Profile cover photo
Profile photo
Michael Kukat
Communities and Collections
View all

Post has attachment
The polarity of AC power in DC systems

Well. In electronics and computing, and especially when both comes together, there is a WTF moment from time to time. And usually, after analyzing it long enough, there is a simple explanation for it.

I can't wait to find out the explanation for this WTF.

Carefully look at the photos. Too bad i shot them in different angles, but that's not the problem. Look at the orientation of the wall wart. It's a standard unearthed wall wart providing 15V DC for the hybrid synth prototype. In theory, there is no coupling at all between the AC input and the DC output, but usually, there are 2 capacitors from those life inputs to DC GND. But that's a symmetric thing, so it also doesn't depend on the "polarity" of the AC signal.

Besides this, the proto board is connected to nothing else. WLAN doesn't count, that's a good galvanic isolation. Okay, the ESD mat is grounded via several megaohms.

But there is a difference with the orientation of the wall wart. With one orientation, my firmware update mechanism for a slave microcontroller works fine, with the other orientation, it doesn't work at all. I analyzed this yesterday evening when i wasn't aware that PSU orientation matters, and it really doesn't work at all, even with slower speeds, more delays, higher timeouts. Not sure what exactly happens on the STM32F446 side, but i can tell that the SPI communication and NRST signals work absolutely stable, because the whole prototype works in normal operation, where the link is used to transport all the audio data. So no issues with the 42MBit/s SPI link plus handshaking signals, it just starts behaving incredibly strange on the 2MBit/s firmware update mechnism with some slow signaling around. But only with one direction of the PSU, with the other direction, everything works stable.

Yes, maybe there is a connection problem with an additional signal, veroboard constructions are not the most reliable setups, but the effect of this pure DC powered board with not other connections except power depending on the orientation of the power supply AC side is really interesting.

A while ago, a friend had the same effect with a Logitech Harmony IR receiver for a PlayStation, so such problems are not totally uncommon. But that IR receiver has another connection besides power, which i don't have.

That will be an interesting session during the weekend to find out what's happening here. I'm happy enough that the problem is 100% reproducible in the lab, close to the oscilloscope. Usually, problems go away if i dare to come closer with debugging tools or soldering irons :)
2 Photos - View album

Post has attachment
Desktop computing 2018

Meanwhile, the Acer Switch 7 Black Edition fully replaced my Dell XPS 15 (even if it has a bit less CPU power, likely due to thermal throttling) and today, i tried the last interesting feature. Not because i need it, but i wanted to know how far we can go.

It was a requirement that this 2-in-1 can serve as a desktop replacement and that worked out fine. Thanks to the quad core i7 and the 16GB RAM, even virtualization with several VMs is not a problem. And with the modern GPUs, even the CPU internal ones, you usually can drive up to 3 monitors without any special hardware. For notebooks, the simplest ways is to connect a display supporting MST to the DisplayPort connector and daisy-chain the secondary to this primary monitor. Plus the internal display -> voila, 3 screens.

And while not unexpected, it was a bit surprising how good this works with this small 2-in-1. With a cheap USB-C to DisplayPort adapter, it just worked out of the box. Okay, with 4k, you get just 30Hz on the external screens (60Hz with just one of them), but for most tasks, that's enough. I will not use the setup very often this way because i meanwhile prefer one single large screen. I was a big fan of multi-head setups in the past (with a peak in 2 horribly expensive 30" Apple Cinema Displays replacing 4 cheap 19" 4:3 displays), but no matter how many monitors i have around me - most of the time i look at just one of them, so desktop switching is a space-saving alternative to multi-head setups. There still are use cases where it makes sense, when you want to have some documentation on the side screen while drawing schematics or hacking some code on the primary screen. Maybe i want to integrate the second 4k monitor in the studio setup, i don't really need it in the lab, i bought the second one more for redundancy, if one fails (not that this ever happened with a Dell monitor here).

In addition, the main screen of the Acer can serve as a drawing tablet with it's stylus interface. So you don't need to buy an expensive Wacom Cintiq in addition to your computer :) And in this studio context, this also makes it nice for DAW applications with the touch screen, you can have the mixer on the internal screen, the arrangement on the large main screen, yes, i like those new possibilities.

And when you're done with your work, just pull the plugs and enjoy some games, the Geforce MX150 is quite okay for some less demanding games. For traveling, snap the keyboard cover on and just take it with you.

Yes, i really love this device. It was a great choice and my XPS 15 is already sold, so at the end it wasn't that horribly expensive.

By the way - even the smaller Acer Switch Alpha 12 can handle this setup. And with a Thunderbolt 3 dock, i could get 60Hz on both screens. Plus charging, Ethernet, several USB3 ports, all via just one single connector.

Post has attachment
Long term compatibility

What does a maybe-1990s Wacom ArtPad II with ADB connector (KT-0405-A) have to do with a brand new 2017 2-in-one computer?

The digitizer technology.

The Acer Switch 7 BE uses Wacom EMR for the stylus and this teechnology is also used in many other Wacom products. But since when do they have it? I recalled that i have this old drawing tablet in the box of vintage Mac stuff, so i just tried.

The stylus of this old tablet works fine with the Switch 7, including pressure sensitivity and the button. This gives me hope that the range of compatible styli is very large, as long as they support Wacom EMR. Yesterday, i found ot that the Intuos 5 uses something different, the stylus of this table doesn't work.

Post has attachment
Keyboard intensive care

I had a different plan for today, but first, i wanted to replace the rubber feet of my Amiga 2000 keyboard because they dissolved already and left a mess here and there. Shouldn't take long.

Replacing the rubber feet in fact didn't take too long, but longer than expected due to the goo not wanting to be removed. During this procedure, the numeric block 0 fell off, and i noticed that the Help key also is stuck somehow. Okay, i need to fix this.

I quickly found out that this was not the first time the 0 fell off, i already fixed this a while ago, and as the rubber spring was also missing already, i stole it from the least used key - Help. So the first try to glue it back together already failed and the second will also not be so much more successful. Wait, i have this old, dirty, ugly A2000 keyboard in the basement, let's see if i can steal the key there.

I quickly found it but then i noticed the C= key in place of the left Amiga key. That's something special. And it feels completely different than the other one, much harder feel, not the dead octopus of the regular Amiga keyboards. Having a closer look revealed it's one of the very early mechanical Amiga keybaords, an NMB Hi-Tek. Not only don't fit those keycaps to the cheaper Mitsumi keyboard, it would also be a shame to not restore this great keyboard. Even better than the Cherry, which has some trouble with some games not using the OS to talk to the keyboard but bringing their own routines for this, violating some timings.

So i took it apart, removed all the keycaps, which was not that easy sometimes, i ripped out the whole key body of two keys, but no damages. While the ultrasonic cleaner had it's portion of work (and this time, it really did it's job right), i cleaned the rest, removing plenty of debris, still leaving some rust behind, but for now, it's okay. I had some more fun putting the metal supports for the larger keys back in place and finally had to unsolder 4 of the keys to avoid damage by forcing those parts back in place.

Anyway - besides some yellowish tint of the enclosure, this keyboard looks and works like new now. Wonderful. I hope the glue of the other keyboard is also okay now. Ah, and it also got new rubber feet, they started to dissolve already. I'm still not sure if the cable is original because not only is the soldered DIN connector a bit strange, in the keyboard itself, the connector also looks a bit suspicious.

Now i can't sell my fully-loaded A2000, because it has such a great keyboard :)
9 Photos - View album

Post has attachment
The Raspberry Pi's grandfather

While scanning through some old floppies and tapes to image them, i found another set of floppies for this nice thing. As i'm pretty sure i never had documentation for this board, it might be as complete as it can get now.

Datakamp EPC II Z80 board, Data Research CP/M 3.0 for it, plus backup disks. I'm sure it was quite expensive back then, 30 years ago. At least more expensive than a Raspberry Pi (or whatever modern ARM-based embedded toy you prefer).

I want to test this thing one day. Too bad there is not much documentation available. As this board even has a video output, it might need a keyboard, which makes things a bit more difficult than just hooking up a serial terminal. But let's see what happens. Okay, might take some more months or years until i find the time for this :)

Oh, wow, just found out that Datakamp still exists:

Post has attachment
SGI Indy first power-up after many years

Okay, i carefully checked the monitor, removed some dust, had a look at the capacitors, everything looks good. Still a good adrenaline kick after powering it up and hearing the typical noises of the degauss coil and the high voltage building up. But - just works.

The hard drive seems to be close to death, it took half a minute to complete the first seek, but it boots. And i think i had my home directories of an NFS server back then, so without network this won't work. I think i should give it a fresh installation and then carefully think if i want to sell it or keep it.

Post has attachment
Too useless to keep, too nice to give away

I bought too many cameras and now something else has to go to make room :)
My SGI Indy was not a candidate and still isn't. I mean, i got rid of all CRTs (except oscilloscopes and, well, cameras :), but this one (and another beige 17" SGI screen) was kept, i even gave away all the 20" SGI CRTs i had (5 of them, IIRC). I didn't turn on this machine since many years, and even if i would, i likely might install a fresh IRIX and put it back to the archive. So there is no logical reason to keep it.
But an Indy with a 24 bit frame buffer, 256MB RAM, original IndyCam, matching CRT, keyboard and mouse, this nice setup just lacks the silent Sony PSU for perfection. I think i even fixed the NVRAM some years ago.

The speakers and additional microphone belong to my Octane, but as this machine is FUBAR, the Indy inherits them.

I'm too cowardly to power it up right now, i need to have a closer look at the innards, especially of the CRT, first. I don't trust anything that uses capacitors. Every piece of electronics gear that wasn't powered up for some years needs a closer inspection before waking it up again. Half of the times i don't respect this, stuff reminds me with a big bang and an ugly smell. And some evil work afterwards to clean a PSU from capacitor guts.

Post has attachment
Let there be knobs&sliders

So i used the long weekend to hack a bridge between this Crumar GDS keyboard console and the DIY hybrid synth prototype. Even if the DIY synth is heavily inspired by the original GDS, it's hard to bring them together in a useful way without having the computer screen in addition, and i didn't want to go that far because at the end, this bridge is just a throwaway module not required for the final user interface.

The GDS keyboard (without the GDS system, which i don't have) is connected to USB and/or MIDI using my DIY interface box. I added a simple way to control the LEDs of this console (program changes, 64 LEDs + 1 on/off bit = 7 bits, so it was the simplest way at the moment) and added another provisoric client to the synth engine control interface. It's not great, but enough for some testing.

You can add/remove any of the 24 sounds to the engine, up to 8 layered sounds are supported. You can tweak the most important parameters of those sounds also, basically what you can do with a DK Synergy minus what's not yet implemented.

In addition, you can edit all 16 oscillators (okay, most sounds only use 3-4 of them), again the most important parameters, especially the filter (which is a kind of 32 band graphic equalizer, controlling the output level of the oscillator based on it's frequency) and the envelopes. I think i should add some special handling to the "frequency factor" knob. Because FM synthesis alone already is a bit challenging, but tweaking FM sounds without having an idea what parameter values you really dial in is, uhm, interesting. But lots of fun :)

Besides this, some code cleanup, "efficiency" leftovers converted to more readable, modern C++ style, because they are no longer necessary now, after lots of other stuff is optimized. So i could reduce the CPU load on the embedded system another little bit, giving back some more headroom for some still missing features and other stuff, that might make it into the final engine.

Post has attachment
Uses for obsolete surveillance cameras

In this case, a Panasonic WV-CP240EX. The sixpack was cheap :) I think those make sense as helpers for soldering, drilling and other work where my old eyes could use a bit of support and the large lense magnifier is not handy enough.

Another plus of those cameras is the light sensitivity. Even with this very bad lighting, the image is really okay.

Post has attachment

I couldn't resist. Small surveillance cameras using Vidicon tubes! I found them on eBay and after seeing i decided that i want them for the collection.

I'm not sure about the age, the serial numbers of 2 start with 77, another one with 83, maybe 1977-1983 isn't that implausible for them. Panasonic WV-71. You might notice that they have just one single BNC connector. And exactly this makes it a little bit difficult to get a video signal from them.

After searching the web for a while, i found a service manual of a WV-BM90 monitor supporting those cameras, so i can look at the interface. No schematics of the cameras around there.

Technically, it's an old of up-the-coax implementation, so this single coax cabel delivers power to the camera, video from the camera back to the monitor and in addition, the monitor delivers vertical sync pulses to the camera. According to the service manual, powering the camera is constant current 220mA (which makes sense, because this way the cable length doesn't matter), you can decouble the rest with an electrolytic capacitor from this. The vertical drive signal are 5V pulses on top of the normal power.

My standard CVBS monitor doesn't really like this way of non-standard signal, so i need to build something to remove the vertical drive pulses from the signal and likely add real vertical sync signals instead. This needs some more experimenting and research. I even don't get any video output on the coax at the moment, but all 3 cameras show a signal after the target preamp, so in general, they work fine, the might just need the right signals to throw out an image. Which is pretty amazing after likely several decades of 24/7 operation.
Okay, one has a leaking capacitor, but this was somewhat expected. Even if it's rare with Panasonic gear, maybe they use Panasonic capacitors, which i also like as replacements for repairs.

Just wanted them for my collection, but when i have some more time, i'll dive deeper into this to make them work :)
Wait while more posts are being loaded