Terrible Fire Accelerators - Page 163

CRP · 21 May 2018, 21:14

So can anyone confirm whether the -p parameter refers to items in the JTAG chain in inverted order, i.e. 0 is the second item in the chain, while 1 is the first?
I have applied the instructions but it does sound a bit weird. If inverting the jed does not cause any damage I might try inverting the jeds.

Robban · 21 May 2018, 22:08

Quote:

Originally Posted by plasmab

It’s unlikely that any damage would occur with the wrong Jed file in the cpld

That is reassuring. It is strange that i have exactly identical behaviour on two cards though. But it could of course be bad soldering or cheap ebay parts (only CPLDs and memory were from DigiKey). I just wanted to share my experience since it was identical to CRPs

Quote:

Originally Posted by CRP

So can anyone confirm whether the -p parameter refers to items in the JTAG chain in inverted order, i.e. 0 is the second item in the chain, while 1 is the first?
I have applied the instructions but it does sound a bit weird. If inverting the jed does not cause any damage I might try inverting the jeds.

When i re-read the wiki I found that the Raspberry Pi part was copied directly from the TF328 wiki. And on that card it seems the CPLDs are the other way around in the jtag-chain.
I just noticed the wiki has been updated 8 days ago, but to me it really looks like it still states them in the wrong order? I could be wrong, and i'm sorry for the confusion in that case (compare with the stated order in tf530/bin/README.md).

chue · 21 May 2018, 22:53

CRP: the bus CPLD is the first in the chain.

I just thought I'd chime in on the black screen issue. I was having the same thing until I tried the RC2 binaries from the dev branch. I had given up on my build until I tried these bins. Thanks to MattyMuc for pointing to those.

Now I'm able to get Diagrom to post and run some tests, but randomly hit either an "unimplemented instruction" or if testing memory I get a failed test. I'm still looking into this. I seem to get the same thing when using a simple relocator board with my 68K.

CRP · 21 May 2018, 23:25

I understand that bus is first in the chain and ram second. What I do not get is why to write the first element in the chain with xcs3prog I need to specify the -p 1 parameter, and to write the second element I need to use -p 0. If the wiki did not say otherwise, I would use -p 0 on the bus and -p 1 on the ram.

Anyway, I tried to program the CPLDs inverting the two parameters. In both cases I get a black screen, but:
- bus on 1 and ram blank: green led gets bright and yellow led turns off
- bus on 0 and ram blank: green led stays dim, and yellow stays off

So if I interpret the leds correctly, then bus on 1 should be right. But this is all very much guesswork for me...

CRP · 22 May 2018, 00:47

Also, does anyone else encounter "flakiness" when programming the CPLDs with the Raspberry Pi (model 3B)?
Sometimes it works flawlessly, other times I can't get it to write successfully, lots of fuse errors, verifying multiple times gets different results etc.
So I am not *entirely* sure that whatever I am writing to the CPLDs is correct, apart from JTAG position issues...

TimT · 22 May 2018, 02:06

Hi Everyone

Received my TerribleFire 530 today! started soldering the parts I had. Installed the XC9572XL's with no problem. Then I lingered to long with a CAY16 removing the solder mask. So now the copper traces are covered in solder. So my question is do the traces connect with the 4 pads anyway? if not I will have to remove all the solder and use epoxy.

Also I see the board is marked for component orientation but some of the parts I got from digikey are not marked. is there a diagram that shows the orientation?

Thanks

chue · 22 May 2018, 02:50

Quote:

Originally Posted by TimT

Hi Everyone

So my question is do the traces connect with the 4 pads anyway?

Thanks

That looks ok to me. I have a hi-res rendering of the board and it shows that the 4 pads are connected together. See attached image in the upper left.

TimT · 22 May 2018, 03:14

Quote:

Originally Posted by chue

That looks ok to me. I have a hi-res rendering of the board and it shows that the 4 pads are connected together. See attached image in the upper left.

Thank you chue! anyone have a high-res image of a completed rev3 board? would be helpful for component orientation.

chue · 22 May 2018, 03:58

Well here's mine - As I recall all parts were marked, so I'm unsure where you are having trouble. As a side note, I do see some potential solder bridges on my board, so I will have to check it out.

8 Bit Dreams · 22 May 2018, 06:26

Quote:

Originally Posted by chue

Well here's mine - As I recall all parts were marked, so I'm unsure where you are having trouble. As a side note, I do see some potential solder bridges on my board, so I will have to check it out.

As for me - i see several bridges on ram cpls and also second RAM ic has bridges too

plasmab · 22 May 2018, 14:28

Quote:

Originally Posted by 8 Bit Dreams

As for me - i see several bridges on ram cpls and also second RAM ic has bridges too

Agreed. One of the ram pins is shorting power to ground too. Microscope time. Even a cheap usb one

chue · 22 May 2018, 19:36

I can say that the soldering looks much better under a scope. I would attribute the bridges in the prior image to camera shake.

Here is an image of the chips under a usb scope. The image is a composite of multiple images, therefore you may see color changes. Also because of the compositing, the RAM chips are not all the same size.

I did check the RAM chips and RAM CPLD with a meter and it does not indicate shorts.

chue · 22 May 2018, 19:48

After checking the board again, as indicated above, I decided to do something radical.

I hinted before that I was having trouble with Diagrom when using my 68K with a relocator board. I was also seeing the dreaded "unimplemented instruction" error in this scenario.

I once again tested the 68K + relocator combination to confirm. Diagrom would again randomly show "unimplemented instruction".

After this I pulled out the Diagrom and put in my 1.3 ROM.

I put in the TF530 and was able to boot Workbench, as well as play a couple of games from floppy (Turricane II, Stunt Car Racer, Agony).

The TF530 is not "perfect" with the 1.3 ROM, as sometimes it will boot to a black screen. A simple reset will bring up the "insert disk" image.

I suspect my Diagrom setup is bad - maybe a bad eprom.

In any case, more testing needs to be done. I may try some testing from within the Workbench environment, and/or install the IDE header and Classic Workbench onto compact flash.

8 Bit Dreams · 22 May 2018, 20:44

Quote:

Originally Posted by chue

After checking the board again, as indicated above, I decided to do something radical.

I hinted before that I was having trouble with Diagrom when using my 68K with a relocator board. I was also seeing the dreaded "unimplemented instruction" error in this scenario.

I once again tested the 68K + relocator combination to confirm. Diagrom would again randomly show "unimplemented instruction".

After this I pulled out the Diagrom and put in my 1.3 ROM.

I put in the TF530 and was able to boot Workbench, as well as play a couple of games from floppy (Turricane II, Stunt Car Racer, Agony).

The TF530 is not "perfect" with the 1.3 ROM, as sometimes it will boot to a black screen. A simple reset will bring up the "insert disk" image.

I suspect my Diagrom setup is bad - maybe a bad eprom.

In any case, more testing needs to be done. I may try some testing from within the Workbench environment, and/or install the IDE header and Classic Workbench onto compact flash.

If You still have Your double-wipe CPU socket - it can be the issue. It is better to swap it out to gold plated one with turned pins

nogginthenog · 22 May 2018, 21:49

Quote:

Originally Posted by plasmab

Agreed. One of the ram pins is shorting power to ground too. Microscope time. Even a cheap usb one

I bought a cheap jewelers loupe from aliexpress recently. Great for for inspecting small stuff.

plasmab · 22 May 2018, 23:52

These pins look shorted to me

chue · 23 May 2018, 04:50

That looks like the 2nd RAM chip from the right in my composite image? I checked those with a meter and they are ok. Thanks for looking though!

solidcore · 23 May 2018, 11:49

Chue what flux are you using btw?

chue · 23 May 2018, 13:24

I am using "MG Chemicals 8341-10ML No Clean Flux Paste, 10 ml Syringe," from Amazon: https://www.amazon.com/gp/product/B0...?ie=UTF8&psc=1

plasmab · 23 May 2018, 15:09

Note that i noticed that some CPU PGA sockets (the cheap ones) do not work. They dont connect all the power pins. You can tell these ones because they have a cut out in the centre.

22 May 2018, 02:06	#3246
TimT Registered User Join Date: May 2018 Location: Savannah Ga USA Posts: 32	Hi Everyone Received my TerribleFire 530 today! started soldering the parts I had. Installed the XC9572XL's with no problem. Then I lingered to long with a CAY16 removing the solder mask. So now the copper traces are covered in solder. So my question is do the traces connect with the 4 pads anyway? if not I will have to remove all the solder and use epoxy. Also I see the board is marked for component orientation but some of the parts I got from digikey are not marked. is there a diagram that shows the orientation? Thanks Attached Thumbnails

22 May 2018, 03:58	#3249
chue Registered User Join Date: Feb 2018 Location: United States Posts: 81	Well here's mine - As I recall all parts were marked, so I'm unsure where you are having trouble. As a side note, I do see some potential solder bridges on my board, so I will have to check it out. Attached Thumbnails

22 May 2018, 19:36	#3252
chue Registered User Join Date: Feb 2018 Location: United States Posts: 81	I can say that the soldering looks much better under a scope. I would attribute the bridges in the prior image to camera shake. Here is an image of the chips under a usb scope. The image is a composite of multiple images, therefore you may see color changes. Also because of the compositing, the RAM chips are not all the same size. I did check the RAM chips and RAM CPLD with a meter and it does not indicate shorts. Attached Thumbnails

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Listening to Amiga music with too much treble makes it sound terrible!	Foebane	Amiga scene	4	08 October 2016 06:30
Terrible usb drive performance?	lovinggames	support.FS-UAE	5	11 March 2015 09:06
Broken MASPlayer - Terrible hissing.	SS454	support.Hardware	8	17 November 2013 01:41
Accelerators	pmc	MarketPlace	9	04 November 2009 20:23
Problem with a terrible, simply effect!	Camionsauro	Coders. General	13	19 April 2009 08:53

21 May 2018, 21:14	#3241
CRP Registered User Join Date: May 2018 Location: Milano Posts: 59	So can anyone confirm whether the -p parameter refers to items in the JTAG chain in inverted order, i.e. 0 is the second item in the chain, while 1 is the first? I have applied the instructions but it does sound a bit weird. If inverting the jed does not cause any damage I might try inverting the jeds.

21 May 2018, 22:53	#3243
chue Registered User Join Date: Feb 2018 Location: United States Posts: 81	CRP: the bus CPLD is the first in the chain. I just thought I'd chime in on the black screen issue. I was having the same thing until I tried the RC2 binaries from the dev branch. I had given up on my build until I tried these bins. Thanks to MattyMuc for pointing to those. Now I'm able to get Diagrom to post and run some tests, but randomly hit either an "unimplemented instruction" or if testing memory I get a failed test. I'm still looking into this. I seem to get the same thing when using a simple relocator board with my 68K.

21 May 2018, 23:25	#3244
CRP Registered User Join Date: May 2018 Location: Milano Posts: 59	I understand that bus is first in the chain and ram second. What I do not get is why to write the first element in the chain with xcs3prog I need to specify the -p 1 parameter, and to write the second element I need to use -p 0. If the wiki did not say otherwise, I would use -p 0 on the bus and -p 1 on the ram. Anyway, I tried to program the CPLDs inverting the two parameters. In both cases I get a black screen, but: - bus on 1 and ram blank: green led gets bright and yellow led turns off - bus on 0 and ram blank: green led stays dim, and yellow stays off So if I interpret the leds correctly, then bus on 1 should be right. But this is all very much guesswork for me...

22 May 2018, 00:47	#3245
CRP Registered User Join Date: May 2018 Location: Milano Posts: 59	Also, does anyone else encounter "flakiness" when programming the CPLDs with the Raspberry Pi (model 3B)? Sometimes it works flawlessly, other times I can't get it to write successfully, lots of fuse errors, verifying multiple times gets different results etc. So I am not entirely sure that whatever I am writing to the CPLDs is correct, apart from JTAG position issues...

22 May 2018, 19:48	#3253
chue Registered User Join Date: Feb 2018 Location: United States Posts: 81	After checking the board again, as indicated above, I decided to do something radical. I hinted before that I was having trouble with Diagrom when using my 68K with a relocator board. I was also seeing the dreaded "unimplemented instruction" error in this scenario. I once again tested the 68K + relocator combination to confirm. Diagrom would again randomly show "unimplemented instruction". After this I pulled out the Diagrom and put in my 1.3 ROM. I put in the TF530 and was able to boot Workbench, as well as play a couple of games from floppy (Turricane II, Stunt Car Racer, Agony). The TF530 is not "perfect" with the 1.3 ROM, as sometimes it will boot to a black screen. A simple reset will bring up the "insert disk" image. I suspect my Diagrom setup is bad - maybe a bad eprom. In any case, more testing needs to be done. I may try some testing from within the Workbench environment, and/or install the IDE header and Classic Workbench onto compact flash.

22 May 2018, 23:52	#3256
plasmab Banned Join Date: Sep 2016 Location: UK Posts: 2,917	These pins look shorted to me

23 May 2018, 04:50	#3257
chue Registered User Join Date: Feb 2018 Location: United States Posts: 81	That looks like the 2nd RAM chip from the right in my composite image? I checked those with a meter and they are ok. Thanks for looking though!

23 May 2018, 11:49	#3258
solidcore Registered User Join Date: Sep 2017 Location: birmingham, uk Posts: 548	Chue what flux are you using btw?

23 May 2018, 13:24	#3259
chue Registered User Join Date: Feb 2018 Location: United States Posts: 81	I am using "MG Chemicals 8341-10ML No Clean Flux Paste, 10 ml Syringe," from Amazon: https://www.amazon.com/gp/product/B0...?ie=UTF8&psc=1

23 May 2018, 15:09	#3260
plasmab Banned Join Date: Sep 2016 Location: UK Posts: 2,917	Note that i noticed that some CPU PGA sockets (the cheap ones) do not work. They dont connect all the power pins. You can tell these ones because they have a cut out in the centre.

Currently Active Users Viewing This Thread: 1 (0 members and 1 guests)