Posted by majsta on Wednesday, March 31 2021 @ 17:05:44 UTC (234 reads)
(comments? | Score: 5)
Vampire 1200: Vampire 1200 wave soldering
Soldering by hands 150 pins card edge connector on Vampire 1200 was something that took lot of time for me. It is easy when you have to do it once but it become pretty exhausting over time when you have to deal with much larger numbers of cards. Next logical step was to try to do it more professional with help of Wave soldering technique. Initial setup was semi-automated with possibility to solder two connectors same time. For next batch of cards soldering will be completely automatic.
It is not so hard to tell what was the most exciting moment between Apollo-team members during work on Vampire 1200 accelerator cards. For sure, yesterdays results regarding enabling fast Vampire 1200 IDE port is on top of everything.
In the process of designing card all of the facts needed to make this type of accelerator were known, all but one. Is it possible to keep regular Amiga 1200 IDE port and have one more, same time, much faster mounted on V1200 card. In theory we did know what needs to be done, after all we did recreate Gayle chip and its registers needed for IDE to work on Vampire 500 and made very fast IDE port on that card. But here situation was different, much more complex to the point that we could only enable either Amiga IDE port or Vampire IDE but not both same time. By enabling Vampire IDE port situation got even worse, not only by disabling regular Amiga 1200 IDE port but PCMCIA got disabled too. Yeah, yeah, you may say that it is all about address decoding, Amiga memory map, appropriate IRQ handling and drivers... OK, it is, but it was very hard to make perfect combinations to not cause conflicts between Vampire 1200 and Amiga.
On last conference call between Apollo-team members we decided that only solution to this problem is to have two cores for people to chose. One with working Vampire IDE or one with working IDE and PCMCIA found on Amiga motherboard. In my own desperation I was thinking constantly that there must be some hardware flaw I did by accident during PCB design. Things started to change few days ago when we pushed harder and finally made it work :) After all of this work is polished some more all of the new features will be presented in next Vampire 1200 core release.
Vampire 1200 IDE in action
Posted by majsta on Monday, July 13 2020 @ 19:25:28 UTC (655 reads)
(comments? | Score: 0)
Vampire 1200: Redesign, once more!
Working on 3rd Vampire 1200 accelerator revision took every piece of energy I had in my body. It was so time consuming and complex that I can't even explain. 2nd redesign revealed that we have mechanical problems. And yes that happened because I have designed accelerator without having Amiga 1200 plastic housing. In fact I have designed accelerator without even having complete Amiga 1200. I was thinking that pictures and documentation found online should be enough to calculate everything. Oh boy, could I be more wrong. Result, missed accelerator size by 3mm. It was too wide, hitting case and very hard to insert into trapdoor. It took me over one month to fix that problem. So, 3mm not much you may say, but in PCB design that's huge and had to move all components to make them fit to new PCB size, rerouting again, and again, who knows how many times.
Finally some time ago everyone jumped in to help when they heard what problem I have. Well, they were polite enough to not say that I was stupid, but I was. Like you try to make window frame on the house without measuring hole on the wall or even seeing the house. That's how designing this accelerator was. Apollo-team members send me some plastic cases for A1200. Some famous Amiga shops and case makers also did help providing me old and new, redesigned cases. I was able to measure and this time accelerator fits perfectly!
When I was redesigning the card, we, inside team talked a lot about certain features or problems connected to Amiga 1200 design. We took special attention to Amiga 1200 "dirty" clock problems and I hope that we solved that in this revision. Also, card has some other features now for potential support for other popular expansions.
Soldering first prototype took 2 days under abnormal conditions created by very high summer temperatures in my city and the fact that there has been work on the house connected to mine. Imagine soldering 0402 resistor in the situation when everything is vibrating from power drills used to destroy walls. Surrounded with hot air, heat from iron stations, vibrations and workers who are thinking that they know how to sing, it was nightmare.
Posted by majsta on Tuesday, October 30 2018 @ 17:47:56 UTC (907 reads)
(comments? | Score: 0)
Vampire 1200: Amiga 1200 edge connector pursuit
In the process of making Vampire 1200 I have found only one problem. Card edge connectors with 150 pins are extremely hard to find. Fujitsu made them and then discontinued in 2000. Now they don't have replacement part. Original part number is: FCN-225J150-G/A. Connector itself is right angled with 1.27mm pitch and two separated sides. One side consists of 2x25 pins and another from 2X50 pins. Even building footprint for it was challenging. As I see it here we have only two options either we search for NOS or we manufacture those connectors. So far I have sent anywhere between 200-300 mails and made lot of phone calls with every person who had in his hands connector at least similar to what we need here. Searched over old newsgroups, tracked down people who owned Amiga shops 20 years ago, talked with every card edge manufacturer or reseller in the World. Even talked to Fujitsu directly. Most challenging was to read Japanese and Chinese web pages to be able to search their NOS. So far result is like this. In the existence there are 174 NOS connectors with insane prices and tooling fees with Sullins and Taiwan companies are huge for Amiga hobby projects. What can we say about connector itself. It is not even standard connector as Fujitsu sees it, it is not even listed in their datasheet where all versions of FCN-225J were listed. This decision Commodore made created pure nightmare for any potential hardware developer. Even back then, when Amiga 1200 was designed official proposal was to use two connectors, one with 50 and one with 100 pins, but even that is not an option now because also those are hard to find at normal prices. I keep asking myself after one week of constant searching for those connectors why did they make decision like that when lot of those pins are unused and any card edge connector with 120 pins could be used. Those connectors was standard then and they are standard now and probably they will be standard type of connectors in next 30 years. Another question is why they didn't put just simple 2.54mm pitch male header. All of this only tells me that someone tried to play monopoly game on the market, to create distance from other projects but like history teach us in some areas this is not smart to do. In reality put whatever you want inside box but for talking to outside world use exactly what everyone else uses. Talking about that makes no sense now, what is done it's done. I don't want to use PCI connectors and cut them because that won't look professional and it is time consuming if you consider that we will probably need to make at least 1000 Vampire 1200 cards. It would be also too stupid to have DDR3 on board and in the same time connector who is butchered and then glued. I ll probably find a way how to solve this problem like I solved many others in past years but situation where you need to pay for a connector same price as you are paying FPGA is not an option for now simply because makes no sense. IMHO those connectors shouldn't cost more than 5USD. Quantity we have now is enough to do one small series of cards but what happens later? Some drawings done by kipper2k and actual connector picture, datashet can be found in in Read More, Pictures, Files... section of this article.