Things that can't be understood

My "fancy" homebrew computer (not pictured here) has been running nicely for about 8 years. It was big and powerful in 2009, using the new Intel Core i7-920 chip, with lots of RAM and lots of cooling.  Nowadays, Gimli (as I call it) runs mostly Windows 10, with a touch of Linux.

It runs up to 8 background threads simultaneously, running up some nice statistics with BOINC -- the science "@home" brokering site. Maybe we've discovered a pulsar or two, who knows?

This was alongside my "real" work which has been operating my ham radio station -- running SDR radios, logging, digital mode operation, etc.

That was until a few weeks ago, when Gimli started failing -- stopping suddenly with no blue screen of death or software logging, just a blank screen.  I thought maybe my checkered past as an overclocking experimenter was catching up with me.  I tried going back to the absolute minimal BIOS settings just in case.

I starting checking out what I might need to buy to upgrade this "ancient" machine through Newegg or Amazon -- and the discussions I'd need to have with the XYL.  Yes, there are many new options on the market.  The biggest one might be the advent of SSD technology (Solid State Disk).

As a last stand, I thought I'd check out the computer as best I could.  It's practically impossible to find a fault of the kind I was having.  You'd just have to replace key components one by one until it started working again.

Opening the box and blowing out the dust, I started wiggling the cables and re-seating the connectors. I pulled out the two RAM modules and "cleaned" the contacts (wiping with fingers), and I plugged them back swapping their positions.

Naturally when I put things all back together, Gimli came to life and hasn't crashed for a week or so now.  Maybe we're go for another 7 years?

Or maybe we're running on fumes.

VAIO inverter surgery

 

My Sony VAIO (VGN-FS710/W) gave up the ghost last month. The screen went dark, and it emitted some screechy noises.  I figured what a hardware service call would likely cost, and I decided I quickly needed a new laptop to support my work at ARRL and other things. So I sprang for a new inexpensive Toshiba (Satellite L6460), which did not cost much more than that hypothetical service call.

Now, after my computing situation has stabilized, I had a chance to open up the VAIO to see what could be done.  There were some helpful disassembly hints on the Internet, and they were essential to avoid breaking things.  The culprit almost leaped out at me.  It's the tiny inverter card (above) that makes high voltage to run the fluorescent LCD back light.  The Internet (again) seems to be saying this is a frequent source of problems.  The HV rectifier was charred and broken, so obviously this was the problem.  It's a $50 part, and I hope replacing it will solve the problem.

I suppose there's a moral here.  I could have saved money, but incurred opportunity cost and risk, if I simply repaired the VAIO.  I'm happy to have a new, more powerful laptop. Still, the VAIO is handy because it has some critical features (like Firewire) that make it very useful for some of my work.

Update (11/12/2010): Thanks to lcdparts.net, I have a new inverter, and the VAIO is back in operation!

It Begins Again... (AT&T)

I've been enjoying stable "Elite" DSL service (6 Mb/s download) for about 5 years now.  Time to update to AT&T's "Uverse" service?  It just arrived in my neighborhood.


Image via WikipediaMy initial DSL service was a pretty grim story. SBC took over the Connecticut territory from SNET not long before.  (SNET, of blessed memory, was a member of the original Bell network, having been Connecticut's own phone company for a very long time.  Sigh!)


Surely, AT&T would have figured out their new Uverse TV/Internet/Voice technology by now. Nothing can go wrong, go wrong, go wrong.

Actually, they've made 3 errors so far just scheduling the installation, and I've had to talk to 4 or 5 people on the phone.

The interesting technical issue is that their internal database shows that I am supposed to be 733 feet from the "node" (VRAD) in my neighborhood.  But, tracking real phone cables, I come up with some long zigs and zags that may add up to triple that number.  So there's a chance the tech will show up (at the newly appointed hour, we hope) and declare that my connection would be substandard.  And the whole thing is off.

I'm not canceling my Comcast TV just yet.

Orion Surgery

My Ten-Tec Orion transceiver was 5 years old in February. It's getting to be an old timer -- in computer years, at least. It needed a few chores done.

1. The LCD panel was flickering. Internet scuttlebutt says this may be because of impending failure of filter capacitors that were undersize (in terms of ripple current handling).
2. With the current version 2.xx firmware, the LCD contrast is close to zero after a hardware reset. This can be a serious problem, depending on room temperature. Contrast is normally set via a menu adjustment -- that requires you to navigate through the LCD display. The display could be so faint that you couldn't make the adjustment. There is an internal bias control that should fix this.
3. Finally, I have been seeing the "RIT freeze" problem that others have noted. At times, the RIT control loses the ability to make any adjustments. Many people have reported this problem, but the cause and cure have not been well understood. It seems that you can fix it temporarily by grabbing the RIT/XIT encoder control and twisting hard in odd directions -- as if there were a bad solder joint on the PC board. A recent report suggests the trouble is actually with one of the board-to-board connectors making unreliable contact.

All the above have been discussed on the helpful Ten-Tec reflector, which I recommend to other Orion users. I have been saving up these problems for some time now. Hopefully, I only have to disassemble the Orion once!

Making a long story short, I pulled the rig out of my operating position, unplugging many cables, unscrewed many screws, and finally spread out the works on my bench. The RIT fix requires disassembling the front panel and the LCD/computer board. That's a fairly big operation, particularly since a wrong move with a tool could lead to a very costly repair job.

I located and replaced 3 electrolytic capacitors on the A9 board with higher voltage, low ESR units, per the helpful analysis of N6IE. Soldering was tricky, to prevent inadvertent short circuits and to get enough heat into the ground connections, which were not thermally isolated from the big ground planes on either side of the board. (That could have been avoided by a better PCB layout!) OK. That's problem #1 done.

The email instructions say that to solve #3, you need to look for bent or misaligned connecting pins on the computer / LCD board. I looked very carefully, but saw no misalignment or other problem. Still, it's possible there was some oxidation or dirt. I cleaned the pins and the sockets as best I could. So problem #3 might be resolved.

The LCD bias issue (#2) turns out to be easy. There is a control on the LCD/computer board that sets the default bias for the display. You just adjust it for a good display with the LCD menu set at the default 50% level (and of course with the power on and the rig having a chance to warm up).

Accomplishing all this, I reassembled the Orion and put back all those screws, and put the rig back in the operating position. Everything appeared to work, except no transmit power. Oops! There are a lot of reasons why this could happen. Fortunately, I did not smell smoke, and that eliminated some of the more expensive possibilities. Still, I'm not sure how much bench testing and repair I would be able to do, if this turned out to be a non-trivial problem. Thinking of where that shipping box might be, and how much Ten-Tec charges...

So, what did I really do in that repair process? I changed out the capacitors, true, but the fact that everything works in receive mode indicates that the power board was probably working. The other thing I did was to unplug many internal cables, disassemble the LCD boards, and reassemble. There could be a loose connection or a misplaced cable.

The problem was "fixed" by opening up the box all the way again, and carefully reinstalling all the cables and screws. We have transmit power for now. I hope it stays that way!

The triple repair job is done. The LCD is behaving much better than before, and at least for now, the RIT/XIT controls work smoothly.

One thing I hadn't counted on was the trouble of reconnecting my ratsnest of cables to the Orion. I have a linear, a transverter, computer audio and controls, and other gadgets that need to be connected. All these just got hooked up over several years, and of course I had nothing documented on paper. Figuring out how to reconnect took a while, but it should be better next time. Now the cables have labels on them. Documentation is still pending...

LOTW troubles

Steve K9ZW has a point.

LoTW bad news: Way too complicated to set up, especially compared with other "secure" services like e-banking. (And, I might add, poorly supported under Linux.)

LoTW good news: Works well once set up.

Home science? No way!

This blog post is really worrying. How many of us in science and technology did not get our start experimenting in the basement?

Will ham radio be the next "unsafe" activity to be banned? (FCC pre-emption or not.)

Photo troubles!

Some photos from earlier postings seem to have vanished. I hope I can sort this out quickly.

Update: Some photos from before 2007 got lost in a server shuffle. I think I've restored them all now. Blogger's photo archive is a lot easier to use than what I had in those old days -- having to put photos on a private server.

So no one reported this...?