Superbrain #3 : The Superbrain II board

[jonb]

Using a hastily cobbled together partial implementation (just the LS86, no resistors) of Mark's sync mixer fed into one of my own video mixer boards, I can see the SBII board is trying to display some video.

Flickering like crazy but the boot prompt is visible.

From this I can draw a few conclusions:

• Video chip is OK, able to read from the data bus.
• Character ROM is probably OK
• Video interrupt service routine is getting called, and it's working

This is very unusual for a Superbrain (in my experience) where there are RAM problems, because the boot code normally gets corrupt after being copied to Bank 3.

Updated to say: The flickering was caused by the boot ROM setting 60Hz mode. This can be reset by the ICE using simple OUT 68h, 20h. I'm now seiing this:

It's rolling slowly from bottom to top, but way more legible.

[jonb]

On a hunch I decided to try connecting the SBII board to the floppies on my Superbrain, as one of the first things I noticed was that on boot it appeared to be waiting for a ready signal. What if the drives I'd plugged into it didn't generate a ready? This sounded similar to the problems you get when interfacing 3.5" drives to Amstrad computers that are quipped with 3" floppies.

I'm surprised to say it actually booted! So perhaps there is less wrong with this board than I feared. I can't see much on the screen though; this is because the video mixer circuit is not working and think the boot disk I am using is a SBI disk (and hence, would not set the video chip up correctly). I can see that if I do DIR and look in the screen RAM, there is a directory listing (corrupted). So, it has booted into CPM and can access the disk..

[jonb]

After success with the Superbrain 1 drives, I've switched over to the HxC emulator and tried all the disk images I have. I found one that appears to work, albeit in reverse video.

The screen can be altered to normal video with an escape sequence but the image collapses, maybe because I am not using any of the discreet components in Mark's VSYNC-HSYNC to CSYNC design. It is pretty solid like this; the only problem is I cannot see the top or bottom line and there is no adjustment on the TV I'm using.

[1024MAK]

The resistors are included to make interfacing more compatible with existing equipment. Not all are needed in every application.

If you stick the Hsync, Vsync and final composite video signal into a ‘scope, it should be possible to see what, if any tweaking is needed. Or it may just be the levels going into your mixer / combiner. The logic family chip used also can affect things (74LS / 74HC / 74HCT ).

Mark

[jonb]

Well.. first off the logic gates are all LS7400 series. No HC HCT or other types. I feed the output of the H/Vsync mixer into the CSync of my composite mixer. With the resistors in place it makes no difference (still have signal collapse when we go to white letters on a black background).

I did have issues when I prototyped my own video mixer on breadboard. Perhaps I should implement on Veroboard for shorter paths and better connections? The SB mixer looks like this:

The blue board in the middle is my proven PCW CSYNC/Video to Composite mixer.

Lots of scope for noise here.. maybe that is the problem?

[jonb]

Scope traces: Cyan=VSYNC, Magenta=HSYNC, Blue = CSYNC

Zoomed out

Closer look..

There's some noise on the VSYNC, but - in general - is this what you'd expect to see?

[hoglet]

There's some noise on the VSYNC, but - in general - is this what you'd expect to see?

I think the problem is that VSYNC is active low, and HSYNC is active high, so CSYNC is a bit confused.

Edit: Can you invert CSYNC?

Dave

[jonb]

Probably...

There are 2 spare gates on the 74LS86. I need to read the datasheet!

Other thing is if I disconnect either sync or the CSYNC it doesn't alter the picture (so for a while I thought the '86 wasn't working).

[1024MAK]

I agree with Dave. The line / horizontal sync looks inverted. If you have a look at the more complex version of my sync combiner (same thread as the link I gave earlier), the two gates along with the resistors and jumper shunts / links enable you to select whether that gate inverts the input signal (or not).

Mark

[jonb]

Nice one guys!

I built the full on circuit and after setting JP3 got this:

Looking closer the picture isn't great, but I put that down to using breadboard.

Returning to the first picture, there are still problems:

• Top line missing
• Half of the bottom line is missing
• The monitor sometimes struggles to lock onto the signal (despite me fiddling with the trimmer on the little composite board)

Mark, you mentioned on your other thread that the position and size of the output could be adjusted. I guess that is not possible with this setup and one would need the GBS-8200 board as well? I do have one lying around but I was saving it for the Amiga 1200.

[jonb]

And so...

This is the plan: Use the PCW 9512 CRT and video circuit pictured as a monitor and enclose it and the SBII board / PSU in a home made wooden case. The CRT deflection is a bit off but I should be able to fix that with a lot of fiddling (there's no real adjustment apart from the magnets on the yoke). Alternatively, if I could find a mono CRT monitor somewhere, I could use that.

Either way, things are looking up for the SBII!

[1024MAK]

Returning to the first picture, there are still problems:

• Top line missing
• Half of the bottom line is missing
• The monitor sometimes struggles to lock onto the signal (despite me fiddling with the trimmer on the little composite board)

Mark, you mentioned on your other thread that the position and size of the output could be adjusted. I guess that is not possible with this setup and one would need the GBS-8200 board as well? I do have one lying around but I was saving it for the Amiga 1200.

That was by using the settings / controls on the GBS-8200 board.

With all analogue video systems, keep in mind that television (4:3 aspect ratio) always has an image that is biggger than the available screen area, so that the viewer does not see any black borders. So all analogue CRT sets have appropriate technicans controls (and sometimes user controls) to set it up like this (assuming the received signal is within the specification).

But for computer monitors, the size of the generated video picture should either be the exact right size, or have border areas. But not all computer manufacturers stayed within the specification for broadcast video. So the monitor has to be matched to the system. Most CRT monitors also have appropriate technican controls or user controls, or both. But most LCD displays have limited controls, as they were designed to work with signals that comply with broadcast video specifications.

So if the display you are using does not have enough range in the controls available, the first step is always to try another display. The only other solutions are to see if the timing of the computers video system can be changed, or to use a scan converter like a GBS-8200.

Note that some LCD displays don’t display the top line or bottom line of a BBC or Master. And nearly all don’t display the left hand and right hand side of the image generated by a QL in monitor mode.

Mark

[jonb]

Hi Mark

Seems like the main issue with this is the Superbrain's video frequency. I couldn't find a way to get it to save the 50Hz setting to the boot image.

Until now.

I found a Superbrain configuration utility amongst the few images I have that is compatible with the only SB II boot disk image, altered the scan frequency (to 50Hz) and it is working fine (also, saved the setting to the boot track).

So, now that I have a booting SB II board, HxC emulator, correct boot disk and CRT (previous post), I am nearly ready to start building a case for it. There's one more thing to do, and that is to connect the main serial port up. On a complete SB II there is a small (so-called "intermediary") board which is an adapter for the big 50 way header on the main board and the two RS232 ports. I'll have to recreate it.

I guess I'd better order a new battery for the RTC as well.

[hoglet]

This all looks very nice!

One question....

Early on in this thread you reported possible memory problems.
viewtopic.php?p=224672#p224672

Did you end up replacing the DRAM?

Or is the memory test in your ICE now suspect?

[1024MAK]

@Jon

Mark

[Talywain]

Well done on getting the board up and running. I’m pleased to know it’s gone to a good home.

DDT and PIP, it’s been a while since I saw those acronyms, that takes me back a bit.

I’m sorry to report that I finally got to the back of my attic without finding any more goodies that would be useful.

I wonder if in another 20 years people will be lovingly restoring 386 computers? Personally I can’t see it happening, but then again, it wouldn’t be the first time I was wrong.

[jonb]

This all looks very nice!

One question....

Early on in this thread you reported possible memory problems.
viewtopic.php?p=224672#p224672

Did you end up replacing the DRAM?

Or is the memory test in your ICE now suspect?

I think it's the ICE

By the time I realised that, I'd already socketed up Bank 3 and spent time swapping DRAMs about. I only found this out after I'd played a hunch about the disk ready wait loop (attaching an HxC to it, which I should have done from the start). Suddenly it booted and I had to cobble together a display for it. I think now that there was never anything wrong with it and I'd been deceived by the (apparently) defective ICE.

@Jon

Mark

Not so fast...!

Further testing indicates that the image is stretched wider than the LCD is expecting, so while it looks like a good display, the right most two characters cannot be seen. Moreover, the CRT monitor (part of an Amstrad 9512 I have in pieces) is not able to display the full image, either. In the screen shot I posted, the program running is centre justified, so edge display problems are not visible. The Amstrad CRT circuit has no adjustability for image width or position (although it has V-Height, V-Lin, V-Hold and H-Hold). The machine is usable, but on the LCD monitor only. Now I need to find a monochrome monitor, somewhere...

I do have the guts of an Amstrad GT65 here (that's the old green screen monitor) but it is broken, apparently. I bought it for the tube, to repair an 80 column PET. Hmm, another rabbit hole!

[hoglet]

I think it's the ICE

By the time I realised that, I'd already socketed up Bank 3 and spent time swapping DRAMs about. I only found this out after I'd played a hunch about the disk ready wait loop (attaching an HxC to it, which I should have done from the start). Suddenly it booted and I had to cobble together a display for it. I think now that there was never anything wrong with it and I'd been deceived by the (apparently) defective ICE.

Interesting....

I think it's very hard to build an ICE that is 100% guaranteed to work in every system. Particularly when the timing of many original designs is quite close to the edge.

My weapon of choice now for diagnosing a broken machine is to capture a 16-bit logic analyzer trace (data bus + misc control) of the reset sequence. This can then be decoded back into instructions (using the decode6502 or decodez80 C programs). The reason being that this is pretty much non-intrusive (just one extra load on each signal). It's usually pretty obvious from this where the problem lies.

Dave

[jonb]

Interesting, Dave. The ICE works on the Superbrain I (not tested it recently) so I would have thought it'd be OK on the II. Still, as you say, the timings may be different. It looks a bit like my 8080ICE, which cannot read data reliably on the MDS-230 I talked about a while back. You give me hope (that both are fine, but upset by the systems they are in).

Now, back to the monitor issue. Recall I said I thought the GT65 guts were allegedly broken. Not so:

Banzai!

It's a bit hummy, but the GT65 PCB is more compact than the PCW's and has more adjustment capability. We will soon see if it can be driven by the Superbrain. The 5v output is still working, too, but I think it doesn't have enough power to run the drive motors (this 6128 is modded for 5v only operation).

(About this time I am wishing I had a DIN plug that fitted its flying lead!)

[jonb]

Some judicious bodgery later...

It's still too far to the left and none of the trimmers have made a difference. There's a single ferrite cored choke that moves the image left to right but it hasn't got enough range to correct this. It's a real pity because (as before) the centre justified config program looks grand:

I'm now wondering if the GT65 board can be modded to provide some more adjustability. I have the schematics... just need some advice!

[jonb]

In the hope that someone has an idea, here is the section of the GT65 board where the choke is located (circled in red):

Moving this choke has a small effect on the left - right positioning of the image.


The corresponding section on the Superbrain's video board is shown:

The trimmer adjusts the image centre position. So I'm wondering if the H-SYNC ("drive") is being used to set the timer, and the GT65 circuit is doing something functionally similar. Maybe adjusting one or more of those resistors near the choke (R719, R720, R721) would have the right effect?

[jonb]

Hmm, possibly not R271. Here it is at ~68 ohms, down from 82.

..perhaps using trial and error here is not such a good use of time..

[jonb]

Alternatively, maybe I can alter the SBII video circuit to give a more compliant output.

Pardon my ignorance. Here are two monostables acting on the HSYNC and VSYNC signals.

If I alter the timing components of one of them (R18/C11 for horizontal, R19/C12 for vertical), might I be able to alter the image positioning, or perhaps the "stretch"?

[1024MAK]

Jon, it’s the relationship between the horizontal sync pulse and the actual video data in the video signal.
If the video data is being sent too soon after the horizontal sync, unless you can alter the circuit to produce a slightly earlier sync (or maybe a slightly shorter sync pulse), no messing with it will sort out the problem.

Of course, if you do manage to get an earlier or shorter sync, the next problem could be the loss of the image on the right hand side of the screen...

So in the Superbrain II, where does the horizontal sync originate from?

Mark

[jonb]

Both sync signals are generated by an 8350 CRTC. I've attempted to capture the schematic with my phone. The resolution is good enough to read from.

SB II Mainboard Schematic

It looks like the CRTC (Z57) has everything hard wired. No adjustability without modification. I would have tried this board in the Superbrain 1 chassis (with its CRT display) but the connectors are different and I'm loath to dismantle it again - it likes to breakdown. Maybe a good idea could be to swap the CRTCs over, but I would have thought that if the SBII one was faulty in some way, it'd not generate an image at all.

When a CRT monitor adjusts an image for width, what is it doing? At a guess, I'd say speeding up the horizontal scan (to make the image narrower) or slowing it (to make the image wider). Is that correct? If so, somewhere on the GT64 video board is a group of components that sets this scan speed. It's hard wired but there must be a way of making it adjustable...

[jonb]

Hmm, I may have found it. H-POS is indeed adjusted by the choke at L703. H-WIDTH appears to be controlled by another choke, L704, but it isn't adjustable. However, placing something magnetic near its top widens the image, and doing the same on its underside narrows the image.

I've been able to further adjust L703 but the ferrite core is starting to disintegrate. I think it's been glued into place as it's not really rotating; I'm guessing that I've removed some of the ferrite.

So I am here now:

Good old MBASIC. I'm missing 10 characters to the right of the screen. I'll see if I can shrink it with a magnetic "adjustment" of L704. Failing that, I guess I'd need to replace L704 with something adjustable.

[jonb]

L704 = 100uH inductor. To be on the wise side I'd be looking for something that adjusts either side of 100uH.

I've ordered a small adjustable inductor 43-100uH but it has a secondary coil that should allow me to get it over 100uH by connecting it in series.

oLBbv3VF_o.png (8.4 KiB) Viewed 1165 times

The adjustable bit is across pins 3 & 4. It wasn't too expensive (few quid) so we'll see what difference it makes. This is all very experimental and I won't be surprised if it doesn't work!

[1024MAK]

In most CRT displays, the drive to both the horizontal (line) and the vertical (field) stages are generated by free running sawtooth oscillators that are synchronised by the incoming synchronisation signal(s). Of course they have very different time periods, hence use different designs of circuit.

Due to the frequency required, the horizontal (line) may well use an inductor. But there should also be a capacitor. It may be easier to try different values here.

Do you have a wide view of the monitor schematic showing the circuit? If yes, please attach it as compressed zip file, as the forum software now reduces the size of large images, which removes details and makes schematics blurry

Mark

[jonb]

The service manual is here: http://www.cpcwiki.eu/manuals/service.m ... 0.gt64.zip

It covers the CPC464 and GT65 monitor. This does say there is some adjustment near the yoke (magnetic disks) but these are not enough to fix the image.

Oh, I see what you mean. The SB II schematic I posted is ruined! Still, I think it is better to add adjustability to the GT65 board rather than hack the SB II board up. One of them is way more valuable than the other (I'll leave it up to you to guess which )

It's definitely "stretch" that is the problem. I nearly have a good image, just missing 10 characters on the right hand side. Interestingly, the LCD TV image is missing 6 characters, so I would say the SB's video image is way out of spec.

Incidentally, you should see the total lash-up I have on my desk... this'll make you laugh!

Who said you shouldn't muck about with CRTs outside their cases?

[jonb]

The boot ROM (which has a label on it: "SB II SB4.2", a reference to the ROM version (which matches what it prints in the screen memory at start up). This is the latest ROM I have seen (4.003 being the previous most recent) and I will dump it and upload to VCF, where there is a thread that discusses SB ROMs.

I don't visit VCF regularly, so could you let me know when it's there, or post it here too. Thanks.

Nigel, I posted it on the VCF SB ROM thread: http://www.vcfed.org/forum/showthread.p ... post553918