Price of time ...
Long time for design multilayer PCB, soldering, testing ....
Time is not the only thing; I think that if Ingo would really add his spent time to the costprice, the MB02+IDE would not be affordable!
There are some really rare parts on this interface, some can cost about 20 euro a piece.
The printed circuit board has to be a more expensive four layer one to make the interface stable and this small, instead of a two layer pcb.
And don't forget prototyping.
It's really not expensive if you calculate the costs yourself.
I've done that trying to make a MB02 succesor, but then I decided to buy one from Ingo as it's far more easier and cheaper to get one from him ;)
Time is not the only thing; I think that if Ingo would really add his spent time to the costprice, the MB02+IDE would not be affordable!
There are some really rare parts on this interface, some can cost about 20 euro a piece.
The printed circuit board has to be a more expensive four layer one to make the interface stable and this small, instead of a two layer pcb.
And don't forget prototyping.
It's really not expensive if you calculate the costs yourself.
I've done that trying to make a MB02 succesor, but then I decided to buy one from Ingo as it's far more easier and cheaper to get one from him ;)
Yes, FDD controller WD2797 is rare chip.
If will all logic integrated to one small low price CPLD then can be used small PCB with two layers (Z80DMA + WD2797 + rom + sram + PIO 2855 + RTC + CPLD).
Yes, FDD controller WD2797 is rare chip.
If will all logic integrated to one small low price CPLD then can be used small PCB with two layers (Z80DMA + WD2797 + rom + sram + PIO 2855 + RTC + CPLD).
To me, this would be the way further- to make new small and 100% compatible MB02+, integrated in few IOs, enabling the user to connect also FDD.
But MB02+ users are so rare, hardly anyone could undergo such a painful process of designing this new concept. (Or to implement MB02+ to some existing clone, like Speccy 2010 or ATM- my eternal wish, that MB02+ would spread amongst the guys from the ex CCCP countries).
Hope mine old one will still last for several years.
It seems it is too complex to be produced in small quantity with resonable price.
Maybe making some features optional (like fdd) would help.
But then-how much of MB would be there...
Yes, FDD controller WD2797 is rare chip.
If will all logic integrated to one small low price CPLD then can be used small PCB with two layers (Z80DMA + WD2797 + rom + sram + PIO 2855 + RTC + CPLD).
Yes you are right that might be done. Also it is possible to realize functions of the MB02+ using other equipment like divIDE+ with some adds or as part of clones.
But that is not my business. I as a fan of the original concept of MB02+ am only interested in making perfect clones of that interface. Any (drastic) modification of the original concept like making it without fdd support would create a different interface with different properties and possible incompatibilities to older MB02+ hardware.
The MB02+ is at first a floppy interface (like a +D is one), that can be improoved by IDE support. It is a legend. We discussed some time ago the importance of FDD support today. But I think, if we make it without floppy, future system software will not support floppy, dividing the MB02+ users in different groups. I will not do so.
Using of CPLDs was also discussed. You are right that this could save space on PCB and could improove flexibility of reorganizing hardware functions. But this flexibility is also a danger for compatibility inside the MB02+ user community (if better arrangement will be established software will only support that new device). Also I will not have time for developing, testing of porting hardware functions to CPLD.
So my initial question is still present:
Is there any interest in getting a MB02+IDE interface in future and what would be the maximum acceptable price?
And please use modern 512kB flash eprom instead old 2kB eprom.
Original MB02+ support up to 512kB sram and up to 512kB rom. Original rom memory size is 2kB, but using flash eprom is better (possibility reflash/upgrade rom versions, etc...)
Exist MB02+ versions of flash utilities for flash eprom 64/128 and 512kB, all is tested.
Flash eprom type 29F040 can be used and your MB02 will in full configuration :-)
Work on new MB02+IDE reset feature(OUT 23,192) with +2A/+3 models ? This models need often longer reset (this is problem on original MB02+).
And reset signal from ZX bus must be boosted (74LS244 etc...) for internal IDE interface. Some CF cards use pull-down on reset pin and ZX is unstable. If you will use 40 pin IDE connector then please connect +5V to pin 20. This pin must be unused for new 80 wires ide cable, but old IDE cables use all contacts. Also CFtoIDE and SDtoIDE adapters use this pin as +5V (work without power cable ).
And please design PCB size for any standard plastic box.
And please use modern 512kB flash eprom instead old 2kB eprom.
Original MB02+ support up to 512kB sram and up to 512kB rom. Original rom memory size is 2kB, but using flash eprom is better (possibility reflash/upgrade rom versions, etc...)
Exist MB02+ versions of flash utilities for flash eprom 64/128 and 512kB, all is tested.
Flash eprom type 29F040 can be used and your MB02 will in full configuration :-)
Work on new MB02+IDE reset feature(OUT 23,192) with +2A/+3 models ? This models need often longer reset (this is problem on original MB02+).
And reset signal from ZX bus must be boosted (74LS244 etc...) for internal IDE interface. Some CF cards use pull-down on reset pin and ZX is unstable. If you will use 40 pin IDE connector then please connect +5V to pin 20. This pin must be unused for new 80 wires ide cable, but old IDE cables use all contacts. Also CFtoIDE and SDtoIDE adapters use this pin as +5V (work without power cable ).
And please design PCB size for any standard plastic box.
Velesoft and Ingo, would it be very hard to integrate into the MB02+ you are offering here, also Velesoft's KMOUSE ? And about the IDE devices, do you support and integrate both master and slave? or only master? To support both would be very good.
the 4-layer board 100 x 160 mm can be modified a bit to improove compatibility with some different firmware versions (like recommended by VELESOFT) but to include the mouse interface is not possible I think. There is no space to include an additional jack for the mouse nor the chips and wires on the board.
Also I will not do a complete redesign of the layout. A (commonly used) kempston joystick interface is part of the MB02+IDE.
The IDE port on the board supports master and slave device. To use it one has to insert a double CF card adapter or 2 adapters.
Maybe I coud look if there is a possibility to use an industrial made CF-card adapter.
as I could get some rare chips to make MB02+IDE interfaces I think about making a last lot of it. Some simple TTL chips like the 74LS38 seem to disappear from the market.
I don't think the 74LS38 is in danger of going away soon, Farnell stock them brand new (and have over 1000 in stock right now). In any case subsitutes for that chip can be made with a 74HC part and some MOSFETs.
A lot of about 15 till 20 items had to be made to get an acceptable price for the 4-Layer-PCBs I had to order.
If you're making this sort of number, consider going completely surface mount and having them factory assembled (going full SMD keeps the price down, less manual labour during the assembly, it can all be done with pick-and-place and reflow soldering - and of course you can use a smaller PCB - I cram 6 ICs into a 60x100mm board for the Spectranet, and there's room left over for things like mounting screw holes). I had 50 Spectranets assembled by Quick-Teck ( http://www.quick-teck.co.uk/index.php ), they did all the part sourcing, organized the PCB manufacture (the Spectranet PCB is 4 layer) and assembly - the price was very reasonable. The Spectranet has quite a high component count too (6 ICs and dozens of passive components and a couple of transistors). I just provided them with gerbers for the PCB and a BOM, and they came back with a quote, then I had them assemble them. Even if you don't get the cost down much, at least you won't have to spend hours doing the complete assembly (for the Spectranet, I still do the few through hole components myself, and also the edge connector since they simply aren't made so you have to make them yourself). Incidentally they can also do through hole wave solder assembly, I found out about Quick-Teck from zonadepruebas (a Spanish retro forum), where they had the ZdP Jupiter ACE assembled. I can ask on ZdP how that went, if you like.
There are other PCB fab and assembly services, I've heard that PCB Train are good but I've not used them nor got a quote from them, so I don't know what their prices are like.
two layers (Z80DMA + WD2797 + rom + sram + PIO 2855 + RTC + CPLD).
Or go all the way and stuff the whole lot into an FPGA. The DMA certainly can be implemented in a small FPGA, and I suspect the WD2797 can fit in the same device too with some external components, and of course all the bus decoding logic. It also means you don't have a dependency on an obsolete chip.
Incidentally, Xilinx recommend 4 layer boards for their CPLDs to get a proper ground and power plane, but in reality with many circuits I've found they work fine with 2 layers, especially retrocomputing stuff since they aren't high frequency (ground bounce can still be an issue).
you are right, the expensive Farnell company is probably selling most of TTL chips but I found that most german electronic shops I use as Reichelt, Kessler-Electronic, Segor or Conrad are missing an increasing number of TTL hardware.
On my MB02+IDE board I use SMD for all chips that are available in SMD. Only the DMA, E(E)PROM, PIA, GAL, RTC and FDD controller are DIPs. The E(E)PROMS and GAL are also available in SMD but I use a programmer without such SMD socket. Velesoft told that also DMA is available in SMD but only the RFT DMA UA858D will work with the original BS-DOS and as I told compatibility is most important for me.
Because of the fact that some of the chips are very rare I would order the PCB AFTER I got are chips I need for production. That's why it is a problem to get the PCBs soldered because the manufactures don't like to use my material.
Also I like to test some functions before soldering the next part.
Because of high costs I do not pay for fully tested boards so I have to do that myself.
To realize the MB02+ inside a gate array would be a way for minimizing but it would be a complete new development. I have not the time to do that. And if this is done it works for the device that was used for development. Some years later the chip maybe obsolete and repairing or productoin would require new development.
So I myself will only make a good reproduction of the original interface to be used with the original Spectrum computer. This reproduction only consists of parts that would be principally available while the original Spectrum was manufactured.
Or go all the way and stuff the whole lot into an FPGA. The DMA certainly can be implemented in a small FPGA, and I suspect the WD2797 can fit in the same device too with some external components, and of course all the bus decoding logic. It also means you don't have a dependency on an obsolete chip.
Many people underestimate the effort to be taken to develop an FPGA like you mention.
When someone wants to start such a project he needs:
1) Full knowledge of the original hardware, that is the MB02+IDE; only the 8BC (www.8bc.com/sinclair) members and Ingo have that
2) Advanced experience with FPGA's; some have that for their work. I think the people who have developed for example a ZX Spectrum in an FPGA, were able to do that because they are advanced engineers with those parts for their work
3) Time.
The people who have all three are blessed.
So I guess noone is blessed after all ;)
And of course, it would be a new piece of hardware, incompatible with the current.
If you want it to be compatible, you probably have to do some BSDOS changes too.
Guess that is too much for a human ;)
No offense Ingo, I think you did the greatest job!!!!!!!!
Ow, one more thing: some people were able to combine some of these elements, like in this 'multi-machine' FPGA platform: http://www.worldofspectrum.org/forums/showthread.php?t=36966
But that also costs at least 200 euro, and it's not even a plug-and-play device ;)
So an FPGA does not have to be cheaper.
Of course these developers want to make some profit for the time spent developing, but that's fair.
It seems it is too complex to be produced in small quantity with resonable price.
Maybe making some features optional (like fdd) would help.
But then-how much of MB would be there...
The price is very reasonable, it is a bargain. As Ingo wrote, MB02 is a legend. No other interface offers you all these possbilities. I don?t want list all the advantages here as there are lots of pages already filled about this. But it is more than just a IDE-device, it is the nonplusultra for the real Spectrum users. Who want just play some games can get a Divide and be happy. But for example many original software uses tape for saving. Might be a game or your work in Art Studio, Music, Code in Assembler, etc. All the works that usually you save to tape you can save with MB02 to disk without change the code. You just copy your programs to disk and work with them as on tape, just faster, better.
There are many more things, like DMA, the fastness of the disk at all, IDE support... but all these other interfaces might offer as well. But no interface offers all in one.
I am MB02 user since 1995, when I saw this interface in action the first time, it was totally clear that I have to have one. I had talks with the developer, who made this just for a few friends and I told him, that this interface should be available to more users. Then Jaxon Hollis showed his prototype at a Spectrum show in Germany and many letters came in , everybody who saw it wanted have it. At least there were 40 visitors on the show and when finally a new company called 8BC finally produced the MB02, we sold about 50. Then Ingo made another 20 pcs. 3 years ago, and these are the only MB02 sold outside of Czech Republic. So there are not many users at all, but I am sure that nobody who ever bought the MB02 thought that he invested one Cent too much.
Is there any interest in getting a MB02+IDE interface in future and what would be the maximum acceptable price?
Greetings
Ingo.
Many people ask me every week for mass storage devices for Spectrum. Some buy the Divide or a Plus D Interface, but I think every month is min. 1 guy who I think would take a MB02. Price of 200-250 Euro is not a problem for them.
At least all spend much more on PC hardware every year, and this hardware is outdated before the year ends. So 250 Euro for a computer that brought you fun the last 30 years...it is nothing. That?s what I tell the customer because this is my opinion.
Only the DMA, E(E)PROM, PIA, GAL, RTC and FDD controller are DIPs. The E(E)PROMS and GAL are also available in SMD but I use a programmer without such SMD socket. Velesoft told that also DMA is available in SMD but only the RFT DMA UA858D will work with the original BS-DOS and as I told compatibility is most important for me.
Original BS-DOS contain bugs and code for controll DMA transfer is also incorrect (with this bugs work only with UA858D DMA chip). Other compatible Z80DMA chips need fixed code. Fix in BS-DOS is very easy (only some pokes...). All DMA software is fixed for possibility ude with any DMA chips. :-) http://velesoft.speccy.cz/data-gear.htm
Exist also bug fixed version of BS-DOS (signed as BS-DOS 309) and exist in versions for MB02+(full DMA support),MB02+ version without DMA and FDD support, DIVIDE+(not use DMA and FDD) or for original DIVIDE (not use DMA and FDD).
you are right, the expensive Farnell company is probably selling most of TTL chips but I found that most german electronic shops I use as Reichelt, Kessler-Electronic, Segor or Conrad are missing an increasing number of TTL hardware
74LS37 is only 60p from Farnell in 10+ quantities, that's better than not being available at all. I can't imagine 60p being a deal-breaker on a board expected to cost over 150 eur.
Because of the fact that some of the chips are very rare I would order the PCB AFTER I got are chips I need for production. That's why it is a problem to get the PCBs soldered because the manufactures don't like to use my material.
Have the manufacturer do partial assembly, that's what I did for the Spectranet (I had Quick-Teck do the SMD stuff which was time-consuming for me to do, and I do all the through hole parts like the crystal, the ethernet magjack, the LEDs, and of course the Speccy edge connector which I have to make myself). A huge time saving. Just have the assembler source the common parts (the 74LS37 will be easily sourced by an assembler, basically anything with a Farnell part number, they will be able to find). If you're making 20 boards, it'll probably save you 60+ hours of fairly dull work if you have just the common parts assembled for you.
Also don't skimp on PCB electrical testing. If you are making 4 layer boards, find a board maker who will do the electrical testing for free (many will do it for free on 4 layer boards). PCB CART does very high quality 4 layer PCBs (never had a failed board from them). Quick-Teck's PCB manufacturer does free electrical testing on their PCBs, too. Of course with our stuff we can't have an assembler do the final testing of assembly because you need special hardware (a Spectrum) and connectors that haven't been manufactured for 20 years (the Spectrum edge connector) so you just have to do that yourself; but you'll find very few problems with a reflow soldered board (asking around at Replay 2010, I was told to expect probably a 5% failure rate), and any that you do find can be corrected if you're already able to do SMD soldering yourself.
it is the 74LS38 not 37 I need (together with a lot of other parts). You are right the price for one chip is not that item that rises the price of the board.
But compared with other shops in germany Farnell is in most kinds the most expensive one. In Germany Farnell do not sell to private persons so I have to find a company that do the business for me.
Btw. the MB02+IDE contains 30 chips and about 1700 soldering points.
I did not check the actual prices of PCB-Pool now. But 5 years ago when I ordered the first lot of boards for MB02+IDE I found a company that did the job a lot cheaper in good quality. I decided to get 3 more items (15 while 12 were payed) instead paying the (expensive) check.
Especially because I was not sure if there were additional design errors I had to solder the boards myself (btw. there were some small problems). I made a first prototype that missed some mechanical properties and some functions. Then I did a redesign and I did not do a further test board but ordered the whole lot I needed.
Otherwise the price would be too high.
While looking for a case and wishing more comfort I decided to design an additional board - a special CF-IDE converter with special mechanical design adapted to the design of the MB02+ and containing Reset and NMI switch as well as 2 LEDs. I gave that additional board to the interface without additional payment from the users.
The whole project needed a lot of my spare time. And I do not plan to make a complete redesign with the risk of new problems (espaecially mechanical ones).
I will ask some producers for soldering parts but I like to be sure that the price will pay all costs I have even if there will fail some parts or if I will not sale all items. I have the whole risk so I have to have the controll.
If I would order soldered boards I have to be sure to sell out all boards I make. That is not sure I think.
Comments
Time is not the only thing; I think that if Ingo would really add his spent time to the costprice, the MB02+IDE would not be affordable!
There are some really rare parts on this interface, some can cost about 20 euro a piece.
The printed circuit board has to be a more expensive four layer one to make the interface stable and this small, instead of a two layer pcb.
And don't forget prototyping.
It's really not expensive if you calculate the costs yourself.
I've done that trying to make a MB02 succesor, but then I decided to buy one from Ingo as it's far more easier and cheaper to get one from him ;)
Yes, FDD controller WD2797 is rare chip.
If will all logic integrated to one small low price CPLD then can be used small PCB with two layers (Z80DMA + WD2797 + rom + sram + PIO 2855 + RTC + CPLD).
To me, this would be the way further- to make new small and 100% compatible MB02+, integrated in few IOs, enabling the user to connect also FDD.
But MB02+ users are so rare, hardly anyone could undergo such a painful process of designing this new concept. (Or to implement MB02+ to some existing clone, like Speccy 2010 or ATM- my eternal wish, that MB02+ would spread amongst the guys from the ex CCCP countries).
Hope mine old one will still last for several years.
Hood
Maybe making some features optional (like fdd) would help.
But then-how much of MB would be there...
But that is not my business. I as a fan of the original concept of MB02+ am only interested in making perfect clones of that interface. Any (drastic) modification of the original concept like making it without fdd support would create a different interface with different properties and possible incompatibilities to older MB02+ hardware.
The MB02+ is at first a floppy interface (like a +D is one), that can be improoved by IDE support. It is a legend. We discussed some time ago the importance of FDD support today. But I think, if we make it without floppy, future system software will not support floppy, dividing the MB02+ users in different groups. I will not do so.
Using of CPLDs was also discussed. You are right that this could save space on PCB and could improove flexibility of reorganizing hardware functions. But this flexibility is also a danger for compatibility inside the MB02+ user community (if better arrangement will be established software will only support that new device). Also I will not have time for developing, testing of porting hardware functions to CPLD.
So my initial question is still present:
Is there any interest in getting a MB02+IDE interface in future and what would be the maximum acceptable price?
Greetings
Ingo.
Original MB02+ support up to 512kB sram and up to 512kB rom. Original rom memory size is 2kB, but using flash eprom is better (possibility reflash/upgrade rom versions, etc...)
Exist MB02+ versions of flash utilities for flash eprom 64/128 and 512kB, all is tested.
Flash eprom type 29F040 can be used and your MB02 will in full configuration :-)
Work on new MB02+IDE reset feature(OUT 23,192) with +2A/+3 models ? This models need often longer reset (this is problem on original MB02+).
And reset signal from ZX bus must be boosted (74LS244 etc...) for internal IDE interface. Some CF cards use pull-down on reset pin and ZX is unstable. If you will use 40 pin IDE connector then please connect +5V to pin 20. This pin must be unused for new 80 wires ide cable, but old IDE cables use all contacts. Also CFtoIDE and SDtoIDE adapters use this pin as +5V (work without power cable ).
And please design PCB size for any standard plastic box.
Velesoft and Ingo, would it be very hard to integrate into the MB02+ you are offering here, also Velesoft's KMOUSE ? And about the IDE devices, do you support and integrate both master and slave? or only master? To support both would be very good.
Hood
the 4-layer board 100 x 160 mm can be modified a bit to improove compatibility with some different firmware versions (like recommended by VELESOFT) but to include the mouse interface is not possible I think. There is no space to include an additional jack for the mouse nor the chips and wires on the board.
Also I will not do a complete redesign of the layout. A (commonly used) kempston joystick interface is part of the MB02+IDE.
The IDE port on the board supports master and slave device. To use it one has to insert a double CF card adapter or 2 adapters.
Maybe I coud look if there is a possibility to use an industrial made CF-card adapter.
Greatings Ingo.
I don't think the 74LS38 is in danger of going away soon, Farnell stock them brand new (and have over 1000 in stock right now). In any case subsitutes for that chip can be made with a 74HC part and some MOSFETs.
http://uk.farnell.com/texas-instruments/sn74ls38n/ic-buffer-driver-receiver/dp/1470758
If you're making this sort of number, consider going completely surface mount and having them factory assembled (going full SMD keeps the price down, less manual labour during the assembly, it can all be done with pick-and-place and reflow soldering - and of course you can use a smaller PCB - I cram 6 ICs into a 60x100mm board for the Spectranet, and there's room left over for things like mounting screw holes). I had 50 Spectranets assembled by Quick-Teck ( http://www.quick-teck.co.uk/index.php ), they did all the part sourcing, organized the PCB manufacture (the Spectranet PCB is 4 layer) and assembly - the price was very reasonable. The Spectranet has quite a high component count too (6 ICs and dozens of passive components and a couple of transistors). I just provided them with gerbers for the PCB and a BOM, and they came back with a quote, then I had them assemble them. Even if you don't get the cost down much, at least you won't have to spend hours doing the complete assembly (for the Spectranet, I still do the few through hole components myself, and also the edge connector since they simply aren't made so you have to make them yourself). Incidentally they can also do through hole wave solder assembly, I found out about Quick-Teck from zonadepruebas (a Spanish retro forum), where they had the ZdP Jupiter ACE assembled. I can ask on ZdP how that went, if you like.
There are other PCB fab and assembly services, I've heard that PCB Train are good but I've not used them nor got a quote from them, so I don't know what their prices are like.
Or go all the way and stuff the whole lot into an FPGA. The DMA certainly can be implemented in a small FPGA, and I suspect the WD2797 can fit in the same device too with some external components, and of course all the bus decoding logic. It also means you don't have a dependency on an obsolete chip.
Incidentally, Xilinx recommend 4 layer boards for their CPLDs to get a proper ground and power plane, but in reality with many circuits I've found they work fine with 2 layers, especially retrocomputing stuff since they aren't high frequency (ground bounce can still be an issue).
you are right, the expensive Farnell company is probably selling most of TTL chips but I found that most german electronic shops I use as Reichelt, Kessler-Electronic, Segor or Conrad are missing an increasing number of TTL hardware.
On my MB02+IDE board I use SMD for all chips that are available in SMD. Only the DMA, E(E)PROM, PIA, GAL, RTC and FDD controller are DIPs. The E(E)PROMS and GAL are also available in SMD but I use a programmer without such SMD socket. Velesoft told that also DMA is available in SMD but only the RFT DMA UA858D will work with the original BS-DOS and as I told compatibility is most important for me.
Because of the fact that some of the chips are very rare I would order the PCB AFTER I got are chips I need for production. That's why it is a problem to get the PCBs soldered because the manufactures don't like to use my material.
Also I like to test some functions before soldering the next part.
Because of high costs I do not pay for fully tested boards so I have to do that myself.
To realize the MB02+ inside a gate array would be a way for minimizing but it would be a complete new development. I have not the time to do that. And if this is done it works for the device that was used for development. Some years later the chip maybe obsolete and repairing or productoin would require new development.
So I myself will only make a good reproduction of the original interface to be used with the original Spectrum computer. This reproduction only consists of parts that would be principally available while the original Spectrum was manufactured.
Greetings Ingo.
Many people underestimate the effort to be taken to develop an FPGA like you mention.
When someone wants to start such a project he needs:
1) Full knowledge of the original hardware, that is the MB02+IDE; only the 8BC (www.8bc.com/sinclair) members and Ingo have that
2) Advanced experience with FPGA's; some have that for their work. I think the people who have developed for example a ZX Spectrum in an FPGA, were able to do that because they are advanced engineers with those parts for their work
3) Time.
The people who have all three are blessed.
So I guess noone is blessed after all ;)
And of course, it would be a new piece of hardware, incompatible with the current.
If you want it to be compatible, you probably have to do some BSDOS changes too.
Guess that is too much for a human ;)
No offense Ingo, I think you did the greatest job!!!!!!!!
Ow, one more thing: some people were able to combine some of these elements, like in this 'multi-machine' FPGA platform:
http://www.worldofspectrum.org/forums/showthread.php?t=36966
But that also costs at least 200 euro, and it's not even a plug-and-play device ;)
So an FPGA does not have to be cheaper.
Of course these developers want to make some profit for the time spent developing, but that's fair.
The price is very reasonable, it is a bargain. As Ingo wrote, MB02 is a legend. No other interface offers you all these possbilities. I don?t want list all the advantages here as there are lots of pages already filled about this. But it is more than just a IDE-device, it is the nonplusultra for the real Spectrum users. Who want just play some games can get a Divide and be happy. But for example many original software uses tape for saving. Might be a game or your work in Art Studio, Music, Code in Assembler, etc. All the works that usually you save to tape you can save with MB02 to disk without change the code. You just copy your programs to disk and work with them as on tape, just faster, better.
There are many more things, like DMA, the fastness of the disk at all, IDE support... but all these other interfaces might offer as well. But no interface offers all in one.
I am MB02 user since 1995, when I saw this interface in action the first time, it was totally clear that I have to have one. I had talks with the developer, who made this just for a few friends and I told him, that this interface should be available to more users. Then Jaxon Hollis showed his prototype at a Spectrum show in Germany and many letters came in , everybody who saw it wanted have it. At least there were 40 visitors on the show and when finally a new company called 8BC finally produced the MB02, we sold about 50. Then Ingo made another 20 pcs. 3 years ago, and these are the only MB02 sold outside of Czech Republic. So there are not many users at all, but I am sure that nobody who ever bought the MB02 thought that he invested one Cent too much.
It is really worth the price.
Regards
Thomas
Many people ask me every week for mass storage devices for Spectrum. Some buy the Divide or a Plus D Interface, but I think every month is min. 1 guy who I think would take a MB02. Price of 200-250 Euro is not a problem for them.
At least all spend much more on PC hardware every year, and this hardware is outdated before the year ends. So 250 Euro for a computer that brought you fun the last 30 years...it is nothing. That?s what I tell the customer because this is my opinion.
Regards
Thomas
Original BS-DOS contain bugs and code for controll DMA transfer is also incorrect (with this bugs work only with UA858D DMA chip). Other compatible Z80DMA chips need fixed code. Fix in BS-DOS is very easy (only some pokes...). All DMA software is fixed for possibility ude with any DMA chips. :-)
http://velesoft.speccy.cz/data-gear.htm
Exist also bug fixed version of BS-DOS (signed as BS-DOS 309) and exist in versions for MB02+(full DMA support),MB02+ version without DMA and FDD support, DIVIDE+(not use DMA and FDD) or for original DIVIDE (not use DMA and FDD).
74LS37 is only 60p from Farnell in 10+ quantities, that's better than not being available at all. I can't imagine 60p being a deal-breaker on a board expected to cost over 150 eur.
Have the manufacturer do partial assembly, that's what I did for the Spectranet (I had Quick-Teck do the SMD stuff which was time-consuming for me to do, and I do all the through hole parts like the crystal, the ethernet magjack, the LEDs, and of course the Speccy edge connector which I have to make myself). A huge time saving. Just have the assembler source the common parts (the 74LS37 will be easily sourced by an assembler, basically anything with a Farnell part number, they will be able to find). If you're making 20 boards, it'll probably save you 60+ hours of fairly dull work if you have just the common parts assembled for you.
Also don't skimp on PCB electrical testing. If you are making 4 layer boards, find a board maker who will do the electrical testing for free (many will do it for free on 4 layer boards). PCB CART does very high quality 4 layer PCBs (never had a failed board from them). Quick-Teck's PCB manufacturer does free electrical testing on their PCBs, too. Of course with our stuff we can't have an assembler do the final testing of assembly because you need special hardware (a Spectrum) and connectors that haven't been manufactured for 20 years (the Spectrum edge connector) so you just have to do that yourself; but you'll find very few problems with a reflow soldered board (asking around at Replay 2010, I was told to expect probably a 5% failure rate), and any that you do find can be corrected if you're already able to do SMD soldering yourself.
it is the 74LS38 not 37 I need (together with a lot of other parts). You are right the price for one chip is not that item that rises the price of the board.
But compared with other shops in germany Farnell is in most kinds the most expensive one. In Germany Farnell do not sell to private persons so I have to find a company that do the business for me.
Btw. the MB02+IDE contains 30 chips and about 1700 soldering points.
I did not check the actual prices of PCB-Pool now. But 5 years ago when I ordered the first lot of boards for MB02+IDE I found a company that did the job a lot cheaper in good quality. I decided to get 3 more items (15 while 12 were payed) instead paying the (expensive) check.
Especially because I was not sure if there were additional design errors I had to solder the boards myself (btw. there were some small problems). I made a first prototype that missed some mechanical properties and some functions. Then I did a redesign and I did not do a further test board but ordered the whole lot I needed.
Otherwise the price would be too high.
While looking for a case and wishing more comfort I decided to design an additional board - a special CF-IDE converter with special mechanical design adapted to the design of the MB02+ and containing Reset and NMI switch as well as 2 LEDs. I gave that additional board to the interface without additional payment from the users.
The whole project needed a lot of my spare time. And I do not plan to make a complete redesign with the risk of new problems (espaecially mechanical ones).
I will ask some producers for soldering parts but I like to be sure that the price will pay all costs I have even if there will fail some parts or if I will not sale all items. I have the whole risk so I have to have the controll.
If I would order soldered boards I have to be sure to sell out all boards I make. That is not sure I think.
Greetings Ingo.