20dB 'RF' attenuator - seeking feedback to improve

[enut11]

This is the 4-stage T attenuator after a few mods.    There are 18 resistors on the board with a layer of Kapton, then some copper foil strips that I trimmed very carefully.   It's well within a half dB up to 1.5GHz.   

@joeqsmith, very impressive. Can you post a picture of the pcb?
enut11

[joeqsmith]

This is the 4-stage T attenuator after a few mods.    There are 18 resistors on the board with a layer of Kapton, then some copper foil strips that I trimmed very carefully.   It's well within a half dB up to 1.5GHz.   

@joeqsmith, very impressive. Can you post a picture of the pcb?
enut11

I looked to see if I had taken a clearer picture before I started adding the tape and foil to it.   This was cleanest one I found.   Just a bunch of resistors.

How's that Nano coming along?   Did you join the IOGroups?   I understand there is now a fair amount of information there that may help you get started. 

[ogden]

I have a many cheap Chinese made no-brand-name SMA connectors.  The "Gold" plating is so thin if I attach anything and remove it, metal underneath will show.  Then it will start to turn black and corrode.

Mine do not corrode, yet plating flake off easily and contaminate connectors. So I have different sets of cables and accessories - for flaky SMA and quality (H&S) SMA

[Calvin]

Hi,

these relatively cheap Rs might do quite well especially those in the 50-100R range
MMA0204AC5009FB300 (0,184€@100pcs at Mouser)

regards
Calvin

[ogden]

Hi,

these relatively cheap Rs might do quite well especially those in the 50-100R range
MMA0204AC5009FB300 (0,184€@100pcs at Mouser)

It is 0,184€/piece @100pcs That is *very* expensive 0204 resistor.

[Calvin]

hi,

well then .... which ´cheaper´ RF-resistor could You suggest?

regards
Calvin

[ogden]

well then .... which ´cheaper´ RF-resistor could You suggest?

There is huge difference between "cheap resistors" and "cheap RF resistors". I fail to find any mention of "RF" in your initial post.

[joeqsmith]

Hi,

these relatively cheap Rs might do quite well especially those in the 50-100R range
MMA0204AC5009FB300 (0,184€@100pcs at Mouser)

regards
Calvin

https://www.mouser.com/datasheet/2/427/mmu0102mma0204mmb0207hf-465485.pdf

Looks like digikey offers a kit.
https://www.digikey.com/product-detail/en/vishay-thin-film/764-1558-KIT/764-1558-KIT-ND/7313792

For my home projects, I've had decent results using standard resistors up to a GHz or so.   I played around a bit at higher frequencies but I would say I was chasing my tail at best as I just didn't have the tools.   I used some commercial mixers, YIG and coupler to push the Nano above 3GHz, but no way to calibrate it with any sort of confidence.   

Years ago during a visit to the Dayton Ham show, someone was selling Narda terminators for a couple of bucks or so each.  I picked up a few in hopes I could use them to make a decent SOL standard set for my higher frequency experiments.   These can be taken apart.    Using a lathe and some Teflon, I made up some new internals for the open and and copper for the short.  The parts were phase matched on the old VNA.    Still, the software I wrote for the old VNA is light years better than using a grease pen, it is no where near what a modern VNA would have in terms of calibration.  It assumes the SOLT is perfect which it obviously is not. 

Then again, it's just a hobby.

[enut11]

This is the 4-stage T attenuator after a few mods.    There are 18 resistors on the board with a layer of Kapton, then some copper foil strips that I trimmed very carefully.   It's well within a half dB up to 1.5GHz.   

@joeqsmith, very impressive. Can you post a picture of the pcb?
enut11

I looked to see if I had taken a clearer picture before I started adding the tape and foil to it.   This was cleanest one I found.   Just a bunch of resistors.

How's that Nano coming along?   Did you join the IOGroups?   I understand there is now a fair amount of information there that may help you get started.

@joeqsmith, Still playing with the NanoVNA when opportunity arises.
Will join an IOgroup when I feel I can ask intelligent questions.
At the moment I need to come to grips with RF fundamentals.

I found some PC interface software called nanovna-saver-v0.2.2 which seems to be OK.

I also purchased some more cable and connectors online. Now I am finding out that connector quality may be an issue.
If you have a reliable online source for SMA I would appreciate the info.
enut11

[ogden]

If you have a reliable online source for SMA I would appreciate the info.

mouser.com They are official distributor of Huber+Suhner

[joeqsmith]

I pretty much stay with Amphenol, Johnson or Molex lower end parts from Digi-key.   For the VNA, I use RG400/U, which is far from ideal.  Start pricing some decent cables and you may be in for sticker shock.     I expect a single Gore flex cable would cost more than my old HP VNA with cables cost.   

https://www.gore.com/products/gore-vna-microwave-rf-test-assemblies

The following paper from HP talks about the various connector grades.   I don't think this covers the $0.50 connectors we are seeing from China. 
http://na.support.keysight.com/pna/connectorcare/Connector_Grades.htm

If you are just playing around and learning the basics, using the really cheap parts shouldn't be much of a problem.  Just consider them disposable.   If you stay interested in RF design,  you can invest in better parts, cables and TE. 

Article on connector care.   Things you want to learn before you buy good connectors.   
https://www.microwaves101.com/encyclopedias/connector-care   

I have not tried the Nanovna saver software but see people post about it from time to time.  It sounds like they have done a really nice job with it. 

[Berni]

Id say don't get any fancy RF connectors unless you have a good reason to.

If you have a NanoVNA or a really old network analyzer that only goes to 1GHz any cheep SMA connector will do fine. As long as its not actually out of spec (Ocasonaly chinese crappy connectors are that bad).

If all your connectors are crap to begin with then there is no danger of damaging them. Just use a quality SMA to N adapter on the network analyzer or whatever so that if you wreck the SMA connector you can just replace the adapter.

Its only when you go above 4GHz or start measuring really low down return losses is where you need the fancy Gore brand cables that cost an arm and a leg. So given how much they cost you really want to take care of them

[enut11]

@joeqsmith
In your picture showing the NanoVNA with some attenuators, on the 20dB ones using discrete pigtail resistors, is there a copper pad under the junction of the resistors?
If so, what is the size of the gap in the central copper strip before and after this pad.
enut11

[joeqsmith]

@joeqsmith
In your picture showing the NanoVNA with some attenuators, on the 20dB ones using discrete pigtail resistors, is there a copper pad under the junction of the resistors?
If so, what is the size of the gap in the central copper strip before and after this pad.
enut11

Yes, there is a pad there.  These boards are used to test components on the VNA and to remove the effects of the connectors and such (de-embed).  This particular board is the thru which I have taken a razor blade to.   If you are interested in laying out boards, you could use a free program like AppCAD to get an idea what dimensions you need.     

The number of mate cycles you will get out of the really cheap connectors may not be very good.  You may want to inspect them for any sort of burrs or debris before using them.   

A few articles on cables:
https://www.microwaves101.com/encyclopedias/cable-care
https://www.tek.com/blog/improving-vna-measurement-accuracy-quality-cables-and-adapters

[enut11]

@joeqsmith, what I wanted to know is how much of that central strip did you cut away before and after the pad at the junction of the resistors?
thanks
enut11

[enut11]

@joeqsmith
I played around with the APPcad numbers for my 1.5mm double sided circuit board until I got 50 ohms.
What I do not know is the 'FR' value (dielectric) for my PCB so I assumed a mid-range figure of 4.6

Ended up with a 2.8mm central strip with 2.5mm gaps either side. Looks OK?
enut11

[enut11]

@joeqsmith
In your picture showing the NanoVNA with some attenuators, on the 20dB ones using discrete pigtail resistors, is there a copper pad under the junction of the resistors?
If so, what is the size of the gap in the central copper strip before and after this pad.
enut11

Yes, there is a pad there.  These boards are used to test components on the VNA and to remove the effects of the connectors and such (de-embed).  This particular board is the thru which I have taken a razor blade to.   If you are interested in laying out boards, you could use a free program like AppCAD to get an idea what dimensions you need.     

The number of mate cycles you will get out of the really cheap connectors may not be very good.  You may want to inspect them for any sort of burrs or debris before using them.   

A few articles on cables:
https://www.microwaves101.com/encyclopedias/cable-care
https://www.tek.com/blog/improving-vna-measurement-accuracy-quality-cables-and-adapters

Yikes! What a minefield cables can be. May be better off sticking to the lower HF area for the time being...

[joeqsmith]

@joeqsmith, what I wanted to know is how much of that central strip did you cut away before and after the pad at the junction of the resistors?
thanks
enut11

Again, I'm using a razor blade so it's not exact.  Basically, I used the length of the axial body.   In hindsight, using a smaller gap may have helped improve the performance by adding a little more capacitance.   Then again, maybe a little foil and some tape....    You may notice the gap is similar on the 0603 parts.  It's certainly not optimum but as you can see, it can yield some decent results at lower frequencies. 

I have ordered a few of the low cost attenuators.   These may be a good starting point for your experiments. 

@joeqsmith
I played around with the APPcad numbers for my 1.5mm double sided circuit board until I got 50 ohms.
What I do not know is the 'FR' value (dielectric) for my PCB so I assumed a mid-range figure of 4.6

Ended up with a 2.8mm central strip with 2.5mm gaps either side. Looks OK?
enut11

At lower frequencies, the software should do fine but I doubt you would want to pay to have a PCB made from better materials and controlled impedances.  I would check with your board house before you settle on a design.    Maybe even try a couple of different dimensions to hedge your chances. 

Yikes! What a minefield cables can be. May be better off sticking to the lower HF area for the time being...

I've had a lot of fun playing with RF without having to resort to anything too crazy.   There's no reason you can't use the Nano to experiment in the GHz range.   I would like to see the Nano V2 3GHz unit.  If you can believe the internet, they are now working towards 6GHz.   That will be impressive.   

[enut11]

Is the V2 a larger screen for around x2 the price?

[joeqsmith]

From the little bit I have seen, I thought it was a 3GHz version that doesn't use harmonics.   I've seen some posts about it going to 3.5.   I wouldn't count on anything until they start shipping and we can have a look. 

[enut11]

While waiting for my SMA connectors and adapters to arrive I started playing around with a hand-carved PCB (using my old BNC connectors) and a star arrangement of resistors.
The paralleled series resistors are 82 ohm each. Attenuation is 20dB.

Question: When paralleling resistors, is it better to separate them (as I have) or to mount them close against each other?
enut11

[joeqsmith]

What's that one look like on the Nano if you sweep to 40MHz, compared with your first one?   Take a few pictures of the Nano's screen if you don't have the software working yet. 

[enut11]

Hi joeqsmith
No way to check any of my attenuators on the NanoVNA until I get my SMA bits and pieces.
I did check with my DSO and it is slightly better compared with the original attenuator up to 50MHz.
My DSO is -3dB at 170MHz.
enut11

[joeqsmith]

Hi joeqsmith
No way to check any of my attenuators on the NanoVNA until I get my SMA bits and pieces.
I did check with my DSO and it is slightly better compared with the original attenuator up to 50MHz.
My DSO is -3dB at 170MHz.
enut11

As far as I can measure, it is -1.5dB at 45MHz.

I had assumed that you meant it was 20dB minus the 1.5 or 18.5dB at 45MHz and the new one is now 17dB at 170MHz.  But maybe you are actually reading 3dB at 170MHz.   If it really is this far off, I wonder how you validated your setup using the scope.   Are your scope and source both 50 ohms?  If you use a cable to connect the two, is it flat?  If not, are you accounting for this error in your measurements?   

It will be interesting to see how your results compare when you place the same attenuators on your Nano. 

[duak]

I don't think the spacing between the paralled resistors matters too much but I expect closer to be better.

The cutout in the ground plane concerns me.  i think I see a significan loop area because of the cutout.  What happens when you lay a ground plane on top of the circuit?  Make sure there are at least four connections to the existing ground plane and that it is as close as possible to the resistors without shorting them out.  i use really thin one side copper FRP for test shielding.

Also, are your test cables in good shape?  Maybe the shielding braids and connections are not solid and the attenuator is being bypassed by leakage between the input and output cables.

Cheers,