A DIY 1Ghz 30dB 250W attenuator design

[noreply]

I thought it’s time to share details of an attenuator I made for use in testing high power RF devices with a SSA


A 1Ghz 30dB 250W attenuator - DIY design


I have recently acquired a SSA (Siglent SSA3021X PLUS)

With the help of some fellow forum members my SSA is now a SVA (Siglent SVA 1032X)

If you are interested – here is the link …

https://www.eevblog.com/forum/testgear/siglent-ssa3000x-spectrum-analyzers/


I have a personal interest in RF design and to ‘test’ these devices – some of them > 150W of RF power – it’s something you DON’T WANT to ‘plug-into’ a SVA without the proper protection.


So, in my case the device under test (DUT) will be producing about 150W of RF at about 146MHz (centre freq).

To do a frequency vs. gain characterization – I need to drop the power to a reasonable level BEFORE it hits the input of the SVA.


This is why I decided to make my own 30dB 250W attenuator.


My design is very simple (see the various photos attached) and is based on a monolithic RES-NET 30dB attenuator device.


The biggest problem is to be able to dissipate the 250W

 
The attenuator circuit is most likely a PI network – so its symmetrical – there is no specific input or output connectors.


In my case I have used 2 x high quality N female connectors which are machined  (not ‘cast’ like most cheap connectors) which have direct PCB stripline ‘stubs’.


I also got a 120mm(l) x 70mm(w) x 45mm(h) aluminium heat sink capable of dissipating at least 250W with a small low CFM fan.


So, now that you have all the major components to make the 30dB attenuator, what’s the best way to do this?

– to produce a ‘flat’ response for at least 1GHz? – this is the specification for the RES-NET 30dB device.


My objective was to keep the ‘connections’ to ABSOLUTE minimum – and I managed to achieve this without major problems.


You will be able to see the complete construction via the attached photos (1 to 37).


Please note – the RES-NET device needs to be ‘soldered’ to N connectors BEFORE mounting on the heat sink.



You will need to drill and ‘tap’ 4 x M3 x 16 mm deep holes

– centred on the heat sink, the rest is very easy

– because the RES-NET and the N Connector geometries all worked out well (by design selection and not luck)


The N connectors are left ‘floating’ at one end (see photos)

– simply because the 4 x M3 Stainless Steel bolts have such great strength and bonded the RES-NET attenuator to the heat sink with absolute rigidity

– more than adequate to make a tight N connector fitment to the associated Male Plug when screwed into the attenuator's connectors.


One notable thing – is that the N connectors are in direct contact with the RES-NET attenuator.


This is good for additional thermal dissipation

- as 90% (educated guess) of heat dissipated will go to the heat sink

– because due to the great thermal bond of the flat / polished surfaces with thermal paste

– the thermal resistance is VERY low.


If we are pushing 250W and more

– the N connectors might get warm

– but should not affect the transmission.


I decided the benefit of direct solder connection to the RES-NET device was more important

– to maintain a flat frequency response

- rather than to use some offset fittings so that the N connectors would not be in direct contact with the RES-NET attenuator.


I have not performed a frequency / load response as yet – still arranging equipment in lab.


As soon as I manage to do this will post results here.


I do expect a fairly ‘flat’ response at least to 1GHz


This project will provide me with a 30dB of attenuation

– so will drop, say 200W of RF to 200mw whilst dissipating most of the energy into the heat sink

– at the output – I have an N to SMA connector with another 20dB inline attenuator which is rated at 2W (well above any power level which might connect to it)

– the total output attenuation will be 50dB

– so my SVA will receive a 2mw signal

– which I can further attenuate by another 30dB – using the SVA’s internal (programmable) attenuator.


This is the SAFEST approach when testing high power RF devices – a must have device for any lab which has SVA / SSA instrumentation


Professional 30dB 250W attenuators can cost well above US$600


The total cost of this device was about US$150 (this includes the exorbitant shipping costs for the components)
The RES-NET 30dB attenuator device alone was US$40 – not sure if it is a currently manufactured part – despite various suppliers having inventory.


I hope you enjoy the design.


It’s my ‘THANK YOU’ to various forum members who helped me with their knowledge!


Please follow the Pictures 1 to 36 for pictorial design sequence 



NEXT POST - has remaining 12 images (thought it best to show inline images rather than zip file

[noreply]

Remaining 12 images (thought it best to show inline images rather than zip file (which will be too big)

[noreply]

Attached is the sweep response for the 30dB 250W attenuator

It appears to be well within the 1Ghz Freq response specification 

[Gyro]

It would probably be better to dedicate this thread to your Attenuator (and title it accordingly). It looks as if it deserves one. Threads tend to get very confused when you have various projects being added and tying up feedback with each.

[Gyro]

From the mechanical perspective, it looks as if it would be a good idea to use standoffs to support the outer mounting holes of the connectors. I know N connectors are pretty strong, but there's potentially quite a lot of stress (and flex?) on the screwed corners if you get an accidental cable tug.

[Elasia]

From the mechanical perspective, it looks as if it would be a good idea to use standoffs to support the outer mounting holes of the connectors. I know N connectors are pretty strong, but there's potentially quite a lot of stress (and flex?) on the screwed corners if you get an accidental cable tug.

+1 when its cheap and easy to do... couple standoffs and drill / tap a few more holes

do you have a bom list of what connector model / resistor / heat sink part number that is?

[noreply]

It would probably be better to dedicate this thread to your Attenuator (and title it accordingly). It looks as if it deserves one. Threads tend to get very confused when you have various projects being added and tying up feedback with each.

Done 

[noreply]

From the mechanical perspective, it looks as if it would be a good idea to use standoffs to support the outer mounting holes  :-+of the connectors. I know N connectors are pretty strong, but there's potentially quite a lot of stress (and flex?) on the screwed corners if you get an accidental cable tug.

+1 when its cheap and easy to do... couple standoffs and drill / tap a few more holes

do you have a bom list of what connector model / resistor / heat sink part number that is?

My initial design WAS going to use the 'standoffs' for the N connectors 

After testing the mechanical aspects - I found that the existing mounts (2 bolts via RES-NET device into heat sink) were so strong - the N connectors are solid machined stainless 2mm thick - there was no way this could be bent under stressed condition - without pulling out the cable in the process from the plugged-in N connector 

BUT

You are perfectly correct - no problem in fitting some extra standoffs if desired 

(I simply did not have any at the time and in my case you need a tank to break the thing without the standoffs – so I let it slide)

I am going to get specifics for ALL the parts I used and generate a BOM - will post shortly