Rebuilt tubes

As broadcasters, we don’t really hear that much about ceramic power vacuum tubes these days, as more and more broadcast transmitters migrate to solid state devices.  Once upon a time, however, power tubes where the engine that drove the entire operation.  Tubes had to be budgeted for, stocked, rotated and replaced on a regular schedule.  Some of those dern things were expensive too.

Take the 4CX35,000A which was used in the Harris MW50 transmitters.  The transmitter used two of these tubes, one in the RF section and one it the modulator.  As I recall, new tubes cost somewhere north of $8,000.00 each from EIMAC.  Plus, in the A models there were two 4CX1500A driver tubes.  All of which could add up to an expensive maintenance cost every two years or so.

The next best option was to buy rebuilt tubes.  Rebuilt tubes were about half the cost of brand new ones.  Some people complain that rebuilds don’t last as long, or only last half as long as the new tubes.  I never found that to be the case.  I often found other factors effected tube life far greater, such as filament voltage management, cooling and by extension, cleanliness.

I can say I never had a warranty issue with ECONCO tubes.  I cannot say that about EIMAC, as during the late 90’s and early 00’s (or whatever you call that decade) I had several brand new 4CX3500 tubes that were bad right out of the box.  These days, ECONCO and EIMAC are both owned by CPI.

I spoke with John Canevari from ECONCO who had a lot of information.  For example, as the tube ages, the filament gets more flexible, not less.  Most ceramic power tubes use a carbonized tungsten filament containing some small amount of thorium.  As the tube ages, the filament can no longer boil off enough electrons and the emission begins to drop off.  That is the normal end of life for a power tube.  Occasionally, some catastrophic failure will occur.

There are many steps in the rebuilding process:

  1. Dud is received from the field, the serial number is recorded and the tube is tested in.
  2. The tube is prepped by sand blasting the sealing rings
  3. It is opened
  4. Filament is replaced.  In 60-70% of the cases, the grid is replaced.  In those tubes that have a screen assembly, 20-60% of those will be replaced.
  5. Interior of the tube is cleaned
  6. Tube is resealed and tested for leaks with a gas spectrometer
  7. Tube is placed on the vacuum machine.  Tubes are evacuated hot, smaller tubes take 12 to 24 hours, very large tubes can take up to one week.
  8. The tube is nipped off of the vacuum while still hot.  When the tube is fully cooled the vacuum scale is normally around 10-12
  9. Exterior of tube is cleaned and replated.  Silver for tubes that are socketed and Nickel for tubes that have leads.
  10. Tube is retested to manufacture’s original specification or greater.

After that, the tube is sent back to its owner or returned to stock.  John mentioned that they are very proud of there vacuum tube processing machines, so I asked if he could send along a picture.  They certainly look impressive to me, too:

vacuum tube processing machine
Vacuum tube processing machine, photo courtesy of ECONCO

Not exactly sure which tube type these are, but they sure to look like 4CX15,000:

vacuum pump on rebuilt ceramic power tubes
Vacuum pump working on rebuilt ceramic power tubes, photo courtesy ECONCO

Econco has been in business since 1968 and rebuilds about 600-1,000 tubes per month.  In the past, broadcasters used most of the larger tube types.  However, with the majority of broadcast transmitters shifting to solid state, other markets have opened up such as industrial heating, military, research and medical equipment.

Nautel YouTube channel

Nautel, Ltd., which always seems to be a forward thinking company, has their very own YouTube channel. Okay, that is not such a big deal, I have my own YouTube channel too.

There are lots of video on how to configure a NV transmitter, which, as I have said before, fancy GUIs are all well and good, I am more concerned with the MOSFETS generating the RF.  I would forgo the GUI in favor of more more reliability, but that’s just me.

I would put the DRM+ video up on the blog, but the embed function is disabled.  Anyway, I enjoyed watching many of these, you might too.

Filament Voltage Management

4CX35,000C ceramic vacuum tube
4CX35,000C ceramic vacuum tube

There are still many hollow state (AKA tube type) transmitters floating around out there in the broadcast world.  High power, especially high power FM transmitters are often tube types and there are many good attributes to a tube transmitter.  They are rugged, efficient and many of the well designed tube units can last 20-25 years if well maintained.

There downside of a tube transmitter is tube replacement.  Ceramic tubes, like a 4CX20,000 or 4CX35,000C cost $6-9K depending on manufacture.  A well maintained tube and last 3-4 years, I have had some lasting 8 years or more.  My personal record was for a 4CX35,000C that was a final PA tube in a Harris MW50A transmitter.  The tube was made by EEV (English Electrical Valve, now known as E2V)  and lasted approximately 84,000 hours, which is 9.58 years.  When it finally came out of service it looked like it had been through a fire, the entire metal plate body was dark blue.  I took it out because the power was beginning to drop a little and it was making me nervous.

This was not accident, I did it by maintaining the filament voltage, keeping the tube and transmitter clean.  The tube filament supplies the raw material for signal amplification.  Basically, the filament boils off electrons, which are then accelerated at various rates and intensity toward the plate by various control grids.  The plate then collects the amplified signal and couples it to the rest of the transmitter.  When a tube goes “soft,” it has used up its filament.

I had a long conversation about this one day with Fred Riley, from Continental Electronics, likely the best transmitter engineer I have ever known.  At the time, the consensus was to lower the tube filament voltage no more than 10%.  On the 4CX35,000C, the specified filament voltage is 10 volts, therefore, making it 9 volts was the standard procedure.  What Fred recommended was to find the performance “knee,” in other words, where the power began to drop off as the filament voltage is lowered.  Once that was determined, set the voltage 1/10 of a volt higher.  I ended up running that EEV tube at 8.6 volts, which was as low as the MW50’s filament rheostat would go.

The other important thing about tubes is the break in period.  When installing a new tube, it is important to run only filament voltage for an hour or two before turning on the plate voltage.  This will allow the getter to degas the tube.  New tubes should be run at full filament voltage for about 100 hours or so before the voltage is reduced.

Tube changing procedure:

  1. Remove power from transmitter, discharge all power supply caps to ground, hang the ground stick on the HV power supply.
  2. Remove tube, follow manufacture’s procedures.  Most ceramic tubes come straight up out of their sockets (no twisting).
  3. Inspect socket for dirt and broken finger stock.  Clean as needed.  Finger stock, particularly in the grid section, is important for transferring RF.  Broken fingers can lead to spurs and other bad things
  4. Insert new tube, follow manufactures recommendations.  Ceramic tubes usually go straight down, no twisting.
  5. Make all connections, remove grounding stick, half tap plate voltage supply if possible, close up transmitter
  6. Turn on filaments and set voltage for manufactures recommended setting.  Wait at least 90 minutes, preferably longer.
  7. Turn on plate voltage and tune transmitter.  Tune grid for maximum current and or minimum reflected power in the IPA.  PA tuning should see a marked dip in the PA current.  Tune for dip, then load for maximum power.
  8. Turn off transmitter, retap plate supply for full voltage
  9. Turn on transmitter and plate supply, retune for best forward power/efficiency ratio.
  10. After the 100 hour mark, reduce filament voltage to 1/10 volt above performance knee.

Of course, every transmitter is slightly different.  There may not be a dip in the plate current if the transmitter is running near it’s name plate rating, in which case one would tune for maximum forward power.

This system works well, currently one of the radio stations we contract for has a BE FM20T with a 4CX15,000A that has 9 years on it, still going strong.

Broadbanded AM antennas

Partly for my own edification, partly just because, here is some information about AM antenna systems and their bandwidth. An AM tower is a radiator which, simply by the physical constraints of the tower structure itself, is pretty narrow banded, even under the best conditions. Add to that, antenna tuning units, transmission line phasing, antenna phasing units, diplexing units and things can get very squished outside of the immediate carrier frequency. This seems to be a particular problem with directional antennas, which most AM stations employ.

WGY 810 kHz, Schenectady, NY transmitting tower w/open feed line
WGY 810 kHz, Schenectady, NY transmitting tower with open transmission line

As an engineer, you can get some idea of how narrow an antenna system’s bandwidth is by looking at the base impedance measurement.  Every AM station is required to keep the latest impedance measurement on file.  When looking at these measurements, there will be on curve which indicates base resistance (R) and another curve that indicates reactance ( X, although often noted as + or -j).  If the resistance and or reactance curve is slopped steeply at the carrier frequency and out to 20-30 kHz, it is a narrow tower.  Add to that the differing phase shifts of an ATU and or Phasor and things will be compounded.  That is why it takes a professional to design and tune up these things, a poor design will never sound right.

Another way to get some idea of bandwidth requires a field strength meter.  Modulate the transmitter with a 10 kHz tone at 50% modulation.  Then, away from the near field, measure the carrier and 10 kHz +/- the carrier frequency on the log scale.  The side bands should be symmetrical and about 1/4 the carrier level.

Generally speaking, antenna systems need to be designed for low VSWR across the entire side band range (+/- 10 kHz from carrier) as well as symmetrical distribution of radiated energy across the lower and upper sidebands.  Several factors influence these conditions:

  1. Electrical tower height, perhaps the hardest thing to change once a tower is constructed.  Short towers (less than 80 electrical degrees), or very tall towers, (taller than 200 electrical degrees) present problems.  If one where constructing an AM station and could choose any tower height, something between 120 to 190 electrical degrees would be ideal.  Existing towers can be top loaded to add electrical height for an additional 30 degrees or so.  Beyond 30 degrees it becomes difficult to physically attain and therefore impractical in most situations.  Top loading and bottom loading a tower can reduce bandwidth if done improperly.   Bottom loading an AM tower is almost never done due to the very high voltage and current as the electrical length approaches 180°.
  2. Antenna matching networks can greatly improve or degrade bandwidth, depending on how they are designed.  A T matching network has more parts and is more expensive, however, it allows for optimum control over the R and jX phasing.  This becomes much more difficult with directional antenna where phase considerations are a part of the stations antenna field pattern development.
  3. Phasors present the biggest challenge, particularly in the power divider sections.  A tank circuit power divider is the worst choice, a shunt circuit power divider is the best bandwidth choice, however, it is the hardest to conceptualize.

Obviously, the more complicated the antenna system, the harder it will be to keep the bandwidth open over 20 kHz of spectrum.  This is especially true on lower frequency AM signals, where the bandwidth is a much larger percentage of the frequency.   Multiple pattern, multiple tower DAs are a nightmare.  Single tower non-directional stations are the easiest to modify.

As far as the circuit itself, higher Q circuits have smaller bandwidths.  Simply stated, in an alternating current circuit, Q=X/R.  The better the reduction of X, which also has a lot to do with the relationship of the current and voltage phasing, the better the Q will be.  This is why a T network is the best design for an ATU.  With a 90° or 180° tower, this is relatively straight forward.  In towers that are shorter or taller than that, it becomes more difficult as the value of R becomes less friendly.

In most cases, some sort of L/C network can be deployed to decrease the Q of an antenna system at the base of the tower.  Directional stations also need to have the phasing equipment looked at, because, as noted above, certain designs can created bandwidth bottlenecks.  All in all, it is usually an expensive proposition for a multi tower directional station to broadband it’s antenna system.  This is another reason why IBOC on AM is destined to fail, many AM towers cannot pass the extended sidebands adequately.