1008 K Axlon Compatible Board w/Base RAM and no soldering

[Meadow Glen]

I just love this stuff, it's so cool.

[+Spancho]

I‘m as well interested in 2 sets of boards for the 800 and 2 for the 400.

[reifsnyderb]

I didn't like having the /S4 and /S5 jumper wires with that extra connector at J2.  I found some instructions for the original Atari 400 RAM upgrade and connected /S4 and /S5 to pins R and N, respectively.  So 4 wires will need to be soldered to the bottom of the board as per Atari's original instructions.  The board and installation are now much cleaner.

 

 

[reifsnyderb]

15 sets of boards for the 800 have been ordered.  While I made a couple changes I am extremely confident these will work.  These will be tested, of course.

5 boards for the 400 have been ordered.  The boards for 400's should work but I'll have to test them first.  So I didn't order as many.

 

 

[ClausB]

On 4/21/2022 at 3:55 PM, reifsnyderb said:

But then I'd have to make a jumper that is soldered to the OS board and I'd rather not have to modify Atari equipment, if possible.

You could use clip leads to take S6 and S7 from the ROM ICs, so no solder required. For that matter you could take A15 and EXSEL from the ROM board too, and not need the extra boards in slots 1 and 3.

[reifsnyderb]

8 minutes ago, ClausB said:

You could use clip leads to take S6 and S7 from the ROM ICs, so no solder required. For that matter you could take A15 and EXSEL from the ROM board too, and not need the extra boards in slots 1 and 3.

A possibility would to make a new ROM/personality board but I've avoided looking into it out of copyright concerns regarding the ROMs.  I don't know how others have avoided this issue.  If there aren't any copyright concerns, the ROMs could be put onto a flash chip...which would take up very little space.

[gilsaluki]

GReat! 1st on the List for an 800 board!  

[_The Doctor__]

yes you could put a header on the top of all the card and let that feed the signal from each slot to the other if it's in one slot it can be sent to all other slots that way... no soldering... and this method was used back in the day for certain memory upgrades, you can simply take it a step further.

[reifsnyderb]

1 hour ago, _The Doctor__ said:

yes you could put a header on the top of all the card and let that feed the signal from each slot to the other if it's in one slot it can be sent to all other slots that way... no soldering... and this method was used back in the day for certain memory upgrades, you can simply take it a step further.

I've been using the internal connections the 800 already has.  i.e.  Slot 2 Pin T connects to Slot 1 Pin S.  Since this board already has the 48k base RAM the only thing I need to do is pass /S0 and /S1 to board 2.  But, yes, I've seen instructions for other boards that use extra connections.  For example the 1090 RAM boards had those connections between the boards.

[andymanone]

Very cool project ??!

 

Is there a rough idea for the final price?

 

Cheers,

andY

[reifsnyderb]

I've spent a lot of time considering and exploring the price of completed boards and lots of factors come into play.  So, I'll list them first...

 

1.  4-layer boards.  The 2nd layer is the ground plane.  They are a little more expensive but they are easier to run all of the traces and having the ground plane is a plus in regards to preventing interference.  I kept the length of the board to under 100mm to avoid an "engineering fee" of $35 only to discover that ordering 15 boards returns that "engineering fee".  Go figure.

 

2.  KiCAD reports there are 246 pads.  (I didn't count them to verify.  The count might include the 44 for the edge connector.)  If one pad isn't soldered right the board won't work.  So, I won't ship anything without thoroughly testing it first.  I don't like to receive non-functioning hardware and don't expect anyone else to either.  So, there is assembly and testing time to consider.

 

3.  Shipping is horrible.  The cheapest shipping for the boards doesn't have a reasonable timeframe.  So, I am forced to go with DHL.  They have demonstrated they can get the boards to Harrisburg, PA, but for some strange reason can't always finish the job by delivering the boards the last 30 miles.  So, I may have to go get them.  Or not.  It's hard to say as DHL can either deliver the last 30 miles or keep re-scheduling the delivery indefinitely.  UPS wants an extra $25 to deliver the same package because I live in a rural part of the country...even though they have up to two trucks going by my place every day.

 

4.  Shipping for electronics from a supplier.  It's more reliable than for the boards.  UPS doesn't add that extra $25 fee.  But it's still another cost.

 

5.  Electronics.  I have a concern about buying from some of the cheaper places like AliExpress as people report fake chips.  Also, it takes upwards of a month to receive items purchased from AliExpress...unless I want to pay extra shipping.  So, I almost always buy from DigiKey or Mouser.  (Admittedly, things like connectors are sometimes so overpriced that it makes sense to wait on AliExpress shipping.)

 

6.  Cost of packaging, shipping materials, labels, etc., etc.  All of that little stuff adds up.

 

7.  The cost for me to ship something within the US is going to be close to $10.  The cost to ship international will be at least $25...plus customs paperwork and, depending upon the Post Office, loads of time waiting at the Post Office.  (I've waited up to 45 minutes for the local Post Office to process the paperwork for a single international shipment.)  I do need to check on international shipping costs.  That $25 would be for the cheapest form of international shipping via USPS.  I've shipped a lot of stuff international in the past and can state that from experience, the shipping time is anywhere from 5 days to 5 weeks.  Faster USPS international shipping is around $45 for Priority and around $70 for Priority Express.

 

So, after considering parts, time, and all of the above factors, and after laying everything out in a spreadsheet, I figure a fair price is $75 a kit.  (For the 800, this includes the slot 1 Jumper Board.)  A similar 1056 Board for the 400 would be $72 as a 400 board wouldn't need that extra Slot 1 jumper board.  For 2 or more I'll ship for free within the US.

 

 

 

 

 

 

 

 

 

 

[+mytek]

21 minutes ago, reifsnyderb said:

4-layer boards.  The 2nd layer is the ground plane.

Actually the usual way I go is to have the 2 mid layers be Ground and +5V so that you only have to run logic traces on the top and bottom planes. Also having a sandwich of Ground and +5V creates one large decoupling capacitor as well as the extremely low resistance on the power planes, all of which minimizes noise issues.

 

[reifsnyderb]

1 hour ago, mytek said:

Actually the usual way I go is to have the 2 mid layers be Ground and +5V so that you only have to run logic traces on the top and bottom planes. Also having a sandwich of Ground and +5V creates one large decoupling capacitor as well as the extremely low resistance on the power planes, all of which minimizes noise issues.

 

With boards I've been building, Ground is on the 2nd layer.  The 3rd layer has +5v traces and a minimal number logic traces.  Most of the logic traces are on the top and bottom planes.  Sometimes it's very difficult to make a connection with 2 layers so I've got to go to the third layer.  I think I have logic 15 traces on the 3rd layer.

Edited April 25, 2022 by reifsnyderb

[danwinslow]

Hey, this is custom curated small-batch electronics. Don't worry about the charge. People who do this kind of thing for the community should actually make some money, a little at least. You're doing us a favor. As a matter of fact, I'm planning on paying even more for an assembled board, if you're willing.

[+mytek]

20 hours ago, reifsnyderb said:

With boards I've been building, Ground is on the 2nd layer.  The 3rd layer has +5v traces and a minimal number logic traces.  Most of the logic traces are on the top and bottom planes.  Sometimes it's very difficult to make a connection with 2 layers so I've got to go to the third layer.  I think I have logic 15 traces on the 3rd layer.

That 3rd +5V layer as you've called it, should probably still be created as a full copper plane, and if you need to run traces on it, that can still be done with relief provided between the plane and the logic trace.

 

On  my 4-layer layouts I only use the inner planes as power and ground, and just resort to vias to route the difficult traces by going back and forth between the top and bottom layers. I also try to apply a directional bias when laying traces. So for instance I might have all the top layer traces going in a horizontal direction, and the bottom traces going in a vertical direction. This is pretty standard practice and keeps you from creating too many dead ends early on. However I didn't do this with the 1088XEL, which was pretty hap hazard, and created a lot of issues for me as I got about 2/3 of the way through.

 

As for being able to eliminate decoupling capacitors by using power planes, that would be a no go, since the capacitance between the power planes is fairly weak (it's a poor capacitor). It probably does provide some decoupling, but not enough to eliminate the need for individual de-coupling capacitors whose main purpose is to provide local power storage next to the chips in order to prevent voltage sag during switching of the internal logic on those chips. Having the robust low resistance power planes just insures that ample power can be delivered to these capacitors and chips.

 

Here's a good summary from the article "Eliminating MYTHS about Printed Circuit Board Power/Ground Plane Decoupling"...

 

Quote

Careful measurements have shown that many of the common decoupling myths are false.  While many of these myths are considered ‘rules of thumb’, they have no basis in a science which includes all the real-world and full wave three dimensional effects. The proper decoupling strategy for PC boards need to account for the two purposes of these capacitors.  In order to provide rapid charge delivery to the IC for functionality, a low ESR capacitor should be placed close to the IC’s power and ground pins.  However, this does little or nothing for the EMC problem.

 

The EMC problem requires that a distributed capacitor strategy  be used.  That is, the capacitors should be distributed over the entire board.  This is not inconsistent with the functionality requirement, since a capacitor near an IC can serve both purposes easily.

 

It has also been shown that the actual value of the capacitor is not important.  It is recommended that the largest value of capacitance within the selected surface mount package size be used.

 

It has been shown that a single value of capacitance is better than multiple values.  The multiple values do not help at high frequencies (because of the inductance of the vias, etc.), and can actually decrease the decoupling performance significantly at lower frequencies.  The decreased decoupling performance is due to unavoidable cross resonances between components.

 

And finally, it was shown that lossy, or ‘high ESR’ capacitors can actually improve the decoupling performance at high frequencies.  These lossy capacitors should be used in combination with normal capacitors.

 

It is hoped that this systematic approach to decoupling design strategy will help engineers understand the trade-offs associated with the various options.  It is further hoped that engineers are now better able to discount the many myths and misconceptions that abound.

 

 

[Paul Westphal]

On 4/22/2022 at 7:18 AM, reifsnyderb said:

Here's a board that should, theoretically, work on a 400.  The edge connector has /S0, /S1, /S2, and /S3 connected as per the Atari 400 pinout.  /S4 and /S5 have to be connected via J2.  I used the instructions from tfhh's 48/52k RAM card for the edge connector pinout of RD4 and RD5.  (Two wires need to be soldered on the underside of the board for RD4 and RD5.)  Two resistors (R4 and R5 added) and a couple OR gates (already on the board as U41) were used to duplicate the Atari 800 circuitry for RD4 and RD5.  Once I get these boards in I'll test on my Atari 400.

 

I'd be interested in two of these.

[reifsnyderb]

It looks like the first batch of board should be shipping later this week as only SMD placement of the 100nF capacitors needs to be done.  When they ship the boards I'll order all the parts to complete the first batch as I am extremely confident they will work.  

 

I've been using the prototype board in my 800 and it's still working great...even with the two jumper wires soldered on it.  Since I have Syncalc, I loaded it and it shows 245k free...which I believe to be normal for Syncalc.   (https://atariwiki.org/wiki/Wiki.jsp?page=SynCalc)

  

 

[reifsnyderb]

The bare boards are finished and sitting in China waiting to be picked up.  Hopefully, they'll be in transit before tomorrow as it looks like they close down until May 3rd.  

[reifsnyderb]

Boards are shipped and might be in my hands by Thursday.  We'll see.

 

If they are, I'll have both the 400 and 800 boards tested within a day or receiving them.  Having a 400 with 1056k of RAM will be a nice upgrade.   ?

 

[reifsnyderb]

The new boards arrived today and the electronics should arrive tomorrow.  Tonight I moved the chips from the prototype board to the new 1056 Board for the 800 and it worked!  There is no need to solder anything to the 800 or to the personality card to get these working.  The new Slot 1 Jumper board works beautifully to send /S0 and /S1 to 1056 Board in slot 2.  I will need to test with the electronics as this board uses all 74LS chips and I have 74F chips coming.  However, I don't anticipate any timing problems due to the memory being synchronized with PHI2.

 

Once I put together and test a 1056 board with the 74F chips I'll test out the 400 version of the 1056 board.  It will be neat to upgrade my 400 from 16k to 1056k.   ?

 

[reifsnyderb]

A board with the F chips has been assembled and it is running the all-important M.U.L.E. test with the computer playing itself and is doing great.  All other tests look good.  SpartaDOS also shows the Axlon RAM as well.

 

The 74F08 chips I ordered turned out to have a wide SOIC-14 footprint.  I had to bend the pins straight down and solder them in place on the test board so as to fit the chip to the footprint.  But I wanted to test with as many 74F chips as possible.  (Only the hex inverter chip is a 74LS chip since I couldn't get 74F chips for it.)  Since the 74F08's work great it won't hurt to substitute a 74LS08, with a narrow footprint, on these boards as I know that 74LS chips work too.  So, I'll order 20 74LS08's to assemble these boards as bending the pins to fit the footprint sucks.  The 74F08's I got in had a wide footprint even though they are both listed as SOIC-14 footprints.  There needs to be a special place in hell for whoever decided it would be great to have a narrow and wide footprint of the same name.  ???

 

I do have a custom footprint I made that supports both the SOIC-14 narrow and SOIC-14 wide footprints and will change all boards to use that footprint so as to prevent this nonsense from happening again.   

 

 

[+mytek]

2 minutes ago, reifsnyderb said:

A board with the F chips has been assembled and it is running the all-important M.U.L.E. test with the computer playing itself and is doing great.  All other tests look good.  SpartaDOS also shows the Axlon RAM as well.

 

The 74F08 chips I ordered turned out to have a wide SOIC-14 footprint.  I had to bend the pins straight down and solder them in place on the test board so as to fit the chip to the footprint.  But I wanted to test with as many 74F chips as possible.  (Only the hex inverter chip is a 74LS chip since I couldn't get 74F chips for it.)  Since the 74F08's work great it won't hurt to substitute a 74LS08, with a narrow footprint, on these boards as I know that 74LS chips work too.  So, I'll order 20 74LS08's to assemble these boards as bending the pins to fit the footprint sucks.  The 74F08's I got in had a wide footprint even though they are both listed as SOIC-14 footprints.  There needs to be a special place in hell for whoever decided it would be great to have a narrow and wide footprint of the same name.  ???

 

I do have a custom footprint I made that supports both the SOIC-14 narrow and SOIC-14 wide footprints and will change all boards to use that footprint so as to prevent this nonsense from happening again.   

 

 

Congrats ?

 

BTW, I've been substituting 74HCTxx chips for the 74LS, and then only going to the 74F when I need a bit more speed. This way you get the benefit of the low power requirement of the CMOS, while still having relatively the same speed as LS. Even 74HC usually works fine in most of the applications, although the HCT is more tolerant of being fed TTL level signals.

 

[BillC]

59 minutes ago, reifsnyderb said:

A board with the F chips has been assembled and it is running the all-important M.U.L.E. test with the computer playing itself and is doing great.  All other tests look good.  SpartaDOS also shows the Axlon RAM as well.

 

The 74F08 chips I ordered turned out to have a wide SOIC-14 footprint.  I had to bend the pins straight down and solder them in place on the test board so as to fit the chip to the footprint.  But I wanted to test with as many 74F chips as possible.  (Only the hex inverter chip is a 74LS chip since I couldn't get 74F chips for it.)  Since the 74F08's work great it won't hurt to substitute a 74LS08, with a narrow footprint, on these boards as I know that 74LS chips work too.  So, I'll order 20 74LS08's to assemble these boards as bending the pins to fit the footprint sucks.  The 74F08's I got in had a wide footprint even though they are both listed as SOIC-14 footprints.  There needs to be a special place in hell for whoever decided it would be great to have a narrow and wide footprint of the same name.  ???

 

I do have a custom footprint I made that supports both the SOIC-14 narrow and SOIC-14 wide footprints and will change all boards to use that footprint so as to prevent this nonsense from happening again.   

Mouser carries the Texas Instruments SN74F04N and SN74F08N, which are 14-pin PDIP according to the datasheets

[reifsnyderb]

3 minutes ago, BillC said:

Mouser carries the Texas Instruments SN74F04N and SN74F08N, which are 14-pin PDIP according to the datasheets

Only one problem.  Those are through-hole.  I've already ordered 74LS08's that are of the SOIC-14 narrow variety.  (I double-checked the datasheets, too.)  

[reifsnyderb]

First working non-prototype.    ?