F18A VDP

[Megatron1999]

Hi.

 

Does anyone know if there's a Single Chip version of the F18A or something similar in development as a Drop in replacement to the TMS9918. The idea of a single chip with all logic RAM etc all on one chip is an interesting idea and must have some commercial possibilities/success if developed. The idea of a 'NEW' improved single chip TMS9918 wouldn't just appeal to old Retro computer Geeks but the masses of people using microcontroller needing/wanting a VDP to generate Text & Graphics.

[ti99_forever]

What makes it NOT a drop-in replacement?

I'm confused. It replaces the original chip, and it's RAM is onboard.

[HackMac]

Marvelous: Two threads with the same content...

I posted my answer on the other thread.

[+Lee Stewart]

What makes it NOT a drop-in replacement?
I'm confused. It replaces the original chip, and it's RAM is onboard.

 

Realizing your idea is not possible, because this chip which does the TMS9918 emulation is dump. It needs it's extra memory (an extra chip) to load the configuration. It could be possible to use an Xilinx Spartan-3AN instead of the Spartan-3E for example, which has integrated memory, but nevertheless you need extra chips for adjusting the 5 Volt TTL levels to low voltage (3.3 or lower) TTL/CMOS, or backwards, to communicate with the vintage TI-Chips. Additionally the FPGA has different voltages (3.3, 2.5 and 1.2 Volts) and that's why you need extra voltage regulators and circuits for them. And the only Spartan-3AN with onboard flash memory you can solder by hand is the XC3S50AN. But it seems that this device has not enough capacity to load the configuration for the 9918 emulation. So in any case an extra flash chip is needed.

 

The F18A is, in fact, a drop-in replacement of the TMS9918A, with the added benefits of on-board VRAM and VGA output with 80-column display in text mode.

 

...lee

[HackMac]

I think, what Megatron1999 tries to ask is, if it is possible to replace the whole circuit of the F18A module with a one chip solution. That is the way I understood him.

Edited October 12, 2015 by HackMac

[matthew180]

Does anyone know if there's a Single Chip version of the F18A or something similar in development as a Drop in replacement to the TMS9918.

The F18A *is* a drop-in replacement for the 9918A. Albeit it is about 1.5x wider than the original IC, but it is an electrical and functional direct pin replacement.

 

The idea of a single chip with all logic RAM etc all on one chip is an interesting idea

The only way to make a replacement that is physically "one chip" would be if you are an IC fabrication company, i.e. TI, Intel, etc. Personally I don't believe silicon IC fabrication will be available to a hobbyist any time soon.

 

and must have some commercial possibilities/success if developed.

In my experience this is not the case. Retro computer hobbyists always feel there is this huge market for updated retro hardware and such, especially for ideas they are currently having. I suppose it depends on how you define "commercial possibilities/success", but in most definitions I think you would expect to find something like "make a profit", which is not something you can do focusing on retro hardware.

 

The idea of a 'NEW' improved single chip TMS9918 wouldn't just appeal to old Retro computer Geeks but the masses of people using microcontroller needing/wanting a VDP to generate Text & Graphics.

These ideas only appeal to a very small niche group of people. People using microcontrollers en masse don't have a need for video, they are generally not making retro computers or video game consoles (some do, *most* don't). Serial I/O or a few LEDs are usually all that is needed. For those who do need/want microcontroller video there are plenty of options (of which the F18A would be equally well suited.) People also tend to trip over the ubiquitous availability of modern video output from computers and can't imagine anything less than 1280x960 HDMI 24bit color output (like the Raspberry Pi, etc.) for their 8-bit microcontroller project.

[Megatron1999]

The point I was trying to make is :

Wouldn't it be great to have a NEW TMS9918 Chip (everything in on one piece of silicon) based on the F18A. I'm not knocking the F18A as it's very good but due to small production runs etc this limits it's use.

I was wondering if anyone had even considered approaching a commercial company to see if they'd be willing to MASS produce an F18A type Chip, as I'm sure there must be a reasonable demand ?

Well just a thought !

[Enigma]

I suppose there's several ways to do this..

 

1) Approach several commercial companies that do this sort of thing and see if they are interested ? Though they'll be looking at things purely from a financial MONEY point of view. There's also the matter of Intellectual Property right which are probably still owned by Texas Instruments, so they'd want a cut !

 

2) Once you've got a working prototype (even one just on an Altera DE Board) then you could do Crowd Funding to generate enough Money to get a limited production run.. But who would back it ?

 

3) If you've got the Guts and the Money just do it yourself .. get several quotes from companies doing this sort of thing and do a production run yourself... though this will be VERY expensive !

[Asmusr]

Where would you put the VGA connector on a single chip?

[HackMac]

The TMS9918A also does't have an monitor connector...

[TheMole]

The TMS9918A also does't have an monitor connector...

 

Yes, but by definition you would not be able to make a pin-compatible tms9918a clone with VGA output, you need'd to change the pinout. Even with the 9928 you'll still need two additional pins for the h- and v-sync signals. This means it can never be a drop-in replacement for existing systems and thus would only be useful for anyone looking to build new devices, which would raise the question on what the benefit of 9918 compatibility would be?

Edited October 13, 2015 by TheMole

[Asmusr]

[HackMac]

Ok, it isn't the connector itself, but the signals, what he means.

 

But who wants VGA signals? An ordinary RGB signal with sync on green is enough. By using SDI we can transmit video signals digitally.

Or we keep the existing signals.

[bad robot]

Just make two versions.. one with PAL/NTSC output (analogue TV 50./60Hz) and a VGA monitor chip..each chip having a different designation.. i.e. TMX9918TV (PAL/NTSC TV) & TMX9928VGA (VGA Monitor ) ? There where several version of the Original Chip TMS9918,TMS9928,TMS9929 etc.. But I suppose the main reason to do this (make a one Chip solution) would be cost to get each chip down to below $10 ! Mass producing these would drastically cut the cost and open them up to all sorts of uses both industrial and hobby. So in principal it's a good idea ... but actually doing it would be somewhat more difficult !

[Omega-TI]

The F18A has been around a while and we are just now getting software written specifically for it and it's extended features. I'd hate for a third-party to come in, now, steal the code that Matthew developed with input of other TI'ers, just so they could market a slightly smaller or competing device. I honestly fail to see the point.

 

If they had their own code, I doubt it would be 100% compatible with the F18A, and that would create other issues I would not want to deal with either. In the past there was strife over another old rare display method which caused all kinds of grief. That kind of war going on over an item about half the people currently use might make this forum a place many might not want to be. I believe a new device now could only hurt the TI-Community by fracturing and polarizing people and getting in the way of software development and people enjoyment of the HOBBY. The F18A has become the modern, obtainable and working standard, why screw with a good thing? The thing is simple to use and it works... works great.

 

Why not VGA? It's a practical and clean standard that has been around nearly as long as the TI and this makes it compatible with an obscene number of cheap and easy to obtain monitors without the need for additional converters.

 

From a personal perspective Matthew put in his personal time, effort and money to develop and bring us this great little gadget. I'll be dammed if I'd support some unknown entity that comes in and steals his code just to produce a cheap knock-off.

[TheMole]

The F18A has been around a while and we are just now getting software written specifically for it and it's extended features. I'd hate for a third-party to come in, now, steal the code that Matthew developed with input of other TI'ers, just so they could market a slightly smaller or competing device.

 

I think you're jumping to conclusions here Omega, I don't think anyone is talking about stealing Matthew's work here...

 

On the topic of what it would cost to create a custom ASIC, there's this calculator here that should give some insight: http://asic-cost-calculator.sigenics.com/

I'm not a specialist on this, but after playing around with the config it seems to me that you're looking at an NRE of at least $250k. That gives you a chip that will cost (in a 0 profit type situation) around $20-$25 to manufacture. So if you would somehow be able to guarantee that you can sell at least 10000 units you might be able to sell chips at around $70 (factoring in transport/administrative costs and a small profit margin), which is 8$ cheaper than the current F18A... not sure if it's worth the effort (not to mention that 10k units is ridiculously ambitious...). Even if you sell at cost, you'd still have to charge around $50. Definitely not worth it, imho.

[Omega-TI]

 

I think you're jumping to conclusions here Omega, I don't think anyone is talking about stealing Matthew's work here...

 

You're probably right. Producing 100% compatible code from scratch would add some extra time and money into the equation for sure.

 

 

Definitely not worth it, imho.

 

Total agreement.

[+OLD CS1]

Always "just."

[matthew180]

The point I was trying to make is :

 

Wouldn't it be great to have a NEW TMS9918 Chip (everything in on one piece of silicon) based on the F18A. I'm not knocking the F18A as it's very good but due to small production runs etc this limits it's use.

That's what is sounded like from your first post, but like ti99_forever said, I'm confused about what you gain by having a single-IC package? What are you trying to gain?

 

From a usability standpoint what does it matter if the device your are using is a circuit board with multiple components or a single IC? Your interface with the "component" is the same. Many DIY companies exist today based on exactly this idea, Adafruit, SparkFun, Arduino, etc. making break-out boards and wrapping up microcontrollers and other ICs into easier-to-use packages.

 

I agree that the F18A form-factor is not a 600mil DIP. That would have been great! That was one-of-three initial design goals that I could not manage the first time around and it still bothers me. But I do not see it as a limiting factor in someone using the F18A in any design where they want 9918A-like video.

 

As for the small F18A production runs, they match the demand. The F18A has always been available since the summer of 2012, with the last few months (summer 2015) being the largest period of unavailability where I was having more boards made (even in that time the back-log of people waiting for boards is only about twelve.) It took about *three years* to sell the first batch of 250 boards... Not very good numbers if you are trying to build in bulk to keep manufacturing costs down, which is the main reason these retro-type projects are always going to be more expensive than things like Arduino and Raspberry Pi, etc. If anyone wants to front the cash to build 1000+ F18A boards, let me know!

 

 

...

So if you would somehow be able to guarantee that you can sell at least 10000 units you might be able to sell chips at around $70 (factoring in transport/administrative costs and a small profit margin), which is 8$ cheaper than the current F18A... not sure if it's worth the effort (not to mention that 10k units is ridiculously ambitious...). Even if you sell at cost, you'd still have to charge around $50. Definitely not worth it, imho.

Agreed, very ridiculously ambitious and there is no way you could make that guarantee.

 

If I could make 10,000 F18A units then I could easily get the costs down to about $20..$30 per unit out the door. But so far the global F18A market has proven to be about 262 units... Even if I somehow managed to capture part of the Arduino community, the number of projects that need full-blown video is a fraction of what people are using microcontrollers for.

 

There are already two video boards for Arduino, the Gameduino and Gameduino2, both cheaper, more capable than the F18A, and available via SparkFun, AdaFruit, and SeeedStudio. However, what I do NOT see are tons of web sites where people are showing off their Gameduino-based projects. Video is hard. And making a system that produces video, i.e. writing all the code, a game, etc. is hard. Unless people have a serious personal interest building such a thing, it simply does not happen.

 

This is why I get frustrated and cynical about people complaining about the costs of retro-hardware projects. We would all love it if there was enough of a market to subsidize our retro projects and make them cheaper, worth-while, etc. I would love to build retro-hardware full-time as my day job, but I would be homeless and starving. That era is over, that's why we call it "retro". No matter how much we love our retro-computers and feel everyone should be equally enamored with our technology of choice, the fact is that the world in general does not care and all retro-computer communities are small and fractured.

 

IMO everyone needs to just come to the realization that a retro-computer hobby is expensive, the communities and markets are *very* small, support the hardware and software projects that people are spending their personal time on (usually sacrificing time with family and friends), and just be happy and enjoy their hobby of *choice*.