The end of tubes?

[tubelab.com]

It seems the last bastion of tubes is high power high frequency RF, and that niche may be closing to SS technology.

In many places the solid state take over is eminent. I work as a RF transmitter designer at Motorola. We used to manufacture product at the facility where I work. I started there in 1973 and all the RF stuff was already solid state. There was one product manufactured that used a pair of 12AX7's because it went in aircraft, and the certification process for a new design takes years. It was cancelled in about 1975.

Our transmitters used silicon then GaAs and now GaN. No SiC yet, costs too much and doesn't do 1 GHz efficiently yet.

On the flip side of that coin.......

The USA just converted all high powered analog TV over to digital.

In the early days of TV the lowest numbered TV channels (2 through 6...54 to 88 MHz)were the ones all the broadcasters wanted. Why? Transmitter technology could make good power, feedline losses at the TX and RX were lower, and RX noise figures were about 10db. Next used was channels 7 through 13....(174 to 216 MHz). TX power was lower, feedlines were lossier, and RX noise figures were higher. The UHF spectrum was allocated after the Klystron and the IOT were developed (470 to 890 MHz) but feedline losses and 20 db RX noise figures kept it from widespread use. In the early 60's some school systems used the low UHF channels to broadcast educational programming to schools. Miami was one. 500 kids in a non airconditioned auditorium watching some boring teacher on TV, yeah I skipped that class.....

When the digital TV revolution came, every TV market got the opportunity to refarm all the channels. Guess what happened. In most big city markets the lower channels were abandoned, almost all broadcasters switched to UHF. Why did all the broadcasters switch to UHF? Because of the NEW, and continued advances in VACUUM TUBE TECHNOLOGY.

Note the channel number displayed on your TV bears no resemblance th the actual RF channel in use, there is a mapping function. That's why new TV's have to scan all the channels to build the map.

The FCC authorizes ERP (effective radiated power) up to 1 MEGAWATT in the TV spectrum. ERP is TX RF power, minus feedline losses, multiplied by the antenna gain. So at 54 MHz a 50 foot antenna might have 12 db of gain. To get 1 MW with 2 db of feedline loss, you need a 100KW transmitter. At 500 MHz a 50 foot antenna might have 27 db of gain. To get 1MW with 7 db of feedline loss you need 10 KW. Assuming equal TX efficiency which electric bill would you want to pay.

Back to the TUBES, UHF TV stations used to use klystrons. Both analog and digital TV transmitters require linear transmitters so a linear (AB1) klystron might make 10% efficiency. RCA developed the IOT (inductive output tube) in the 1940's and it saw limited use, but continuing refinement has led to the current design capable of 50% efficiency at 10 KW. Show me the silicon (or whatever) that does that! Now that all the TV stations recently bought shiney new transmitters powered by a vacuum tube output device, I think the tubes will be around for a while, even in the RF world.

[Bigun]

Originally Posted by tubelab.com

Our transmitters used silicon then GaAs and now GaN. No SiC yet, costs too much and doesn't do 1 GHz efficiently yet.

I've had engineers tell me that GaN is likely to be a better option than SiC for many applications and may in fact cut short the life of SiC technology.

[Triodethom]

Tubelab when you speak of kystron I am always reminded of lost planet . Robby the robot would be so proud. I take it they are water cooled ?

[smoking-amp]

Danger! Danger! Will Rogers... We must return to the ship immediately.

[HollowState]

Originally Posted by smoking-amp

Danger! Danger! Will Rogers... We must return to the ship immediately.

That was Lost In Space. I think Triodethom was refering to Forbiden Planet and the klystron frequency modulator that came packed in liquid borne (whatever that is), and was wrecked by the monster from the id.

[tubelab.com]

I take it they are water cooled ?

Some of the newer IOT's are internally oil cooled. They all are cooled by some sort of liquid in a closed loop heat exchanger system. I don't know what the liquid is. Water is OK for cooling your gaming computer. There may be better options for high temp applications.

I've had engineers tell me that GaN is likely to be a better option than SiC for many applications

Gallium Nitride (GaN) devices currently operate in the 28 to 120 volt range. That is the sweet spot for a broadband matching network to a 50 ohm system. This is key to building RF power amps that cover several octaves of bandwidth. Military tactical radio systems often cover 50 MHz to 2.5 GHz. A single GaN device can do this.

Gallium Arsenide (GaAs) devices opreate at low voltages, and tend to operate over a narrow frequency range. GaAs used to be very expensive due to the extreme purity requirements for acceptable yields. LED's drove down the cost of raw material processing. Cellular phones and satellite TV drove down the cost of GaAs RF. There are now several speciallized GaAs RF IC's for those markets. A complete dual band RF power amp for a cell phone cost under $1.

Silicon Carbide devices for high power tends to want a higher voltage which makes a broadband match much more difficult and lossy. This makes them good for big powered amps that cover an octave or less.

There has been a continuous advancement in silicon LDMOS technology so that you can now get a single device that puts out a killowatt of power from 2 to 50 MHz. These are still very low impedance devices requiring a speciallized RF transformer to match the output to 50 ohms. LDMOS is still the most linear technology available for high linearity requirements like LTE and driving those big tubes in a TV transmitter.

So today an RF engineer has a wide pallet of sand to choose from with new stuff appearing every day. Often the technology choice is dictated by the application. The lines get blurrier every day.

[Nielltronix]

People keep saying that there will someday be 500 or 600v GaN devices, but it's all vaporware (excuse the pun). If/when that happens, the TWT/klystron/IOT market will see some competition. But again, yield is low on these devices, so electron tubes are going to be around for a good long while still...

[Triodethom]

Originally Posted by HollowState

That was Lost In Space. I think Triodethom was refering to Forbiden Planet and the klystron frequency modulator that came packed in liquid borne (whatever that is), and was wrecked by the monster from the id.

Thanks it was Forbiden Planet rob the robot was the prototype for the robot from lost in space and did show up in at least one episode .

[tubelab.com]

there will someday be 500 or 600v GaN devices, but it's all vaporware (excuse the pun).

I don't know enough about solid state physics to know if this is even possible, but I do know this.......follow thw money. Seniconductor device makers go where the money is. To a new process like GaN looking for possible homes the best customer is the government, because they have lots of taxpayer dollars to spend. Here you get big bux for your chips and your production volumes are low. A bad lot or low yield will not kill you. This is where most of the GaN volume is sold today. The stuff is so new that computer models don't exist, or are not accurate. If you are making parts for the million a month low margin cell phone market, you better have a stable well characterized and controlled process with good yields on large wafers.

If there is a good sized untapped market for 500 volt GaN devices, somebody who fully understands GaN device fabrication will go after it. Trust me....there is no GaN vendor ready for this today. We are dealing with a new market, with new stuff coming from the leading GaN vendors every month, and we are still seeing 48 volt devices that blow up under realistic operating conditions.

Note. GaN devices need to built on top of a thermally conductive substrate since GaN has poor thermal characteristics. Often that substrate is silicon or SiC.

[7N7]

In 1972 I was privileged to view a Klystron that stood about 7 feet (210cm) tall, which was capable of delivering 8MW pulsed at X-band (1kW continuous). God knows what they must have cost the British taxpayer...

Can anyone tell me what sort of SS device can replace that? (And by that I do not mean ways of pouring money down the drain: governments never need prompting in that regard).

7N7