BBC Master screen interference PiTube direct

for bbc micro/electron hardware, peripherals & programming issues (NOT emulators!)
User avatar
hoglet
Posts: 6695
Joined: Sat Oct 13, 2012 6:21 pm
Location: Bristol

Re: BBC Master screen interference PiTube direct

Postby hoglet » Wed Dec 13, 2017 9:15 pm

Could someone who has this problem try typing in the following program into the 6502 Co Pro:

Code: Select all

10 MODE 0
20 FOR X=0 TO 1279 STEP 2
30 MOVE X,0
40 DRAW X,1023
50 NEXT
60 GOTO 60
RUN

It will make the screen completely white.

I'd like to see if the interference becomes very noticeable.

Dave

User avatar
BigEd
Posts: 1547
Joined: Sun Jan 24, 2010 10:24 am
Location: West
Contact:

Re: BBC Master screen interference PiTube direct

Postby BigEd » Wed Dec 13, 2017 9:22 pm

(Does this work too:

Code: Select all

10 MODE 0
20 GCOL 0,129
30 CLG
40 GOTO 40
?)

User avatar
hoglet
Posts: 6695
Joined: Sat Oct 13, 2012 6:21 pm
Location: Bristol

Re: BBC Master screen interference PiTube direct

Postby hoglet » Wed Dec 13, 2017 9:29 pm

BigEd wrote:(Does this work too:

Code: Select all

10 MODE 0
20 GCOL 0,129
30 CLG
40 GOTO 40
?)

Yes, that would have the same effect, filling screen memory with 0xFF bytes.

dp11
Posts: 716
Joined: Sun Aug 12, 2012 8:47 pm

Re: BBC Master screen interference PiTube direct

Postby dp11 » Wed Dec 13, 2017 9:42 pm

So I think we might need to gate nTube with phi2 to drive oe of the level converter.

My level converter is passive resistors and diodes I will do some tests to see if the problem occurs with this version.

User avatar
marcusjambler
Posts: 147
Joined: Mon May 22, 2017 11:20 am
Location: Bradford
Contact:

Re: BBC Master screen interference PiTube direct

Postby marcusjambler » Thu Dec 14, 2017 8:31 am

From what I've seen the issue is most pronounced on the x86 but is also present on the other processors that I've tried.
There is evidence of it in the 'new zx81' emulator by ROBC using the the native ARM proc, with the issue being inverted.
i.e. the artifacts are white on black.

User avatar
marcusjambler
Posts: 147
Joined: Mon May 22, 2017 11:20 am
Location: Bradford
Contact:

Re: BBC Master screen interference PiTube direct

Postby marcusjambler » Thu Dec 14, 2017 6:48 pm

Both these short basic programs produce the following :
Attachments
IMG_0446a.jpg

User avatar
marcusjambler
Posts: 147
Joined: Mon May 22, 2017 11:20 am
Location: Bradford
Contact:

Re: BBC Master screen interference PiTube direct

Postby marcusjambler » Thu Dec 14, 2017 6:50 pm

shot of ZX81 emulator :
Attachments
IMG_0441a.jpg

RobC
Posts: 1850
Joined: Sat Sep 01, 2007 9:41 pm

Re: BBC Master screen interference PiTube direct

Postby RobC » Thu Dec 14, 2017 11:34 pm

marcusjambler wrote:There is evidence of it in the 'new zx81' emulator by ROBC using the the native ARM proc, with the issue being inverted. i.e. the artifacts are white on black.

That'll be because the ZX81 emulator uses a mode 4 screen with an inverted palette (so that '0' is white and '1' is black in the main display).

crj
Posts: 444
Joined: Thu May 02, 2013 4:58 pm

Re: BBC Master screen interference PiTube direct

Postby crj » Fri Dec 15, 2017 12:25 am

BigEd wrote:(Does this work too:

Code: Select all

10 MODE 0
20 GCOL 0,129
30 CLG
40 GOTO 40
?)

Code: Select all

10 MODE 0
20 COLOUR 129
30 CLS
40 REPEAT UNTIL FALSE
...?

(CLS should be way quicker than CLG, and I see no excuse for a GOTO in this case. (-8 )

User avatar
hoglet
Posts: 6695
Joined: Sat Oct 13, 2012 6:21 pm
Location: Bristol

Re: BBC Master screen interference PiTube direct

Postby hoglet » Fri Dec 15, 2017 11:08 am

Hi all,

Apologies for the long post, but I want to make sure we all understand this issue (and it's ramifications) properly.

I spent a bit of time with Ed looking at this yesterday. Neither of us actually have one of Kjell Sundby's Internal Master level shifter boards, nor have we ever used the internal tube interface, so we had to improvise...

This is my original hand etched level shifter connected to the Master's internal tube interface with patch wires:
IMG_1151.JPG

Doing this, I was clearly able to replicate the problem:
IMG_1154.JPG

There's a good diagram in the Master service manual that helps understand what's going on:
master_dram_timing.PNG

In each 2MHz clock cycle there are two accesses to the DRAM. When Phi2 is low the Video Processor reads the DRAM, and when Phi2 is high the CPU reads/writes the DRAM.

What's different on the Master is that the CPU, Video Processor, DRAM, Internal Tube (and Cartridge port) all share the same data bus. So it;s important that any peripheral responding to a CPU read only drive the data bus when Phi2 is high. Kjell's internal level shifter uses the same circuit as the external one, and so is not respecting this.

To test this theory out, we need to generate an output enable signal for the data bus buffers that is active only when Phi2 is high. The additional gates to do this are shown in the green box.
IMG_1160.JPG

Here's this what this looks like in my improvised test system (the extra chip is a plain old 74LS00):
IMG_1158.JPG

And here's the end result:
IMG_1155.JPG

The bottom line here is that the internal level shifter needs to be re-designed to include this modification. Probably the easiest way to do this is using a 74LVC1G98. This is a tiny 6-pin chip containing the following gates:
1G98.PNG

To get the required logic function for OE, you would connect:
- A (pin 3) to 0v
- B (pin 1) to Phi2
- C (pin 6) to INTUBE
- Y (pin 4) to OE on the data bus 74LVC245A

(as per figure 8 in the data sheet)

Now the important bit....

In the mean time, I think it's wise not to use the current internal level shifter. The 74LVC245A is quite a strong modern driver, and can source/sink at least 24mA. During the periods of contention, this will be putting quite a strain on the Master's TMM41464 DRAM chips, which could result in premature failure.

Sorry to be the bearer of bad news....

Dave

dp11
Posts: 716
Joined: Sun Aug 12, 2012 8:47 pm

Re: BBC Master screen interference PiTube direct

Postby dp11 » Fri Dec 15, 2017 11:22 am

I've been thinking about this a bit more.

I would like the level shifters to always have nOE tied low. The DIR pin then only during the correct phase of the clock and if nTUBE is low and Read is high will DIR be set such that it drives the Masters Databus. The reason for this is that the databus can then be sniffed by the Pi all the time.

User avatar
marcusjambler
Posts: 147
Joined: Mon May 22, 2017 11:20 am
Location: Bradford
Contact:

Re: BBC Master screen interference PiTube direct

Postby marcusjambler » Fri Dec 15, 2017 1:14 pm

Great work guys =D>

I'll stick with the external one until theres a new design.
Hopefully there may be a way to mod the current internal boards [-o<

Marcus

pbw72
Posts: 23
Joined: Tue Mar 14, 2017 8:38 pm

Re: BBC Master screen interference PiTube direct

Postby pbw72 » Fri Dec 15, 2017 2:44 pm

Wow, great work indeed =D>

As advised, I will remove the pitube for the moment as I don’t want to cause any damage. I do not have an external version so hopefully there is a way to adapt the internal one (I would happily purchase an add-on modification if such a thing could be created?)

I must say though I am somewhat dismayed that these have been sold without (what appears to be) enough research and development and could potentially damage our machines.

I can only assume that this is what happens to anyone using the internal version?

Fingers crossed a solution can be found for existing internal boards. Since you guys don’t have an actual board, If it helps I can send my board to you to see if anything is possible?

Again great work guys, I salute you

User avatar
BigEd
Posts: 1547
Joined: Sun Jan 24, 2010 10:24 am
Location: West
Contact:

Re: BBC Master screen interference PiTube direct

Postby BigEd » Fri Dec 15, 2017 2:56 pm

Wouldn't want to downplay the warning, but the potential damage is only a hypothesis - better safe than sorry.

We hadn't realised, Dave and I, that neither of us had ever hooked up our various level shifter options to the internal tube port of the Master. (We don't know what testing Kjell might have done.) It's interesting that the video disturbance has only now come to light - as far as we can tell, it would affect everyone and is pretty obvious, so we're thinking perhaps not many machine-hours of internal Tube use have been racked up yet. Or perhaps there are differences between Masters which we're unaware of.

sundbyk
Posts: 79
Joined: Thu Jun 16, 2016 10:03 am

Re: BBC Master screen interference PiTube direct

Postby sundbyk » Fri Dec 15, 2017 3:31 pm

Hi,

When designing the first version of the master card I had the same problem, but I thought it was noise on the datalines.
Because when I added the inline 33ohm resistors on the data lines the noise was gone.
I will change the design and make new cards.

I will see what we can do with replacing the cards.

Kjell S.
Last edited by sundbyk on Tue Dec 19, 2017 7:04 pm, edited 1 time in total.
***************************
Kjell Sundby
Web: http://www.sundby.com
****************************

User avatar
fordp
Posts: 925
Joined: Sun Feb 12, 2012 9:08 pm
Location: Kent, England

Re: BBC Master screen interference PiTube direct

Postby fordp » Fri Dec 15, 2017 3:31 pm

I have have seen this on my Master with internal PI CoPro.
FordP (Simon Ellwood)
Time is an illusion. Lunchtime, doubly so!

User avatar
fordp
Posts: 925
Joined: Sun Feb 12, 2012 9:08 pm
Location: Kent, England

Re: BBC Master screen interference PiTube direct

Postby fordp » Fri Dec 15, 2017 3:33 pm

sundbyk wrote:Hi,

When designing the first version of the master card I had the same problem, but I thought it was noise on the datalines.
Because when I added the inline 330ohm resistors on the data lines the noise was gone.
I will change the design and make new cards.

I will see what we can do with replacing the cards.

Kjell S.

I suggest you come up with a suitable modification and post the details on your website. There is plenty of room on that board to attach extra stuff.
(Maybe use two gates of a 74HCT00, Intube on pin 12, Phi2 on pin 9, Strap 13 to 14, 10 to 11 and pin 8 goes to OE on the data bus '245)
Last edited by fordp on Fri Dec 15, 2017 4:03 pm, edited 1 time in total.
FordP (Simon Ellwood)
Time is an illusion. Lunchtime, doubly so!

pbw72
Posts: 23
Joined: Tue Mar 14, 2017 8:38 pm

Re: BBC Master screen interference PiTube direct

Postby pbw72 » Fri Dec 15, 2017 3:34 pm

I would prefer to err on the side of caution, so ‘potential’ is good enough for me!

I guess the more people use it the more is learnt but i am surprised this hasn’t come up before. As most may know, there are two versions of the master mobo, issue 1 & 2, mine is an issue 1.

The only difference I know of between the two is the link for the composite connection to make it display colour (I’m sure there are others?)

This has bummed me out slightly, re-visiting the master after years away and finally being able to explore it more fully than when I was a teenager was great (still working on carving out a space in the house for a man cave :lol: )

As I said, if you guys need to borrow the board to see if there is a way to mod it then let me know.

Here’s hoping for a solution (I wish I was more technically gifted with this stuff :( )

Paul

pbw72
Posts: 23
Joined: Tue Mar 14, 2017 8:38 pm

Re: BBC Master screen interference PiTube direct

Postby pbw72 » Fri Dec 15, 2017 3:38 pm

sundbyk wrote:Hi,

When designing the first version of the master card I had the same problem, but I thought it was noise on the datalines.
Because when I added the inline 330ohm resistors on the data lines the noise was gone.
I will change the design and make new cards.

I will see what we can do with replacing the cards.

Kjell S.



Kjell - That would be very welcome indeed, I do think this is a great piece of kit you have made and applaud you time and effort.

User avatar
BigEd
Posts: 1547
Joined: Sun Jan 24, 2010 10:24 am
Location: West
Contact:

Re: BBC Master screen interference PiTube direct

Postby BigEd » Fri Dec 15, 2017 3:55 pm

Ah - so do all your internal adapters have inline resistors Kjell? That would much reduce the current from contention and also shift the net voltage. At which point, whether or not the noise is visible will depend on the input thresholds of the video ULA (probably) - which might vary from part to part.

Edit: it's starting to seem likely to me that only a proportion of Masters will suffer from this video noise. That's a better situation than we thought, only yesterday.

Edit: reports from Marcus, on two Masters, from Paul and from fordp, on one each. Total of four machines.

User avatar
marcusjambler
Posts: 147
Joined: Mon May 22, 2017 11:20 am
Location: Bradford
Contact:

Re: BBC Master screen interference PiTube direct

Postby marcusjambler » Fri Dec 15, 2017 4:43 pm

Edit: reports from Marcus, on two Masters, from Paul and from fordp, on one each. Total of four machines.


One of my masters has a videoNula upgrade.

Actually... thinking about it. The VideoNula master was working fine with the internal tube processor and only began showing these symptoms when the Video Nula was installed... So, chances are that the extracted chip release may give an indication.
Its one of these :
Attachments
IMG_0455a.jpg

User avatar
hoglet
Posts: 6695
Joined: Sat Oct 13, 2012 6:21 pm
Location: Bristol

Re: BBC Master screen interference PiTube direct

Postby hoglet » Fri Dec 15, 2017 5:33 pm

Hi Kjell,
sundbyk wrote:Because when I added the inline 330ohm resistors on the data lines the noise was gone.

Can you confirm whether the resistors are 330 ohm, or 33 ohm as indicated on the schematic I just found:
https://easyeda.com/sundbyk/PITubeDirec ... -xRPpKI4EC

If 330 ohm, I'm a bit less concerned about the bus conflicts causing the DRAM stress, as these will limit the current to 15mA.

Dave

sundbyk
Posts: 79
Joined: Thu Jun 16, 2016 10:03 am

Re: BBC Master screen interference PiTube direct

Postby sundbyk » Tue Dec 19, 2017 7:08 pm

Hi Dave,

The resistors are 33 ohm and not 330 as I previously said.
I have tested my master and it does not have this problem.

Kjell S.
***************************
Kjell Sundby
Web: http://www.sundby.com
****************************

sundbyk
Posts: 79
Joined: Thu Jun 16, 2016 10:03 am

Re: BBC Master screen interference PiTube direct

Postby sundbyk » Tue Dec 19, 2017 7:11 pm

There are a problem with snow effect on some BBC Master's using the internal PiTubeDirect Signal Converter for Master.
The ones connected to the external tube works fine.
It seem like there are different version of chips in the Master causing the problem, but this need to be investigated more.

Will investigate and find a solution for this and hope to get some with help from you guys. There may be more that needs to be fixed.

All people that have bought the Master version and have this problem will get a new card.
Since this is a "hobby" I would just ask to get back the cost of shipping the replacement.
I estimate to have new cards ready some time in January for shipping.

Hope nobody gets any problems with their Masters after this.

Kjell S.
***************************
Kjell Sundby
Web: http://www.sundby.com
****************************

pbw72
Posts: 23
Joined: Tue Mar 14, 2017 8:38 pm

Re: BBC Master screen interference PiTube direct

Postby pbw72 » Tue Dec 19, 2017 10:08 pm

Hi Kjell,

Many thanks for revisiting this issue and hopefully providing a solution via a new card. No problem with covering the postage.

If worst case scenario happens and the internal version still has problems, is there a way to use this version of the Pi externally? Or is it a complete new piece of kit?

Thanks again for your efforts :D

Paul

User avatar
IanS
Posts: 494
Joined: Mon Aug 31, 2009 6:02 pm

Re: BBC Master screen interference PiTube direct

Postby IanS » Tue Dec 19, 2017 10:29 pm

Seem a shame to have to produce completely new boards, can the existing boards not be modified. Is the change just the additional couple of gates listed earlier in the thread by hoglet. How easy is it cut the Phi2 line on the board, is it accessible?

If it just requires the addition of a 74ls00 (or 74LVC1G98), can a few people offer to do a few each, to avoid a potential loss for what is a "hobby" scheme.

(I have a pile of 74LVC1G58, which can't be used directly, would need to use two to invert the output, would the timing still work?)


Return to “hardware”

Who is online

Users browsing this forum: No registered users and 7 guests