GBS-8220, 23DB Connectors and Leads

[Higgy]

Cheers for this. I bought a GBS-8220 the other day. I bought a female SCART connector so just plug in my normal AMIGA SCART into that - but I have white dots.

So I am interested in building your circuit. Have you built it on stripboard? If so can we have front/back photos of it.

I saw all these 74HCT14 and thought what! but I see you buy 1 chip and there are loads inside!

Where do you power/get Vcc - off the GBS-8220 or from AMIGA's 23-pin? I see there are sync stripper circuits that take power off the board.

I power my GBS-8220 from a 5V 2.5A mains adaptor.

Thanks.

[Johan1973]

I have not built it on any board. The IC and all components are inside the 23 Pin DSUB housing. And it takes quite a deal of effort to wire that correctly and isolate it with shrink wrap hose, prevent interference from the IC etc. That said. It can easily be built on a PCB, though it is almost not necessary.

The IC gets power of the 23Pin DSUB (pin 23 is 5V). You could tap into the power from the GBS 8220 as well, if you like. It is not exactly as if hte 74HCT14 are using that much current. The biggest issue is avoiding switching noise from the IC..

The GBS 8220 I run from the same ATX power supply that I use to power my A1200.

[Stedy]

Hi Johan,

Can I simplify your circuit with a few suggestions?

Only use a single 74HCT14 to drive the sync signal and add a series terminating resistor, matched to the cable type. A typical TTL buffer has an output around 10-20 ohms, the cable type varies. If you match the driving impedance to the transmission line, you eliminate high frequency effects, important with the 1 GHz edge rates of an 74HCT14 device.

As a guide use the following resistors:
Cable type Series resistor
50 ohm co-ax 33 or 39 ohms
75 ohm co-ax 47 or 56 ohms
Piece of wire or twisted pair 82 ohms

Get this right and you can eliminate R1, C3 and the ferrite bead.

An AC coupling capacitor for a video sync signal should be at least 100uF to allow for the DC component of the signal. Your current 0.47uF capacitor will distort the signal edges, though the schmitt inverter will clean it up, it can cause a shift of 1-2 pixels, worst case.

Any pictures of the display before and after the circuit was added?

Ian

[Johan1973]

Nice of you to give some feedback Ian.

I have not been able to capture the phenomena on photo, since it occurs randomly and then for such brief periods of time. The symptoms are white dots randomly appearing in the image when CSYNC reach to high levels for the GBS 8220.

The most prominent case for causing them that I have found is running State of the Art demo, which can produce the white dots regulary, allowing me to test the circuit. And that is how the values where chosen, bit by bit reducing the white dots until nothing remained. You are right in that it shifts the picture a bit, though you can easily compensate for this with the settings in the GBS 8220, and that is why I only recommend using this circuit for the GBS 8220, not to hook up directly to any vga display.

I think the problems would vanish if only the GBS 8220 would support RBGHV instead of RGBS, and while it has a RGBHV mode, it refuse to accept anything other than RGBS. Reason for this I assume is in the firmware, though I can not confirm this.

I tried matching the impedance to the cable, and I instantly get the white dots again, which is the sole reason why the R1, C3 and Ferrite bead is there in the circuit at all. Ideally C3 and the Ferrite bead should be removed. And I guess I will do some more testing of this circuit (I am never one to stop perfecting things). So thanks for aiding in this

EDIT: I am reworking it all from the ground op, and will soon have a revision available... Ians insights and some testing with two other GBS 8220 showed me that they are behaving very differently. Even with the same cables hooked up etc.

[Johan1973]

This is the experimental build of the circuit I am using now. Testing takes quite a bit of time, and testing three different GBS 8220 adds to the time.
If anyone feels like building the experimental circuit and adding your findings here, please do so, since it would speed up the development of the design.


Before the Ferrite Bead (Or instead of it) you could use a resistor, impedence matched to the cable as Ian suggested. Right now I am testing with just ferrite, as one GBS 8220 worked fine with my original design, one brown shifted the colours, and one caused green shift, which leads me to the conclusion that the variance in their construction is sufficient to cause issues and that the circuit needs to be redesigned.

Reson I use ferrite bead right now and nothing more beyond the cap and IC, is simply that it seems to do about 90% reduction of the white dots problem with all three GBS 8220 I use.

[tomse]

@Johan
have you tried adjusting the potentiometres, turning down the RGB amplifications to remove the "snow" ?
(this is a long thread, so I haven't read all the posts.)

[Johan1973]

Originally Posted by tomse

@Johan
have you tried adjusting the potentiometres, turning down the RGB amplifications to remove the "snow" ?
(this is a long thread, so I haven't read all the posts.)

Yes, I have. And the snow is not caused by the RGB amplification, it is caused by the composite sync. The GBS 8220 all seems very quirky when it comes to what they want from this signal, what they tolerate etc.

To clearify: I do not have these issues using my ZX +2 with the GBS 8220, its only the amiga. And I have played around with it enough to find that it is the composite sync, not anything to do with RGB that causes it (to summarize it since you did not read this year long thread).

[tomse]

thanks for the sum-up.

I'm using the GBS-8200 myself and I don't experience the problems you have.

I just ordered another GBS-8200, as I want to build it into my SX-1.

I really have no use of dual VGA out, so I'm sticking with the 8200 rather than the 8220. They're almost the same except for the dual VGA out.

[Johan1973]

That is quite interesting. I need to get myself a GBS-8200 and compare it. Do you have a good trusted ebay seller to recommend?

[tomse]

I can't say I have any trusted dealers on ebay.. it's evil bay after all xD

I've had a single report of the "snow" but that was fixed adjusting the RGB amplification, hence my question :-)

[Johan1973]

I understand. Going to try to fix this for GBS 8220, though I probably will get a GBS 8200 asap as well, to test. Do not use the double vga output.

[Solo761]

I ordered one here few weeks ago. Arrived yesterday so I haven't got to try it yet.

I think this might be the cheepest one around, $23.77 with free shipping. There's only one potential issue. Pictures on the site say it's v4.0, the one I got is also v4.0, but the difference is the date below the version. Pictures on the site show 2012.08.17 and mine has 2010.05.12. I'm not sure if this is important or simply the date PCB was made.

There's also v5.0, but that one is not GBS8200 but HD9800, (my guess is ) different brand of the same thing. Except I'm not sure if that one is actually newer than GBS, it could easily be that their numbering is different and it's not actually newer firmware.

[tomse]

mine is a 2012-08-17 too v4.0.

I just think it's the year they're produced when they update the silk marks.

I wouldn't worry too much about it.

I bought my latest from "goodlucksell" for around $26 on ebay
free shipping and all (I think it took them a week before sending, which only happend today).

[Stedy]

Hi.

Just a thought, try a 3.3V TTL device that has 5V tolerant inputs. The logic 1 output of the Amiga CSync is 4.8V yet the GBS 8220 operates off 3.3V. What might be happening is that the higher voltage is activating the ESD diodes on the board, this adds a charge to the device affecting normal operation.

A while ago I looked at the ZX Spectrum, it used normal TTL I.e. 2.4V logic 1 rather than the HCT type logic of the Amiga. This might explain the issues.

Do you have an oscilloscope?

Ian

[Johan1973]

Sadly, I do not have an oscilloscope.

But from what I have observed, it falls very much in line with what you are describing. I need to find a good 3.3V IC that can do this, as that would solve quite a bit.

Btw; does anyone know a supply of the 23 pin DSUB connectors? I have been using 25 pin dsub I have dremeled to 23 Pin. It works, bit it is not looking so nice. Here in sweden I am unable to find any source of the 23 pin dsubs..

[tomse]

Originally Posted by Johan1973

Btw; does anyone know a supply of the 23 pin DSUB connectors? I have been using 25 pin dsub I have dremeled to 23 Pin. It works, bit it is not looking so nice. Here in sweden I am unable to find any source of the 23 pin dsubs..

There's one in the US (eBay) can't find him right now, but do a search there from time to time and he'll pop up.

[ajk]

@Johan1973

I've got a few 23 pin connectors, not enough to open a shop or anything but I'd be happy to send one over. Drop a PM if you can't find any elsewhere.

They were still available in most electronics shops up to maybe 10 years ago, but I suppose new ones have not been made in ages now...

[Johan1973]

Another day of investigating this turned up this;

1. The first GBS 8220 I got and has been doing the main testing on is faulty. I have talked to the seller and confirmed this, it is behaving in a way that it should not be doing.

2. The IC I used for the build was a bit wonky, after having tested another IC I could easily determine that. Seems it inverted the signal in first stage, but both second stages on the csync was not reinverted. And that caused loads of problems.

3. The Ferrite bead remove 90% or so of the white dot problem. However, the RGB lines are very sensitive to noise (in the input of the GBS 8220). And here is where I finish my day on this subject today. Tomorrow I will investigate what I can do about it.

http://sharesend.com/wws9jej2

The above is an archive with schematic, and pictures of how to build this circuit into the 23 Pin DSUB. The Ferrite bead are in the vga connector (not shown on photos. Since I plan on adding something on that end to deal with the RGB issue as described above).

[Johan1973]

Before going to sleep, I had an idea flash through my mind, that perhaps it is perhaps peaks on R, G, B that is causing the remaining white dot artefacts. So I quickly tried 330 Ohm pull down resistors, on R, G, B. Which seems to do _something_. Tomorrow will be the day I have to investigate this further, and see if I can borrow an oscilloscope.

[Higgy]

Keep up the good work.

The GBS-82*0 seem to be the best bet if your computer desk already has a PC & monitor and there is no space for another. Or you are not allowed the AMIGA plugged in to the living room TV :-)

Maybe the Scanjuggler will get released at some point ;0)