Mac-to-VGA monitor adapter struggles

[cheesestraws]

All in all, this is pretty darn good!

I don't know how I missed all this but this is great stuff! Nicely done!

 

[François]

@bigmessowires

I was thinking if you use a microcontroller in your adapter, would there be a way to detect what kind of signals the Mac is sending, and have the microcontroller configure itself, so the adapter wouldn’t need any configuration switches, it would be completely automatic!

 

[bigmessowires]

Thanks! Yes the microcontroller could use some logic that might reduce the number of DIP switches. But you would still need at least 6 DIP switches to set the monitor ID / desired resolution for Macs that don't support Multisync. It might be possible to automate the decision about which Mac sync signals to connect through to the VGA side.

 

[joshc]

<silly pointless idea>Touchscreen with a menu so you can pick the resolution?</silly pointless idea>

Product cost just went up by 500%, but it would be cool...

 

[bigmessowires]

You could maybe have a push button that cycles through a built-in list of possible video modes, and some LEDs or a tiny OLED to display the info. But let me get the basic sync splitting working first. The VG900b must be fully conquered. It sits there, mocking me.

 

[bigmessowires]

Would a voltage boost circuit be feasible?

Maybe. The low voltage is more like a symptom rather than the root problem though... I can probably only draw a few milliamps from self-powering, no matter what the voltage. If I can't make it work this way, I may try to add a 3V coin cell battery and run everything off that.

 

[Phipli]

Maybe. The low voltage is more like a symptom rather than the root problem though... I can probably only draw a few milliamps from self-powering, no matter what the voltage. If I can't make it work this way, I may try to add a 3V coin cell battery and run everything off that.

Is the current draw discontinuous - would a fat cap to draw a constant current, smoothing the spikes in demand, help?

Am I over complicating things? Shall I go back to snoozing in the corner after wearing myself out attacking the garden?

 

[bigmessowires]

No, that's a good idea. I'm using a 22 uF capacitor. The current draw seems smooth enough.

Some interesting results after trying an external 5V supply! It works much better. The VSYNC pulse width is shorter than when self-powered (the pulse lasts for 12 lines instead of about 16) and it's a consistent duration, instead of jumping around like I saw when self-powered. I'm still a little confused because I thought my change to the RSET resistor value would give me a duration less than 10 lines. Hmm.

The VSYNC output voltage is also higher, unsurprisingly, at 3.5V.

The bench supply that I'm using says it's putting out 0 amps, to a resolution of 0.01A. So it's something less than 10 mA.

Subjectively it works better. The EL151fp has a nice stable image now - it looks great. But when I switch off the external 5V supply, the picture immediately falls apart.

The 2001FP also has a good stable image. But when I switch off the external 5V supply, the image stays stable, and I can't reproduce the weird vertical jittering that I saw yesterday.

One strange thing on the EL151fp: at the bottom of the screen, after the last line of the Mac's screen image, there's a black line and then two lines from higher up in the image are duplicated. This seems like a scaling bug in the monitor, probably exposed by something non-standard in my sync timing. I think it may be due to the vsync pulse being longer than expected.

Actually I just discovered something very strange. Without the external 5V supply connected, I get vertical jittering. With the 5V external supply connected, even if the supply's output is switched off, the picture is stable. If I then disconnect the external supply, the picture remains stable until the next power cycle. If the external supply is completely turned off, the image is still jittery.

I have to conclude that my supply is actually putting out a small amount of current even when its output is switched off, or the output stage has a capacitor in it that's affecting things.

TL;DNR - LM1881 works better at 5V than 4.1V.

 

[Phipli]

The bench supply that I'm using says it's putting out 0 amps, to a resolution of 0.01A. So it's something less than 10 mA.

Worth checking the specs on the PSU - low currents are difficult to measure and so while the indicated precision is 0.01A, the accuracy at low currents may be worse. It might not be able to (making up numbers) measure currents below 0.05. I remember this being the case with... Either my multimeter of my bench PSU.

This seems like a scaling bug in the monitor, probably exposed by something non-standard in my sync timing. I think it may be due to the vsync pulse being longer than expected.

One of my nubus video cards shows the bottom most row of pixels at the top of my monitor. I assume the sync timing is poor.

I get vertical jittering. With the 5V external supply connected, even if the supply's output is switched off, the picture is stable. If I then disconnect the external supply, the picture remains stable until the next power cycle. If the external supply is completely turned off, the image is still jittery.

I have to conclude that my supply is actually putting out a small amount of current even when its output is switched off, or the output stage has a capacitor in it that's affecting things.

Or grounding it / coupling it?

 

[bigmessowires]

I'm thinking that self-powering may be too difficult: very little available current, lower than normal voltage, and big questions about variability across different Mac models and video cards. I'm still going to try, but in case self-powering isn't feasible, how else could I get a supply close to 5V? The idea of plugging a LocalTalk pass-through cable or USB power jack into a simple monitor adapter is very unappealing. The only alternative would seem to be a battery, but they're big (compared to this adapter at least), their voltage drops as they're depleted, and they're not 5V. I think the only standard battery that could even fit would be a coin cell. None of the options are great, so I guess I need to try hard to make self-powering work!

 

[akator70]

Self-powering is obviously the best, but I wouldn't object to a USB port for power. I have many video devices that require additional 5V input (switches, adapters, converters, upscalers, etc.). There's a bunch of peripherals for my other old hardware that require an additional 5V input as well.

 

[mdeverhart]

I think someone already suggested this, but what about self-powering by rectifying from the video signals (as in your current design), and then using a boost converter to raise it to 5V? The current load looks really low, so it seems potentially feasible. I suppose you still have the uncertainty about the current sourcing capability and variation between different machines though.

 

[sfiera]

ADB pass-thru sounds like a good enough solution to me. Or even using ADB somehow! Plugging in display+ADB together is not so uncommon. Obviously KVMs do it, and some Apple monitors use both as well: the AudioVision 14 and the original Apple Studio displays, for convenience and additional controls. Maybe you could think of a way to make ADB pass-thru a feature rather than a compromise.

 

[bigmessowires]

This is intended to be a simple dongle the size of a matchbook, so having any extra power cables that plug in at the sides is not desirable. If it were the only alternative then I would probably choose USB, since the cables are smaller and less obtrusive than ADB (I misspoke when I said "LocalTalk" before) and don't require pass-through. A boost converter might be feasible, but the whole boost circuit would almost be more components than the circuit being powered, and would need more PCB area too. Normally there's the boost controller IC, an inductor, diode, and several passive components including some large-ish caps. The inductor in particular is pretty big. I don't know if there are smaller/simpler boost circuits that can be used when you only need a few tens of mA?

A charge pump might be an alternative to a boost converter. I don't know too much about these, but no inductor is required and I think they can be well-suited to a lower-power circuit like this one. But the boost ratio is usually fixed at 2:1, which would give an output around 8 or 9 volts, which would then need to be dropped back to 5V through some means that wastes power (diodes?). If anyone has a brilliant idea for a charge pump circuit that might work here, let me know!

I think the best option is probably self-power or bust. Scope the design and features so that it hopefully can run from self-power, and any feature that would require too much power does not get included. I think it should be possible to achieve self-powering of a simple device using a microcontroller running at ~4V, at least for the Mac IIci and IIsi. For those computers, if you know how many amps i the circuit will draw, you can estimate the VCC you can achieve from self-powering with Schottky rectifier diodes as 4.7V - i*60. So for i = 10 mA you would get 4.1V, which is what I measured in the prototype.

 

[bigmessowires]

Using the new adapter, I've started collecting some data on what types of sync are output by different Mac models and video cards when 13" RGB 640x480 mode is requested, and what voltage is attainable through self-powering. Here's what I have so far. I hope to add greatly to this list tomorrow:

IIci - composite sync, 4.1V
IIsi - composite sync, 4.20V
P475 - both composite and h/vsync, 4.64V
PowerMac 7100 video card (not the built-in video), very tight fit! - both composite and h/vsync, 4.56V
Performa 6214 built-in video - both composite and h/vsync, 3.15V

That last result is interesting, and needs more follow-up.

 

[mdeverhart]

A charge pump might be an alternative to a boost converter.

This one looks promising:

https://www.analog.com/en/products/MAX682.html#product-overview

2.7-5.5V input, 5V output, just requires three capacitors and a resistor. I haven’t done a full datasheet scrub, but it looks promising.

 

[bigmessowires]

Nice find on that charge pump IC!

I had a chance to try the self-powered LM1881-based adapter prototype on some computers and NuBus video cards at Mactobefest. The results were not very encouraging, so I stopped collecting data after a few.

Q700 - both composite and h/vsync, 4.79v
"Toby" video card - no sync at all, no video output
Mac II monochrome P/N 630-4385 - composite sync but no video output, 3.2v-3.5v
Mac 8-24 card - no sync at all, no video output, 0 volts
Mac Hi resolution video card P/N 630-4222? - no sync at all, no video output, 4.35v

I did walk away with a Toby card that I can experiment with more.

 

[Phipli]

Nice find on that charge pump IC!

I had a chance to try the self-powered LM1881-based adapter prototype on some computers and NuBus video cards at Mactobefest. The results were not very encouraging, so I stopped collecting data after a few.

Q700 - both composite and h/vsync, 4.79v
"Toby" video card - no sync at all, no video output
Mac II monochrome P/N 630-4385 - composite sync but no video output, 3.2v-3.5v
Mac 8-24 card - no sync at all, no video output, 0 volts
Mac Hi resolution video card P/N 630-4222? - no sync at all, no video output, 4.35v

I did walk away with a Toby card that I can experiment with more.

What do you mean by no sync? Surely all cards output some kind of sync when active? What I'd pin settings were you using?

The "high resolution video card" is just a revised Toby design. Very little difference at all. The Toby is the most difficult Apple card to get video from.

The 8•24 - I'm confused that you had issues with this, they're a good vanilla card that generally works fine with LCDs. I've never not just used one of these as plug in and go.

 

[bigmessowires]

I'm not sure what happened with the "no sync" cards. My adapter detected neither CSYNC nor H/VSYNC on these. Maybe that's because my adapter doesn't work right, or because the cards only output sync on green. The 0v card was more puzzling, and I'm not sure it wasn't measurement error. If it was truly 0v then it means all three of the sync signals were constantly at 0, or were not even connected.

As I sit here now reviewing the data, I suspect at least some of those NuBus cards only have sync on green, and no other sync signals.

But in short, none of these examples would obviously benefit from a self-powered sync splitter adapter in the same way as the IIci and IIsi. The Q700 doesn't need it since it already has separate h+v sync. Three of the other cards didn't seem to have any sync, so something weird is happening with them. The Mac II monochrome card could maybe benefit from an adapter like this, but the self-powered voltage was quite low around 3.2-3.5v, which is not great. I wish I had more of these cards to experiment with at home, instead of just a brief few minutes of testing at the meetup.

 

[bigmessowires]

I had planned to give away the VG900b at Mactoberfest yesterday, but in the end I decided to hold on to it. It's been so consistently hard to please that it's my most demanding test case for any future adapter designs.