A friend and frequent event-production collaborator has been accumulating
legacy Telex wireless intercom gear from various sources, and while the
equipment arrives in a nominally working state it sometimes has a few
little problems and tends to take a certain amount of wear and tear on
gigs anyway. With enough annoying problems having built up over time
across his pile of gear, he asked me to take a look at a bunch of it to
see what could be fixed/improved/whatever.
Some of it was pretty minor -- broken XLR connector latches but he had spare parts for that; miscellaneous loose stuff; and a lot of dirty/corroded battery-pack contacts which 90+% of the time are behind flakey performance. [You'll see my spraycan of de-ox-it in several of the pictures...] The intercom-connectivity is easy to get wrong with mis-setting of a nine position DIP switch and a level selector, needed to adapt to any of *four* common wired-side standards. But some of the issues were more subtle, such as distorted audio or units that used to work together and since failed; these would require a little more investigation.
One easy thing to miss when systems are set up is the fact that there can
be either dynamic or electret microphone elements in the headsets, and the
little slide switch in beltpacks or base stations needs to be set to match
which type is plugged in. The headsets show absolutely NO external
indication as to which is installed. Mismatches generally result in very low
transmitted audio. The operation manual for the Radiocom TR-200 belt-pack
transceiver specifically states "all headsets used with this intercom are
dynamic microphones", which is clearly wrong since the small collection
provided to me for testing turned out to be a mix of both types.
Fortunately, it's easy to test for without even having to power up any gear. An ohmmeter connected between pins 1 and 2 on the 4-pin XLR headset plug tests the microphone's DC resistance; dynamics are about 150 ohms and electret modules weigh in somewhere over 600. And it should be noted that the electret elements with the proper setting are VERY hot; units using them probably need the mic gain turned almost all the way down.
Here's what's inside a TR-200 wireless beltpack -- or about half of it, as
the transmitter board is on the other side of this. It's all analog, using
generic FM transmission on various VHF bands between 190-ish and 210 MHz --
nothing digital or spread-spectrum about it. Frequencies are
crystal-controlled and basically fixed for each unit and base station,
there's no agility for changing configurations and it seems like trying
to put in different "rocks" would be a royal pain in the ass. Two
flexible antenna wires with physical separation are used instead of
having something like a duplexer inside.
There are several different revisions of beltpacks and base units, each reflecting fairly substantial design changes inside.
|One issue was that the "battery" LED on a pack didn't flash on power-up or warn when the battery was getting low. Here's why. The two panel LEDs have these little brown connectors run up onto the LED leads, which then got bent over for retention after they poke out the bottom. I've circled where one of said bent leads wound up too close and sporadically touching another part. I suppose it was bloody lucky that didn't short something out destructively... The fix was to unbend that bit of the LED lead and clip it short; these little connectors really aren't going anywhere once the unit is assembled.|
Can you hear me?
After dealing with some of the low-hanging fruit it was time to move on
to what seemed like a bigger challenge: two identical base stations on
the same frequency range, neither of which would receive from a beltpack
and one marked "used to work". Well, that could be any sort of issue,
With these systems each base station can handle four beltpack units in a completely full-duplex fashion, so that all the users can talk interactively. The base can also connect to a wired system so that everyone on the production crew can chatter whether they're fixed or mobile. The base thus effectively has four separate radio receivers. That's the numeric limitation -- while all beltpacks can hear the single frequency the base transmits on, they must each send back on unique frequencies to avoid interference since the carriers are always on even when nobody's actively talking. So for as many different receivers can be crammed into the base station, that's how many remote units you have. To add more wireless units, you have to get another base on a different band and *its* four associated packs and link it all together with cables.
Which is a strong driving reason that newer systems have moved to various CDMA or TDMA schemes as adding more units is pretty much a matter of accomodating them in software, not to mention far better immunity to interference.
|So here's the BTR-200 receiver daughterboard. It's an interesting topology -- the receive antenna feeds a single front-end section that I suppose is a slightly more broadband preamp and IF converter, and then that splits to four separate IF sections that pull out each beltpack's audio and then mixes that back together on the intercom bus. It's pretty easy to see where those four similar sections are along the front of the board. Detection of whether a beltpack is transmitting is done in each separate IF strip via noise-threshold squelch, which lights its respective LED on the front panel when the pack is online.|
I fired up a base and a beltpack and sure enough, no receive. But then as
I started poking around the receive board looking for obvious physical
problems I noticed that the LED for the beltpack's channel had lit, and
a quick test confirmed that it was online and I could hear it at the base.
Huh?? How'd that happen...
After a short bit of thrashing I determined that powering the base down and back up again brought back the no-receive problem, and poking around the receiver and pressing on it various ways would make the LED suddenly pop on and restore functionality. After a while I further localized the physically sensitive area to the little metal shielding box under the grey fish paper with a couple of holes in it. That houses a handful of parts, but if the problem was in there it was going to be harder to access.
Along with the units had come a USB stick with various operation and service manuals the owner had collected, and the service manual for the base is pretty comprehensive. The procedure for tuning the RF sections requires test gear that I've barely heard of let alone own, so the last thing I wanted to get into was twiddling with various adjustments while flying blind and possibly de-tuning the thing into a hopeless brick. But while I'm not anything like a radio-circuits wizard I could follow along the schematic and at least try to narrow down what was going on. It seemed like something inside the box was the issue, as the lightest pressure either on the box or on the two variable-capacitor screws under the holes seemed to most easily bring beltpack-receive on -- after which it would *stay* on, so it wasn't like a typical flakey physical connection that makes or breaks as you wiggle it.
|What's inside the metal can is the local oscillator and mixer that creates the first IF from the incoming signal. Its boundary is circled on my schemo printout here. The manual's discussion of the circuit describes it as a harmonic Colpitts oscillator whose overtone frequency must be set at a certain offset from the receive band of interest -- mostly handled by the crystal but could be adjusted by "pulling" over a narrow range. The series inductor is under the third hole in the shield, making it available for tuning. Probing around various downstream stages with a scope finally let me conclude that this oscillator wasn't starting up on application of power, but wiggling the box afterward somehow let it start running after which it would stay running.|
What was really interesting was that the other base-station unit had the
*exact same problem*. Wiggling this local-oscillator shroud or the board
near it would enable reception, and it seemed that pulling the back of
the can toward the front panel was the most desirable motion. I even
made sure it wasn't a conductive scenario to my finger, by pressing on
it with a plastic rod. So the issue was inside there someplace.
The next immediate problem was that the shield box was soldered up so thoroughly and firmly connected to the board's ground plane that none of my soldering irons had enough heat output to even *begin* to melt it free. So apparently getting a look inside there to localize the non-starter was going to be, uh, a non-starter. I didn't want to start trying to cut into stuff, and besides, my old Dremel tool had just shredded its main load bearing three days before so it was out of commission.
Okay, now what??
I posted later to my production-tech-geeks mailing list, hoping to get a few additional ideas:
The problem is apparently that it won't start up by itself on receiving power, so nothing in the IF stages downstream sees any signal. The *really* funky thing is that if I physically squeeze a particular side of the small metal shielding box that's around the oscillator and mixer, I can get it start up and then it *stays running* and everything demodulates just fine. Until the unit gets powered down, and then we play the same game all over again. This is evidently the *same* problem across two almost identical base station units, which makes it even odder. I can't tell if it's actually a physical flakey or that if pressing the side of the box in a little changes some parasitic capacitance to ground just enough to let the oscillator get enough gain to start running. The LO and mixer and one more amp stage are the only parts inside this shield; everything is just right out in the open. Both base stations in question are on the same frequency band, so I'm wondering if the oscillator's needed center frequency for the particular range of interest isn't quite compatible with its design component values but it winds up having higher gain when tuned for on other bands nearby. The third base-station I have for a different band doesn't have any hint of this problem. I've done a couple of superficial checks to make sure signal paths are peaked where they're nominally supposed to be, but without appropriate RF tuning gear on hand I don't want to muck with spinning variable-cap screws and coil tuning slugs any more than trying a momentary deviation. The oscillator has an adjustable coil off the crystal side presumably for "pulling" its frequency, and a variable cap in its output tank which already appears set correctly. Since applying torque to anything also stresses it against where it mounts, that tends to kick the oscillator into running anyway so it's really touchy and difficult to diagnose via any adjustment. And apparently I don't have an iron hot enough to unsolder where the box connects to the circuit board's ground plane to have a look inside, so besides wondering what to do next, I figured y'all might be amused by the progress or lack thereof on this particular set of gear. It's not really usable as is, even if a redneck fix might constitute savaging a finger-size hole into the top of the base-station's cover so you'd power up the box, reach in and dork with the side of the LO/mixer housing until it started seeing units, and then not power it off for the rest of the gig. Sure, it might work, but I don't think that would fly especially when the systems get loaned out to other users. It would be the moral equivalent of hand-cranking a car engine and hoping it never stalled.A couple of folks responded and asked the quite sensible question, "so especially for the 'used to work' unit, what might have changed/degraded between prior functionality and now?" ... which I couldn't really answer, but we often think "bad electrolytic caps" and such when dealing with older equipment. Except that there aren't any of those around this area of circuitry, and this gear isn't *that* old or abused to the point where resins and insulation would start breaking down.
The AHA! moment
|However, that must have gotten me thinking because later it struck me that pulling the LO box forward and upward was the easiest way to trigger the oscillator into running ... and the layout has that box sticking up a bit where it's mounted, but it didn't look like it was on quite straight anyway ... And then I crouched down and sighted along the upper edges of the unit case, and suddenly had my answer to that question.|
|The fish paper was apparently glued onto the box because they knew it was going to be really close to the top cover of the unit. How close? Here's a carpenter's square ruler across the two top edges at the front and back of the enclosure. Except that the RF shield sticks up so far that the ruler *isn't touching* the rear case flange to the left at all. Thus, clearance was negative and the top cover of the unit had been smooshing the local-oscillator can downward this whole time. That's why it looked a little cockeyed, and pulling it back up straighter restored functionality.|
|Further evidence was afforded by how the flat ground strap had a little wiggle in it here, where it might have been straight at assembly but yielded to compressive forces later.|
And the likely reason the unit's condition "degraded" over time was simple.
These things go out on gigs. They get tossed into bins, they get set up
and then other objects get placed on top ... plenty of opportunity for
intermittent higher pressure on the top of the case. Couple that with
poor packaging design that subjects critical internal parts to that
pressure, and there's your answer. Comparing both
similarly non-receiving units showed the same non-clearance for the
top of the shield, and I was beginning to realize that even without
getting the little box open it was likely I could fix both of these
units in a not-too-redneck way.
The answer would be to simply shorten the two nylon spacers holding up that end of the receiver board and let the LO shield area sit down a bit and not contact the top cover.
My soldering station *did* have just enough oomph to remove the three grounding straps to the board and the antenna connection, and then I could pull the whole thing for a better look. I *still* couldn't see inside the LO can at all, but there was clearly a lot of leeway for bending the entire thing back up straighter and anchoring it there with a couple of ground-strap connections. Initial tests showed that yes, simply doing that allowed the Colpitts to start up by itself. I still had no idea what the box might have been internally touching but my primary suspicion was the crystal housing as that would likely be one of the taller parts inside there. Impedance around the crystal and transistor base connection is pretty critical for letting this thing work right.
|One spacer shortened so far, for height comparison. It wouldn't need a lot of drop, even though there was plenty of empty space for it under the RF board. The screws up through the main board simply remained captive in their holes, as the main board mounts pretty close to the bottom of the chassis and there was nowhere for the screws to fall out so all had to do was pop the shortened spacers right back onto them.|
|There was a minor subtlety with the interconnect pins, which at the old spacer height were close to bottoming out in their own connectors. Each set of pins had to be clipped down a little bit, in proportion to how far along the board toward the new low side it sat, forming a gentle overall slope to match the newly introduced tilt.|
|And lo! The LO was now LOwer, LOL.|
More bad puns
On going after the same fix on the second unit, I manged to get one end of
the shield open for a look inside. As I suspected, the crystal is about
the tallest part in there and just about anything touching its housing
would prevent oscillator startup. Even with the shroud not in contact,
just my finger touching the crystal would sufficiently change the
resistance / capacitance / whatever to impede operation. Since I had
the box open on this unit, I slipped a little wisp of gaff tape in
to stick to the underside of the shield and that was enough isolation
to allow running even if the box *did* come in contact. Maybe the
fish paper should have been on the *inside* of the box instead?
Anyway, at this point the case was now crystal clear and I had run the problem into the ground. Okay, that was gratuitously bad.
Both units could now receive beltpacks normally, as the oscillator fired
up immediately and gave me a visibly healthy IF near where it emerges
from the shielded area.
A sharp look along the back edge of this unit case here also shows a couple of thicknesses of gaff tape placed to keep the center of the lid just that much higher -- as the clearance mod may have repaired the units, it still felt like sort of a "bend it just so and hope" fix so the more I could do to prevent stray movement of the LO can the better.
It's certainly one of the stranger fixit jobs I've come up with. But hey, if we're able to talk reliably during our gigs from now on it's all good. Oh, just for the record on that topic, for wired Clear-Com compatibility the internal DIP switches 1 thru 9 in the base should be set to 101100011 and the slide switch to "low". Be careful about the external connection! Switching to "ext intercom" moves system termination from inside the BTR200 to expecting to find it on the external line, and if it isn't present the audio level shoots WAY up and tends to howl feedback into everyone's ears.