159 replies to this topic

[batari]

BillC, on Wed Feb 9, 2011 12:41 PM, said:

batari, on Wed Feb 9, 2011 5:15 AM, said:

Today I tried out a ML6428CS1 IC bought from here (they have over a thousand in stock):

http://www.onlinecom...html?p=11366096

Someone said in another thread that these were "slightly inferior" because their filters are 4th instead of 5th order, but I found the output quality to be quite good, maybe even a bit better than the FMS64xx.

FMS6406:
batarimod_white.jpg

ML6428:
IMG_4097.JPG

So, these look like a suitable replacement if we can't get the FMS chips.

I was the one who said they were inferior when I was pointing out their availability at Online Components. I wrote that because I have no personal experience with the chip, only reading the datasheet and comparing it to the one for the FMS6400, and didn't want to mislead anyone if they didn't produce an acceptable result. I'm happy that they did prove to be an acceptable replacement for the FMS6400.

I notice since my posting, #764 in the Atari A/V mod update thread, about quantities available at Online Components that all 85 of the ML6428CS1 have been sold. IIRC they were only $0.18ea. which was a bargain, I would have ordered some but the minimum purchase for international orders is $100.00.

The link batari posted is for the ML6428CS1X they do still have in stock .

I did get some of the 18 cent chips, but a deal like that couldn't last. Still, the other batch of chips is a dollar or less, and are the same chips (the "X" means it comes in a cut tape instead of a tube).

[LittleJoe]

Here's pics of output of my test build using s-video. No pots yet.

Attached Thumbnails

• 
• 
• 

[batari]

Thanks - your noise seems to be at the pixel clock frequency. Your grayscale looks clean so I'd guess the interference is coming from Chroma.

Try disconnecting the BLK line (pin 6) and see if it goes away. Also, you could try a 0.1uF cap near the 4050.

[Longhorn Engineer]

The 10uF cap near the FMS6400 chip needs to be ceramic. Electrolytic caps won't be fast enough to do the job.

[LittleJoe]

Just so we have a control. Here's screenshots of 2.2f. Same construction techniques. Wired identically. Same grounding, camera, and tv. Both rev 14 4 switch 2600's.

Attached Thumbnails

• 
• 
• 
• 
• 

Edited by LittleJoe, Wed Feb 9, 2011 9:50 PM.

[A.J. Franzman]

Your pics from post #55 look a bit better than those from #52. For example, counting black as 0, the grayscale bars 1 and 2 in post #52 are too dark. I see the vertical banding seems to have gone away in #55 too.

Edited by A.J. Franzman, Wed Feb 9, 2011 9:56 PM.

[LittleJoe]

Here's some wierdness that shows up in composite only. This is with a .1uf cap on the 4050, and with tia 6 disconnected. It's much the same with TIA 6 connected. The .1uf cap didn't help in general, and the two caps (.1uf and 10uf) on the FMS64xxx are ceramic. These 2 caps are also on the 2.2f mod as well. Disconnecting tia6 didn't eliminate the vertical lines.

Attached Thumbnails

• 

[batari]

A.J. Franzman, on Wed Feb 9, 2011 9:52 PM, said:

Your pics from post #55 look a bit better than those from #52. For example, counting black as 0, the grayscale bars 1 and 2 in post #52 are too dark. I see the vertical banding seems to have gone away in #55 too.

The brightness of the bars is irrelevant if they can be adjusted to one's liking using contrast and brightness pots.

The banding is an issue, though.

LittleJoe, on Wed Feb 9, 2011 10:06 PM, said:

Here's some wierdness that shows up in composite only. This is with a .1uf cap on the 4050, and with tia 6 disconnected. It's much the same with TIA 6 connected. The .1uf cap didn't help in general, and the two caps (.1uf and 10uf) on the FMS64xxx are ceramic. These 2 caps are also on the 2.2f mod as well. Disconnecting tia6 didn't eliminate the vertical lines.

This and the banding need to be addressed.

My banding is very minor - only visible on the orange bar and only when viewed at an angle. I just tried adding the transistor back in, and the banding disappeared. However, some red bleed appeared - it's not as bad as when COL is unbuffered but there nonetheless. Let me know how it works for you. Worst case, a second transistor on COL/BLK may balance out the delays and bleed may go away. Here is the schematic with added transistor. If you don't have pots yet, replace the left one with a wire and the right one with a small resistor (50-75 ohms or so.)

Attached Thumbnails

• 

[Longhorn Engineer]

The 2N3904 adds about a 22ns delay in the signal. Try buffering the COL signal as well.

Parker

[batari]

I spoke too soon. The vertical lines did (nearly) disappear with the transistor, but adding the transistor brightened up the picture a bit and when I adjusted the pot to dim the picture back down, the lines came back. So all the transistor does is add bleed

After some more experimenting, I did something by accident that I think revealed the problem. I accidentally disconnected the buffered COL output completely from the circuit and the color still came through. The colors were drab and fuzzy but still there. I think the lines are nothing more than ghosts of COL making its way back into the circuit. If I touched the COL line and moved my other hand over the circuit, I could actually change the intensity of the color.

So I'm taking out the transistor, and going back to the previous circuit I posted. If it's an RF problem, probably the only solution is laying out some boards with bypass caps in the proper places and with ground planes top and bottom, then order boards and hope they work.

[Nathan Strum]

batari, on Wed Feb 9, 2011 11:14 PM, said:

A.J. Franzman, on Wed Feb 9, 2011 9:52 PM, said:

Your pics from post #55 look a bit better than those from #52. For example, counting black as 0, the grayscale bars 1 and 2 in post #52 are too dark. I see the vertical banding seems to have gone away in #55 too.

The brightness of the bars is irrelevant if they can be adjusted to one's liking using contrast and brightness pots.

Ideally, any TV used for testing a mod should be calibrated to NTSC standard color bars first (there are several DVDs available you can do this with - this one is excellent, and also comes in Blu-ray).

Then the mod would be adjusted until the grayscale and colors displayed properly without making any further adjustments to the TV. Besides helping to better dial in the values of a mod design, this should also help weed out discrepancies between different 2600s.

[Nathan Strum]

Duplicate post. Ignore.

Edited by Nathan Strum, Thu Feb 10, 2011 10:41 PM.

[batari]

Nathan Strum, on Thu Feb 10, 2011 10:37 PM, said:

batari, on Wed Feb 9, 2011 11:14 PM, said:

A.J. Franzman, on Wed Feb 9, 2011 9:52 PM, said:

Your pics from post #55 look a bit better than those from #52. For example, counting black as 0, the grayscale bars 1 and 2 in post #52 are too dark. I see the vertical banding seems to have gone away in #55 too.

The brightness of the bars is irrelevant if they can be adjusted to one's liking using contrast and brightness pots.

Ideally, any TV used for testing a mod should be calibrated to NTSC standard color bars first (there are several DVDs available you can do this with - this one is excellent, and also comes in Blu-ray).

Then the mod would be adjusted until the grayscale and colors displayed properly without making any further adjustments to the TV. Besides helping to better dial in the values of a mod design, this should also help weed out discrepancies between different 2600s.

For a moment, I thought this DVD was this:

[Nathan Strum]

Nope, and stupidly - they didn't include that short on it. Really, would it have killed them?

(But it's still a really good calibration disc.)

[batari]

If the problem is RF, there's probably not much I can do to completely eliminate it with a hand-built circuit. So, I've laid out a board, which has bypass caps very close to the two ICs, and the 10uF cap within 0.75 in as per the FMS64xx datasheet. It has ground planes top and bottom to minimize RF, and vias are tented to ease soldering. The jumper is a solder jumper - close it if you are using the FMS6400 or ML6428, and leave it open if you're using the FMS6406 or FMS6410.




I'm going to order some but I'm using batchpcb.com which will probably take about a month, so I'll set this project aside until then. I will attach the Gerber files in case anyone else wants to try these out, but if you do make sure you get solder mask top and bottom or it will be hard to solder. Note that this board is at the alpha stage and may not work at all, though I think I didn't make any mistakes.

Gerbers:
 vidmod0.1a.zip   41.96K   49 downloads

[wood_jl]

Although most of this thread is beyond my comprehension, I pick up little bits and pieces, perhaps enough to realize this is beyond my comprehension! But it is truly fascinating (for the layman) to watch you electronics wizards at work. Wish I could offer something, other than encouragement and applause!

[LittleJoe]

batari your not seeing any problems with composite? I've got good s-video, but composite needs help.

[batari]

wood_jl, on Fri Feb 11, 2011 9:42 PM, said:

Although most of this thread is beyond my comprehension, I pick up little bits and pieces, perhaps enough to realize this is beyond my comprehension! But it is truly fascinating (for the layman) to watch you electronics wizards at work. Wish I could offer something, other than encouragement and applause!

Don't be fooled, I'm actually a bit out of my league with this analog stuff.

LittleJoe, on Fri Feb 11, 2011 9:50 PM, said:

batari your not seeing any problems with composite? I've got good s-video, but composite needs help.

I am seeing vertical lines and a bit of noise like you showed in your pics, but not as bad. I still think it's RF, and hopefully the board will fix it. I suspect S-video is more tolerant of noise, which I think is due to the separation of the signals, where the Chroma line can't corrupt Luma like it does when they are summed inside the chip.

[horseboy]

I found a source with plenty of FMS6400s. I don't know if this changes anything for you guys.

[batari]

horseboy, on Mon Feb 21, 2011 10:14 AM, said:

I found a source with plenty of FMS6400s. I don't know if this changes anything for you guys.

Nope. It's always best to have more options.

[horseboy]

batari, on Mon Feb 21, 2011 12:40 PM, said:

horseboy, on Mon Feb 21, 2011 10:14 AM, said:

I found a source with plenty of FMS6400s. I don't know if this changes anything for you guys.

Nope. It's always best to have more options.

Sounds good to me. I can't wait to see a finished product.

[0100010]

In comparing the color bar images between LittleJoe's post 52 and 55 // it would appear they both have vertical banding. Just that the LHE mod exhibits banding that is half as wide (or double the rate?), as compared to the banding you can see from the new mod.

Just my observation.

In any case, I wish you all the best of luck. I'm just waiting for something to become available so I can buy one.

[akaviolence]

batari, on Fri Feb 11, 2011 10:03 PM, said:

I still think it's RF, and hopefully the board will fix it.

Any news? Did you get a board fabbed yet? Did it help with the interference?
Also, i'm having trouble finding a small 1K pot and wondered if you had homed in on a good enough final value for each pot so I can eliminate them from my board design.

[batari]

akaviolence, on Fri Mar 11, 2011 12:51 PM, said:

batari, on Fri Feb 11, 2011 10:03 PM, said:

I still think it's RF, and hopefully the board will fix it.

Any news? Did you get a board fabbed yet? Did it help with the interference?
Also, i'm having trouble finding a small 1K pot and wondered if you had homed in on a good enough final value for each pot so I can eliminate them from my board design.

I got an email yesterday that the boards had shipped. I expect they will be here early next week. And now that you mention it, I forgot to pick up some of the parts I need, including pots So check back Monday or Tuesday and I may have an answer.

[akaviolence]

Did you use batchpcb? When did you put the order in? Just curious about turn around times.