk on the side (although honestly it's the Trumpcard 500 that looks goofy, not matching the lines of the A500, like literally WTF).
A more electronics-savvy friend of mine helped install the AdSpeed (well, switching the CPU, I could handle) along with a toggle switch on the back of the case (that part, he did, also simple but he had the parts and tools) to enable/disable it. It was a pretty slick mod and it sped things up considerably - IDK, maybe 50-80% depending on the task. To be fair, this was a pretty well-expanded A500 with a fully populated 4MB BaseBoard in the trapdoor. You could feel the difference compared to a stock Amiga. It definitely ran Frontier better but back in the day there was no way to quantify it - no Afterburner or FRAPS or Mangohud or what-not. But honestly that game runs best on a 68030 or better.
While it of course would be best to get it repaired - would it be possible to place a pistorm inside the box? I remember were is an adapter for the PiStorm to connect externaly to an A500, but it is flat and uses much mor space on the side. However, a Pistorm could be more compact if using a Pi zero 2 and both the Pi and the Pistorm having no headers assembled, so you coud connect them as a sandwich.
Would be realy cool to use this original extension case of the time with more modern hardware inside...
You probably could mount a pistorm inside this case but remember its not my accelerator and ultimately has to go back to its owner.
Oh, I forgot it wasn't yours.
However, would you mind to publih detailed photos and some measurenents? Maybe someone would design a replica of this case for 3d printing.
Back in the days I never was a big fan of theese kind of external extensions. Esp. theese heavy ones with a hard drive inside. One of the reasons I didn't bought an A500 but an A2000 back then. I always doubted the mechanical strenght. But theese case is so smal, I realy like it. And while you could do a pistorm inside the A500(+), there is still a compatibility issue with some older software. As the internal version replaces the original 68k, the external just deactivates it. As the case is smal an a pistorm is light, I think putting it in such a case would be the best solution.
I had a Supra Turbo for my A500 back in the day. However, not that speedy one that can seemingly outperform a 25MHz 68030, oh no no. Mine had the blistering speed of a 10MHz 68010. Seriously, what was the point, it barely did anything. I did later get a hard drive and attached it via the passthrough connector. I remember wishing I had a real accelerator instead in one of the Zorro slots in the hard drive enclosure.
I think to make best use of this card you will need some sort of fast ram although I still don't think it's getting near a 25mhz 030.
We all wished for the hardware we could never afford back in the day. My Amiga in the 90s was bare bones throughout that period, it wasn't until the mid 00s when I got it out again that it started getting upgrades.
@@CRGIt's entirely possible that I'm misremembering as I can't find any info on the Supra Turbo other than the 28. I do remember a 68010, maybe it was 14MHz and a separate speed "hack" I tried. I'm getting old. I did have trap door RAM, but if I'm remembering correctly that is slow, not fast RAM. I don't remember at all if my hard drive enclosure came with additional fast RAM (I bought it used).
@@QunMang ICD AdSpeed? That was a 14.3MHz 68000, not a 68010. It did have a bit more circuitry, I think 8K each of cache and cache tag...? It also had a jumper that you could connect a switch to, to enable/disable it but you have to reset the machine for it to change IIRC. Or not. It was over 30 years ago... my memory's fuzzier than my sac nowadays
Wonder is it the same Alan that sold me my A1200 Tower back in 2000 haha. Probably!!
It may well have been, I'll ask him when I next see him.
You should've sysinfo'd the CPU whilst you had it out. I think you'll need to find someone with a working version, so you can dump the pld's and back them up.
If I can get the card working I'll certainly do that but yes I probably should have done it while the chip was off.
@@CRG It would have shown a 68000 at 7MHz, shocker
flip the switch and see if it runs when the CPU is clocked at the slower rate
No it doesn't run, the CPU gets the 7mhz clock ok but it still just sits at a black screen.
one of the resistors looked discolored
Really? I didn't notice that myself but I'll take another closer look. Thanks.
GLENCEPTION
Glen, as stupid and painful as this sounds, what about just remove a component, test it, and if it checks out, mark it off a board print out, and then apply fresh solder and reflow it back in, doing the whole board front and back?
Not a bad idea to be honest and I'll probably try that if swapping out the tag ram doesn't have any effect.
@@CRG I updated my comment to remove a component, test it, and reinstall it if good.
I'm not sure I can properly test the likes of the PLDs or the PAL. I might be able to test the SRAMs using the TL866 although I'd need to make some sort of adapter for it.
Is it not possible to accurately ID CPUs of this era through software? I'd be curious to know what the 68K in that card actually is.
If I can get it going I can run whichamiga, that should give us the detailed info on what the CPU is.
@@CRG IIRC from bitd they overclocked a 16 or 20MHz chip. You can do some research and see if there even was any higher clocks available at that time.
@@NorthWay_no @CRG As far as the clocks of the 68000 are concerned...
The standard 68000s are available up to 16 Mhz and the 68HC000 CMOS versions up to 20 Mhz. Both are still made I think, by Freescale. It was a sure bet that the HC CMOS variant would easily hit 28 Mhz, but some such boards offered up to 40 MHz settings to let you experiment.
I also remember that sometimes those chips needed a little more voltage, like 5.2V, to run a bit faster. Maybe that's where the hole for an external power supply comes into the story. an external PSU could have 5.3V and help overclocking but they never integrated it into the final product. IIRC the CMOS specifications allowed for up to 6V.
@amurtigress_mobile365 thanks for the info. Seems they were overclocking regardless of what chip they used and from what I've since been told its the 16mhz version they fitted to these boards. I do have a spare HC 20mhz chip here that I was considering fitting but since the current cpu started ok at 14mhz in that other card I thought it just best to continue to use it. It must have worked at 28mhz at some point.
Those 2 sanded of RAMs are probably the same type as the regular ones near the CPU, probably the tag memory. Would be cool if you were able to read out the programmable logic contents! I always wondered how the state machines in cache controllers worked.
Lets suppose one of those tag rams are broken - the controller would most likely try to fetch data into data memory and update tags, but tags wont update (because the broken RAM chip) - tag isn't equal to whatever part of address it should be equal to - cache controller tries to fetch same data in a loop?
😂 nice one.
Thanks 🙂
Get the sus chips off the board and compare their Vcc - GND resistances.
Yeah not a bad idea, they need to come off anyway so I will certainly check that.
rip
Haven't given up just yet but so far it isn't looking good.
@@CRG yea, everything i can think of is already mentioned below
Hi muzicmolehead here 1.2 or 1.3 openrating boot on the amiga 500 on the supu speed up bit acccrator thats all it needs best muzicmolehead
I've tried it on my rev 6 Amiga 500 and it doesn't work on that either.
my supra works fine in my a2000 with a 3.1.4 kick and 8 mb fast
Sound like addon for SNES haha. Addons for snes start with super to all time.
no, not Super, this was SUPRA, as in the brand that (in)famously made that 2400bps external modem that everyone (including me, until I got my USRobotics Courier HST) used back in the day, that got hotter than the devil's balls. It had an aluminum billet case that doubled as a heat sink; I wondered what kind of circuitry was being tortured inside of it, because it just got soooo hot... Anyway they got bought by Diamond Multimedia in 1995, which itself was party to mergers and acquisitions... the hardware market was wild in the 1990's.
Would you like to sell it? :)
Made in USA. Shocking.
I didn't even notice that but yep its right there on the box!
Enjoyed that video. What happens if you run the card without its CPU?
I didn't try that but I expect it will do nothing as the accelerator will be disabling the onboard CPU.