1088XEL Atari ITX Motherboard DIY Builders Thread

[+mytek]

Personally I wouldn't be all that concerned about getting each and every component lead to have solder flowed all the way to the top. These are after all 'plated thru' holes. What is important, is to not have cold solder joints, or poor flow.

 

This is how it should look on the all important bottom side of the PCB.

 

 

[+DrVenkman]

1 hour ago, Bikerbob said:

I am using Silver Solder 60/40 and did not use flux on both sides.. hmm never thought of that. OK.. thanks guys.

 

James

I use 60/40 rosin core leaded solder. No need to any extra flux.

[flashjazzcat]

The only other thing I can add (apologies if it is obvious: I do not mean to insult or patronise) is: heat the plate-thru and the component leg, not just the latter. Iron tip should be in the crux of the two.

[Bikerbob]

Well.. I offically do not like disk caps.. phew.. they are done!

 

Getting there.. off to do stuff outside.. can't be a celler dweller all the time.

 

James

[+MacRorie]

4 minutes ago, Bikerbob said:

Well.. I offically do not like disk caps.. phew.. they are done!

 

Getting there.. off to do stuff outside.. can't be a celler dweller all the time.

 

James

It’s all easy from here!

[+Firedawg]

9 minutes ago, Bikerbob said:

Well.. I offically do not like disk caps.. phew.. they are done!

 

Getting there.. off to do stuff outside.. can't be a celler dweller all the time.

 

James

 

Looking good James!  You are asking good questions too.  Take your time and inspect after every solder session for good clean solder joints and look for any solder bridges that may formed.  Reviewing this thread for builder issues to avoid and flashjazzcat's videos will keep your project moving a long nicely.

 

Mike

[+DrVenkman]

Another thing you can do while you take breaks between soldering sessions - especially if you’re a bit unsure of yourself - is set your multimeter to continuity mode and just check connections. Measure from the leg of one component to the next via or component leg you can find. This gets a little easier once you have sockets installed but before you have chips inserted and daughter boards like the UAV and U1MB in the way. You can verify your solder joints are good, or at least spot check several and give yourself a bit of confidence in the rest of the work. 

[+tf_hh]

On 7/31/2019 at 3:25 AM, mytek said:

Jurgen I checked out two of my XELs (both in PAL mode) and one had just barely over 4V going to the color trim pot, and yet the other was a very healthy 9V. Hard to say if both have a PN2222 from the same manufacturer or not, but I was ordering some parts for another project from Jameco.com and added some PN3904s to the cart. When those arrive I'll swap out the poor performing PN2222 and see if that fixes it.

 

Thanks for checking this. I also do more investigations and found out, that the CPU was the culprit. I used a "NOS" CPU from Best Electronics, produced by Signetics, Datecode 8236. This one has an interesting fault: After warming up (approx 2-3 minutes), the width of the "high" phase of PHI1 (2.0 volts ore more) will be reduced from the typical 200 nanoseconds to 150 nS and less. While this happens, the maximum voltage at the color pot will be reduced by something below 7 volts. After 20 minutes of usage the system also get instable.

 

For a good and correct colored display using PAL there must be something around 6.8......7.2 volts be present at the CADJ input of the GTIA. As well known, this varies from each GTIA to the other. When the maximum at the color pot is around 7 volts only, then it´s sometimes not enough - the picture will be too greenish. Using the 2N2222 transistors I got from Mouser the maximum voltage at the color pot (joint of R26 and C36) is always 0.5 volts less than with the 2N3904.

 

So I will revise my post, but on the other hand... maybe some PAL users has a similar thing - one time ?

 

Jurgen

 

[+mytek]

1 minute ago, tf_hh said:

Thanks for checking this. I also do more investigations and found out, that the CPU was the culprit. I used a "NOS" CPU from Best Electronics, produced by Signetics, Datecode 8236. This one has an interesting fault: After warming up (approx 2-3 minutes), the width of the "high" phase of PHI1 (2.0 volts ore more) will be reduced from the typical 200 nanoseconds to 150 nS and less. While this happens, the maximum voltage at the color pot will be reduced by something below 7 volts. After 20 minutes of usage the system also get instable.

Interesting. I'll have to try a CPU swap on my system with the low color adjust voltage and see if that fixes it.

 

1 minute ago, tf_hh said:

For a good and correct colored display using PAL there must be something around 6.8......7.2 volts be present at the CADJ input of the GTIA. As well known, this varies from each GTIA to the other. When the maximum at the color pot is around 7 volts only, then it´s sometimes not enough - the picture will be too greenish. Using the 2N2222 transistors I got from Mouser the maximum voltage at the color pot (joint of R26 and C36) is always 0.5 volts less than with the 2N3904.

Since the application uses the transistor as merely a switch, the voltage difference between the two different transistors has got to be due to the diode like voltage drop of the device. Looks like the PN2222 has a greater voltage drop than the PN3904. The additional 0.5 V is significant, so I think might revise my BOM for that particular transistor to be a PN3904 instead of the PN2222. Although my other XEL is at nearly 9 VDC using a PN2222 which is odd.

 

[Bikerbob]

Sorry I need to ask again since no one answered me.. Where do I do the vsync mod.. since I read it is on the solder side of the board.. and my 1.1 board has no markings on the solder side for any components.

 

James

[+Stephen]

1 hour ago, Bikerbob said:

Sorry I need to ask again since no one answered me.. Where do I do the vsync mod.. since I read it is on the solder side of the board.. and my 1.1 board has no markings on the solder side for any components.

 

James

Sorry BB - I am not sure.  My 1088XEL is the beta PCB.  I never populated the final version.  If nobody answer, send @mytek a PM and he will get you sorted.

[+DrVenkman]

1 hour ago, Bikerbob said:

Sorry I need to ask again since no one answered me.. Where do I do the vsync mod.. since I read it is on the solder side of the board.. and my 1.1 board has no markings on the solder side for any components.

 

James

Well, I have a production board and never did the mod, mostly because I'm perfectly happy with the Composite and S-video out of this box for now. My back panel does have a cutout for a planned DVI jack for the SOPHIA DVI board I have yet to buy. Sorry I can't be of any help there.

[+MacRorie]

1 hour ago, Bikerbob said:

Sorry I need to ask again since no one answered me.. Where do I do the vsync mod.. since I read it is on the solder side of the board.. and my 1.1 board has no markings on the solder side for any components.

 

James

So, I am sure this photo is somewhere in this thread and/or some other XLD thread (and somewhere in my HDD), but the easiest way for me to find it was to take a picture of the printout in my "build book."  Hope this helps, let me know if you have questions!

 

 

[Mr Robot]

I never bothered with that either, I use that 13 pin connector for my VBXE now

 

[+mytek]

6 minutes ago, Mr Robot said:

I never bothered with that either, I use that 13 pin connector for my VBXE now

It was only an issue with certain monitors that really wanted a VSYNC.

 

On the 1088XLD, that's been cured by having an LM1881 sync separator chip on board, which also serves to buffer the sync outputs.

 

[Bikerbob]

So build without it for now then?

 

Ok dokie

 

James

[+MacRorie]

2 hours ago, Bikerbob said:

So build without it for now then?

 

Ok dokie

 

James

All depends on how you are going to use that port and what with.  IIRC, a JVC SC1224 does NOT need the modification, so there's that.

[Bikerbob]

my plan is to use the computer on a monitor with S-video.

 

James

 

 

[+DrVenkman]

8 minutes ago, Bikerbob said:

my plan is to use the computer on a monitor with S-video.

 

James

 

Ah, the same way I do, and I venture to guess, many or most XEL builders do.  In that’s case, no need for any mods to the DIN13 connector. You should be good as-is.

[+mytek]

3 hours ago, MacRorie said:

All depends on how you are going to use that port and what with.  IIRC, a JVC SC1224 does NOT need the modification, so there's that.

JVC version NO, GoldStar Version YES, Samsung ???.

 

3 hours ago, DrVenkman said:

Ah, the same way I do, and I venture to guess, many or most XEL builders do.  In that’s case, no need for any mods to the DIN13 connector. You should be good as-is.

Correct ?

 

[wildstar87]

On 8/3/2019 at 9:23 AM, mytek said:

Interesting. I'll have to try a CPU swap on my system with the low color adjust voltage and see if that fixes it.

 

Since the application uses the transistor as merely a switch, the voltage difference between the two different transistors has got to be due to the diode like voltage drop of the device. Looks like the PN2222 has a greater voltage drop than the PN3904. The additional 0.5 V is significant, so I think might revise my BOM for that particular transistor to be a PN3904 instead of the PN2222. Although my other XEL is at nearly 9 VDC using a PN2222 which is odd.

 

Should I just use PN3904 instead on the build?

[+mytek]

48 minutes ago, wildstar87 said:

Should I just use PN3904 instead on the build?

It certainly won't hurt. However it's not certain that the difference being seen isn't due to a defective PN2222. I've been a bit busy, lately so haven't verified if this is so.

 

[Bikerbob]

OK...here we go.. So this is where I am at now.

[Bikerbob]

Now my issues.. Can someone point me on the board things that I still need in order to fire this up and test it?

 

My U1MB which I did first.. is a bit of Sh%t show.

 

According to the pins I think I have messed up.. I talked with Lotharek and in the below image I need to go from that pin to pin 15 on the Xill chip.. not easy to do.. ANY TIPS???  One other wire tap I need to do from a pin to a resistor.. easy.

 

OH.. and the FTDI I bought does not fit.. so if you think they are universal.. think again.. all the same pins etc... but about .25in narrower in width.. blows...   So I will be talking to McRorie to get another.

 

ANY and all help, suggestions etc.. welcome.. I can solder.. but it would seem not desolder well.

 

James

Edited August 13, 2019 by Bikerbob

[flashjazzcat]

I take it the entire trace from the CPLD to the connector has not been vaporised, so perhaps you could scrape off the solder mask from the trace near the connector and just patch across the gap.

 

Note the pin in question goes to the second leg from the top on the left edge of the CPLD as viewed. Does it also travel to the CPLD leg marked with an arrow? I only get continuity between the problem pin and the aforementioned chip leg (second one down, left hand side).