The little matter of the Variac.

A lot of FM broadcast transmitters (and who knows, maybe even TV, AM….?) use a tetrode or pentode tube based power amplifier. Control of the final output power is achieved by metering the output and adjusting the voltage to the screen of the tube, thus adjusting the amplifier’s gain.

A woman loves a man with a great big old Johnson.*

On a lot of transmitters this gets done electromechanically. In this case this is done using a Variac or variable transformer, but in some smaller rigs (2.5, 2KW, and below?) a rheostat voltage divider may be used.

In the above picture, the screen voltage is all the way up for maximum gain. This occurred as the result of the tube wearing out and its cathode emission going soft, so the automatic power control kept trying to run the gain up higher and higher to maintain the desired power output until the poor little motor tanked.

Now a word on folksonomies and genericized trademarks. You may notice I capitalized Variac. This was originally a trademark held by the General Radio Company with first use in 1933. It was allowed to expire, probably by the end of 1994. Variac, however, kind of became a household word (if you could call a variable autotransformer a household item?) like Band-Aid, Xerox copy, Jello, and the like. Within the folksonomy of electronics there are a few things that have become genericized like this and it may be easier to categorize information involving them by an old brand name. Other examples would include “Black Beauty” capacitors which were a Bakelite encased oil and paper cap, Vactrol for light variable resistor type optocouplers usable for audio and analog signals…. Sorry for geeking out worse than usual here. Anyway, with the Variac / variable autotransformer…

Shown here, a Harris HT 25 FM. This is very very similar to the FM 25 K series and was made like this for years, because once Harris had a design that worked very well, there was no reason to make big changes. I bet they made thousands of these transmitters and many of them are still happily thrumming along to this day.

The mechanism is simple but prone to issues, especially if the automatic power control is used. Here’s the problem: Every change in line voltage, even a change in output impedance caused by rain or ice hitting the antenna, may cause the power to fluctuate. The controls compensate by moving the mechanism, and eventually something gives.

The original mechanism used this unusual bidirectional synchronous motor geared down to 0.5 RPM, running on 120 volts AC. This motor is discontinued by Hurst, though they’re still around and still have a somewhat similar product. However, it has to be custom manufactured with a lead time of 14 weeks, and nobody just has stock of ’em.

The solution: get to hackin’. First, a control system to step this down to a commonly available DC gear motor:



And now comes the fun part. The DC motor’s shaft is fatter and shorter, but it’ll work! Here’s the original shaft coupling it goes into and the Variac. Off to the right is the transformer that steps 240VAC up to 1000VAC. Or maybe it’s 707VAC if it just uses the peaks. I dunno, man. It Just Works.

There’s also a silvery looking band wrapped around it with one of the two end stop screws. More on that later. I removed all the grubscrews for safe keeping.


The bronze coupling drills like a hot knife going through butter, albeit with a disturbing squeaky sound.

This is how the stop screws are installed from the factory. They hit the micro switches under the mounting plate to limit the motor’s travel.

Now, you’ll notice the new motor’s shaft is shorter. Under the silver band there’s another pair of set screws that clamped the old motor’s shaft. However, the hole the stop screw is in is tapped exactly the same as that set screw hole.

Flippity flop ’em.

The silver band will just cover up the set screw down the upper hole once it’s all assembled.

Trust me— it works fine.

And no, of course the silly thing didn’t spin in the vise, what ever made you think that? No, of course not! DURRRRRRRRR

Here’s the mounting plate. I expanded the hole where the original shaft went down to 1/2 inch, and wound up just drilling two new mounting holes and using the side mounting holes in the gearbox.

This motor also provided three tapped screw holes in the bottom, but I didn’t wind up using them. If I had this to do over I’d probably do it though. They’re Metric and I couldn’t tell you off the top of my head what the screw sizes are. M2? M3?

And it’s up and running.

If I had it to do over I might slightly revise the motor mounting as I don’t entirely like the amount of slop the rubber grommets induce in the system, but it works– and it doesn’t hunt back and forth. The Variac requires a very small amount of torque to turn it (I could grab that axle with two fingers gently and turn it throughout its full range). The DC motor I used is impossible to turn by hand so I brought a 9v battery to hold against the terminals as an assembly aid to reach all the set screws. 😉


And now I need to redo my nails.


So I’m trying to get SSL certificates working for and am currently having little luck. I’ll revisit that later. For now you can get here by https but get a huge certificate mismatch error all up in yo’ grill and I don’t even know– probably has something to do with the fact certbot can’t figure out my vhosts. According to all known laws of aviation, there is no way a bee should be able to fly. Its wings are too small to get its fat little body off the ground. The bee, of course, flies anyway because bees don’t care what humans think is impossible.

Tapped out….

This Ryobi pack wouldn’t take a charge. After removing the “tamper proof” Torx screws I had my answer as to why: one cell had gone out of whack and fallen to 1.2 volts or so.

I tried boosting it up a little to see if it’d come back but the charger only ran for about a minute. I know, terrible idea, yada yada…

The protection/balancing board is neat. It’s got some kind of serial data connection on it (which doesn’t go to the tool, it’s for factory test/calibration) and two BIIIIG MOSFETs on a heatsink that I’ll be stealing for use in other adventures. But alas, this pack, it’s pining for the fjords.

Curious condition in which the battery sees the charger but the charger doesn't see the battery.

Lithium ion batteries are kinda known to have a shelf life, so it’s not entirely unexpected that it’d go after a few years.

If tubes could speak…

This one’s probably been around the block a few times. I was initially kinda worried because the fins looked roughed up but after some unscientific testing to make sure I could blow through them and fit a cable tie down between the fins to ensure there was a good pathway, I went ahead and installed it, and it came up just fine with a perfectly good stack temperature…






These holes seem to be a common feature of some part of the remanufacturing process

The old tube went soft pretty quick. Don’t they have pills for that now?



And here’s what that goes in, a Harris/GatesAir 25,000 watt FM transmitter.



To swap it, you release a hose clamp holding a big collet on the movable top plate of the cavity so you can slide that plate blocker/chimney up, then release the hose clamp holding it to the tube (bottom), slide it up then clamp it again out of the way. Then you can carefully disconnect the plate voltage supply cap and somersault the tube upside down above the socket and remove it. I wish I’d taken a video of how bizarre this all looks.

Cooling air is forced both through holes in the socket and out of the cavity through the tube’s plate cooler after entering from the blower duct.

The two big fat orange wires – GatesAir / Gates / Harris Broadcast really likes that fat orange wire – are the filament supply, 10vdc Max at 150 amps. The HV, ~9500vdc at about 3 amps and change, comes in on a piece of RG-213 coax from this off board power supply.

First person shorting stick wielding view
Rectifier modules

A little dust but not too bad. The dust seemed to really like gathering on anything live with plate voltage. Eww. A more through cleaning will occur soon.

And the award for silliest front panel design goes to….

The QEI Model 691 FM modulation monitor!


It always makes me think of Pokey The Penguin comics.

Please feel free to use this photograph to torture graphic designers.

Sorry about the stray marks, I’ve been asked to respect the secret identity of my client. You’re not cleared for that. Fnord!

Power Line Harmonics

So you may have seen me yelling about power line harmonics… here’s what I was looking at earlier this morning. This is the power at a facility I was doing some work at earlier today. The same power has laid waste to two variable frequency drive units and an Omron 24v power supply used to run a Programmable Logic Controller (PLC).

I used the FFT mode on my Tektronix DPO 2012 oscilloscope to better detail what’s going on. The yellow trace is the waveform coming in (you can see it’s not a perfect sine wave). The red trace is a plot of the frequency vs. amplitude after a Fourier transform. This same kind of plot may already be familiar to you: it’s displayed on many audio players and stereo systems as a spectrum analyzer visualization.

If this were perfect clean 60 hz power, I would have only the tall peak at the left which is 60 hz, then it would roll off to the baseline. What I got, however, is this (and a realization I accidentally set my scope’s clock 12 hours off *before* DST kicked in…)

Harmonic Horrors


The FOUR other peaks are the harmonics. Initially I thought I was looking at a second, third, fourth, and fifth harmonic, which would put the last one at 300 hz– reading this again though, it’s scaled 500 hz/division so what I’m looking at is the odd order harmonics at up to 540 hz– a NINTH harmonic.


Now let’s look at that third harmonic, the most prominent. That one’s about 20dB down from the fundamental (the scale is 20dB per vertical division). That’s a voltage ratio of 0.1… so the third harmonic, if you were to isolate it, would be 12 volts at 180 hz. That just plain doesn’t belong!!!

A textbook square wave is the fundamental frequency plus all of its odd-order harmonics– 3, 5, 7, 9…. I suppose this makes sense, as if you smashed this AC waveform very very badly you would get a square wave.

Power line harmonics are an annoying effect of loads with a poor power factor. Want an example of a load with a poor power factor? You’re looking at it. The wall chargers for smartphones, power supplies for computers, and even LED, florescent, and compact florescent light bulbs are guilty due to the nature of their power supplies. Without getting too much into the theory I’ll say this: the way they work is that they draw power right at the peaks of the usual sine wave power. This is why the peaks in the above screenshot are getting smashed: that’s when EVERYTHING draws its juice.

Some *nicer* devices incorporate power factor correction circuitry to mitigate this. This usually takes the form of a low pass filter that acts on the amount of current drawn by the device, in an effort to keep it from just suddenly grabbing only the peak. Note that I say— “some”—

The harmonics tend to overstress everything by causing high currents to flow in wires, transformers, power supplies, etc– they are not only harsh to the equipment but they are a waste of energy.

And here, boy, do we ever have ’em.

The solution, ultimately, will probably be to have the local power company install a bank of capacitors for power factor correction on their poles.

Utillity PFC capacitors. Photo from a Cooper Power Systems press release.

Then, hopefully, the power will stop being quite so…….. hungry for electronic snacks……

Another “South Florida Sucks” post, explosively inspired, plus other ramblings

I find the following amazing:

Things you can buy locally in South Florida without waiting a week for shipping!

  • Three phase 240V 2 horsepower rated variable frequency drive controllers (albeit at Grainger’s ~250% markup)
  • Excessively overpriced fuses

Things you cannot buy locally in South Florida without waiting for shipping:

  • Fuse holders
  • Single phase variable frequency drive controllers (thankfully the machine I’m working on has 3 phase available)
  • Anything that has to be delivered via UPS
  • Fine stranded 16 gauge electrical wire (as opposed to the thick stranded THHN type stuff you’d run through conduits in a wall)

I had a fun time calling around this morning to find this out– Grainger only has two locations left around here, and even the sales lady on the phone was surprised by this. She was used to South Florida having a ton of them a few years back. I first learned of the Richmond Heights one having closed up shop when my last brush with high energy explosion happened… Luckily the one remaining location happened to have a VFD in stock I could use. Phew.

Grainger were the only ones who would be able to get the fuse holder here next day because they had them in a warehouse in Orlando (they also have their own trucks, so I’m confident it didn’t get melted down in Hialeah in the UPS warehouse fire). They had TWO in stock, dashing my plans to get three matching ones for the 3-phase VFD— oh well, different fuse holders but same fuses, whatever works for now. A 3 pole fuse holder would have had to ride the slow boat from China, just like if I’d wanted three of the single pole ones.

What bugs me is I USED to have the exact fuse holders I would have needed but threw them out last year, trying to reduce the size of my collection of parts I thought would be easy to buy again if needed.


Anyway, here’s the beautiful mess that inspired me to get very mad this morning while standing in front of a voltageless robot.


The problem: VFD controller smoked out. I didn’t take any pictures of the outside of the controller for it looked unremarkable save the “MOFF” error flashing. It looked like “ПOFF” due to the limitation of it having only a 7-segment readout, more rambling on that later. It might as well have said “F.OFF” though. The error meant it wasn’t getting sufficient voltage on the internal DC buss.


A quick primer on how VFD drives work:

On this inverter, the high energy semiconductor devices for the rectifier and the inverter were neatly built into one easy to deploy little module by Infineon, an EasyPIM series module.

I’ve seen similar modules in several VFD drive products. They contain multiple transistors and diodes coupled thermally (but not electrically) to a backing plate for heatsinking purposes. To avoid undue stress to the semiconductor junctions, they are filled with a curious sort of…. snot.

Sorry if I’m grossing you out, but touch it sometime, it’s really reminiscent of a sticky booger.

A similar product to that is Raychem’s “GelGuard” which I’ve seen used in outdoor telephone connectors. Very weird stuff.

Here is a diagram of the EasyPIM, borrowed off the Infineon website.

Here’s… what was left of the module… This is 180 degrees to the layout of the diagram above, more or less, and I picked away the blackened snot to reveal what…. DID NOT remain…

The obliterated parts are the rectifier diodes. Even though only four of the six were used in circuit due to this being a single phase drive, the remaining two were blasted to smithereens by excessive reverse voltage. The IGBTs at left (right in the diagram) that control the outputs look pristine and would probably even test good. Sorry, too lazy to whip out the ESR meter. (Does that even work on high voltage IGBTs?)

The CPU block in the above diagram is the timekeeper of the operation. It reads in whatever form of commands the unit is set to accept as to which direction and speed to run the motor. It also monitors several analog values throughout the unit – voltages, current levels, and temperatures – and can adjust its operation to protect itself and the motor if necessary by throttling back the amount of power it sends to the motor or shutting down if necessary. In this case, it was successfully detecting the lack of high voltage DC to run the drive and just wouldn’t do nuffin’.

Now if the damage had been confined to the module – that’s actually a standard part from Infineon and could probably just be ordered, buuuut—-

Fomp. I like to imagine this failure went… “Fomp”.


Yeah, this let it all right. It almost looks like the conformal coating tried to contain the Magic Smoke, and maybe even flashed over? The two line fuses for the drive were blown so something fun must have happened on the board itself. If the rectifier had just failed short, I would have expected ONE of the two to have gone, not both simultaneously. Flash to ground probably got the other one.

The client has been having a drive blow up every three months or so!! I’m going to retrofit the machine for him with some BIG HONKIN’ SURGE PROTECTION. After his first VFD blowout he called in Florida Plunder & Loot who did *something* to adjust his line voltage, but it’s still mega nasty. I’ll put the scope across it before installing the new drive and likely cringe so hard my face freezes that way for life.

Amazingly the drive still powered up, scrolled “HELLO” before going to “MOFF”, and I was able to get all the settings from it via the front panel user interface. See the small transformer at top left on the board? I suspect there’s a separate switcher power supply just for the CPU and logic that allowed for this even though the HVDC system was blown to bits.

And herein is another bit of rambling: The user interface on VFDs, for the most part, royally sucks. You’re presented with a menu system on the front panel which gives you anywhere from about a dozen to *HUNDREDS* of parameters, identified only by number. You have to break open the manual and go through dozens if not hundreds of pages of crap if your setup is in any way complicated. Fortunately THIS ONE IS NOT. This one was simply being driven by a +24vdc trigger to turn the motor on, and 0-10 volts DC to control the frequency between 0 and 60 hz. But still— Nice features I’ve seen in some units: Danfoss and Cutler-Hammer VFDs have a small control module that speaks plain English to you. Some other controllers (I’m unsure what brands off the top of my head) save the parameters to a small serial EEPROM or even an industry standard memory card, so if your drive does— this— you just swap the card into the new one and you’re ready to spin again.


An important word on the “Death Caps”:


DEATH CAPS. No bleeder resistor is provided.

The capacitors inside a VFD drive will retain potentially lethal voltages FOR A VERY LONG TIME after power off. Even if they are discharged, they may exhibit a peculiar self-recharging phenomena where a stored charge in the electrolyte partially recharges the cap to dangerous levels again. Always discharge caps and test before touching! On a small self contained drive you cannot get in contact with the caps without disassembling the unit; however, I’m aware of there being larger units where the DC section is a separate unit and those can probably really get you. TEST FIRST. Also, if you are storing a spare unit, be sure to check the manufacturer’s directions on maintaining the capacitors!! Most say to power the drive up every year or so to keep the capacitors “formed”. If you suddenly connect power to a drive that’s been stored too long, EXPLOSIVE capacitor failure may occur. If you’ve got one that’s been sitting, consult the manufacturer’s directions for reforming. If unsure, use a Variac to slowly bring the voltage up, or put a small incandescent light bulb in series to limit the current. (The drive should draw almost nothing if it’s just powered up and not doing anything.)

Oops I think I accidentally made myself a rave flyer

Every time I start messing around thinking I want to make business cards for myself, THIS KIND OF THING HAPPENS


but believe me the music will be more interesting than you’d expect
1 2 3 4 63