Bird Termaline RF dummy load teardown

[Bud]

Got this puppy some years back for almost nothing. The seller said it was sitting on the shelf for 25 years or something. Turned out the oil was totally gone and the load was dry.  The resistor measured just fine at DC but RF performance was not good. Took it apart to clean possible rust but the guts were very clean. After filling the load with mineral oil from the drug store it came back to life as per the specs well into a few GHz  . Attached are also the Smith charts before and after putting the oil in. The design  of the load accounts for oil's dielectric constant, so SWR was poor with no oil  .

[TheSteve]

So clean inside and out! I have some ancient Bird loads and attenuators too. No lab is complete without them. Now you need a nice kilowatt or even the 2.5 kilowatt model.

[w2aew]

Cool - I never had one of those apart, so it's interesting to see what it looks like.  I have one that has leaked a bit, so I need to service it.  How much oil did it take?

[Bud]

Approximately 300mL. There was plenty left in that 500mL bottle.

Edit: found in my records that Bird uses "Shell Diala S2 ZX-A" transformer oil. But they refer to it by their part number "Bird #5-30 dielectric coolant", as per the datasheet (attached).

[G0HZU]

The big old 1kW Bird attenuator I have here is the 1000-WA model. It is quite easy to take apart but it does contain little attenuator segments that look to be formed on white Beryllium Oxide carriers. I don't think these are meant to be messed with.

https://birdrf.com/en/Products/Test%20and%20Measurement/Attenuators/1kW/1000-WA-Series_RF-Attenuators.aspx

It is quite interesting to see inside it. See the image below. This one was a custom version with 40dB attenuation rather than the usual 10, 20 or 30dB catalogue versions. To achieve the 1kW rating each segment along the line has a slightly higher attenuation than the one before it. I think the first attenuator segment is only about 0.4dB. This means the attenuator is directional, the high power always has to be connected to the same end or the attenuator can be damaged. If used correctly the attenuator should heat up evenly along its full length. The excess solder blobs you can see in places are caused by the operator who fitted each segment. The segments are soldered into the copper heat spreader sections like a huge surface mount component. I guess if any excess solder forms a blob they just leave it there. It is quite messy in places in this respect.

The performance of this one is extremely impressive, the input VSWR is better than 1.01:1 up to about 250MHz. It degrades to 1.03:1 by 433MHz but this is still a really good attenuator considering it is rated at 1kW and is absolutely huge and very heavy.

[TheSteve]

Now I might have to measure the return loss of my Bird 30 dB attenuator

[0culus]

It's good to know you can charge these with drug store mineral oil and get good impedance results though. 

[Bud]

Yes, that means the dielectric constant of the mineral oil is close to that of the original oil. I actually bought two bottles - one "Heavy Oil" that I used to take the measurements, and the second one was "Lite Oil". I wanted to try both but have not got to testing the Lite one yet. They differ in viscosity and I planned to strap a thermal sensor to the load's barrel , apply 100W of RF and measure the temperature rise ramp for the heavy and then for lite oils. I'd think the oil which transfer heat quicker from the resistor to the heatsink would be better to use.

[G0HZU]

I looked up the specs for your 81B load and it has an impressive frequency range and can manage a VSWR of <1.2:1 at 4GHz. That was a surprise for me.

The WA-1000 1kW attenuator shown in my earlier post is a monster size at about 50cms (20 inches) long and although it achieves really low VSWR up to 500MHz it starts to degrade above this. By 1000MHz the VSWR is climbing through 1.1:1 and although it is spec'd to 2.5GHz the VSWR plot starts to look like a rollercoaster above 1.5GHz. It just manages to stay below 1.2:1 by 2.5GHz but it's obvious that those huge resistive pi and tee attenuator sections are limiting the performance above 1.5GHz. I've not tried to measure the VSWR up at 4GHz but I think the result would not be good.

[Bud]

Perhaps parallel capacitance builds up, and fringing capacitance, due to large physical size of the structure.