High power attenuator

[radiogareth]

Needing something to provide some protection for my R&S CRTU-RU test unit (nominally 50 watts in max.....) I found this...
https://www.ebay.co.uk/itm/Lucas-Weinsc ... Sw3ANb1ssL
As it looked nice and solid and with N-types I took a chance and bought one.
Picture shows the insides. With a spec as shown I'd guess that its pulse rated (radar??) but for me it will be CW or DATV. Fortunately 100 watt tabbed thick film resistors are available (ali-express for me) and with a little drilling and tapping they will fit nicely under the input/output wires (on RHS of picture. A bit of heatsink+fan on the bottom side and I should have a -20dB attenuator that can cope with high power. It might even improve the bandwidth spec??
Gareth

2020-05-18 20.38.52.jpg (104.21 KiB) Viewed 3224 times

[g0mjw]

It looks nicely made. It will be OK at HF/VHF but clearly not designed for microwave frequencies. Did you measure the performance? It's coaxial and may be quite good.

Pulse ratings are common for radar components. Peak pulse power ratings of several kW are obtainable from quite small components, especially valves, as long as peak voltage and current ratings are not exceeded. I think the venerable 4CX250K was rated to 8 kW.

If you want a high power, high value, VHF attenuator and have not got one, a high power dummy load and resistive tap works fairly well. That unit you have there might be a good starting point if you change one of the resistors for a 100W tabbed 50 ohm load and the coupling resistor for something appropriate.

Mike

[radiogareth]

So I got round to doing the work on this unit today. I bought a couple of 150W 'FloridaRF' 50R loads and filed the ends to fit a bit better. Drilled and tapped and then soldered in the new units.
I have attached the before and after plots from SATSAGEN showing I THINK somewhat variable attenuation vs frequency but very good VSWR 50MHz-6Ghz.
If I assume that SATSAGEN is 'reasonable' at a specific frequency, I can dial that amount of attenuation into my R&S and get 'about right' readings.
At least It can handle 150W now, although the case may get HOT at that level but nothing another heatsink would't solve. Next project is to mount some attenuators (100watt-20dB) on an old CPU heatsink along with some BNC in & out. Seen on a Russian 2m amp test, looked like a good way for HF/VHF.
Guidance and comments welcome....
Gareth

2020-06-26 12.59.52.jpg (125.86 KiB) Viewed 3042 times

Lucas_Stock_VSWR.jpg (124.47 KiB) Viewed 3043 times

Lucas_Moded_VSWR.jpg (120.7 KiB) Viewed 3043 times

Lucas_Stock_dBm.jpg (124.74 KiB) Viewed 3043 times

Lucas_Moded_dBm.jpg (116.69 KiB) Viewed 3043 times

[radiogareth]

Some band specific measuring this morning (satsagen). These are for the modified version, other than the traces published above I did not check each band. Of possible relevance is the presence of a small 'feed-through 100k resistor between the two N-type centre pins - it feeds through, to dissipate ?static charges??
all numbers in dBm
50Mhz -59!!
70MHz -56
144Mhz -46.4
435MHz -39.4
1240MHz -30 ?leakage?
2200MHz -27 2300-26 2340 -25 2370 -23 2380 -22 2400 -21 2420 -20 2440 -19 More leakage??
3400MHz -38
5600MHz -42 5700 -43 5800 -42.

Not exactly 'HP' standard then......

Gareth

[g4eml]

The feedthrough resistor is setting your attenuation.
With 100K the theoretical attenuation would be about 75 dB.
As you are not achieving that it would appear the coupling between the two sides is dominating.

It might be worth reducing the value of the resistor. This will reduce the attenuation but will probably give a flatter response. You could probably go as low as 1K which should give you about 30dB. It won’t be a perfectly matched attenuator but it would be usable.

Colin.

[radiogareth]

OK, but why the power resistors on the in and out side? They are both 50R 'input' (in to out of each socket) but if you meter through centre pin to centre pin I see 100R?
So I THINK I have -40dB power attenuator? With leakage, might try without the 100K 'feed-through' and see if it improves the apparent isolation, especially if I block the tiny hole with a screw).
Gareth

[g0mjw]

It can't work with a normal 100k resistor as it wouldn't look anything like a 100k resistor at 6 GHz. Better use a lower value. Match will not be too bad. You could go to 750 for a 30 dB attenuation perhaps but it's not going to work to very high frequencies due to series inductance/capacitance. For 6 GHz you need something else.

[g0mjw]

OK, but why the power resistors on the in and out side? They are both 50R 'input' (in to out of each socket) but if you meter through centre pin to centre pin I see 100R?

Gareth

100R centre pin to centre pin is what I would expect as you have two 50 ohm resistors in series via ground in parallel with a very high value resistor.

[g4eml]

If you take out the series resistor and block the hole you won’t have an attenuator, you will have two dummy loads in the same block of metal. There may be some coupling between them but that will be unpredictable and very frequency dependent.

In a Pi attenuator the series resistor provides the coupling between input and output. An ideal resistor would give a flat response but as Mike says any leaded component is going to have capacitance and inductance which will affect the frequency response.

Colin.

[radiogareth]

Thanks everyone It wasn't until I unsoldered the original resistors that I 'discovered' the third member. Everyday is a school day......I was thinking it was like the dual tabbed attenuators, in-out more than 50R but in-to-ground and out-to-ground both 50R.
I'll try something smaller as suggested and run another plot. Other than the lack of high accuracy, I still have a high power attenuator.....meanwhile I'm mounting this (pictured) on a heatsink. I'd guess it needs a neat machined cavity to be any use at high microwave frequencies....

20dB.PNG (375.56 KiB) Viewed 2918 times

Gareth