Life is full of surprises


Andy G4JNT
 
Edited

100dB attenuator made up from six PI  stages of ~16.7dB each using standard 0805 resistors
I measured its frequency response and was more than a bit surprised how good it was, maintaining to within 1dB of the design value.


Have a guess what it's upper frequency limit is - ie where the loss has changed by more than a couple of dB.


DD1US
 

1.3 GHz

 

Regards

 

Matthias

 

www.dd1us.de

 

 

Von: UKMicrowaves@groups.io <UKMicrowaves@groups.io> Im Auftrag von Andy G4JNT
Gesendet: Sonntag, 16. Mai 2021 15:14
An: UK Microwaves groups.io <ukmicrowaves@groups.io>
Betreff: [UKMicrowaves] Life is full of surprises

 

100dB attenuator made up from six PI  stages of ~16.7dB each using standard 0805 resistors

I measured its frequency response and was more than a bit surprised how good it was, maintaining to within 1dB of the design value.

 

 

Have a guess what it's upper frequency limit is - ie where the loss has changed by more than a couple of dB.

 

Andy

 

 

 


Andy G4JNT
 

higher - higher



On Sun, 16 May 2021 at 14:49, DD1US <matthias.bopp@...> wrote:

1.3 GHz

 

Regards

 

Matthias

 

www.dd1us.de

 

 

Von: UKMicrowaves@groups.io <UKMicrowaves@groups.io> Im Auftrag von Andy G4JNT
Gesendet: Sonntag, 16. Mai 2021 15:14
An: UK Microwaves groups.io <ukmicrowaves@groups.io>
Betreff: [UKMicrowaves] Life is full of surprises

 

100dB attenuator made up from six PI  stages of ~16.7dB each using standard 0805 resistors

I measured its frequency response and was more than a bit surprised how good it was, maintaining to within 1dB of the design value.

 

 

Have a guess what it's upper frequency limit is - ie where the loss has changed by more than a couple of dB.

 

Andy

 

 

 


Peter Hamilton
 

I've used a similar layout, but with 0603s and only up to 30dB attenuation, up to 5.8GHz. I was surprised too.

--

Peter (G4VLC)



On Sun, 16 May 2021 at 14:55, Andy G4JNT <andy.g4jnt@...> wrote:
higher - higher



On Sun, 16 May 2021 at 14:49, DD1US <matthias.bopp@...> wrote:

1.3 GHz

 

Regards

 

Matthias

 

www.dd1us.de

 

 

Von: UKMicrowaves@groups.io <UKMicrowaves@groups.io> Im Auftrag von Andy G4JNT
Gesendet: Sonntag, 16. Mai 2021 15:14
An: UK Microwaves groups.io <ukmicrowaves@groups.io>
Betreff: [UKMicrowaves] Life is full of surprises

 

100dB attenuator made up from six PI  stages of ~16.7dB each using standard 0805 resistors

I measured its frequency response and was more than a bit surprised how good it was, maintaining to within 1dB of the design value.

 

 

Have a guess what it's upper frequency limit is - ie where the loss has changed by more than a couple of dB.

 

Andy

 

 

 


Peter Hamilton
 

 ... so, based on my 5.8GHz surprise, my guess is 10GHz.

--

Peter (G4VLC)



On Sun, 16 May 2021 at 14:59, Peter Hamilton via groups.io <lists=g4vlc.net@groups.io> wrote:
I've used a similar layout, but with 0603s and only up to 30dB attenuation, up to 5.8GHz. I was surprised too.

--

Peter (G4VLC)



On Sun, 16 May 2021 at 14:55, Andy G4JNT <andy.g4jnt@...> wrote:
higher - higher



On Sun, 16 May 2021 at 14:49, DD1US <matthias.bopp@...> wrote:

1.3 GHz

 

Regards

 

Matthias

 

www.dd1us.de

 

 

Von: UKMicrowaves@groups.io <UKMicrowaves@groups.io> Im Auftrag von Andy G4JNT
Gesendet: Sonntag, 16. Mai 2021 15:14
An: UK Microwaves groups.io <ukmicrowaves@groups.io>
Betreff: [UKMicrowaves] Life is full of surprises

 

100dB attenuator made up from six PI  stages of ~16.7dB each using standard 0805 resistors

I measured its frequency response and was more than a bit surprised how good it was, maintaining to within 1dB of the design value.

 

 

Have a guess what it's upper frequency limit is - ie where the loss has changed by more than a couple of dB.

 

Andy

 

 

 


militaryoperator
 

 
100dB attenuator made up from six PI  stages of ~16.7dB each

What am I missing? why would anyone want 100db of attenuation?

B


Andy G4JNT
 

No  not that good.   
But it held 100dB to around 2.8GHz.   If I'd used vias close to the resistors, it  would have gone higher, I suspect.  But relying on edge grounding and wrap-around set te limit I suspect.



On Sun, 16 May 2021 at 15:19, Peter Hamilton <lists@...> wrote:
 ... so, based on my 5.8GHz surprise, my guess is 10GHz.

--

Peter (G4VLC)



On Sun, 16 May 2021 at 14:59, Peter Hamilton via groups.io <lists=g4vlc.net@groups.io> wrote:
I've used a similar layout, but with 0603s and only up to 30dB attenuation, up to 5.8GHz. I was surprised too.

--

Peter (G4VLC)



On Sun, 16 May 2021 at 14:55, Andy G4JNT <andy.g4jnt@...> wrote:
higher - higher



On Sun, 16 May 2021 at 14:49, DD1US <matthias.bopp@...> wrote:

1.3 GHz

 

Regards

 

Matthias

 

www.dd1us.de

 

 

Von: UKMicrowaves@groups.io <UKMicrowaves@groups.io> Im Auftrag von Andy G4JNT
Gesendet: Sonntag, 16. Mai 2021 15:14
An: UK Microwaves groups.io <ukmicrowaves@groups.io>
Betreff: [UKMicrowaves] Life is full of surprises

 

100dB attenuator made up from six PI  stages of ~16.7dB each using standard 0805 resistors

I measured its frequency response and was more than a bit surprised how good it was, maintaining to within 1dB of the design value.

 

 

Have a guess what it's upper frequency limit is - ie where the loss has changed by more than a couple of dB.

 

Andy

 

 

 


Andy G4JNT
 

Test equipment, my good sir, test equipment!

If you need to look at the output of that Tx driver or beacon source via a receiver antenna input, to decode it down in the noise , needs 100 - 150dB of attenuation.   I've got switchable attenuators that go to 100dB, but needed  the rest without having to cascade piles of 20dB N and SMA attenuators.
Besides, it was a challenge.

Back in my last year of full time employment at HMG, I had some test kit that could achieve appreciably more than 200dB isolation at 2GHz, feeding one radio transmitting several watts into another on Rx separated by just a few metres.   Being military kit, their screening was good enough - but making cables that didn't leak at those levels needed care.  Semi-rigid everywhere, top-spec N connectors and HP switchable/remote programmable attenuators that cost close on £250k they did go to 40GHz)
And a digital/FFT spectrum analyser that could display and actually see close on 200dB of dynamic range.   Took a lot of effort to get that - but I regarded it as a pre-retirement challenge to see 200dB.    IIRC, had to go to a few mHz resolution bandwidth.

The Aeroflex Digital Radio test set I have here can manage a little over 150dB dynamic range at its minimum 1Hz resolution bandwdith, but the Systron Donner synth used as the test source leaks from its casing,  so best I've managed to see at home is a mere 120dB at GHz frequencies.  An ADF4351 in a biscuit tin with SMA feedthrough, lid stuck all around with self-adhesive copper tape and powered from an internal battery may be worth a punt.



On Sun, 16 May 2021 at 15:22, militaryoperator via groups.io <Military1944=aol.com@groups.io> wrote:
 
100dB attenuator made up from six PI  stages of ~16.7dB each

What am I missing? why would anyone want 100db of attenuation?

B


Peter Hamilton
 

Ah, I didn't look close enough at the pics. I had 3 vias around each ground pad, as close as possible, and the smaller Rs would have helped. I was expecting to tweak it to compensate for the error at 6GHz, but it turned out to be unnecessary: less than half a dB at 30dB.

--

Peter (G4VLC)



On Sun, 16 May 2021 at 15:24, Andy G4JNT <andy.g4jnt@...> wrote:
No  not that good.   
But it held 100dB to around 2.8GHz.   If I'd used vias close to the resistors, it  would have gone higher, I suspect.  But relying on edge grounding and wrap-around set te limit I suspect.



On Sun, 16 May 2021 at 15:19, Peter Hamilton <lists@...> wrote:
 ... so, based on my 5.8GHz surprise, my guess is 10GHz.

--

Peter (G4VLC)



On Sun, 16 May 2021 at 14:59, Peter Hamilton via groups.io <lists=g4vlc.net@groups.io> wrote:
I've used a similar layout, but with 0603s and only up to 30dB attenuation, up to 5.8GHz. I was surprised too.

--

Peter (G4VLC)



On Sun, 16 May 2021 at 14:55, Andy G4JNT <andy.g4jnt@...> wrote:
higher - higher



On Sun, 16 May 2021 at 14:49, DD1US <matthias.bopp@...> wrote:

1.3 GHz

 

Regards

 

Matthias

 

www.dd1us.de

 

 

Von: UKMicrowaves@groups.io <UKMicrowaves@groups.io> Im Auftrag von Andy G4JNT
Gesendet: Sonntag, 16. Mai 2021 15:14
An: UK Microwaves groups.io <ukmicrowaves@groups.io>
Betreff: [UKMicrowaves] Life is full of surprises

 

100dB attenuator made up from six PI  stages of ~16.7dB each using standard 0805 resistors

I measured its frequency response and was more than a bit surprised how good it was, maintaining to within 1dB of the design value.

 

 

Have a guess what it's upper frequency limit is - ie where the loss has changed by more than a couple of dB.

 

Andy

 

 

 


John Fell
 

Bin there , tested that .
Under 100dB before leakage with +3dBm at 1296MHz .
Tea was not that special either .

73
John
G0API

On Sun, 16 May 2021 at 15:42, Andy G4JNT <andy.g4jnt@...> wrote:
Test equipment, my good sir, test equipment!

If you need to look at the output of that Tx driver or beacon source via a receiver antenna input, to decode it down in the noise , needs 100 - 150dB of attenuation.   I've got switchable attenuators that go to 100dB, but needed  the rest without having to cascade piles of 20dB N and SMA attenuators.
Besides, it was a challenge.

Back in my last year of full time employment at HMG, I had some test kit that could achieve appreciably more than 200dB isolation at 2GHz, feeding one radio transmitting several watts into another on Rx separated by just a few metres.   Being military kit, their screening was good enough - but making cables that didn't leak at those levels needed care.  Semi-rigid everywhere, top-spec N connectors and HP switchable/remote programmable attenuators that cost close on £250k they did go to 40GHz)
And a digital/FFT spectrum analyser that could display and actually see close on 200dB of dynamic range.   Took a lot of effort to get that - but I regarded it as a pre-retirement challenge to see 200dB.    IIRC, had to go to a few mHz resolution bandwidth.

The Aeroflex Digital Radio test set I have here can manage a little over 150dB dynamic range at its minimum 1Hz resolution bandwdith, but the Systron Donner synth used as the test source leaks from its casing,  so best I've managed to see at home is a mere 120dB at GHz frequencies.  An ADF4351 in a biscuit tin with SMA feedthrough, lid stuck all around with self-adhesive copper tape and powered from an internal battery may be worth a punt.



On Sun, 16 May 2021 at 15:22, militaryoperator via groups.io <Military1944=aol.com@groups.io> wrote:
 
100dB attenuator made up from six PI  stages of ~16.7dB each

What am I missing? why would anyone want 100db of attenuation?

B


Paul G8AQA
 

I am told that mounting the resistors on edge may help by reducing capacitance.

Paul G8AQA


Andy G4JNT
 

Mmm... perhaps
But it will also increase inductance.   When they are lying flat, and have a comparable width to the that of the track - as is the case here with 0.8mm FG board, they appear more as a continuation of the microstrip than having stray C or L

Biggest contributor to that design wrt upper frequency limit is the grounding of the shunt elements.   The way I did it, with no vias and just fold-around copper tape, is going to force an upper limit when the distance from R to the edge of the PCB is getting on for  λg/4




On Mon, 17 May 2021 at 10:06, Paul G8AQA via groups.io <paulnick=btinternet.com@groups.io> wrote:
I am told that mounting the resistors on edge may help by reducing capacitance.

Paul G8AQA


Brian Howie GM4DIJ
 

On 16/05/2021 15:24, Andy G4JNT wrote:
No  not that good.
But it held 100dB to around 2.8GHz.   If I'd used vias close to the resistors, it  would have gone higher, I suspect.  But relying on edge grounding and wrap-around set te limit I suspect.
Andy
www.g4jnt.com <http://www.g4jnt.com>
I think it's limited by the capacitance of the resistors. I hit this problem in using chip resistors for feedback in hybrid transimpedance amplifiers. It helped splitting the resistor into 2 or more.

https://www.vishay.com/docs/60107/freqresp.pdf has data and some models

It should be possible to simulate it.

--
Brian


Andy G4JNT
 

Think I'll stop playing attenuators.   Only wanted performance at low frequencies, 144MHz maximum, so it was a happy bonus to get low-microwave performance

Now the real challenge is to get broadband gain of 100dB in a single box without it taking off.
I did it with a chain of MAR-4 and MAR-3 devices at 144MHz but with a helical filter and other filter stages  to give 1MHz bandwidth for a noise meter - and the output was only used internally into a diode detector



On Mon, 17 May 2021 at 11:39, Brian Howie GM4DIJ <brian@...> wrote:
On 16/05/2021 15:24, Andy G4JNT wrote:
> No  not that good.
> But it held 100dB to around 2.8GHz.   If I'd used vias close to the
> resistors, it  would have gone higher, I suspect.  But relying on edge
> grounding and wrap-around set te limit I suspect.
>
> Andy
> www.g4jnt.com <http://www.g4jnt.com>

I think it's limited by the capacitance of the resistors. I hit this
problem in using chip resistors for feedback in hybrid transimpedance
amplifiers. It helped splitting the resistor into 2 or more.

https://www.vishay.com/docs/60107/freqresp.pdf  has data and some models

It should be possible to simulate it.

--
Brian






Dave G8KHU
 

An iteresting read, thanks for posting

Dave