jasonsbeer / amiga-2000-eatx Goto Github PK

Hello Jason,

The termination from C675 on pin 96 must be on pin 95. The orginal schematic from A2000_R6 has no termination on pin 96 but C678 on pin 95.

cheers Marco

This project looks amazing, I bet there are quite a few out there, including me, that don't have the skills,. nor the time free to learn the skills needed to succesfully order a board and make it a reality, i.e ready to put in a case. I fully understand that you the creator might rather want to perfect the design than to build and ship pre-populated boards, but have any of the regular amiga dealers( the few that ar active) showed any interest in this.Or is there any other way to obtain a mostly complete board, or plans for a more user ready viersion?

Hello,

I have downloaded the Kicad project. What you have done is really great !
There are some holes which are not connected like the Int6 signal onto the Zorro slots.
The 3v3 pins of the power connector are not connected also.

Is it normal that there are some missing tracks as shown in the pictures below ?

Int6 signal:

3v3 signal:

Kind regards,
Franck

ISSUE: Does not start up when 20pin PSU is used and SMD oscillator has been installed

Pin 12 on the ATX connector is 3.3v, and supplies power to the SMD oscillator. When a 20-pin PSU is used, pins 11/12/23/24 are missing and although 11/23/24 are connected to their equivalents elsewhere on the connector (i.e. 11 is connected to 10, 23 connected to 22, 24 connected to all other GND) pin 12 isn't connected to any other 3.3v pins so the oscillator doesn't get power.

(Thanks to LIV2 on Discord for working out where Pin12 was going)

TEMP SOLUTION 1: Short Pin 12 to 1 to provide power to the oscillator (verified - it boots like this)
TEMP SOLUTION 2: Use TH oscillator (not verified, can't see why it wouldn't work)

I have finished routing Rev 2.0, which is intended to support both 8372A and certain 8375 Agnus chips on the same board. I'll probably send this off for fabrication soon.

My next quest is to either buy or borrow some different 8375 Agnus part numbers. Please let me know if you can help out with any of the Agnus part numbers below.

UPDATE: I have obtained a 390554-01 Agnus. Still looking for the other part numbers.

390554-01
390554-02
318069-12
318069-13
318069-10
318069-11
390544-01
390544-02

I also added a discord channel for this project...https://discord.gg/NU7SPYfNFj

I have ben having some trouble using my 2 flickerfixers / scandoublers on the board.
A little backstory.
I have 2 scandoublers / flickerfixers which both works out well on my stock A2000 rev6

I have ben comparing the picture and signals from my org. A2000 and the A2000ATX board on my bench - I will post the pictures.
I'm writing because I'm at a dead end in this.

My setup is the same on both boards: Kick 1.3 - up and running, but not booted into anything.
I'm using a Lighttech Flickerfixer.
My second Flickerfixer is a microway: https://bigbookofamigahardware.com/bboah/product.aspx?id=374
This part do not work in the port for some reason.
Both machines are PAL machines, and the Xtal is 28,37516 MHZ

And here's some pictures of a simultaneous comparison of the videoslot.
The Burst signal on both looks like this

:

The hsync signal looks like this:

When on kickscreen, I got this:
Original A2000

ATX amiga

It is noticeable that there is a white barre just below the kick picture.

Second: When trying to change screenmode to interlaced, a thing the flicker card can handle, the ATX board doe not handle this at all.
Two pictures to show this.:
Org amiga 2000:

ATX amiga:

Also a part of the picture (topmost) is out of the picture.

I do not know what causes this issue. All the signals is there, but they seems wrong in some way.
Maybe a misconfiguration of the videoengine??
Idunno.

Putting this here for posterity. The Amiga B2000 schematics have an error related to the connections of CN300 (the keyboard DIN socket). The pins are mirrored on the Commodore schematic. Using the Commodore connections will cause R914 to burn up. These bad pin outs were carried on to the A3000 schematics.

Correct, as used in the Amiga 2000 EATX.

Incorrect, as in the Commodore schematics.

Paging the Amiga RAM team! @texneus @LIV2

I desire to make an alternate revision of the Amiga 2000 EATX that directly supports the 8375 2MB Agnus. I have attached a proposed schematic of the RAM circuit. Not everything is connected, but all the parts relevant to this discussion should be. Can I impose upon you guys to review the schematic and let me know what you think?

Thanks!

CY62167E_16_Mbit_1_ M_ 16_Static_RAM.pdf

file6121153C-Sheet6121153C.pdf

As I do not have a bridge card, I was unable to test the ISA slots. If you happen to build this project and have a bridge card, I would appreciate a message indicating whether they are working as expected.

Hi! Finally I finished assembling the a2000 and it started up almost immediately! :) But as it turned out, the reset practically doesn’t work... every third or tenth time, or even the twentieth time, it works... I can’t understand the reason? When inspecting the reset generation circuit, the following was noticed - if you put your finger on the side 8-9-10-11 of the U805 leg and press the reset button, then the reset works well; removing the finger also starts to work poorly... I measured the esr of capacitors C813 C814, everything is fine normal... where to look?

I have been comparing the A500 and B2000 internal floppy connector circuits. The A500 only supports one internal drive, because it turns the motor on only when DF0 is meant to spin. (Incidentally, can we agree that _MTROD is a typo? It's called _MTRON on the A500.)

The B2000 drives the _INUSEx lines as you would expect the motors, accounting for the twist in the cable before DF0. Then it logically ORs those into _MTRON, so both motors spin if either drive is in use, it's just the LEDs that differ. (Ouch, my antique floppies, they will wear out faster!) Despite the cable twist, the drives still need jumpers set for which _SELx line they respond to.

Since we are aiming at a 100% new parts build, I figure I'll use PC drives. Either modded to work like Amiga drives, or using an adapter right at the drive. I'd do a bit more twisting on the cable, so both drives only have to listen to one of the _SELx lines, and they become interchangeable.

Then you can modify the drive to spin the motor when it is _INUSE, which restores the behaviour of external drives, i.e. all drive motors are independent. The common _MTRON signal can be dropped.

And since ATX cases usually have two bays, I'll install two drives. But then I'd also like a Gotek, hmm... and there's no external drive port any more...

One approach is to just bring out _SEL2 and _SEL3, as well as _MTRx and _DKRST, so the internal floppy port has the needed signals to supply drives using external floppy controllers. But, it's the same number of pins if we supply _SEL2, _SEL3, _INUSE2, _INUSE3, and that would make the case layout so much neater. All drives are treated the same.

So, I suggest that instead of one jumper, for DF1, there would be three. And a couple more logic chips to derive the extra four signals.

You can free up pins 14 and 16 straight away. Make them _SEL2 and _SEL3, respectively. It's a neat solution, the cable shifts the wires by two places after each drive. (Pull out the sharp metal part of the IDC connector, for any new signals that you don't want to send to a given drive.)

And pin 3 is usually cut off, but it could be used. Perhaps as an _INUSEx, because those signals don't really need twisting with ground anyway. E.g. wire the motherboard as:

• 4 - _INUSE0
• 5 - _INUSE1
• 6 - _INUSE2
• 7 - _INUSE3

Similar to above, the wires move one across after each drive.

As I've specified the above pins, the drive closest to the motherboard would be DF0. This means you can make a cable for only one drive, without any twists in it. If you want DF0 at the far end of the cable, you twist the _SELx and the _INUSEx groups, between the motherboard and the first drive. The drive ordering is now reversed, with DF3 being closest to the motherboard.

I think that's pretty close to optimal, really. Simple cable wiring, take some photos and anyone can copy it. It does mean changing the drive behaviour, so the motor spins on _INUSE. (You can do this in the cable, if you really want. But if you're modding PC drives anyway, what's another wire.)

I noticed I have been using a 3.3v SRAM chip in the BOM. Although it runs rock solid in all my prototype builds, I think it is best to replace the 3.3v part with a 5v part.

THE BOM HAS CHANGED--

OLD PART -
Digikey part 428-3574-1-ND
Manfuacturer part CY62157EV30LL-45ZSXIT

NEW PART
Digikey part 428-2077-ND
Manufacturer part CY62157ELL-45ZSXI

The BOM files have been updated. I'll leave this open for any comments or input for a day or two.

ISSUE: The XCLKEN pin is lacking a pullup resistor, which could result in the 74HC258 switching to the XCLK signal when there is not one present.

SEVERITY: There have been several reports of this happening.

SCOPE: Rev 2.x boards only.

PROPOSED FIX: In the event you do not get a 28MHz clock from pin 12 of U109, add a 10k pullup to pin 1 of U109. You can attach the pullup to any +5V source. C109 is nearest. Target future revision for fix.

History
Jul-13-2022 Original report.
Nov-27-2022 Updated severity.

ISSUE: Some A2000 EATX boards present a black screen when using accelerators designed for the MC68000 socket. Examples: Spitfire, TF534, TF536.

SEVERITY: High

SCOPE: All revisions.

FIX:

1. In the event you have black screen issues with this style of accelerator, you should change U103 and U105 to the 74F245. This change fixed the system of user sparxUK, as discussed on Discord.

2. Change the schematics and silkscreen to indicate this change for all machines in the next revision.

3. Change the BOM to 74F245 logic at U103 and U105. This change will be implemented immediately.

OTHER COMMENTS: 74AHCT245 will likely work as well, but has not been specifically tested.

Hello Jason - what a nice project you did here!
How is the tick signal dealt with ? I see you recommend a standard ATX PSU, and you mention how the -5V is generated but what about the tick signal ? Is it onboard too ? Thanks ! (might not be an issue, after all... haha)

The silkscreen shows 1000pF. The correct value is 0.1uF.

ISSUE: Add internal header for CD Audio, such as found on the Amiga 4000.

SEVERITY: Enhancement request.

SCOPE: Future revision.

PROPOSED FIX: Review A4000 schematics and include similar solution.

Starting with revision 6.2, the Commodore A2000 board incorporated a pull down data bus. Due to the way it was implemented, this added a pull down to the direction signal on the Zorro 2 data bus buffers. This pull down is not necessary, but apparently has not caused any issues (?). Consider removing it.

Hi
I'm in the middle of the build and I've just gotten to the point where I've put voltage on the mainboard for the first time to take some measurements regarding the power rails. That's were i have noticed that the voltage on C408 (-12V) is reversed. C408 is not populated yet, so it isn't a problem. But i'am a bit hasitant to put it in like marked on the silkscreen. C412 (-5v) is marked the other (correct) way. Am i wrong ?

ISSUE: A A314 expansion can not be used on the B2000 EATX board.

SEVERITY: Enhancement request.

SCOPE: Future revision.

PROPOSED: Add the option for a pin-header similar to the A500 ram expansion, on the B2000 board, potentially just above the ram-chip - leave it as an option so people can choose not to populate this pin-header if they have no desire for this feature.

BACKGROUND/MOTIVATION:
The A314 expansion is a game-changer in what is possible with an Amiga, since it has direct access to the chip-ram bus it can do some truly amazing stuff - chip-ram on steriods!
However a normal A2000 has no way of using this great new expansion, which means that it is limited to only being used in the A500/600.
The B2000 could change this, since the A2000 essentially is an A500 architecture, it should not be difficult to find the pins required for the ram-slot - I imagine they are all near the agnus and/or ram-chip, and a simple un-populated pinheader on the pcb with these connections would make it possible to "insert" the A314 vertically ontop of the board and essentially take over the chip-ram handling of the computer... A314 & Zorro slots what a combo.

Hi @jasonsbeer, Great project! I can't wait to build one!

Just thought I'd note that the published circuit for Kai's SRAM board is incorrect and doesn't work properly, I poked it a bit to troubleshoot why and got it working in the end but the changes ended up slipping through the cracks and weren't upstreamed at the time - I'll try to sort that out.

I've attached the schematic I came up with after bodging up the board to make it work, if you've already submitted this rev for manufacturing an easy way to work around this would be to use something like my 2MB Agnus adapter and the relevant agnus with this board as it provides it's own ram.
I'd be happy to donate an assembled adapter to help with this project if you'd like, you'd just need to get your hands on an 8375 Agnus

sram-sch.pdf

Hi,
first of all, @jasonsbeer thank you for this wonderful project ! I'm going to build one but I see that those EMI filters (W3F11A1018AT1F) are not available. Can someone suggest a 1206 form factor alternative ? I'm wondering if the 0805 form factor ‎101X15N101MV4E‎ would fit ?
Kind regards

The RTC-62423 datasheet specifies the frequency accuracy to be 20 ppm for the 62423A version, and 50 ppm for the 62423. This is really lousy compared to what you can achieve using a trimmer capacitor and a frequency counter. It might be even worse if the chip is decades old.

To put this into perspective, a 1 ppm error is one second over 11.6 days. So the non-A version could be out by one minute per fortnight. If you're using your A2000 only occasionally, it will be really noticeable every time you go to use it. And there's no way to adjust that.

I'm sure Commodore would have trimmed them all in the factory. This substitute is really a step down from the original hardware. If you need access to a frequency counter, go to an amateur radio club and ask nicely, someone will be happy to trim it to <1 ppm for you. (And now you have a new friend.) Or you could monitor whether it is running fast or slow compared to other clocks, and tweak it as needed.

Could you please include the component footprints needed for the capacitors and watch crystal, for those who want to use a part with a trimmable external oscillator? I am looking at a "parts only" A501 on eBay, and thinking that a through-hole footprint for the chip would be a helpful addition, alongside the SMD version.

Hi everyone,

I've made a calculation for production on PCBWay and seems like it is possible to make a small batch for a price of ~300 USD(all populated will be ~400 USD) per board. If anyone is interested we can make sort of crowdfunding.
I'm located in Russia though so there can be some shipping issues because of all of these things happenning... (EMS works worldwide though, ~70 USD shipping) but we can make a direct shipping from China if needed. If anyone if interested I think we can make a collective order and split parcel to small parcels with direct delivery. Payment is also a question, however we can discuss it here. I now have 7 customers in Russia so I'm almost ready to order 10 boards. Maybe someone else wants to organize the order(for a larger batch), I can pay for boards.

jasonsbeer, sorry for bothering you with this message. I think that anyway collective procurement and crowdfunding is more efficient than separate ordering and will help more people to try the project.
I can share assembly order files(BOM, Pos) if needed - it will help many people and save time&money.
And of course thank you for your amazing work, looks impressive!!!

BR,
Fedor

ISSUE: Through holes for the slots are somewhat small. This makes them slightly difficult to install.

SEVERITY: Enhancement request.

SCOPE: Future revision. Slot footprints only.

PROPOSED FIX: Review actual pin sizes of slots in common use and adjust footprints, as necessary.

Hi, Great work, i would like to give this board a build but TBH i don't want to fiddle with the small components. Is there a BOM and CPL file with the gerbers available so i could get the board partially assembled. Without the .Brd file i cant generate the X and Y coordinates, the only way i could do it is to put in gerber viewer and manually do it (Ouch). Even basic CPL with the coords listed would be a headstart. If i generated the assembly files i would upload to this project page for others to use.

thanks Brian (Kipper2k)

When J9004 is on pos. 1-2, the POST process breaks.
That only happens when my Philips LCD monitor is connected.
This is caused by an unbuffered HSYNC to pin 13 on the VGA connector.
I figure that it is caused by a power drain on HSYNC, done by the monitor itself.

By using pin 8,9 and 10 on U303, which are redundant pins, this has ben fixed from my part.
It is done by lifting one leg from R203B1, running the HSYNC into Pin 9 and10 and out of pin 8 and down to the board toward pin 13 on the VGA.

There is a 23pin to 15pin adapter from commodore (390682-01) that comes with the A4000, that contains a 74HCT08 buffer.
I think mostly to avoid any monitors to be connected almost directly to any custom chips.