Were MSX1 designed with too hard cost constraints?

By mcolom

Expert (125)

mcolom의 아바타

17-09-2020, 23:06

I have the impression that, at least MSX1, were produced under perhaps too hard cost constraints.
For example, tape modulation perhaps could have been improved by detecting changes in frequency (or other better encoding than simple not counting the time the signal is up or low, which is prone to error), but that would have meant to add more circuitry to the boards.
The same for the fact that the sub-slot selection is done by writing to address 0xFFFF instead of using I/O ports, probably because that moves the cost of having extra circuitry to the connected device.
And some MSX1 were fried (my poor VDP, for example, then replaced) by inserting a cartridge when the computer was ON, when it seems that a simple detection mechanism could have been implemented to avoid damage.
And probably there are more examples. And I don't know much about MSX2, that's an odd modernity for me Smile
What do you think? Did they try to save up to the last cent, or those were fine design decisions at the time?

Login or 등록 to post comments

By ducasp

Champion (371)

ducasp의 아바타

17-09-2020, 23:52

mcolom wrote:

I have the impression that, at least MSX1, were produced under perhaps too hard cost constraints.
For example, tape modulation perhaps could have been improved by detecting changes in frequency (or other better encoding than simple not counting the time the signal is up or low, which is prone to error), but that would have meant to add more circuitry to the boards.
The same for the fact that the sub-slot selection is done by writing to address 0xFFFF instead of using I/O ports, probably because that moves the cost of having extra circuitry to the connected device.
And some MSX1 were fried (my poor VDP, for example, but then repaired) by inserting a cartridge when the computer was ON, when it seems that a simple detection mechanism could have been implemented to avoid damage.
And probably there are more examples. And I don't know much about MSX2, that's an odd modernity for me Smile
What do you think? Did they try to save up to the last cent, or those were fine design decisions at the time?

I don't think that I/O for subslot selection would be more expensive, after all, you are decoding 16 bits of memory address and probably the slot select signal as well, and also having to invert whatever you received to set the register to return as a response... I can think of several means of doing it cheaper (not sure if as good as), and having it memory mapped on an area that is not to be used by any application anyway makes it a lot easier to access it and check if extended slots are available or not...

I'll not comment much on the tape interface because my analog knowledge sucks, but if seeing the technology in use for data transmission at that time with modems up to 1200bps, I just think that it was what the technology at that point allowed with reasonable costs, I don't recall any computer in the similar time frame MSX was launched that used something much more different, but again, it is analog and I suck at it, so there is a high chance I'm wrong...

As far as cartridge slots goes, I think this is up to each manufacturer, several consoles do not have any protection whatsoever, and the MSX I used first, Expert by Gradiente, would turn the computer off if the slot lid was open with the computer on and would turn on again only the lid would be back a little after inserting the cartridge... Also, the standard had SW1 and SW2 in the slot, and one of it being shorter in the cartridge is meant just for that, again, Expert by Gradient if I'm not mistaken would not feed any power to the cartridge unless SW1 and SW2 were short circuited. Yeah, not all manufacturers used that, but SW1 and SW2 were there for that.

I don't think that there is any evidence that they tried to save up to the last cent at the expense of what they viewed as important. The system contained several means to be expanded, the Slots system, everything was well designed and founded a really great foundation. If they have cut costs, I would say costs were cut when choosing the VDP, probably better options (and way more expensive ones) were available at that time, but also I tend to think they've chosen the 9918 just because there were many systems with it, good bang for the buck at that time...

At least this is my thinking... It is by no means an expensive and luxury system, but for its proposal of affordable home computing, I think they have put some thought on ideas that do not feel cheap at all...

By NYYRIKKI

Enlighted (5595)

NYYRIKKI의 아바타

18-09-2020, 02:02

I also think your theory about cost constraints is a bit off...

Better tape modulation does not need more hardware but better algorithm. They selected Kansas city standard FSK because that was a common agreed standard and standard was exactly the thing they wanted to come up with... so this fitted well.

I do agree that they could have added something like Zilog CTC to handle tape and act as a programmable timer and that would have improved things and raised a cost a bit... and there are also lots of other things that could have happened differently, but didn't. How ever I do not get a feeling that important features would have been butchered because of cost. I feel MSX is quite balanced entirety when comparing to other computers of that era.

That sub-slot selection mechanism is not simple or cheap, so I can't imagine that it has anything to do with cost reduction. To me it looks more like this feature was just something that management wanted to cram in to the plans on a last minute and it then ended up to be bit of a headache for both hardware and software side of the project... but this is just my imagination.

What comes to the "saving up to the last cent", I would say that it very much depends of the machine you are looking at. If you are looking Casio or Spectravideo then yes... They probably used quite a bit of time to find how to save those last cents... On the other hand, if you look ie. some Sony or Daewoo models, the impression you get is quite different. You see lots of steel, connectors, extra features, expansion possibilities etc. that do not back up this money saving theory.

By mcolom

Expert (125)

mcolom의 아바타

21-09-2020, 12:07

ducasp wrote:

I don't think that I/O for subslot selection would be more expensive, after all, you are decoding 16 bits of memory address and probably the slot select signal as well, and also having to invert whatever you received to set the register to return as a response... I can think of several means of doing it cheaper (not sure if as good as), and having it memory mapped on an area that is not to be used by any application anyway makes it a lot easier to access it and check if extended slots are available or not...

Yes, it could be that in the end it was not an economical decision, but also a move to try to enfore the adoption of this procedure by the programmers. After all, there were a lot of big companies behind and probably they thought they could force the adoption. But game programmers didn't always follow the rules and applied "duck typing": if it looks like a normal address as in the other systems, why shouldn't it be?

ducasp wrote:

I'll not comment much on the tape interface because my analog knowledge sucks, but if seeing the technology in use for data transmission at that time with modems up to 1200bps, I just think that it was what the technology at that point allowed with reasonable costs, I don't recall any computer in the similar time frame MSX was launched that used something much more different, but again, it is analog and I suck at it, so there is a high chance I'm wrong...

Well, in the 80s we had already V.22 models, which used PSK modulation. Here I really think it was an economical choice. since I guess PSK would be really more robust., and at the same time requiring a more expensive decoding hardware.
But it's true that most of the microcomputers at the time used a similar tape modulation.

ducasp wrote:

At least this is my thinking... It is by no means an expensive and luxury system, but for its proposal of affordable home computing, I think they have put some thought on ideas that do not feel cheap at all...

I think that (at least for MSX1) it was a combination of trying to have something cheap (ok, perhaps not that much) copying the modulation of the microcomputers at the time, but with newer ideas (a separate VDP, subslot expansions, etc...)

NYYRIKKI wrote:

I also think your theory about cost constraints is a bit off...

Well, it's not even a theory, it's just some arbitrary thoughts Smile

NYYRIKKI wrote:

Better tape modulation does not need more hardware but better algorithm. They selected Kansas city standard FSK because that was a common agreed standard and standard was exactly the thing they wanted to come up with... so this fitted well.

I think it was chosen because that's what others had. But it's done simply by counting the time the signal is up or down. Yes, it works, but it's prone to errors. From the point of view of the software, using a better hardware decoder shoudn't be more complex. It's just reading which symbol the hardware has decoded. Seems simpler that what the BIOS does, trying to compensate jitter in the tape, and things like that.

NYYRIKKI wrote:

That sub-slot selection mechanism is not simple or cheap, so I can't imagine that it has anything to do with cost reduction. To me it looks more like this feature was just something that management wanted to cram in to the plans on a last minute and it then ended up to be bit of a headache for both hardware and software side of the project... but this is just my imagination.

That's interesting. Perhaps they tried to enforce adoption, or perhaps they had some kind of deadline for define the standard. Perhaps someday we'll know.

By the way (off-topic), NYYRIKKI, I use almost everyday your code in SofaROM. It's great! Smile

By gdx

Prophet (3761)

gdx의 아바타

21-09-2020, 15:04

About sub-slots I think they just wanted to economize an I/O port.

By djh1697

Paragon (1485)

djh1697의 아바타

21-09-2020, 15:47

mcolom wrote:

And some MSX1 were fried (my poor VDP, for example, then replaced) by inserting a cartridge when the computer was ON, when it seems that a simple detection mechanism could have been implemented to avoid damage

My Goldstar FC200 had a microswitch in it, so that power is switched off as soon as a cartridge is inserted, i would change ROM's without worries, although I did it with a disk drive and it damaged the computer. A good idea? Maybe!

By gdx

Prophet (3761)

gdx의 아바타

21-09-2020, 16:18

The micro-switch doesn't disconnect the power of the drive. The best is to use components that can withstand electric shocks.