20dB 'RF' attenuator - seeking feedback to improve

[ogden]

The plating is a bit off and note a resistor kicked up and solder to the backside.

Those on the back are spare resistors 

[enut11]

Hi @joeqsmith
My attenuator set never arrived and they gave me my money back. May be a bit of a crisis over in China.

I was going to compare the commercial unit to my own home built attenuators.

It will be interesting to see your VNA test results.
enut11

[joeqsmith]

Again, from their ad:
Direct: 0dB (reference)
Level 1: 10 0.8 dB
Level 2: 20 1.1 dB
Level 3: 30 1.5dB
Standing wave ratio: 1.20


I fired up the old HP8754A.   The screen is calibrated 1:1.   Keep in mind that I use home made standards to cal it.   My calibration routines assume the standards are perfect.  At some point I will have a better VNA and we can remeasure them.   

I don't have any plans to try and measure them with the nano because of the higher frequencies. 
   

20dB_1p3GHz: 20dB attenuator swept to 1.3GHz 
30dB_1p3GHz: 30dB attenuator swept to 1.3GHz 

20dB_intstep_2p6GHz:  This is the HP8748A's internal step attenuator set to 20dB, with the doubler, sweeping to 2.6GHz
30dB_intstep_2p6GHz:  This is the HP8748A's internal step attenuator set to 30dB, with the doubler, sweeping to 2.6GHz

20dB_2p6GHz: 20dB attenuator swept to 2.6GHz
30dB_2p6GHz: 30dB attenuator swept to 2.6GHz

MWM_18GHzTerm_SWR_2p6GHz:  This is a Midwest Microwave 18GHz 50ohm terminator.  We are looking at the SWR.  At 160MHz, 1.07:1 and 2.2GHz, 1.2:1. 

The MWM terminator was then attached to the attenuators while measuring their SWR.

20dB_SWR_2p6GHz:  SWR 160MHz 1.03:1, 2.12GHz 1.14:1
30dB_SWR_2p6GHz:  SWR 160MHz 1.05:1, 2.2GHz 1.18:1


I then connected the three attenuators in series. The HP8501A was set to average eight sweeps.  40dB of offset was added on the thumb wheel switches so the top of the graph is now 0dB. 

[ogden]

Can't you show 1dB/div instead of 10dB/div vertical scale? Your NanoVNA software is able to show 0.3dB/div or something like that.

[enut11]

@joeqsmith
Looks OK to me. Some brief comments on what you see on the charts and your opinion on how good the attenuator set is would be appreciated.
enut11

[joeqsmith]

@joeqsmith
Looks OK to me. Some brief comments on what you see on the charts and your opinion on how good the attenuator set is would be appreciated.
enut11

If I set the VNA to 1dB/division, the software is also then 1dB/div.  Then use the VNAs internal step attenuator with a thru to calibrate to.  For each attenuator the center line represents the nominal value.  So, the center line is 20dB for the 20dB attenuator and 30dB for the 30dB attenuator.  In all cases we are sweeping to 2.6GHz.   

20dB_2p6GHz_1dBdiv & 30dB_2p6GHz_1dBdiv  are the low cost attnuators.    For a comparison, I have included a 30dB Midwest Microwave 18GHz part.  There is a substantial difference between the cost of these parts.     

While the HP8754A can read down to 0.25dB/division, notice that even after calibrating the unit to the 30dB internal attenuator that the system has a fair amount of noise.   Like the $50 Nano, this old system certainly has it's limitations.   

This first set of low cost attenuators performs pretty much how I would expect at these lower frequencies.  Once I am setup, we can see how they perform at higher frequencies.   The noise on the newer boat anchor is much improved over the 1970s system as well.     

[joeqsmith]

Scam or real?  $2600 USD, seems way too good to be true!!

https://www.ebay.com/itm/Agilent-E8358A-PNA-Network-Analyzer-300-kHz-to-9-GHz-Opt-010/352986648287?hash=item522fa4cedf:i:352986648287

[hendorog]

Not sure any conclusion can be drawn from the higher res traces. 
Without the cal kit characterised I mean.

[enut11]

Scam or real?  $2600 USD, seems way too good to be true!!

https://www.ebay.com/itm/Agilent-E8358A-PNA-Network-Analyzer-300-kHz-to-9-GHz-Opt-010/352986648287?hash=item522fa4cedf:i:352986648287

I just checked my piggy bank and I am a little short on funds...

[ogden]

Not sure any conclusion can be drawn from the higher res traces. 
Without the cal kit characterised I mean.

Without characterised cal kit you shall avoid any other conclusion except one: calkit is not characterised  When we assume that calibration is ok, now at 1dB/div we have better chances to estimate accuracy of attenuators which BTW is surprisingly good considering price.

[PA0PBZ]

Scam or real?  $2600 USD, seems way too good to be true!!

https://www.ebay.com/itm/Agilent-E8358A-PNA-Network-Analyzer-300-kHz-to-9-GHz-Opt-010/352986648287?hash=item522fa4cedf:i:352986648287

Totally legit: https://www.ebay.com.8641224.com/CT/c1875.html (*)

(*) Yes, sarcasm (for people that don't see it)

[joeqsmith]

Not sure any conclusion can be drawn from the higher res traces. 
Without the cal kit characterised I mean.

Odd.  The higher res traces are S21 only and I am using a good thru with the HP's internal attenuator to calibrate the system.  This is a 4GHz HP branded attenuator.   Consider that with the Midwest Microwave part installed, it is within +/-0.5dB.   Now, when we install the low cost attenuator, we can clearly see it's not near as flat and has a bit of an offset.   I have far more confidence with the Midwest Microwave matching with the HP part when using a good thru, than I do with the low cost attenuator.  You may see it otherwise.

Scam or real?  $2600 USD, seems way too good to be true!!

https://www.ebay.com/itm/Agilent-E8358A-PNA-Network-Analyzer-300-kHz-to-9-GHz-Opt-010/352986648287?hash=item522fa4cedf:i:352986648287

Totally legit: https://www.ebay.com.8641224.com/CT/c1875.html (*)

(*) Yes, sarcasm (for people that don't see it)

Gone! That was the deal of the year right there. 

[PA0PBZ]

Gone! That was the deal of the year right there.

Well, the second link is still active, so you can still pay if you want

[joeqsmith]

Gone! That was the deal of the year right there.

Well, the second link is still active, so you can still pay if you want

Click on your area.   When I did this, I get a 404 Not Found error where it would come up before.  I assume this means it was sold. 

[enut11]

What's wrong with my NanoVNA. This is what I see with nothing connected.
Also, cannot calibrate it?? Is there a way to reset it to factory defaults?
enut11

I was trying to calibrate my NanoVNA and the OPEN test (far right marker) would not work.
There is a lot of conflicting advice out there on how to calibrate this device, whether to reset or not, whether to save or not, etc.
I came across this website and all is good now. Change the language top left and scroll down several pages for the Cal notes.
enut11

http://ha3hz.hu/hu/home/top-nav/12-seged-berendezesek/15-nanovna

[joeqsmith]

Were you able to get it working or are you still having problems with it?   

I was using the Reset prior to running the calibration.  The save is optional.  Save 0 would be loaded on power up.  The other have to be manually loaded. 

[enut11]

Yes, my Nano is now working.
What I am doing is learning how to use it and what it is telling me...

[joeqsmith]

Yes, my Nano is now working.
What I am doing is learning how to use it and what it is telling me...

Good to hear.   You may find that making a simple test jig like that one I linked earlier with the solder braid, may be helpful.  Something so you can easily try different components and such. 

This is the cal kit I am working on for the higher frequencies.  I also now have an SMA along with a few sections of shorted coax and the Ts.     
https://www.eevblog.com/forum/rf-microwave/get-ideal-networks-from-cal-kit-parameters/msg2921098/#msg2921098

[joeqsmith]

It took FireFox a while to catch up.    

Shown are the new standards  along with some of the test boards I made for the Nano resting on the new boat anchor.   These new standards were characterized using a set of real standards.       

Like that really bad home made attenuator I show, at these lower frequencies we can get away with a lot of errors and still get reasonable results.  Well, at least for my hobby experiments....  Using the Nano's standards which I have no data for at all,  running a section of coax to 900MHz, we can see the peak to peak ripple is about  0.4dB, which gives us a source match of 33dB.   Running the T-check over the same frequency span, we can see its around 10% error at 800MHz.   

Even without having any data for these, they are not the limiting factor when used with the Nano.   Running these same tests on the Nano would look like total garbage as the noise is really poor above 400MHz.   Keep it below that and you should get some pretty decent results. 

I have looked at a few filters in the GHz range using Radiolistener's firmware which supported 1.5GHz.   Even at these frequencies, the Nano has enough dynamic range to get a feel for what things are doing.   

https://www.eevblog.com/forum/rf-microwave/nanovna-custom-software/msg2703812/#msg2703812

[joeqsmith]

The second set of attenuators has yet to arrive but at least they claim to have shipped.  Looks like mid month. 

I have been getting some help from Mario with my calibration questions.
https://www.eevblog.com/forum/rf-microwave/get-ideal-networks-from-cal-kit-parameters/

So far, getting a system to produce any sort of meaningful results, to measure these low cost attenuators, has been a bust.  There are a lot of variables to juggle with this new boat anchor.  Hope to have things sorted out by the time the second parts arrive.

[joeqsmith]

Attached is the T-check for the uncompensated 3.5mm home made standards up to 2GHz. 

I also had a look at the low cost attenuators at 4GHz and am confident they do not come near meeting that 1.5dB flatness at 30dB.  It doesn't appear to be even close when compared with a commercial part. 

While I have made some progress, the new standards are not good enough to have any sort of confidence in the SWR at these frequencies.   

[joeqsmith]

MWM vs the low cost attenuator swept to 4GHz.  Also shown is the home made 4-stage 50dB attenuator that I had trimmed.   Fairly flat to around 2GHz, then it takes off.   Last plot shows a commercial 50dB attenuator.   

I'm actually impressed the homemade 50dB attenuator is this good. 

[joeqsmith]

The second type of low cost attenuators arrived.   And while they do not have the protective caps, the design looks more what I would expect for something spec'ed at 4GHz.

Plating finish at least looks more uniform.   

I still have no way to measure the SWR but I will go ahead and sweep the S21 and see how it compares.   

[joeqsmith]

Yellow is the MWM commercial part,  purple is the green PCB, red is the blue PCB.    It looks like it may actually meet the stated specs for flatness.  Pretty impressive for the price. 

Once I get the old boat anchor fully up and running, I will retest them.   

[David Hess]

Would performance be better with a more symmetric layout that doubles the number of shunt resistors?