looking for trouble

part one: when stuff breaks

tshoot_040717

de Bill, K7WXW

While tuning through the 40 meter band, I notice the SWR meter spike and return to zero. Ten minutes later, it happens while I am not touching the rig. That’s weird, I think, but the net starts in twenty minutes. I don’t want to start taking things apart. I decide look into it later.

I drop to 80 meters, lower the RF power to five watts, and switch to AM. When I key up, the SWR meter goes to max and stays there. Adjusting the matchbox makes no difference. I move back to 40 meters. The same thing happens. Weirder. It worked ten minutes ago. Looks like I am taking things apart. With the meter out of line, I match the dipole on both bands. Putting it back in, I don’t.

Bad cable? I swap in a new one and check the old one with a ohmmeter. The new cable doesn’t make a difference and the old one is good, at least at DC. While I am switching the cables, I bump the meter and the receive audio gets a lot louder. I try to match the antenna again. Success. Could my third hand meter, an Autek WM-1, be the source of my woes?  I take it out of line and get ready for the net. For now, I can use the rig’s SWR meter.

I am embarrassed to admit this, but in an earlier life, I would have chucked the WM-1 and ordered something shiny and new. I was a busy guy.  Who had time to figure out why something didn’t work, right? In this life, I finish the net, put the meter on the bench and start taking the case apart.

I look for obvious problems: loose wires, cracked components, bad solder joints. The solder work is questionable and the sensing toroid isn’t glued down but there are no glaring (that is, easy to see) faults. I make a note to reflow the solder on all the connections and glue the toroid into place.

The instruction manual notes that unexplained SWR spikes aren’t anything to worry about. I speculate for a moment about career path of whoever approved the manual. There’s a schematic for the detection circuit but none for the main board and no explanation as to how it works. Not helpful.

I do a quick web search and turn up a few posts about the odd behavior I am seeing. Unfortunately, none include a fix or schematics of the base unit. I can’t call customer support; Autek disappeared years ago. I am not sure what to try other than cleaning up the bad soldering, so I go to plan usual: asking for help from people who know a lot more than me.  I send an email to the QRP list and set the meter aside for the night.

Grounding Bars and Bolts

Fine Tuning My Amateur Radio Station

de Bill, K7WXW

The ground bar is bolted to the shelf. The grounding straps of tinned copper braid between the bar and my rig, power supply and tuner, are in place. I am holding the last one, which will connect the case of the AC line filter to the bar, in my hand. The filter is on the floor near the wall outlet. But I can see there is a problem with the strap. Being four feet long, it will make a good antenna for any stray RF lurking around. Last thing I need is another entry point for noise, so I put it down and look for my notebook.

Other design decisions aren’t working out either. To get behind my rig, I have to move the desk. To move the desk I have to get out a quarter-inch socket wrench and unhook the copper strap that ties the ground bar to the rod outside.  Inevitably I need to crawl under the desk to retrieve the bolt after dropping it.  When I am done, I have to reverse the procedure, hopefully without the dropping-the-bolt step. This is not, as they say, an optimized process.

Sidenote: My friend Scott, N7NET, offers a tip for starting nuts in hard-to-reach places: dab a little Vasoline in the back of the socket to hold the nut in place until it catches the threads.

Apparently good paper designs often fail when they encounter real world voltages, screws, and transistors.  Spectacularly sometimes, too, as in what could you have possibly been thinking when you designed it that way? Mostly though, designs fail in small ways: a hard to reach screw or an oscillating amplifier.  Even good designers – and I am not in the category – have these problems. It is really impossible to know beforehand exactly how the thing we are designing is going to be used until someone is actually using it.

A piece of gear can’t get much simpler than a copper bar with holes drilled in it, a couple of standoffs, and a fistful of 4-20 nuts, lockwashers, and bolts.  Bolt the bar to the desk, secure each ground strap with a bolt, lockwasher, and nut, done. What could be easier? Something other than a ground bar, apparently.  The way I designed my system, undocking the main ground to get behind the desk is a giant hassle. The ground strap to the AC line filter is too long. And with a single bolt, nut, and washer per strap it takes two hands and two tools to make or break a connection. It all adds up to a list of changes that I didn’t know needed to be made until I actually put the ground bar together and tried to use it. And it also reflects a simple fact: paper designs, however carefully done, almost always need a tweak or three.

My first inclination is fix it now! Giving into this impulse almost always turns out, even in the simplest cases, to be a bad idea. My first unvarnished solution to a problem is usually (never) my best one. Sometimes it is not even my second or third best one. I don’t trash the first thing that comes to mind, but I don’t stop with it, either. I just see it as the start to a list of potential fixes to particular problems.

Even well-designed commercial gear doesn’t get through the design-build-use cycle without lists like this being written. When I was an engineer, we called them engineering change orders, ECOs for short. ECOs are another version of my lab notebook: what is the problem, what is the fix, what does it affect, how much does it cost? In the commercial world, an engineer has to write an ECO and get it approved by the people who live with the product and modifications made to it: manufacturing, marketing, testing and so on. Here’s a truth: those folks don’t like change. It mucks up their day. To get an ECO approved that engineer has to think through the problem he or she is trying to solve in a way that makes sense. Improvisation rarely wins the day.  Whether I was the designer or the design team manager, ECOs were my antidote for the spontaneous fix.

I don’t have a marketing department and I manage all the manufacturing and testing around here. The only person who has a bad day due to a poorly thought-out fix is me. That said, I don’t enjoy re-fixing a fix that didn’t work, so I address problems by following my homebrew version of the engineering change order process.

I start by looking at my original design. What was I trying to do? Did I miss something? I make a list of the problems and ask if any of them arise out of the original requirements. For the ground bar, ground straps should be as short as possible and in a star configuration. That requirement limits, among other things, where I can place the ground bar. If a requirement causes a problem and I can’t come up with a fix, I either live with it or I change the goal. Other problems come up because how I thought things would work is different from how they actually work. And, finally, there are the things I just get wrong. Now I have a to-be-fixed list, based on actual use, that reflects my original design goals.

Next, figuring out possible solutions. There are rarely more than three or four to consider. I just don’t have that many good ideas. As always, I write all this stuff down. Why? Mostly I don’t like having the same bad idea more than once.  The end product of this process is my homebrew ECO: a detailed lab notebook record of the what, why and how of whatever it is I am fixing, without all the check boxes and signature blocks one finds in the commercial version.

When I actually don’t know what might work, I go off to do some research. Maybe (almost always) someone has had this problem before. Or maybe a data sheet or application note will give me ideas. And sometimes simply setting the list aside for a bit does the trick. The list that comes out of this process will always be better than randomly trying something and hoping everything will turn out okay.

Now I have fixes I like. I start working through the list. The bolts are installed from the bottom of the bar with a lock washer and nut on the top side so they don’t have to be removed when I disconnect a ground strap. The  strap that runs outside is held with a wing nut instead of a nut now, so I don’t need a wrench or nut driver to take it off. The AC line filter is mounted to the underside of the desk with a six-inch, rather than four foot, ground strap. I can’t do much about needing to disconnect the outside ground every time I need to get behind the desk; no changes on that front.

After the fact, the fixes seem pretty simple and might hardly worth all the record keeping. Bill, that sure seems like a lot of work for a small problem. Couldn’t you just try something and see how it worked? Perhaps. Maybe though, simple fixes are simple because they are thoughtful and thoughtfulness is the result of writing down what one is doing and why.  It took me longer to write this post about the process than it did to actually do it.  That isn’t always the case. The chance of getting something wrong in a design goes up as the square (or worse) of the complexity. More parts, more chances for error. If that’s true, a process that includes time for making, testing and documenting changes seems prudent, even for a ham building gear in the basement.

I disconnect one of the ground connections to re-route it so it won’t block access to one of the antenna feeds. After checking everything against my wiring diagram, I plug gear into the new AC filter and the filter to the mains, then push the desk back against the wall. Getting behind the rig is a lot easier than it was two hours ago. Fix accomplished and later, when I am trying to remember why I did what I did on this project, I have my notes. It works.