Topics

Poor Quality 50 Ohm Load - Where to get accurate ones?


Glen Jenkins WB4KTF
 

The 50 OHM load (SMA-Male) that came with my nanoVNA-H4 measures 51.13+ ohms. Not a good start for calibration. Where is a good source for GOOD loads that are accurate?
--
-----
Glen Jenkins, WB4KTF, Austin, TX


Bob Albert
 

Loads are widely available.  Check ebay for example.  But why do you need an accurate load?  51 Ohms represents an SWR of 1.02.  And how are you measuring it?

On Thursday, August 13, 2020, 05:52:44 PM PDT, Glen Jenkins WB4KTF <wb4ktf@pipeline.com> wrote:

The 50 OHM load (SMA-Male) that came with my nanoVNA-H4 measures 51.13+ ohms.  Not a good start for calibration.  Where is a good source for GOOD loads that are accurate?
--
-----
Glen Jenkins, WB4KTF, Austin, TX


AB6BT
 

What are you using to measure the load resistance?

At 05:52 PM 8/13/2020, you wrote:
The 50 OHM load (SMA-Male) that came with my nanoVNA-H4 measures 51.13+ ohms. Not a good start for calibration. Where is a good source for GOOD loads that are accurate?
--
-----
Glen Jenkins, WB4KTF, Austin, TX


Jim Lux
 

On 8/13/20 5:52 PM, Glen Jenkins WB4KTF wrote:
The 50 OHM load (SMA-Male) that came with my nanoVNA-H4 measures 51.13+ ohms. Not a good start for calibration. Where is a good source for GOOD loads that are accurate?
that's a VSWR of 1.023:1 and a S11 of -39dB


55 ohms is 1.1:1


How much are you willing to pay?

Pasternack has a SMA Male load (PE6002) for $38

VSWR specified at 1.17:1, which is worse.


For $1900 they'll sell you a 3.5mm load for a cal kit
up to 4GHz they claim S11 mag <-40dB

That's about the same as the load you have.
It will come with a certificate, though.

They have a cheaper one for $485 PESTR1001 - it's actually better up to 3GHz, -44dB for S11 mag, worst case, typical is -46

The chart in the data sheet actually shows better than -50dB at 1 GHz

https://www.pasternack.com/images/ProductPDF/PE5TR1001.pdf


Jim Lux
 

On 8/13/20 6:04 PM, AB6BT wrote:
What are you using to measure the load resistance?
This is a nice example of the rabbit hole (or prairie dog village) one can go down..

Is the DC resistance the RF resistance? Maybe, maybe not. Generally, though, DC resistance is less than AC resistance.

What's the temperature coefficient of
a) the load
b) the ohmmeter

What is the lead resistance on the ohm meter (or was it done with a 4 terminal Kelvin measurement)? First hit on google brings up Klein test leads that are 41" long, but they don't say what gauge the wires are. let's say they're something like AWG 20 - that's 0.01 ohms/ft or about 0.07-0.08 ohms total - although usually, the meter is calibrated to read zero with the leads shorted.

Looking up something like a venerable Fluke 113 - the accuracy specification is 0.9% + 2 counts with a resolution of 0.1 ohms

1% of 50 ohms is 49.5-50.5 ohms


This is not meant to beat up on Glen, but it points up the challenges in making accurate measurements - you have to worry about "everything" - just because the NanoVNA reads down to -100dB doesn't mean that it's *accurate* at that kind of level. I spent a while about 20 years ago at work trying to accurately measure a 100 dB attenuator to 0.1dB uncertainty - it's an ordeal.

Fortunately, most of us are making "relative" measurements - tuning a filter, checking pass band attenuation, looking for a good match on an antenna.

The people who worry about getting a 40dB match (1% voltage) are people running a LOT of power (S11 of -30dB from 200kW is 200W reflected) or doing precision calibration of things. Mismatch uncertainty becomes the dominant error source in precision RF power measurement - I built a 13.402 GHz precision noise source for amplitude calibrations, and it had lapped and pinned waveguide flanges, with an mate/remate uncertainty (measured) of, I think, 0.0001 in the reflection coefficient (that's -80dB) - because we knew it's noise temperature was about 8400K, +/- <2.7K (i.e. 0.1%)

These two papers (especially the second one, which has more details) describes the kind of painstaking effort it takes to drive uncertainties below 1%

https://tsapps.nist.gov/publication/get_pdf.cfm?pub_id=22465
https://tsapps.nist.gov/publication/get_pdf.cfm?pub_id=3392

And this is where the $100k VNA with the $20k cal kit earns its keep - when the engineering cost on the thing you're measuring is $500k.


Ultimately, be happy with your 51 ohm load!



At 05:52 PM 8/13/2020, you wrote:
The 50 OHM load (SMA-Male) that came with my nanoVNA-H4 measures 51.13+ ohms.  Not a good start for calibration.  Where is a good source for GOOD loads that are accurate?
--
-----
Glen Jenkins, WB4KTF, Austin, TX


Gufo Rosso
 

use 6 @ at 300 ohm smd resistor 1% or
https://hforsten.com/improved-homemade-vna.html (middle page)
https://www.instructables.com/id/23GHz-SMA-Calibration-Kit/ (23ghz)
hi,
Red Owl


Il giorno ven 14 ago 2020 alle ore 02:57 Bob Albert via groups.io
<bob91343=yahoo.com@groups.io> ha scritto:

Loads are widely available. Check ebay for example. But why do you need an accurate load? 51 Ohms represents an SWR of 1.02. And how are you measuring it?
On Thursday, August 13, 2020, 05:52:44 PM PDT, Glen Jenkins WB4KTF <wb4ktf@pipeline.com> wrote:

The 50 OHM load (SMA-Male) that came with my nanoVNA-H4 measures 51.13+ ohms. Not a good start for calibration. Where is a good source for GOOD loads that are accurate?
--
-----
Glen Jenkins, WB4KTF, Austin, TX





Stephen Laurence
 

Surely, as I recall, it is not the accuracy of the dc resistance of the load that is most important, but the ac impedance of the whole device as the frequency increases. Inductance of the resistor(s) in the load, capacitance across them, poor connector dimensions are FAR more important if you want good results above hf frequencies. Construction of these reference loads is critical, which is why the professional ones useable above 6 ghz cost many times the cost of the nano. Making a good short and open is a walk in the park by comparison.

There are several sites on the internet which show how to make your own sma standards which can be quite good up to several ghz. I have made several sets which are almost as good as some cheap “professional” ones I have, but I have not managed to get them assessed on a professional vna. Unfortunately, owners and users of really good kit do not like “ foreign” stuff to come near their gear in case dodgy connectors damage the ones on their £20,000 plus vna. Likewise they never loan their standard out.


Steve L. G7PSZ


Stephen Laurence
 

A further thought on accuracy of standards.

I suspect the transmission impedance of most coax cables which claim to be 50 ohm probably has a 2-5% tolerance between reels, and possibly 1% variation along its length of a single reel. You only need to gently tread on the coax to distort it and change the impedance in the piece under your foot, which will cause a slight reflection of power going to your aerial. A moderately sharp bend in the cable will do likewise. There is a reason why professional installations use fairly rigid cable like Heliax.

It is far more important that your calibration load presents a constant resistance across the range of frequencies you wish to study.

Steve L


aparent1/kb1gmx <kb1gmx@...>
 

The problem of 51.13 ohm load vs perfect 50 is one of chasing small
numbers far to the right of the decimal point.

IF the antenna or load is Zero reflected energy at 51.13 ohm than
at 50 ohms the resulting SWR would be 1.02. Is that a problem
in any real world??

For development work at Ghz frequencies I had a strip of 0802 SMT
resistors of 49.9 ohms 1% I leave it to some to calculate the error
introduced by that. Other times i'd use 50.6 and 51 ohms as thats
what was handy, allways SMT as lead length was more of an issue.
I'll leave the answer as small, maybe insignificantly so.

The bigger issue with loads is not what real is (in this case 51.13 ohms)
but what is the imaginary part due to reactance.

Allison
-----------------
No direct email, it goes to bit bucket due address harvesting in groups.IO


Roger Need
 

On Thu, Aug 13, 2020 at 05:52 PM, Glen Jenkins WB4KTF wrote:


The 50 OHM load (SMA-Male) that came with my nanoVNA-H4 measures 51.13+ ohms.
Not a good start for calibration. Where is a good source for GOOD loads that
are accurate?
--
-----
Glen Jenkins, WB4KTF, Austin, TX
I am really surprised your 50 ohm load is that far off. Is it quite short and stainless steel finish (the other two are brass) Mine is quite close to 50 ohms.

I made my own SOL Kit from PCB board mount female SMA connectors. Work well at the ends of the cables.

Roger


Jerry Gaffke
 

It's quite possible that you could build something better than the 50 ohm calibration standard
that gets thrown into the box along with a $40 VNA.
I just use the one I found in the box.

If building a calibration standard, I'd use the very common 1% metal film surface mount resistors.
Thin film surface mount resistors could be better, but watch for power handling capability
and any fancy trimming that might give significant inductance.
A female SMA connector has room for four 200 ohm 0805 surface mount resistors
coming in from four different directions to minimize inductance, this also increases
the amount of power it can handle.
A 1/2 Watt 5% carbon composition resistor clipped from a 50 year old radio should work well enough at 30mhz.
But you definitely do not want any kind of wire wound resistor.

I know very little about the various resistor types, here's a quick google hit that apparently does:
http://www.resistorguide.com/thin-and-thick-film/

Jerry, KE7ER

On Fri, Aug 14, 2020 at 04:05 PM, aparent1/kb1gmx wrote:

The bigger issue with loads is not what real is (in this case 51.13 ohms)
but what is the imaginary part due to reactance.


Glen Jenkins WB4KTF
 

Hello to All,
I measured the resistance at DC with my DVM.
I like Gufo's nice response and his link to instructions to build my own 50 ohm load. I looked at Amazon for other 50 ohm loads.
There I saw reviews from several folks commenting that the loads that they purchased there were also way off 50 ohms, so they returned them for refunds. So maybe I am not exaggerating or going into down a rabbit hole. In any case I now have some instructions to build my own 23cm capable 50 ohm loads.
Thanks to everyone who provided input to my concern.
--
-----
Glen Jenkins, WB4KTF, Austin, TX


Bob Albert
 

Unless you have Kelvin leads, your DVM reading is probably high.  What do you read with the leads shorted?

On Friday, August 14, 2020, 05:05:27 PM PDT, Glen Jenkins WB4KTF <wb4ktf@pipeline.com> wrote:

Hello to All,
I measured the resistance at DC with my DVM.
I like Gufo's nice response and his link to instructions to build my own 50 ohm load.  I looked at Amazon for other 50 ohm loads.
There I saw reviews from several folks commenting that the loads that they purchased there were also way off 50 ohms, so they returned them for refunds.  So maybe I am not exaggerating or going into down a rabbit hole.  In any case I now have some instructions to build my own 23cm capable 50 ohm loads.
Thanks to everyone who provided input to my concern.
--
-----
Glen Jenkins, WB4KTF, Austin, TX


KV5R
 

W0QE shows building of fairly high precision cal-kits on youtube:
https://www.youtube.com/watch?v=NSAQ2iQNX2I
https://www.youtube.com/watch?v=-PK9Bn7Ixnw
These might give you an idea, if you want to try making a better load.
Interesting to note he found that using 2 100 ohm 1% SMT resistors is more accurate than 1 50 or 4 200's.
73, --KV5R


Jim Lux
 

On 8/14/20 5:16 PM, Bob Albert via groups.io wrote:
Unless you have Kelvin leads, your DVM reading is probably high.  What do you read with the leads shorted?
I always have problems with the probes making good contact on small surfaces, too.
But really, 1% is probably good enough

On Friday, August 14, 2020, 05:05:27 PM PDT, Glen Jenkins WB4KTF <wb4ktf@pipeline.com> wrote:
Hello to All,
I measured the resistance at DC with my DVM.
I like Gufo's nice response and his link to instructions to build my own 50 ohm load.  I looked at Amazon for other 50 ohm loads.
There I saw reviews from several folks commenting that the loads that they purchased there were also way off 50 ohms, so they returned them for refunds.  So maybe I am not exaggerating or going into down a rabbit hole.  In any case I now have some instructions to build my own 23cm capable 50 ohm loads.
Thanks to everyone who provided input to my concern.


Bob Albert
 

Actually 1% is overkill.  Even 5% is close enough for all but the most serious of measurements.  That represents an SWR of 1.05, far better than most systems.
Having said that, I have measured several nano loads and some others I have acquired over the years and they are very close to 50 Ohms, better than 1%.
Bob

On Friday, August 14, 2020, 06:06:27 PM PDT, Jim Lux <jimlux@earthlink.net> wrote:

On 8/14/20 5:16 PM, Bob Albert via groups.io wrote:
  Unless you have Kelvin leads, your DVM reading is probably high.  What do you read with the leads shorted?
I always have problems with the probes making good contact on small
surfaces, too.
But really, 1% is probably good enough

      On Friday, August 14, 2020, 05:05:27 PM PDT, Glen Jenkins WB4KTF <wb4ktf@pipeline.com> wrote:
 
  Hello to All,
I measured the resistance at DC with my DVM.
I like Gufo's nice response and his link to instructions to build my own 50 ohm load.  I looked at Amazon for other 50 ohm loads.
There I saw reviews from several folks commenting that the loads that they purchased there were also way off 50 ohms, so they returned them for refunds.  So maybe I am not exaggerating or going into down a rabbit hole.  In any case I now have some instructions to build my own 23cm capable 50 ohm loads.
Thanks to everyone who provided input to my concern.


Jerry Gaffke
 

I'm not in a position to watch W0QE's youtube videos,
does he say why two at 100 ohms is best?
I'd assume two 100 ohm resistors are a good tradeoff between the higher inductance
of a single 50 ohm resistor and the higher capacitance of four 200 ohm resistors.
Was he using 0805 metal film resistors?

Jerry, KE7ER

On Fri, Aug 14, 2020 at 06:06 PM, KV5R wrote:
W0QE shows building of fairly high precision cal-kits on youtube:
https://www.youtube.com/watch?v=NSAQ2iQNX2I
https://www.youtube.com/watch?v=-PK9Bn7Ixnw
These might give you an idea, if you want to try making a better load.
Interesting to note he found that using 2 100 ohm 1% SMT resistors is more
accurate than 1 50 or 4 200's.
73, --KV5R


Shirley Dulcey KE1L
 

I think it's mostly that 100 ohms is easier to find than 50 ohms because
it's part of the 20% progression, so everybody has them on hand. 51 ohms is
on the 10% progression. 50 ohms isn't on either, so it's not a value you
will find in resistor assortments, you have to order it specially as a 1%
or closer tolerance part.

On Fri, Aug 14, 2020 at 9:35 PM Jerry Gaffke via groups.io <jgaffke=
yahoo.com@groups.io> wrote:

I'm not in a position to watch W0QE's youtube videos,
does he say why two at 100 ohms is best?
I'd assume two 100 ohm resistors are a good tradeoff between the higher
inductance
of a single 50 ohm resistor and the higher capacitance of four 200 ohm
resistors.
Was he using 0805 metal film resistors?

Jerry, KE7ER

On Fri, Aug 14, 2020 at 06:06 PM, KV5R wrote:
W0QE shows building of fairly high precision cal-kits on youtube:
https://www.youtube.com/watch?v=NSAQ2iQNX2I
https://www.youtube.com/watch?v=-PK9Bn7Ixnw
These might give you an idea, if you want to try making a better load.
Interesting to note he found that using 2 100 ohm 1% SMT resistors is
more
accurate than 1 50 or 4 200's.
73, --KV5R



Dick
 

You are assuming the measuring device is better than 1% accuracy.
That's not that simple.

73, Dick, W1KSZ
________________________________
From: nanovna-users@groups.io <nanovna-users@groups.io> on behalf of Shirley Dulcey KE1L <mark@buttery.org>
Sent: Friday, August 14, 2020 7:14 PM
To: nanovna-users@groups.io <nanovna-users@groups.io>
Subject: Re: [nanovna-users] Poor Quality 50 Ohm Load - Where to get accurate ones?

I think it's mostly that 100 ohms is easier to find than 50 ohms because
it's part of the 20% progression, so everybody has them on hand. 51 ohms is
on the 10% progression. 50 ohms isn't on either, so it's not a value you
will find in resistor assortments, you have to order it specially as a 1%
or closer tolerance part.

On Fri, Aug 14, 2020 at 9:35 PM Jerry Gaffke via groups.io <jgaffke=
yahoo.com@groups.io> wrote:

I'm not in a position to watch W0QE's youtube videos,
does he say why two at 100 ohms is best?
I'd assume two 100 ohm resistors are a good tradeoff between the higher
inductance
of a single 50 ohm resistor and the higher capacitance of four 200 ohm
resistors.
Was he using 0805 metal film resistors?

Jerry, KE7ER

On Fri, Aug 14, 2020 at 06:06 PM, KV5R wrote:
W0QE shows building of fairly high precision cal-kits on youtube:
https://www.youtube.com/watch?v=NSAQ2iQNX2I
https://www.youtube.com/watch?v=-PK9Bn7Ixnw
These might give you an idea, if you want to try making a better load.
Interesting to note he found that using 2 100 ohm 1% SMT resistors is
more
accurate than 1 50 or 4 200's.
73, --KV5R



KV5R
 

It's been a few months since I watched those 2 W0QE videos, but I seem to remember the 2 100's had a wider frequency response than using either 1 or 4 resistors. He duplicated the results of another engineer with the same results. He measured with a high-end VNA and expensive cal sets. So yes, I think that using a pair of SMT resistors, mounted symmetrically on the back of the connector, provided the best balance of inductance and capacitance, i.e., flat frequency response over the widest range. I don't remember exactly what resistors he used.

On Fri, Aug 14, 2020 at 07:35 PM, Jerry Gaffke wrote:

I'm not in a position to watch W0QE's youtube videos, does he say why two at 100 ohms is best?
I'd assume two 100 ohm resistors are a good tradeoff between the higher inductance
of a single 50 ohm resistor and the higher capacitance of four 200 ohm resistors.
Was he using 0805 metal film resistors?

Jerry, KE7ER