TF536 help getting cards to work

[Viserys]

So I bought 3x TF536 PCBs around 2 years ago from Exxos, and have been building them slowly, only working on them from time to time.
First mistake I did was buying from Aliexpress/ebay, but I have later bought from Mouser/Digikey/Reichelt that should be serious suppliers.

I have swapped different components so many times I am not sure what components are used on card 1 and 2.
Card 3 is still made with Aliexpress/ebay parts, as I could not program it(JTAG missing if I remember correctly).
The other two I managed to program successfully with Xilinx chips from Aliexpress(Raspberry Pi method).
I have therefore abandoned working on card 3 for now, and concentrated on getting 1 and 2 working.

I did change the Xilinx on card 1 from one I bought from Reichelt.
It is the C variant, as I read on a post here that it should work and it is a lot cheaper than the I variant.
Programmed it successfully, but it made no difference. Same behavior as before.

If I use Kickstart 3.1.4 on an Amiga 500+ I get black screen. The same if I use Diagrom 1.2.1.
However if I use Diagrom 1.3 I do get a picture although it stops at checking ram.

Card1:

TF536_card1_top.jpg

TF536_card1_bottom.jpg

TF536_card1_screen.jpg

Card2:

TF536_card2_top.jpg

TF536_card2_bottom.jpg

TF536_card2_screen.jpg

Card3:

TF536_card3_top.jpg

TF536_card3_bottom.jpg

I have tried to press left mouse button as suggested in Diagrom 1.3, but I never got it beyond what is shown in the pictures.
My impression is that the system crash at this point.

None of the chips get very hot, the only problem I have had is with the LM117MP on card 1 that after several times(weeks/months apart) of testing the card got burning hot, and maybe that has happened on card 2 as well, but it has been so long I can't clearly remember.
I replaced it with a new one, and I can now touch it without getting burned.
I have measured 5v on PWR2, and 3.3V on JTAG header, and they are fine.

Power supply is a new modern one, so should also be fine.

I have tried inserting a 68000 cpu, and Diagrom works as it should.

I am reasonably sure its the same fault with both card 1 and 2, but my experience at finding faults like this is pretty much non existences.

I have a multi meter, but have not used it beyond measuring voltages and continuity.
I do not have a Oscilloscope.

Any ideas what could be the issue, or what I can try getting these cards to work.

[go0se]

The socket you are using in picture "Card3_top" is a PGA121 however the CPU you are using is a PGA128. I assume you are using the same type of socket for all three cards due to the unsoldered inner socket pin holes on the other two cards?

Your socket is therefore missing power pins to the CPU. That seems likely it will cause supply problems.

Maybe that doesn't fully explain your issues as you are getting some response from Diagrom.

Perhaps @terriblefire could confirm?

[alenppc]

The regulator will get hot if there is a short somewhere, also if you place the crystal upside down. This will kill the crystal.

The pictures are not very clear so it's not easy to see if there are any bridges but this is something you should check under a microscope. However, I notice the following:

- Card 2 has a reprinted CPU that might not be able to boot at 50 mhz.
- It seems two pins are bridged on the IDE connector.
- As go0se points out, the CPU socket might not be the correct one.

You should solder the turned pin headers the other way round (thick part on the bottom) as this will provide a much more stable connection to the socket. For the rest, assuming you know you have good parts, check for bridges under a microscope and make sure the orientation of every component is correct.

[terriblefire]

The socket you are using in picture "Card3_top" is a PGA121 however the CPU you are using is a PGA128. I assume you are using the same type of socket for all three cards due to the unsoldered inner socket pin holes on the other two cards?

Your socket is therefore missing power pins to the CPU. That seems likely it will cause supply problems.

Maybe that doesn't fully explain your issues as you are getting some response from Diagrom.

Perhaps @terriblefire could confirm?

Yes indeed. He has the wrong sockets.

They should look like this

IMG_0054.jpg

Without this socket the CPU will have 4 power pins not connected and is unlikely to run at 50Mhz like that .

[Viserys]

Well that can explain it.
I will try to desolder the socket on one of the cards and replace it to see what happens.
These where sockets that I already had lying around, and as they fit the PCB, and I could fit it with the CPU with no problem it has not occurred to me that they are the wrong type.

Will look if I have the correct socket somewhere, and if not I will have to order some.

I will keep in mind that CPU on card 2 may not work, so will source a new one if that is the case.
Don't think any of the pins on the IDE connector is shorted.
There was a pit of fluff after cleaning the worst flux off, and one of the pins have a "tap" when there is too little flux so will need to reflow it.

[PhilC]

Don’t try desoldering the sockets, carefully solder in some female pin headers where you have missing ones.

[Viserys]

I already desoldered the socket for card 2, but as it was a bit of a pain to do, I will try to solder some extra pins on card 1 and 3.

It was a success in that it boots normally with Diagrom 1.3.
Haven't tried with Kickstart 3.1.4 yet.

However no ram is detected, so something is still not as it should.

[Viserys]

Ended up desooldering socket on card 1 as well as I could not get it aligned properly.
The Amiga booted up fine, and this time ram is detected as it should.

Have not tried anything apart from running memory test 1 time on the 64mb.
It went without issue, so finally looks like I have at least one working card.

[terriblefire]

Ended up desooldering socket on card 1 as well as I could not get it aligned properly.
The Amiga booted up fine, and this time ram is detected as it should.

Have not tried anything apart from running memory test 1 time on the 64mb.
It went without issue, so finally looks like I have at least one working card.

Seems the mystery is solved :)

[Viserys]

Looks like it :D
I will try to swap out the ram on card 2, and see if I get two working cards even if that is a bit greedy of me :twisted: