Breaking Input Stage of Spectrum Analyser with Radio TX

[CatalinaWOW]

My comment was actually more directed at the requirement for a chiller.  But I agree with Joe Smith.  Unless there is some voltage or current limit on the attenuator it is difficult to understand why the power capability of an attenuator is required to be more than the power dissipated in the attenuator.  And the wire sizes and insulation strengths require to deal with 100 W at just about any impedance seem unlikely to be a design limit.

[xrunner]

My comment was actually more directed at the requirement for a chiller.  But I agree with Joe Smith.  Unless there is some voltage or current limit on the attenuator it is difficult to understand why the power capability of an attenuator is required to be more than the power dissipated in the attenuator.  And the wire sizes and insulation strengths require to deal with 100 W at just about any impedance seem unlikely to be a design limit.

Well all I can suggest to you is to do more reading on the matter.

[joeqsmith]

Using your example of 100W and 2X 10dB, assuming T types, you have a 26 input, 35 to ground and 25 out.   Two in series and terminated to 50.   Working backwards, thats 76 in parallel with  35 or 24.  Thats in series with the 2X 26 or 52, or 76 total.  Again, in parallel with the next 35.   So 25 in series with the 24 or 49 total.   Sure, lots of rounding.   

100W or 70.7V /49 or 1.44A   I^2 R or 52.8W on the first resistor.  The other on the remaining parts.   Consider the attenuator is not a bunch of resistors but screened on to some BeO, down to copper...  Doubt there is much or a thermal gradient across it.   

Still the parts do need to handle the voltage and current. 

****
Consider the same two 10dB attenuators in series but unterminated.   Now we have a 35 in series with the 2X 26 all in parallel with the 35.   Or 87//35 or 25 ohms.  Thats in series with the 25 ohm input or 50 ohms total.   

So yes, 2 unterminated 10dB attenuators in series would appear to make a decent load.   

****
Added closeup view of a damaged 100W JFW attenuator I bought off eBay, sold as working.    

****
Added a few photos showing temperatures after replacing the attenuator with a part from Barry.   

[joeqsmith]

Most attenuators will have a temperature derate curve.  Shown is the repaired 100W attenuator in series with a few others where I am applying 300W in order to get an idea of temperature rise from some cheap Chinese UHF connectors.   The larger attenuator is only rated to 300W.