the official Channel F thread!

[tnt23]

Here are shots of BURST and COLOR signals (cyan channel 2) sampled at pins 9 and 7 of 74153. Yellow channel 1 is Composite output.

CHROMA + zoom:

 

 

Chroma looks good. I believe different shapes of chroma carrier pulse trains are due to phase shift for Red and Green vertical lines in the beginning of the scan line.

 

BURST signal + zoom:

 

 

Now there's something strange about BURST. I would expect the burst to be active only in the beginning of scan line, on back porch of HSync. Yet it seems that burst is also active on visible part of the line. Can it be that these extra pulses (with 0 phase shift) get mixed to CHROMA ones (with non-zero phase shifts for different colors) resulting in that odd color distortion? Exception is Blue, third vertical line of the test.

 

I have also removed the RF modulator shielding and was able to locate and disconnect L2. Not much improved on color picture, but BW-wise extra shades of grey to the left and to the right of dark vertical lines are almost gone.

 

 

burst.bmp

[e5frog]

I was under the impression the burst signal was active the 9 cycles in the back porch and then everywhere on the line where there's color... 

 

The second pair of images look odd, the third signal not having a corresponding burst train. 

It's not impossible signals reach the end at different times, messing the image up but that one looks like something is missing. 

[tnt23]

You are right, chroma indeed shall be everywhere on the line. I got some comments from a friend, it appears that BURST and CHROMA in fact are I and Q components from NTSC theory, and get mixed or combined down the line.

Will try replacing the 74153 mux however.

 

BTW this is the picture without Chroma (L7 lifted). I'd say everything is just about perfect on Luma side of things.

Edited July 22, 2019 by tnt23

[tnt23]

(sorry for extra noise in the thread) Another combination of shots showing CA0 and CA1 lines with respect to output composite signal (yellow).

 

 

I don't quite like the CA1 to be honest.

[e5frog]

That looks bad, is it supposed to have pulses or is it supposed to be zero?

[tnt23]

Judging by schematic, CA1 should be 1 for Red, just like CA0 for Green and Blue.

 

CA0..CA2 are produced by 9322 mux (also marked as 74157 on Luxor schematic). CA1 is either FG1 or BG1 depending on FG ENB signal, and the latter is active if FG0,FG1 pair is non-zero.

[e5frog]

Looks quite glitchy anyway, the 9xxx is Fairchild's own branded IC:s IIRC. 

[tnt23]

13 hours ago, e5frog said:

Looks quite glitchy anyway, the 9xxx is Fairchild's own branded IC:s IIRC. 

Yes, on my PCB 74157 is soldered, not 9322.

 

As far as I understand, test points marked A5-xx and C6-xx relate to empty sockets on the PCB. Handy if you want to check certain signals, but also can be useful for debugging.

For example, TP A5-14 can be grounded, and should select 8V and 16V clocks to define BG0 and BG1 colors instead of those coming from VRAM. (Did not work in my case, don't know if it was me or 74LS298)

 

Interesting to see I1C and I1D inputs routed to VRAM but QC and QD outputs not used. Some rudiments for more colors perhaps?

 

Similarly, TP A5-14 can be grounded to disable pixel data coming from VRAM, forcing all zeroes.

 

TP A5-15 disables Burst Gate flip-flop D4 (and probably ruins video sync), which also disables palette data storing in C4, and so on.

[e5frog]

If you should come up with anything there are unused ports that could be used to control things. 

Port 0 has bit 4 and 7 not connected and when not reading the controller there's also bit 7 on port 4.

 

What about the 7445/74145 with all the colors marked out, nothing connected to bit 0,1 and 4.

[tnt23]

On 7/23/2019 at 9:22 AM, tnt23 said:

TP A5-14 can be grounded, and should select 8V and 16V clocks to define BG0 and BG1 colors

Did a better grounding, and indeed it results in wide horizontal color bars.

21 hours ago, e5frog said:

Port 0 has bit 4 and 7 not connected and when not reading the controller there's also bit 7 on port 4

Cool for hacking, don't know what I could use these for

 

Regarding empty bits 0, 1, and 4 on 74145, I guess they default red background to black maybe?

[e5frog]

You could control something from software with the free bits, either in or out - or both.  

[tnt23]

So I decided to try S-Video: lifted C28 and fed it via spare THS7316 channel to pin 4 (C), and connected Composite output to pin 3 (Y):

 

 

Blue looks absolutely gorgeous. Red calls for help, and white background is decorated with some green hints.

 

 

Vertical lines test is also great on Blue. I think that's because the top half of BURST mux E3 outputs zero when Blue is to be displayed:

 

While for Red and Green (and black) 3.58MHz subcarrier is present on BURST line. Still I don't understand this.

[e5frog]

Looks great apart from the stripes. 
So when E3 has 0 out it's fine but the others cause interference? I guess if interference is green and red there would be none when they're off... 

I had some ideas about s-video on my first version Luxor, couldn't get more than an inverted black/white rolling an image with the amplifier that works great with Channel F II, SABA Videoplay 2 as well as two first version Fairchild NTSC boards (though they had a lot of color bleed - on the TV I used anyway). 

S-video would be the best, if you don't have s-video input it's easy to get composite with an adapter or make one yourself with a small capacitor. 
When I installed the Longhorn s-video mod in my Vader (A2600), there was a composite video out option as well but when using an adapter on the S-video output the image was better. 

 

BTW have you found any errors in the schematic, sometimes changes are made and schematics not updated.

[+atari2600land]

Just got backgammon in the mail today. It's a great version of it. My grandpa taught me how to play backgammon, so I knew what to do. If only I could play against the computer. Or if Grandpa was still alive. We still have his actual backgammon set. It's in the den closet. And with that game now added, I need a bigger box to store my games in.

[tnt23]

On 7/25/2019 at 8:25 PM, e5frog said:

S-video would be the best, if you don't have s-video input it's easy to get composite with an adapter or make one yourself with a small capacitor. 

Problem is, I cannot get decent composite in the first place. Splitting to Chroma and Luma did not help much.

 

On 7/25/2019 at 8:25 PM, e5frog said:

BTW have you found any errors in the schematic, sometimes changes are made and schematics not updated.

No I haven't. so far everything I've been checking did comply with the schematic.

 

By the way, the +5V bridge rectifier was getting rather hot, and at some point +5V had just gone. I have desoldered both the bridge and the 7805 for replacement.

[e5frog]

A friend has a similar problem with hot bridge and rectifier. He has swapped diodes and rectifier without success. Must be a short somewhere... 

Not much to do but to start checking IC:s (desolder). Measure current and unplug the socketed chips first I guess. 

Caps can short, little brown disc type as well, even if it's unusual. 

[tnt23]

10 hours ago, e5frog said:

A friend has a similar problem with hot bridge and rectifier. He has swapped diodes and rectifier without success. Must be a short somewhere...

Could be, still would replace the bridge as it was getting too hot. I've found compatible BR310 replacement, 3A continuous, will give it a go.

10 hours ago, e5frog said:

Caps can short, little brown disc type as well, even if it's unusual.

This is what I'm really tempted to do, as +5V was rather messy on the scope. Besides, those brown disc capacitors are used here and there in video generating circuitry, so worth checking, too.

[ubersaurus]

This evening I had the pleasure to interview Vilas Munshi, a former Fairchild programmer who worked on the Channel F from 1976-1978. Some highlights:

 

While there, Munshi said he worked on the Hockey game (converting Lawrence Haskel's prototype program to the F8 processor), Desert Fox, and the two Math Quiz titles. He spent a good chunk of time working on an idea of 7-8 unique games on one cartridge, but the idea never quite came together. Incidentally, he said he had tagged along with Jerry Lawson to Connecticut during the initial phases of the Channel F's conversion from an Alpex prototype to Fairchild product.

 

Lawson was his direct supervisor, and Munshi says he was great to work with - added that Lawson allowed them to play the various competitor game systems. Said it was very much a learning and experimental environment, which he enjoyed. Noted the Channel F was difficult to work on since there was no high level programming language, and the carts tended to be around 1k of memory. Fairchild had its own internal computer systems that they did all the development on.

 

He was laid off in 1978 during one of Fairchild's down periods, and ended up going to Atari to work on calculation/tax software for the 8-bit computer line. He stayed there until 1982, when he left to IBM.

 

All in all it was an illuminating conversation! I may circle back around to him if some more questions emerge... and I certainly hope to use what I have now to produce a mini-doc about the Channel F in the near future, especially if I can get in touch with some other former staff.

[e5frog]

Cool, any documentation left? 

Brad Reid might be possible to locate, I was in contact with him a few years ago after mailing all Brad Reid in the US I could find an email address for - and it worked. 

[ubersaurus]

36 minutes ago, e5frog said:

Cool, any documentation left? 

Brad Reid might be possible to locate, I was in contact with him a few years ago after mailing all Brad Reid in the US I could find an email address for - and it worked. 

Unfortunately no, he said he doesn’t have anything left from his Fairchild days - just some of his hardware from Atari.

 

I think I’ve tracked down Don Ruffcorn and Reid-Selth, and sent them letters. I might follow that up with a call soon though. No luck finding Michael Glass or Rick Maurer (if he’s still alive) but I’m trying my best.

[Ashevent]

@ubersaurus that’s awesome! Thanks for sharing and good luck on those other fellas getting back to you! I’ve been enjoying this entire thread just didn’t have a lot to add but excited about all this FCF info. 

[tnt23]

On 7/19/2019 at 7:19 PM, e5frog said:

First soldered version. 
Had to buy a couple of other components than planned due to not selecting a large and expensive company... 
So 4.7µF instead of 1µF on the video input (not sure about that one), 22µF on the audio (which is fine but it barely fit the footprint), sandwich of two 150 Ohm resistors (parallel) instead of a 75 Ohm and I used a potentiometer I already had.

Did you get a chance to try it? I've built this circuit to try something different while still getting frustrating results with THS7316 amplifier. It did not get me any better results though.

I will be trying a couple of other ideas, one is to make a copy of delay circuit that shifts 3.58MHz carrier phase for Green and Red:

Just to check if everything is fine on this path.

 

Warning: bit of tech whinning on THS7316 inside!
 

Spoiler

By the way, THS7316 seems to be great for modding most retro stuff out there, provided the original signals are of decent quality. It could well be that my particular Channel F is rather noisy especially on chroma, and THS7316 is an HDTV video amp. Looking at its specs, there's 6dB gain (great) and ability to drive 75-Ohm loads (cool), and it also has more than 30MHz bandwidth. Excellent for HDTV, but a bit too much for good old SDTV with its bandwidth of under 6.5MHz I believe. THS7316 also has a nice "5-th order 36-MHz (–3 dB) Butterworth Filter" which once again is great for HDTV and fast DAC sampling clocks:

 

 

In Channel F, color signal is not pure sine wave, but rather a 2-level something mixed of two phase-shifted digital signals (I still don't get the idea but obviously engineers from the past did that for reason). One would expect a bunch of 3.58MHz harmonics to be present in output signal spectre, like 7MHz, 10MHz etc, as well as all sorts of additive or subtractive or whatnot frequencies. These most likely shall be filtered out, and I believe HDTV amps are just useless here.

 

I have then googled a bit for SDTV and NTSC and filtering, and got some quick reading on 'reconstruction filters'. Then I have found this little gem, NCS2553. Said to be "a 3−channel high speed video amplifier with 6th order butterworth standard definition reconstruction filter", it seems to be a better fit for TV outputs modding:

 

 

Looks like it should do a great job of filtering out anything over 6MHz. I have ordered a handful of these (4GBP for 10pcs at RS Components), and will give them a try.

 

 

And as an absolute last resort, I will try making RGB output by means of 256x4 EPROM (82s126 or similar) and a bunch of emitter followers. EPROM is there to translate BG0, BG1, FG0, FG1 signals directly to RGB and (I)ntensity to make light hues of Green and Blue and Grey.

Edited July 31, 2019 by tnt23

[e5frog]

Yes, it worked very well on the Channel F II, but not on the Luxor V.E.S. Just got a B/W and rolling image there. Will try bypassing chroma and add it in after. 

[Mikebloke]

On 7/26/2019 at 1:22 AM, atari2600land said:

Just got backgammon in the mail today. It's a great version of it. My grandpa taught me how to play backgammon, so I knew what to do. If only I could play against the computer. Or if Grandpa was still alive. We still have his actual backgammon set. It's in the den closet. And with that game now added, I need a bigger box to store my games in.

Backgammon has been one of those games I just never understood. I actually have it on a number of consoles now, including the Videopac and the CD-i, but which one was it that actually helped me learn the rules? Yep, the Fairchild Channel F version! I don't know why but my mind was always so fixed on the triangles of the board and I could never understand the significance. The fairchild version it is simply lines, and ironically that was all I needed to understand it.

 

I recently started reading some articles on Jerry Lawson so I'd be interested in anything else Fairchild Channel F related. Was there ever much information on Zircon when they took over sales and technical issues? I find that tends to be a bit of a dead zone of information too (I've seen some of the promotional material during the transition, but that's it).

[ubersaurus]

So as far as I am aware Fairchild liquidated the video game stuff to Zircon sometime in 1979 - I haven't found much contemporary reporting on it in the industry trades, though.