Potential new hardware

[HiassofT]

Thanks for the reply. It helped a little, now I can see something on the screen without doing so many resets. But it's still garbage, black, or green frozen screen.

This is not so good... I just reprogrammed my DE1 to default (DE1 demo in EEPROM), flashed atariroms3.bin@0x0000, programmed the experiment_201301002.sof (I usually create a pof from it and program it to the eeprom) and everything was fine here.

 

When I got my DE1 I also had a lot of problems, lockups etc (similar to yours) and bugged foft quite a lot about it. In the end it turned out to be a bad SRAM chip on my DE1. After I replaced it everything worked fine...

 

Not sure if it's the SRAM on your board too (could be something else of course), but thanks to foft we have quite some tests to narrow it down:

 

Program experiment_20130813.sof and set sw4-8 to up, all others to down. This config uses neither the external flash nor sram but uses 8k of internal FPGA RAM and an OldOS also internal to the FPGA. You should at least get the memopad screen with it.

 

If not, something fishy is going on. Maybe check you power supply - my DE1 works fine when either powered from USB or with the external PS that came with the board.

 

If memopad works try setting sw5 to down (to use the SRAM) and setting sw6 down (to use OS from flash). You might need to powercycle the DE1 or reprogram the sof after changing the switches.

 

Note: if you have sw5 up und sw6 down (8k FPGA RAM, XL OS from flash) the DE1 will boot directly into selftest with a strange looking screen. This is normal, the XL OS isn't prepared for only 8k of RAM (16k is minimum).

 

so long,

 

Hias

[NML32]

@Hias, here are my results from your suggested test using experment_20130813.sof.

 

SW4-8 up displays memo pad.

 

 

SW 5&6 down with a fresh flash.

 

 

SW 5 up and SW 6 down.

 

Thanks for the help. Not sure where I stand now, bad hardware?

[HiassofT]

Thanks for the help. Not sure where I stand now, bad hardware?

Might be a bad SRAM, too.

 

The selftest screen looks exactly identical on my DE1 - so I guess the flash chip should be OK.

 

Just for completeness: could you test with sw5 down and sw6 up (OS from FPGA with SRAM)?

 

The next thing I did (and recommend trying) is testing the SRAM using the DE1 control panel.

 

I couldn't get the DE1 control panel from my DE1 cdrom working at all, but the older control panel works (almost) fine:

 

Download DE1_CD_v08.zip from here http://www.terasic.com/downloads/cd-rom/de1/, program the DE1_USB_API.sof from the DE1_control_panel directory, start DE1_Control_Panel.exe and click Open->USB port 0 (you might need to close quartus first).

 

Then you could try the SRAM test, but this didn't report any errors on my board: switch to the "Tools" tab, click on "Board Test", select "SRAM Test" and click "Start Test".

 

So I used a 512k file with random data, uploaded it to the SRAM and read it back: switch to the "SRAM" tab, select "Entire File" in the "Sequential Write" section, click on "Write a File to SRAM" and select the 512k random file. Then select "Entire Sram" in the "Sequential Read" section and use "Load SRAM Content to a File" to read the data back to another file.

 

Now compare the files. Ideally, they should be identical, but it seems the control panel is buggy. Depending on the (random) input data I got mismatches at a few locations (1-3 words / 2-6 bytes) at word offsets 0xAA83 and/or 0x1AA83 and/or 0x283A5 (byte offsets 0x15506/7, 0x35506/7, 0x5074A/B). If you get a lot more mismatches (I got some 70 differing bytes with my bad SRAM), the SRAM is most likely faulty.

 

so long,

 

Hias

[NML32]

SW 5 down and the rest up.

 

Going to try the other test.

 

Edit: I didn't have much luck with the control panel Board Test. Both RAM test fail and the Flash Test times out.

 

Not sure if this means anything but I'm able to flash a MSX (de1_170513.sof) just fine.

Edited November 5, 2013 by NML32

[HiassofT]

SW 5 down and the rest up.

image.jpg

The screenshot looks like there are 64-byte blocks of garbage. Most certainly parts of the SRAM are faulty or something strange is going on with the FPGA.

 

Edit: I didn't have much luck with the control panel Board Test. Both RAM test fail and the Flash Test times out.

This is no good sign. If you haven't soldered on your DE1 (like I did back then...) I guess it's time to send back your board and get a replacement.

 

Since this is the second broken board in quite a short time, where and when did you buy it? BTW: could you take a picture with the markings on the SRAM and flash chips?

 

Not sure if this means anything but I'm able to flash a MSX (de1_170513.sof) just fine.

Could well be that this one doesn't use the SRAM at all. Back then I also tried minimig and it worked fine, guess it only uses the DRAM.

 

so long,

 

Hias

[NML32]

I bought my de1 from Digi-Key

Edited November 5, 2013 by NML32

[HiassofT]

I bought mine from Digi-Key, too, back in July this year. Maybe they got a batch of bad boards...

[NML32]

Well, I'm sending my DE1 back to Digi-Key and they're going to run some test to determine if there is anything wrong with the board. In the mean time I ordered another DE1 while Digi decides how to handle this. I hope I have better luck with the next board.

 

-Mark

[foft]

Please let me know what Digi-key say. 90% sure it's faulty sram but potentially my core has issues with the EDBLL sram chips. I bought another board myself but received a second AL sram that worked. The data sheets for both parts have identical timing constraints...

[HiassofT]

I'd also be interested in the response from Digi-Key.

 

BTW: Is there anyone else out there with a new board?

 

If yes, could you please try to write a 512k file to the SRAM with the new control panel (from the CD or from the latest CD image on the terasic site), read it back and compare the contents?

 

I couldn't get the new control panel to work at all on my WinXP machine and VM - it'd always complain that Quartus is not installed.

 

so long,

 

Hias

[NML32]

Ok, got my DE1 board today and it too came with a EDBLL sram chip. Well same results as my last board.

[foft]

Also from digikey? What country are you in?

[NML32]

From Digi-Key.

US

 

Also from digikey? What country are you in?

[foft]

Little far to swap. I have a spare de1 I bought to track this down before we decided it was a bad sram chip...

 

If its also from Digi-key it ties in with the bad batch theory. Perhaps try to find another core that uses sram? I tried much less aggressive timings on Hias' board with no improvement.

[HiassofT]

Ok, got my DE1 board today and it too came with a EDBLL sram chip. Well same results as my last board.

This is bad

 

Could you please try writing/reading the SRAM with the control panel from the new DE1 CD?

 

I just compared the datasheets of the old and new SRAMs but couldn't find anything obvious. Actually, the datasheet of the new IS61WV25616EDBLL chip says it can be used as a drop-in replacement for other standard SRAMs. I'd say the bad batch theory is more and more likely.

 

so long,

 

Hias

[NML32]

I did a quick test and the two files were different. I'll try it again when I get back home. It took me a while to find a program that would create a 512K text file.

 

Thanks,

Mark

 

This is bad :(Could you please try writing/reading the SRAM with the control panel from the new DE1 CD?I just compared the datasheets of the old and new SRAMs but couldn't find anything obvious. Actually, the datasheet of the new IS61WV25616EDBLL chip says it can be used as a drop-in replacement for other standard SRAMs. I'd say the bad batch theory is more and more likely.so long,Hias

[Beetle]

Hi there,

 

my board has IS61WV25616BLL -10TL SRAM chip and seems to work (Memopad shown and Self Test looking normal).

Next job is Cart port and Joystick. After that, i think, keyboard and SIO.

 

And, of course, make the core stay permanent.

 

Greets,

Beetle

 

[HiassofT]

Could you please try writing/reading the SRAM with the control panel from the new DE1 CD?

Just realised that the new control panel doesn't seem to have an option to access the SRAM at all - only Flash and SDRAM are supported.

 

So I guess this rules out the option to use the new control panel as an SRAM tester...

 

so long,

 

Hias

[foft]

 

my board has IS61WV25616BLL -10TL SRAM chip and seems to work (Memopad shown and Self Test looking normal).

Next job is Cart port and Joystick. After that, i think, keyboard and SIO.

 

And, of course, make the core stay permanent.

 

Great to hear that my core works on at least one EDBLL sram:-) I guess the others are broken, but hard to say with only one data point:-)

 

Good luck adding the cart, joystick etc. I tested them all except joystick port 2. Note that the clock is not output by the current core, so any carts that latch on the clk edge will not work yet. An easy fix but I'm surrounded by boxes still so not in a position to write any vhdl.

 

Are you going to add a real XL keyboard using 4051s? If so please double check my schematic! I've not tried this yet. Just tried the XEGS one.

 

If it helps for prototyping I still have plenty of 5V->3.3V level conversion boards (24CHF each + p&p). Otherwise be careful not to exceed 3.3V on the GPIO inputs or you risk the DE1.

[foft]

Just realised that the new control panel doesn't seem to have an option to access the SRAM at all - only Flash and SDRAM are supported.

 

So I guess this rules out the option to use the new control panel as an SRAM tester...

 

I guess they removed it since it is a bit buggy - i.e. write/read always created a set of errors - even on a working sram.

 

I really need to merge my sdram code, then that could be an option for those with broken sram. Two weeks now until I start a new job and I think all my time will be swallowed with post-move jobs. Perhaps I'll grab a day or two...

[HiassofT]

Great to hear that my core works on at least one EDBLL sram:-)

I'm afraid that's not the case. It's a BLL sram, which is more or likely identical to the AL/ALL chips (i.e. just 256kx16 without internal ECC), not an EDBLL with internal ECC. Beetle sent me a closeup of the board, it has the old flash and the "0844" on the very left could be a datecode of the board (as the 0826 on the SRAM). So I'd say the board was manufactured about 5 years ago.

 

I guess we are back at the start...

 

so long,

 

Hias

[Lynxpro]

Holy BLEEP is that awesome looking!

 

Stretch goal 47 -- dual pokey (if not quad core). heh.

Repeat after me...AMY!

[NML32]

I really need to merge my sdram code, then that could be an option for those with broken sram. Two weeks now until I start a new job and I think all my time will be swallowed with post-move jobs. Perhaps I'll grab a day or two...

 

That would be great!!

[foft]

Repeat after me...AMY!

Heh, that would be fun! Are any specs available?

 

edit: I see they are on Atari Museum, with the net list:)

Edited November 11, 2013 by foft

[jhusak]

Heh, that would be fun! Are any specs available?

 

edit: I see they are on Atari Museum, with the net list:)

Hello.

I've just have got the fresh minty super-duper smelly MIST platform produced by Lotharek. How do you think, is it very hard to put the a800 into this board?

Atari ST and Amiga are done already .

 

http://lotharek.pl/product.php?pid=96

Edited November 11, 2013 by jhusak