20dB 'RF' attenuator - seeking feedback to improve

[joeqsmith]

Seems like you may return four of six terminators as faulty because specs say that upto 1GHz RL is no worse than -43dB. Why didn't you include NanoVNA "standard" in this comparison?

You may be confusing typical with the guaranteed minimum return loss.  Typical numbers are not guaranteed by the manufacture.   

With it's ridged design, I have no plans on ever using the terminator supplied with my Nano but was also curious.   Included is the data comparing it with ANNE #3,#4  and the Agilent standard. 

[ogden]

With it's ridged design, I have no plans on ever using the terminator supplied with my Nano but was also curious.   Included is the data comparing it with ANNE #3,#4  and the Agilent standard.

Thank you for investing your time and effort to confirm what I told long ago in another thread here.

[joeqsmith]

Thank you for investing your time and effort to confirm what I told long ago in another thread here.

What post are you referring to?

[ogden]

Thank you for investing your time and effort to confirm what I told long ago in another thread here.

What post are you referring to?

One where I state that NanoVNA users shall consider replacing supplied 50ohm load "standard" because even 11$ "plug" like ANNE-50+ is better.

[joeqsmith]

You would need to provide a link to it. 

That said, the ANNE has a guaranteed return loss of 30dB for DC-4GHz.   At 1GHz, the part that was included with my Nano is 36.5.  Better than the performance that the ANNE is guaranteed for.   Of course, if you have someone that can test and screen parts for you then sure.     

I doubt many people would be using the Nano beyond 300MHz.  At these lower frequencies, the return loss for the terminator supplied with mine would be fine.    The real benefit IMO is getting rid of that ridged construction.         

[ogden]

You would need to provide a link to it. 

Nah. If you need it - search yourself.

That said, the ANNE has a guaranteed return loss of 30dB for DC-4GHz. At 1GHz, the part that was included with my Nano is 36.5. Better than the performance that the ANNE is guaranteed for.

LOL nice try.

It is pointless to nitpick guaranteed performance figure of ANNE while same graph you are reading NanoVNA figure (36.5) from, tells that both compared ANNE plugs are better, -52 and -66 @1GHz. Not to mention that spread of NanoVNA load is unknown, specs are not published and you perhaps got best NanoVNA load ever made.

[joeqsmith]

You would need to provide a link to it. 

Nah. If you need it - search yourself.

I'm not the one who brought it up or made reference to it.  If you want me to read it, provide the link otherwise I will ignore it.

That said, the ANNE has a guaranteed return loss of 30dB for DC-4GHz. At 1GHz, the part that was included with my Nano is 36.5. Better than the performance that the ANNE is guaranteed for.

LOL nice try.

It is pointless to nitpick guaranteed performance figure of ANNE while same graph you are reading NanoVNA figure (36.5) from, tells that both compared ANNE plugs are better, -52 and -66 @1GHz. Not to mention that spread of NanoVNA load is unknown, specs are not published and you perhaps got best NanoVNA load ever made.

Learn to read a datasheet.  Making a blanket statement that using the ANNE will somehow improve the NANO's measurements over the standard that was supplied is just being naive. 

[ogden]

If you want me to read it, provide the link otherwise I will ignore it.

Wishful thinking?  I rephrased it already: "I state that NanoVNA users shall consider replacing supplied 50ohm load "standard" because even 11$ "plug" like ANNE-50+ is better."

Learn to read a datasheet. 

Learn to use your head, if any. You can't compare datasheet figures of ANNE to nonexistant datasheet figures of Nano load standard. You have to compare apples to apples, in this case - measurements. In your data ANNE clearly wins as load standard. That's my whole point. You can't compare guaranteed datasheet figure of one device to measurement of another. That is not even naiive to say it politely.

Making a blanket statement that using the ANNE will somehow improve the NANO's measurements over the standard that was supplied is just being naive.

Wishful thinking again. You are putting words in my mouth. When you find where I say something like "ANNE will somehow improve the NANO's measurements" - let me know.

[joeqsmith]

I'm not  sure why you are resorting to insults.  I get you like to argue but this is pretty basic stuff.   Without having a means to verify the standards there is no way of knowing how you have effected the performance.   It's as simple as that. 

[ogden]

I'm not  sure why you are resorting to insults.

Just returning a favor.

I get you like to argue but this is pretty basic stuff.

It is you who is arguing against my statement. Already forgot?

Without having a means to verify the standards there is no way of knowing how you have effected the performance. It's as simple as that.

That would be absurd for everybody to have means of comparing their standards. Funny that you chose to measure/compare NanoVNA load and ANNE terminator in 1..6 GHz range, considering:

I doubt many people would be using the Nano beyond 300MHz

[joeqsmith]

I'm not  sure why you are resorting to insults.

Just returning a favor.

I get you like to argue but this is pretty basic stuff.

It is you who is arguing against my statement. Already forgot?

Without having a means to verify the standards there is no way of knowing how you have effected the performance. It's as simple as that.

That would be absurd for everybody to have means of comparing their standards. Funny that you chose to measure/compare NanoVNA load and ANNE terminator in 1..6 GHz range, considering:

I doubt many people would be using the Nano beyond 300MHz

Sorry you take offense to being called naive.  That's not what I would consider an insult, just a fact based on your statements when you suggest I return used parts that meet the manufacture's specification.   I am used to people on this forum who have little knowledge to resort to personal attacks when they have little else to use to argue their point. 

I'm only trying to help you understand the difference between typical and specified.  Sadly, it seems your ego prevents you from learning something this basic and I suspect you will continue being ignorant and combative about the subject.     

I'm not sure why you would be surprised that I am assembling this cal kit for use at frequencies beyond a GHz.  It was never my goal to use it with the Nano in hopes of improving its accuracy.    Had I planned on continuing to work below a GHz, I would have stayed with my PCB based standards.   

Again, I did not buy the Nano and it has been basically been used as an educational tool.   It's very low cost and simple interface make it a very good fit for learning the basics on how to use a VNA.  I certainly wouldn't suggest anyone waste their money on buying a replacement terminator for it that I didn't know would improve their results. 

I understand that Kirkby uses sorted terminators for the standard kits he sells.  It may be possible to procure just the terminators from him if you suspected the one that was supplied with the Nano.       

[ogden]

I'm not sure why you would be surprised that I am assembling this cal kit for use at frequencies beyond a GHz.

LOL no. I am not suprised at all. Initially you compared ANNE plugs in 0..6GHz range. Then after I mentioned NanoVNA load standard to ANNE comparison, for some "strange reason" you omitted 0..1GHz (of NanoVNA range) and picked 1..6GHz. Could be accident could be not. Anyway you did not manage to disprove my point - that ANNE plug is better than supplied NanoVNA load. Also please let others decide - 11$ for terminator (as load standard) which at least have datasheet figures is waste or not. Unless you show comparison in 0..6GHz range, I have nothing else to discuss with you.

[joeqsmith]

I'm not sure why you would be surprised that I am assembling this cal kit for use at frequencies beyond a GHz.

LOL no. I am not suprised at all. Initially you compared ANNE plugs in 0..6GHz range. Then after I mentioned NanoVNA load standard to ANNE comparison, for some "strange reason" you omitted 0..1GHz (of NanoVNA range) and picked 1..6GHz. Could be accident could be not. Anyway you did not manage to disprove my point - that ANNE plug is better than supplied NanoVNA load. Also please let others decide - 11$ for terminator (as load standard) which at least have datasheet figures is waste or not. Unless you show comparison in 0..6GHz range, I have nothing else to discuss with you.

Sorry that the data I had collected for my one Nano terminator doesn't meet your needs.  I didn't run that test for you but out of my own curiosity.  I know you feel the world evolves around you and you are entitled to my time but I assure you, it's not the case.  I collected that data a few months back and when you asked, I took the time to post what I had.

There are no published spec's for the Nano's terminator. Each will be different (even if they were supplied by the same manufacture).   However, we do have spec's for the ANNE which is 30dB return loss in the range the Nano is capable of being used.   It's obvious that a person would need to measure them both if they want to know what the effects are.  It's not a feeling or please let others decide but a simple fact.   I can understand you not like being called out and it hurts your feelings and you feel its a personal attack and I should just agree with you.  All of that has nothing to do with misrepresenting the data, especially when you start misleading the public.   

In singles, the ANNE will cost $15 or more.  You need to consider sales tax and shipping.  I imagine I could hand select a couple of 0805 100 ohm parts for the frequency range of the Nano, if it were a concern.   History has shown I can't do this once I get much above a GHz and it requires a commercial part.   

The cost of the Agilent cal kits exceed what I would be willing to pay.  I am not interested in used parts that some clueless person has damaged.  Making the shorts and opens were not a simple task and required several attempts at machining parts.  From the data I've shown using rf-messkopf's script and Agilent's standards, the results are reasonable.  The loads are really the problem and it seems the least costly approach was to buy several and sort them. 

Consider they all met the manufacture's spec and can't be returned as defective.   I ran 6pcs.  At 11 + 0.66(tax)  and say $8 in shipping, about $78.  #3 also looks pretty decent.  For my home hobby, good enough.

[ogden]

I can understand you not like being called out and it hurts your feelings and you feel its a personal attack and I should just agree with you.  All of that has nothing to do with misrepresenting the data, especially when you start misleading the public.

Oh gimme a break. *You* are misleading public by comparing measurement @1GHz of one device to guaranteed figure @ 0..4GHz range of another.

[joeqsmith]

I can understand you not like being called out and it hurts your feelings and you feel its a personal attack and I should just agree with you.  All of that has nothing to do with misrepresenting the data, especially when you start misleading the public.

Oh gimme a break. *You* are misleading public by comparing measurement @1GHz of one device to guaranteed figure @ 0..4GHz range of another.

What are you whining about now?   You asked for the data and I provided you with what I had.   I have not made any recommendations to the public in regards to buying a new terminator for the Nano.   It's not my fault that you were clueless about what typical meant.   Saying the part is specified to 30dB is not misleading the public.  Nor is suggesting that the two parts would need to be compared to know for sure.  Again, all basic things. 

[TheoB]

I picked up a few ANNE terminators to see if I could hand select something a little tighter for my calibration kit.    #1 is the part that was shown in the previous graphs.  I have also included the load from the Agilent 85033E.

You hand selected the best one. Isn't that the one that measures closest to 50 Ohm DC as well? For these terminators the quality of the mating surface is probably more important and I trust all those Anne terminators to work fine. After over 100 mate cycles I expect the cheap cal kit that came with the VNA to be worn completely. I have a new one (perhaps 10 times used) and see already metal chips on the connector interface (I do not clean as you showed, but I guess that needs to change). I also don't have a torque wrench.
Good (and reproducible) contact is probably more important than absolute LF impedance. With a MM it's easy to determine 50 Ohm within 1%. That means the S11 would be better than -46dB. I expect the return loss to be fairly flat until ~ 500MHz. Your measurements seem to confirm that (at least for the Anne terms). I plan to simply check that the calibration remains stable. As soon as that's no longer the case, I'll invest in a new (proper) call kit and new connectors for the VNA. My usage is light (hobby).

[G0HZU]

Some of the attempts on here to make a decent 20dB attenuator seem a bit poor to me. It's as if little thought or effort went into it.

Many years a go I made a precision 20dB attenuator to use with my old 1GHz HP8405A vector voltmeter. The aim was to get very good performance to 1GHz. I dug it out and measured it the other day for someone and the plots may be interesting. The first plot to 3GHz was taken using an Ecal module to do the VNA calibration. The second plot to 100MHz was taken using precision attenuators to minimise mismatch uncertainty to a level even better than the Ecal module.

I designed this using custom chip resistors measured on a 6.5digit DMM and the aim was to get as close to 20.00dB as I could with very low VSWR. It's easy to get the attenuation but it also should be achieved with very low VSWR. I think I could better this for VSWR if I made another today (better resistors available now) but the aim at the time was to get really good performance to 1GHz for the vector voltmeter.

I've just measured the VSWR using a the Ecal direct at the analyser and see below for a better VSWR plot with less ripple. This shows the VSWR is extremely low and only degrades to 1.02:1 by 1GHz which was the upper frequency it was designed for.



 

[TheoB]

Some of the attempts on here to make a decent 20dB attenuator seem a bit poor to me. It's as if little thought or effort went into it.

Don't forget that the OP was a beginner that wanted to learn. It's nice to see how he progressed with the help from several members on this forum. You step in at the other end of the spectrum to show how to build a perfectly matched attenuator 
But the OP proposed components values calculate to exactly 50 Ohm. So that's also perfectly matched (at low frequency ignoring resistor tolerances).
In general you can only do this properly if you have decent measurement equipment (a reference, you have access to an Ecal, I don't even dare to ask the price).

[G0HZU]

I was mainly referring to the efforts of others who joined the thread later offering examples of their own. The attempt on youtube by W2AEW was quite poor considering he used chip resistors and there were other poor examples in this thread.

In general you can only do this properly if you have decent measurement equipment (a reference, you have access to an Ecal, I don't even dare to ask the price).

You can make a decent 20dB attenuator if you buy the correct SMD resistors and have NO test equipment. Having a VNA or the Ecal doesn't improve the performance of the attenuator.

[TheoB]

Yeah of course, measuring does not affect the performance but the point was to see what the performance actually has become. Especially above a GHz you can be up for a surprise.

[joeqsmith]

Don't forget that the OP was a beginner that wanted to learn.

Exactly.  It was an educational effort.  They were using a scope and didn't realize it need to be terminated at 50 ohms.  It took me a while to find that one.  Once that was sorted out,  things progressed fairly smoothly.   They were experimenting at very low frequencies and we played a lot with TH parts. 

[David Hess]

Exactly.  It was an educational effort.  They were using a scope and didn't realize it need to be terminated at 50 ohms.  It took me a while to find that one.  Once that was sorted out,  things progressed fairly smoothly.   They were experimenting at very low frequencies and we played a lot with TH parts.

The precision 1 GHz switched attenuator I have uses through-hole parts, but it is built inside a block of aluminum with machined cavities and has two tuning screws for each section.

[joeqsmith]

Exactly.  It was an educational effort.  They were using a scope and didn't realize it need to be terminated at 50 ohms.  It took me a while to find that one.  Once that was sorted out,  things progressed fairly smoothly.   They were experimenting at very low frequencies and we played a lot with TH parts.

The precision 1 GHz switched attenuator I have uses through-hole parts, but it is built inside a block of aluminum with machined cavities and has two tuning screws for each section.

Mario had posted some data for an old HP part:
https://www.eevblog.com/forum/projects/20db-rf-attenuator-seeking-feedback-to-improve/msg2964070/#msg2964070

[joeqsmith]

Of course,  read the OPs first post to get some idea how things start out.

My first 50 ohm 'RF' attenuator built on a 25x65 double-sided pcb.
20dB attenuation using metal film resistors.
As far as I can measure, it is -1.5dB at 45MHz.
I would like to do better than this and get to about 300MHz.
Is this possible using conventional axial resistors.

20dB target and measuring 1.5dB....  Something was obviously wrong and it was not construction.