English Amiga Board


Go Back   English Amiga Board > Support > support.Hardware > Hardware mods

 
 
Thread Tools
Old 27 September 2017, 01:26   #2001
Akira
Registered User

Akira's Avatar
 
Join Date: May 2001
Location: New York
Posts: 19,265
Oh I think we need as many opinions as possible. As I said, mainly to know if there's anything that could be avoided in the future, not because my particular setup isn't working all the time.

Stedy, what do you think mate?

Quote:
Originally Posted by plasmab View Post
Well unless something even stranger is happening thats what the data sheet says will happen.




I still feel there's a goblin hiding here somewhere. The stability when the IDE is connected is TOO GOOD to be true.


Quote:
Originally Posted by kipper2k View Post
here is the info of the chip...
kipper, can you remove the CF adapter and test the games I uploaded, please? You are the only one running a board with the same chips I have. We need your test.
Akira is offline  
Old 27 September 2017, 01:34   #2002
solidcore
Registered User

 
Join Date: Sep 2017
Location: birmingham, uk
Posts: 547
Well lesson learned from this one then, look the datasheets before buying any ram.

I will sure take this on board when i go utsource some ram this week.
solidcore is offline  
Old 27 September 2017, 01:41   #2003
kaffer
Registered User
 
Join Date: May 2011
Location: Cambridge
Posts: 579
Quote:
Originally Posted by plasmab View Post
Well unless something even stranger is happening thats what the data sheet says will happen. The A10 and A11 pins will be floating so they may randomly give different blocks.

Simple RAM tests will pass unless they checked for repeating blocks. Not sure what the Kaffer tests do that way.
SysTest does a pseudo-random fill of all free RAM, then checks it, so it's wise to repeating blocks.
kaffer is offline  
Old 27 September 2017, 01:42   #2004
plasmab
Banned
plasmab's Avatar
 
Join Date: Sep 2016
Location: UK
Posts: 2,917
Quote:
Originally Posted by Akira View Post
I still feel there's a goblin hiding here somewhere. The stability when the IDE is connected is TOO GOOD to be true.

.

So we have two floating address pins with IDE data passing right by them.... cross talk is v possible here.



Sent from my iPhone using Tapatalk
plasmab is offline  
Old 27 September 2017, 01:45   #2005
plasmab
Banned
plasmab's Avatar
 
Join Date: Sep 2016
Location: UK
Posts: 2,917
Quote:
Originally Posted by kaffer View Post
SysTest does a pseudo-random fill of all free RAM, then checks it, so it's wise to repeating blocks.


Weird. Does it report as repeated or just give a bit error?


Sent from my iPhone using Tapatalk
plasmab is offline  
Old 27 September 2017, 01:47   #2006
plasmab
Banned
plasmab's Avatar
 
Join Date: Sep 2016
Location: UK
Posts: 2,917
Expansion board for the CD32

Random thought... The ram may just be binned as 4K because it failed memory tests at other geometrys

I.e there may be more ram in there


Sent from my iPhone using Tapatalk
plasmab is offline  
Old 27 September 2017, 01:49   #2007
kipper2k
Registered User

 
Join Date: Sep 2006
Location: Thunder Bay, Canada
Posts: 3,838
k,

done some testing...

-CF in, CD32 boots from CF no problem.

- CF out, no CD in - CD32 boots to splash after 14 seconds (normal IDE search delay)

- CF out, boot from my turrican mix CD, CD starts to boot after 5 seconds, main menu appears normally

suggestions, to possibly prevent slowdowns when TF not required, put in a kill switch.

as a thought, if the Vcc rail (which supplied all components had a switch on it and was open would this be enough to kill the entire board and make it invisible. Reason for asking is because on the a1200 i done the entire Vcc rail (including 5v and 3.3v supply) so that it can easily be opened and i am thinking this would make the board invisible to the a1200 (or CD32)

Last edited by kipper2k; 27 September 2017 at 01:57.
kipper2k is offline  
Old 27 September 2017, 01:55   #2008
kaffer
Registered User
 
Join Date: May 2011
Location: Cambridge
Posts: 579
Quote:
Originally Posted by plasmab View Post
Weird. Does it report as repeated or just give a bit error?
Data bit error, it's not clever enough to deduce addressing errors.
kaffer is offline  
Old 27 September 2017, 02:21   #2009
solidcore
Registered User

 
Join Date: Sep 2017
Location: birmingham, uk
Posts: 547
@kipper2k I was thinking about adding this in rev2 of my riser board. Which vcc line(s) do you think it is at a bare minimum? I think last time I looked at the tf328 design it had quite a few. I was hoping for a firmware clock buffer jumper to do this if possible though.

Regards
Chris.
solidcore is offline  
Old 27 September 2017, 02:25   #2010
Akira
Registered User

Akira's Avatar
 
Join Date: May 2001
Location: New York
Posts: 19,265
Kipper, i don't know why you ignore me, could you test the games I uploaded to the FTP without a CF on the thing? Not just any random disc.

One disc "booting" does not dictate works/does not work, deeper/lengthier testing is needed to get to the bottom of the issue, and if you have the same RAM chips I do, there is no way it can perform differently unless we're talking magic and unicorns here, which we might, but without the exact same tests, we cannot determine it.



We need consistent and solid results for Stephen to be able to work with

Last edited by Akira; 27 September 2017 at 05:08.
Akira is offline  
Old 27 September 2017, 06:13   #2011
kipper2k
Registered User

 
Join Date: Sep 2006
Location: Thunder Bay, Canada
Posts: 3,838
I am not ignoring you, without any blank CDs i cannot write a disc so i used a couple of ones i already had, i will have to buy some


Quote:
Originally Posted by Akira View Post
Kipper, i don't know why you ignore me, could you test the games I uploaded to the FTP without a CF on the thing? Not just any random disc.

One disc "booting" does not dictate works/does not work, deeper/lengthier testing is needed to get to the bottom of the issue, and if you have the same RAM chips I do, there is no way it can perform differently unless we're talking magic and unicorns here, which we might, but without the exact same tests, we cannot determine it.



We need consistent and solid results for Stephen to be able to work with
kipper2k is offline  
Old 27 September 2017, 06:50   #2012
supaduper
Registered User
supaduper's Avatar
 
Join Date: Oct 2012
Location: Birmingham UK
Posts: 879
sweating guys I needed to check the Samsung ram...and its.......wait for it.........1k.....Phew

So to comfirm by the data sheet the Samsung K4E151612C-TC60 from utsource is 1K

Click image for larger version

Name:	TF328 samsung ram 1k.png
Views:	59
Size:	127.5 KB
ID:	54776


I see there is an inquiry option for the 5v 1k ram Number :Samsung K4E151611C-TC60 (as per data sheet its TSOP) so asked utsource for a quote

Data Sheet attachment below

SAMSung tf328 ram-S-A0001037904-1.pdf

Last edited by supaduper; 27 September 2017 at 07:04.
supaduper is offline  
Old 27 September 2017, 09:12   #2013
plasmab
Banned
plasmab's Avatar
 
Join Date: Sep 2016
Location: UK
Posts: 2,917
Quote:
Originally Posted by Akira View Post
Kipper, i don't know why you ignore me, could you test the games I uploaded to the FTP without a CF on the thing? Not just any random disc.

We need consistent and solid results for Stephen to be able to work with
I dont keep many blank DVDs and CD-Rs are a thing of the past. @akira nobody is ignoring you or things that you are ungrateful but you do sound like you are ordering people about a bit. Patience please.

I dont need any more results as I have a solid understanding of whats happening now with the HY ram. I am 99% convinced that these chips started life as 4096x1024 columns x rows and then got binned to different sized when defects were found. 1K or 4K refresh depending on yield.

I can add patch wires to my muxcpld and test easily enough but wife leaves for India tomorrow morning and I'm going to struggle to find that time. Alternatively i can make up a firmware that will use the existing setup as 2Mb for HY RAM. THat will take me 10 mins.
plasmab is offline  
Old 27 September 2017, 09:25   #2014
plasmab
Banned
plasmab's Avatar
 
Join Date: Sep 2016
Location: UK
Posts: 2,917
Hastely thrown together CPLD jeds before breakfast..

https://www.dropbox.com/s/vdyz80w1os...BTest.zip?dl=0

Should give reliable 2Mb. You need to program the MUXCPLD also.

EDIT: Ops i didnt modify the decode addresses so it wont work. Will sort later

Last edited by plasmab; 27 September 2017 at 10:25.
plasmab is offline  
Old 27 September 2017, 09:31   #2015
supaduper
Registered User
supaduper's Avatar
 
Join Date: Oct 2012
Location: Birmingham UK
Posts: 879
Quote:
Originally Posted by plasmab View Post
Hastely thrown together CPLD jeds before breakfast..

https://www.dropbox.com/s/vdyz80w1os...BTest.zip?dl=0

Should give reliable 2Mb. You need to program the MUXCPLD also.
Awesome ,Hopefully Alen or Brian can test

I also had words with the `Ram Fairy` and told her to get her friggin wings in order !
the result

Click image for larger version

Name:	snow-fairyRam2.jpg
Views:	151
Size:	78.5 KB
ID:	54779
supaduper is offline  
Old 27 September 2017, 10:26   #2016
plasmab
Banned
plasmab's Avatar
 
Join Date: Sep 2016
Location: UK
Posts: 2,917
Expansion board for the CD32

I made a mistake with the firmware fix. to be fair i did it in under 90 seconds so it was rushed.

@Akira regardless of what you "feel" is wrong with the TF328 the simple fact is that we are operating the Hynix RAM outside of what the data sheet says will work. So results will be unreliable. Which is what we are seeing.

The fact that IDE plugged in or not makes a difference for you does not conclusively indicate electrical issues. It could simply be that the presence of IDE allocates RAM for buffers in such a way that the bad RAM is not hit.

Last edited by plasmab; 27 September 2017 at 11:08.
plasmab is offline  
Old 27 September 2017, 11:23   #2017
solidcore
Registered User

 
Join Date: Sep 2017
Location: birmingham, uk
Posts: 547
@akira lets just take a good breath and realize that Stephen has the technical knowledge and built the damn thing, also let us all remind ourselves this is free and open source - the fact Stephen has a communication channel isn't for our benefit but for his own and we have to respect the manner in which he'll want us to report our issues. And between you and me mate - I know what it can be like as I have the same conversation with my dad who has over 40 years of experience in electronics, I get shouted at a lot but in the end you have to take a step back and realize there's more to it than you know - of course, it doesn't hurt to learn more about it reading and researching to get on the same line as everyone else but it's their profession and they will know more until we can demonstrate we get what we're talking about hehe.

@plasmab it's all good buddy you've nailed the issue and we look forward to some resolution for kipper/akira's ram I'm sure akira isn't trying to boss anyone around just feels like he's helping but I can get why it may come across that way as people set their own unrealistic expectations of others and like myself at times can inject ideas that don't quite add up to reality - woops!, but after all Akira is now a tf328 tester - WOOHOO! :P and just wants to help, which is what you wanted lots of testers you sure do have all that support now mate/ I myself want to help as much as I can

@Everyone I think all of us must not alienate each other tester or programmer alike and come to some harmony whereby if at all a problem exists, report it plain and simply on the github as Stephen asked and wait patiently for an answer - We're all excited and it's only going to get better with improved WHDLoad support to come and firmwares for the tf328 , new revisions and all that beautiful soldering ahead... we can't thank stephen enough for it, now we can stop watching youtube videos and have fun ourselves .

What a time to be alive guys Amiga in 2017.
solidcore is offline  
Old 27 September 2017, 11:32   #2018
plasmab
Banned
plasmab's Avatar
 
Join Date: Sep 2016
Location: UK
Posts: 2,917
I'm happy. Just trying to explain the route I'm going. Ram type A works no issues (even with the wrong voltage.). Ram type B does not and the data sheet for it clearly shows why.

My comment about Akira was for his benefit. Its easy not to notice how things come across in the excitement.


Sent from my iPhone using Tapatalk
plasmab is offline  
Old 27 September 2017, 11:48   #2019
supaduper
Registered User
supaduper's Avatar
 
Join Date: Oct 2012
Location: Birmingham UK
Posts: 879
@ Chris wise words young man

And its great we have all learned something from this situation and its great.people all joining in and helping, its exactly how it should be and supporting Steve at the same time, Data Sheets are becoming even less of a mystery to me lol, so we are all moving forward.

One thing we all agree on the Terriblefire TF328 is F@@##+G Awesome
supaduper is offline  
Old 27 September 2017, 13:03   #2020
plasmab
Banned
plasmab's Avatar
 
Join Date: Sep 2016
Location: UK
Posts: 2,917
I just checked the datasheet for the AS4C1M16E5-60TC..

Its a 1K refresh so it will likely work.
plasmab is offline  
 


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

Similar Threads
Thread Thread Starter Forum Replies Last Post
Hello and help identifying my expansion board! edd_jedi New to Emulation or Amiga scene 7 04 April 2010 13:52
Can anyone identify this RAM expansion board?? PowerPie5000 support.Hardware 66 02 August 2009 14:04
WTB: A1200 Expansion Board robheaton MarketPlace 17 11 June 2009 10:49
an expansion board toolkit --might be useful to you giannis_vassili support.Hardware 1 14 June 2008 21:51
Wanted: A1200 expansion board Freckle MarketPlace 0 30 August 2004 17:36

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off

Forum Jump


All times are GMT +2. The time now is 13:53.


Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2019, vBulletin Solutions Inc.
Page generated in 0.13688 seconds with 14 queries