I'm using two RF upconverters (for SDR). They convert about 0..60 MHz up to 100..160 MHz.
I would like to measure the real bandwidth of these upconverters. How good do they perform with low frequencies (ELF), where is the upper limit of the bandwidth and how large is the loss?
Is there a "simple way" to do this job with a NanoVNA?

There is no simple way to use a VNA to sweep an upconverter. You could use a VNA as a signal source and check the output using your SSD though.

or use a pluto and satsagen software ... grin

dg9bfc sigi

Am 17.02.2022 um 02:13 schrieb Dave W6OQ via groups.io:

> There is no simple way to use a VNA to sweep an upconverter. You could
> use a VNA as a signal source and check the output using your SSD though.
>

I'd suggest you use the S11 port as a signal source for the input
frequency(ies) of your transverter and couple the output of the transverter
(in receive, of course) to an o'scope using a 50-ohm termination on the
input of the o'scope. If you happen to have a spectrum analzer, that would
be even better for the output. The TinySA is good to some 950 MHz, would
get the job done, and is rather inexpensive.

https://www.tinysa.org/wiki/

Dave - WØLEV

On Thu, Feb 17, 2022 at 8:30 PM Siegfried Jackstien <
siegfried.jackstien@freenet.de> wrote:

> or use a pluto and satsagen software ... grin
>
> dg9bfc sigi
> Am 17.02.2022 um 02:13 schrieb Dave W6OQ via groups.io:
>
> There is no simple way to use a VNA to sweep an upconverter. You could use
> a VNA as a signal source and check the output using your SSD though.
>
>
>
>

--
*Dave - WØLEV*
*Just Let Darwin Work*

Oops, spell check did me in. My comment should have said SDR, not SSD. So use your existing SDR to check the upconverted nanoVNA signal.

Sigi,

how could I measure the bandwidth in my case (upconverter 0..60 MHz -> 100..160 MHz) with a Pluto and Satsagen?
The lowest frequency at which the Pluto starts transmitting/receiving is about 70 MHz (1st version) or even 325 MHz (2nd version).
Perhaps I could try with my RTL-SDR stick or with the HackRF One with Satsagen to check at least the upper limit of the bandwidth. I'll try.

Dave W6OQ,

good idea.
Perhaps I can post a picture of the "curve above the noise" here later ... ;)

Dave W0LEV,

I also have an old analog o'scope without sweeping abilities.
The TinySA seems to be a good choice, I'll look into it.

On 2/16/22 5:58 AM, Dirk wrote:
> I'm using two RF upconverters (for SDR). They convert about 0..60 MHz
> up to 100..160 MHz.
> I would like to measure the real bandwidth of these upconverters. How
> good do they perform with low frequencies (ELF), where is the upper
> limit of the bandwidth and how large is the loss?
> Is there a "simple way" to do this job with a NanoVNA?
> _._,_._,_


This is tricky because the input and output frequencies are not the
same.  One way to do it is to use a second mixer.  Your upconverter
converts up to some frequency, and then a mixer, fed with the same LO as
the upconverter, converts it back down. You want a nice high level mixer
for this (so less IM products), and you'll need a fair amount of drive
(it's a high level mixer), so you might need an amplifier for the LO. 
You might want a filter on the input to your downconverter (maybe your
upconverter has a filter, so not an issue)) in case the input to the
upconverter leaks through to the output.

If you're buying parts from minicircuits, you're probably looking at
$100-150.  If you're scrounging, a bit less.

The other approach is to use a spectrum analyzer with a tracking
generator that is synchronized.  Something like the TinySA has a
tracking generator, but it's at the sweep frequency, as I recall, so
you're back to the "convert the tracking generator down to 0-60" which
is sort of the same thing with the VNA and a mixer.

If you have a computer, then you can use TWO Tiny SA (or NanoVNA) - set
them into single frequency mode, use one as a signal generator at
frequency X, the other as a detector at frequency X+100, and have the
computer step through the frequencies. I've not fooled with the
interface on the TinySA, but it might not be too hard.  On the NanoVNA,
you'd be limited at the bottom end by the frequency.   Two widgets at
$50 might be a better approach than $200 of minicircuits parts to build
a downconverter.

https://tinysa.org/wiki/pmwiki.php?n=Main.USBInterface

In fact, it looks like you could do a cool thing such as sweep the input
(from one TinySA, as a source) and collect sweeps from the other TinySA,
and then stack it in a waterfall plot.  That would be really powerful if
you used TWO controllable sources, because you could do automated
intermod testing, too.

Jim Lux,

using a mixer looks like a good way. If I don't want to buy Minicircuit parts: would such a module work?
https://www.ebay.de/itm/124925034223
... and if yes: what would be the best method to "get the LO out of the upconverter" for the mixer?
There is a TXCO on the upconverter PCBs (no external access) and I don't like to make soldered contacts on these PCBs (especially only for measuring purposes!),- but if I would do it anyway, the LO from the TXCO would have to be amplified for the mixer?

On 2/19/22 4:07 AM, Dirk wrote:
> Jim Lux,
>
> using a mixer looks like a good way. If I don't want to buy
> Minicircuit parts: would such a module work?
> https://www.ebay.de/itm/124925034223


That *is* a minicircuits mixer

https://www.minicircuits.com/pdfs/ADE-25MH.pdf
( Minicircuits says it's non-catalog, which means it's either a special
order or obsolete or something.. but they've got the datasheet, and
you've got a source for it, so all good)

You generally want the input to the RF port to be at least 10dB less
than the LO input (which is +13dBm for that mixer) -


> ... and if yes: what would be the best method to "get the LO out of
> the upconverter" for the mixer?
> There is a TXCO on the upconverter PCBs (no external access) and I
> don't like to make soldered contacts on these PCBs (especially only
> for measuring purposes!),- but if I would do it anyway, the LO from
> the TXCO would have to be amplified for the mixer?

You could probably use *any* oscillator as the LO, as long as it's the
right frequency.  If you're using a VNA, then you really do need to
match pretty closely - the detection bandwidth of the NanoVNA is <5 kHz,
which, for an LO of 100 MHz is 50 ppm. You'd be better off if you can
pick up the LO off the upconverter.  Otherwise, the apparent phase shift
will continuously rotate at the difference frequency between the two LOs.

You'll need +13dBm drive for the mixer - so an amplifier might be
needed.  And now the whole system is getting more complex. Really, a
pair of TinySAs (or since you already have the NanoVNA2, it can serve as
the source), and you can use a TinySA as the power measurement, is
probably easier.


While it's easy to draw the block diagram for a simple downconverter
mixer, when you start getting into it, the complexity starts to climb:
you need an amplifier for more LO drive, so now you need a power supply;
Oh, maybe you want to put 3dB pads on the mixer ports to improve the
match; Do you need a filter to remove the "other" mixing products. 
Pretty soon you're spending more in miscellaneous glue than you thought
(unless you have a fairly well stocked microwave junk box).


A computer to control the NanoVNA as a source and a TinySA as the
detector and do single frequency measurements seems easier (the NanoVNA
can be set up to put out CW tones, and I believe the NanoVNA2 can too).

>

I just made my first attempt:

NanoVNA sweeping from 10 kHz to 80 MHz, CH0 connected to Upconverter IN.
Upconverter OUT connected to NESDR SMArTee (an RTL-SDR stick, 25..1750 MHz) IN.
PC: with the program SDRsharp watching the spectrum at 100 MHz an up.

The picture shows the result:
With every NanoVNA sweep (about 2 Hz) the "noise level" goes up a few dB and lowers again.
I watched this flat curve up to 200 MHz with no visible upper end of a bandwidth curve (e.g. at 170 MHz as expected). Also no flattening of the curve at the lower end (e.g. below 100,05 MHz or so).

Doesn't look like a working solution ... :(

On 2/19/22 6:10 AM, Dirk wrote:
> I just made my first attempt:
>
> NanoVNA sweeping from 10 kHz to 80 MHz, CH0 connected to Upconverter IN.
> Upconverter OUT connected to NESDR SMArTee (an RTL-SDR stick, 25..1750
> MHz) IN.
> PC: with the program SDRsharp watching the spectrum at 100 MHz an up.
>
> The picture shows the result:
> With every NanoVNA sweep (about 2 Hz) the "noise level" goes up a few
> dB and lowers again.
> I watched this flat curve up to 200 MHz with no visible upper end of a
> bandwidth curve (e.g. at 170 MHz as expected). Also no flattening of
> the curve at the lower end (e.g. below 100,05 MHz or so).
>
> Doesn't look like a working solution ... :(


that should work.

Set the NanoVNA to CW to, say, 30 MHz, and see if you see the output at
130 MHz.

I'd also pad down the output of the NanoVNA to about -30dBm - 0dBm is a
LOT to put into a converter designed for receive usage.   You may also
want to go for even more, maybe -50 or -60dBm.

You could probably just radiate from a wire connected to the NanoVNA to
a wire connected to the upconverter.




When it's sweeping, the NanoVNA puts out little pulses at each
frequency, so your SA needs to be on maxhold, and you also need to do a
lot of sweeps, because the RTL-SDR  also sweeps in steps (the RTL-SDR
only has ~15-20 MHz total bandwidth). You need to catch the signal when
it's sweeping at the same place. Of course, it looks like you're set up
to look for only 2.5 MHz BW in the plot, so that should work.

Jim and all, thank you very much!

My 2nd try:

Same setup as described below. Only three changes:
1. Sweep from 10 kHz to only 30 MHz
2. NanoVNA OUT (CH0) attenuated by 30 dB
3. Because the program SDR# (SDRsharp) is no SA, it has no true peak hold function. But I set the decay to zero in the FFT display settings. This is a kind of simple peak hold.

Result:
a) Base noise much lower (-100 dB, before -68 dB)
b) I can well see the NanoVNA stimulation peaks distributed over the spectrum
c) I can see some wider and more intense peaks (meaning?) in between
d) The higher part of the spectrum (120..130 MHz) shows less and weaker peaks

Better try than the 1st?

On 2/19/22 9:30 AM, Dirk wrote:
> Jim and all, thank you very much!
>
> My 2nd try:
>
> Same setup as described below. Only three changes:
> 1. Sweep from 10 kHz to only 30 MHz
> 2. NanoVNA OUT (CH0) attenuated by 30 dB
> 3. Because the program SDR# (SDRsharp) is no SA, it has no true peak
> hold function. But I set the decay to zero in the FFT display
> settings. This is a kind of simple peak hold.
>
> Result:
> a) Base noise much lower (-100 dB, before -68 dB)
> b) I can well see the NanoVNA stimulation peaks distributed over the
> spectrum
> c) I can see some wider and more intense peaks (meaning?) in between
> d) The higher part of the spectrum (120..130 MHz) shows less and
> weaker peaks
>
> Better try than the 1st?
> _._,_._,_

Yes..

Especially because you don't have peak hold, I'd set your NanoVNA up for
CW - that way you get a *single* peak - not a weird comb.

On 2/19/22 9:30 AM, Dirk wrote:
> Jim and all, thank you very much!
>
> My 2nd try:
>
> Same setup as described below. Only three changes:
> 1. Sweep from 10 kHz to only 30 MHz
> 2. NanoVNA OUT (CH0) attenuated by 30 dB
> 3. Because the program SDR# (SDRsharp) is no SA, it has no true peak
> hold function. But I set the decay to zero in the FFT display
> settings. This is a kind of simple peak hold.
>
> Result:
> a) Base noise much lower (-100 dB, before -68 dB)
> b) I can well see the NanoVNA stimulation peaks distributed over the
> spectrum
> c) I can see some wider and more intense peaks (meaning?) in between
> d) The higher part of the spectrum (120..130 MHz) shows less and
> weaker peaks
>
> Better try than the 1st?
> _._,_._,_
I suspect those peaks you see are the individual step frequencies from
the NanoVNA's sweep. How many points and what frequency range are you
set to?

There's something that looks like a BPSK/QPSK modulated signal in the
middle (it has that sin x/x spectrum) at about 100.470.

You're getting close.

I set the sweep range to 10 kHz to 30 MHz, 101 points.
What can I try next?
I will try CW mode, but this is a single frequency mode without sweep, isn't it?
In CW mode I then would have to change frequency in small steps manually and write down every corresponding gain in SDR#?
Thats hard work ...

On 2/19/22 10:45 AM, Dirk wrote:
> I set the sweep range to 10 kHz to 30 MHz, 101 points.
So roughly a point every 300 kHz, then?
> What can I try next?
> I will try CW mode, but this is a single frequency mode without sweep,
> isn't it?
> In CW mode I then would have to change frequency in small steps
> manually and write down every corresponding gain in SDR#?

You could write a little script that sends the frequency commands, one a
second or something like that. Then run your SDR# as a waterfall, and
you should see the peak move across the waterfall diagonally.

You might check around - I think someone has a command line version that
uses the NanoVNA-Saver software for the interface.

When I made my suggestion, I was expecting a manual operation with CW from the nanoSNA and tuning the SDR to this frequency. To evaluate the upper and lower frequency limits, you could take bigish steps toward the expected limit and then take decreasingly smaller steps to narrow in on the 3 dB point.

In the past, I've found a good broadband noise source works for
measurements of this sort.

Dave - WØLEV

On Sat, Feb 19, 2022 at 10:06 PM Dave W6OQ via groups.io <david.hostetler=
ieee.org@groups.io> wrote:

> When I made my suggestion, I was expecting a manual operation with CW from
> the nanoSNA and tuning the SDR to this frequency. To evaluate the upper and
> lower frequency limits, you could take bigish steps toward the expected
> limit and then take decreasingly smaller steps to narrow in on the 3 dB
> point.
>
>
>

--
*Dave - WØLEV*
*Just Let Darwin Work*

On 2/19/22 2:15 PM, W0LEV wrote:
> In the past, I've found a good broadband noise source works for
> measurements of this sort.
>
yeah, but a good noise source, with enough output power to make the
measurement, might cost more than a NanoVNA <grin>


Let's say you want to check 50 MHz BW, and the detection BW of your
spectrum analyzer is 10 kHz.   You want a signal that is ~30dB over the
noise floor, and for a typical spectrum analyzer that might be 10-15 dB NF.

I've not measured it for VHF, but the RTL-SDR3 has a noise floor of
about -152 dBm/Hz (NF=22dB) for the HF band.


So doing a bit of math, we want -120dBm/Hz out of the converter into the
RTL-SDR.  I don't know if the OPs upconverter has gain, but let's assume
not.


So we need -120dBm/Hz from the noise source.  A typical HP noise source
has 13 dB ENR, or -174+13 = -161 dBm/Hz so you'd need 40dB of gain.  The
ARRL handbook noise source is about 22 dB ENR, so you'd need about 30 dB
gain.

The total power for 50 MHz BW would be 77dB greater than the "per Hertz"
level, or about -43 dBm, so not a big amplifier.   Then you have all the
issues of knowing whether it's flat or not.


Upshot - if you happen to have a broadband noise source with some
amplifiers, then yeah, that's a quick way to see the bandwidth of a
converter.  If you don't, using the NanoVNA as a source, using some
scripting, is probably the easier way.

> Dave - WØLEV
>
> On Sat, Feb 19, 2022 at 10:06 PM Dave W6OQ via groups.io
> <http://groups.io> <david.hostetler=ieee.org@groups.io> wrote:
>
> When I made my suggestion, I was expecting a manual operation with
> CW from the nanoSNA and tuning the SDR to this frequency. To
> evaluate the upper and lower frequency limits, you could take
> bigish steps toward the expected limit and then take decreasingly
> smaller steps to narrow in on the 3 dB point.
>
>
>
> --
> *Dave - WØLEV*
> /*Just Let Darwin Work*/
>
>

Jim, I have a noise source that is rated at +24 dB ENR. Yes, I usually
feed that to an amplifier to obtain enough "signal". But most transverters
(the receive section) have some gain. Those I've built typically have 15
to 25 dB of gain. Using modern DBMs with a conversion loss of nominally 7
dB, a reasonable amount of quiet gain is required before anything hits the
RF port of the DBM (unless you're strictly an FM'er).

Also, if I just want to look at the -3 dB points on a filter or transverter
or whatever, I use a Zener diode as the noise source with a -5 dB
attenuator to stabilize the impedance (w/c RT of 10 dB). Some are pretty
darn noisy - well beyond my calibrated noise source!! If one has an
o'scope, one can even select the greatest noise output from a selection of
Zeners. For most of my Zener sources I don't need an amplifier. For my
radio astronomy hobby, I use an attenuator between the noise source and the
EUT.

Long ago and far away, Ham Radio, the publication, featured a gated noise
source using a Zener diode and attenuators for impedance stabilization. I
built one. It was almost the most useful piece of test equipment I had at
the time. With that, you can obtain a good idea of gain, bandwidth, and
noise figure all in one display on an o'scope.

Dave - WØLEV

On Sun, Feb 20, 2022 at 12:27 AM Jim Lux <jim@luxfamily.com> wrote:

> On 2/19/22 2:15 PM, W0LEV wrote:
> > In the past, I've found a good broadband noise source works for
> > measurements of this sort.
> >
> yeah, but a good noise source, with enough output power to make the
> measurement, might cost more than a NanoVNA <grin>
>
>
> Let's say you want to check 50 MHz BW, and the detection BW of your
> spectrum analyzer is 10 kHz. You want a signal that is ~30dB over the
> noise floor, and for a typical spectrum analyzer that might be 10-15 dB NF.
>
> I've not measured it for VHF, but the RTL-SDR3 has a noise floor of
> about -152 dBm/Hz (NF=22dB) for the HF band.
>
>
> So doing a bit of math, we want -120dBm/Hz out of the converter into the
> RTL-SDR. I don't know if the OPs upconverter has gain, but let's assume
> not.
>
>
> So we need -120dBm/Hz from the noise source. A typical HP noise source
> has 13 dB ENR, or -174+13 = -161 dBm/Hz so you'd need 40dB of gain. The
> ARRL handbook noise source is about 22 dB ENR, so you'd need about 30 dB
> gain.
>
> The total power for 50 MHz BW would be 77dB greater than the "per Hertz"
> level, or about -43 dBm, so not a big amplifier. Then you have all the
> issues of knowing whether it's flat or not.
>
>
> Upshot - if you happen to have a broadband noise source with some
> amplifiers, then yeah, that's a quick way to see the bandwidth of a
> converter. If you don't, using the NanoVNA as a source, using some
> scripting, is probably the easier way.
>
>
>
>
> > Dave - WØLEV
> >
> > On Sat, Feb 19, 2022 at 10:06 PM Dave W6OQ via groups.io
> > <http://groups.io> <david.hostetler=ieee.org@groups.io> wrote:
> >
> > When I made my suggestion, I was expecting a manual operation with
> > CW from the nanoSNA and tuning the SDR to this frequency. To
> > evaluate the upper and lower frequency limits, you could take
> > bigish steps toward the expected limit and then take decreasingly
> > smaller steps to narrow in on the 3 dB point.
> >
> >
> >
> > --
> > *Dave - WØLEV*
> > /*Just Let Darwin Work*/
> >
> >
>
>
>
>
>
>
>
>

--
*Dave - WØLEV*
*Just Let Darwin Work*

Dave,

How about this one?

www.mgte.com/jpg/sanders-5440c.jpg

This is an old unit from back in my surplus days. The pic is still up on
my site. If there is interest, at least in the noise source, then I can
try contacting some old buddies and see what I can find. No promises,
that unit dates back to the 80's.

Mike C.



On 2/19/2022 7:44 PM, W0LEV wrote:

> Jim, I have a noise source that is rated at +24 dB ENR.  Yes, I
> usually feed that to an amplifier to obtain enough "signal". But most
> transverters (the receive section) have some gain. Those I've built
> typically have 15 to 25 dB of gain.  Using modern DBMs with a
> conversion loss of nominally 7 dB, a reasonable amount of quiet gain
> is required before anything hits the RF port of the DBM (unless you're
> strictly an FM'er).
>
> Also, if I just want to look at the -3 dB points on a filter or
> transverter or whatever, I use a Zener diode as the noise source with
> a -5 dB attenuator to stabilize the impedance (w/c RT of 10 dB).  Some
> are pretty darn noisy - well beyond my calibrated noise source!!  If
> one has an o'scope, one can even select the greatest noise output from
> a selection of Zeners. For most of my Zener sources I don't need an
> amplifier.  For my radio astronomy hobby, I use an attenuator between
> the noise source and the EUT.
>
> Long ago and far away, Ham Radio, the publication, featured a gated
> noise source using a Zener diode and attenuators for impedance
> stabilization.  I built one.  It was almost the most useful piece of
> test equipment I had at the time.  With that, you can obtain a good
> idea of gain, bandwidth, and noise figure all in one display on an
> o'scope.
>
> Dave - WØLEV
>
> On Sun, Feb 20, 2022 at 12:27 AM Jim Lux <jim@luxfamily.com> wrote:
>
> On 2/19/22 2:15 PM, W0LEV wrote:
> > In the past, I've found a good broadband noise source works for
> > measurements of this sort.
> >
> yeah, but a good noise source, with enough output power to make the
> measurement, might cost more than a NanoVNA <grin>
>
>
> Let's say you want to check 50 MHz BW, and the detection BW of your
> spectrum analyzer is 10 kHz.   You want a signal that is ~30dB
> over the
> noise floor, and for a typical spectrum analyzer that might be
> 10-15 dB NF.
>
> I've not measured it for VHF, but the RTL-SDR3 has a noise floor of
> about -152 dBm/Hz (NF=22dB) for the HF band.
>
>
> So doing a bit of math, we want -120dBm/Hz out of the converter
> into the
> RTL-SDR.  I don't know if the OPs upconverter has gain, but let's
> assume
> not.
>
>
> So we need -120dBm/Hz from the noise source.  A typical HP noise
> source
> has 13 dB ENR, or -174+13 = -161 dBm/Hz so you'd need 40dB of
> gain.  The
> ARRL handbook noise source is about 22 dB ENR, so you'd need about
> 30 dB
> gain.
>
> The total power for 50 MHz BW would be 77dB greater than the "per
> Hertz"
> level, or about -43 dBm, so not a big amplifier.   Then you have
> all the
> issues of knowing whether it's flat or not.
>
>
> Upshot - if you happen to have a broadband noise source with some
> amplifiers, then yeah, that's a quick way to see the bandwidth of a
> converter.  If you don't, using the NanoVNA as a source, using some
> scripting, is probably the easier way.
>
>
>
>
> > Dave - WØLEV
> >
> > On Sat, Feb 19, 2022 at 10:06 PM Dave W6OQ via groups.io
> <http://groups.io>
> > <http://groups.io> <david.hostetler=ieee.org@groups.io> wrote:
> >
> >     When I made my suggestion, I was expecting a manual
> operation with
> >     CW from the nanoSNA and tuning the SDR to this frequency. To
> >     evaluate the upper and lower frequency limits, you could take
> >     bigish steps toward the expected limit and then take
> decreasingly
> >     smaller steps to narrow in on the 3 dB point.
> >
> >
> >
> > --
> > *Dave - WØLEV*
> > /*Just Let Darwin Work*/
> >
> >
>
>
>
>
>
>
>
>
>
> --
> *Dave - WØLEV*
> /*Just Let Darwin Work*/
>
>

Oh, and here's a link to an explanation of the Sanders Unit:

https://radioschlock.ecrater.com/p/9884612/sanders-associates-5440c-w-18ghz-source-all

Mike C.


On 2/19/2022 10:50 PM, Mike C. wrote:

>
> Dave,
>
> How about this one?
>
> www.mgte.com/jpg/sanders-5440c.jpg
>
> This is an old unit from back in my surplus days. The pic is still up
> on my site. If there is interest, at least in the noise source, then I
> can try contacting some old buddies and see what I can find. No
> promises, that unit dates back to the 80's.
>
> Mike C.
>
>
>
> On 2/19/2022 7:44 PM, W0LEV wrote:
>> Jim, I have a noise source that is rated at +24 dB ENR.  Yes, I
>> usually feed that to an amplifier to obtain enough "signal".  But
>> most transverters (the receive section) have some gain.  Those I've
>> built typically have 15 to 25 dB of gain.  Using modern DBMs with a
>> conversion loss of nominally 7 dB, a reasonable amount of quiet gain
>> is required before anything hits the RF port of the DBM (unless
>> you're strictly an FM'er).
>>
>> Also, if I just want to look at the -3 dB points on a filter or
>> transverter or whatever, I use a Zener diode as the noise source with
>> a -5 dB attenuator to stabilize the impedance (w/c RT of 10 dB). 
>> Some are pretty darn noisy - well beyond my calibrated noise
>> source!!  If one has an o'scope, one can even select the greatest
>> noise output from a selection of Zeners.  For most of my Zener
>> sources I don't need an amplifier.  For my radio astronomy hobby, I
>> use an attenuator between the noise source and the EUT.
>>
>> Long ago and far away, Ham Radio, the publication, featured a gated
>> noise source using a Zener diode and attenuators for impedance
>> stabilization.  I built one.  It was almost the most useful piece of
>> test equipment I had at the time.  With that, you can obtain a good
>> idea of gain, bandwidth, and noise figure all in one display on an
>> o'scope.
>>
>> Dave - WØLEV
>>
>> On Sun, Feb 20, 2022 at 12:27 AM Jim Lux <jim@luxfamily.com> wrote:
>>
>> On 2/19/22 2:15 PM, W0LEV wrote:
>> > In the past, I've found a good broadband noise source works for
>> > measurements of this sort.
>> >
>> yeah, but a good noise source, with enough output power to make the
>> measurement, might cost more than a NanoVNA <grin>
>>
>>
>> Let's say you want to check 50 MHz BW, and the detection BW of your
>> spectrum analyzer is 10 kHz.   You want a signal that is ~30dB
>> over the
>> noise floor, and for a typical spectrum analyzer that might be
>> 10-15 dB NF.
>>
>> I've not measured it for VHF, but the RTL-SDR3 has a noise floor of
>> about -152 dBm/Hz (NF=22dB) for the HF band.
>>
>>
>> So doing a bit of math, we want -120dBm/Hz out of the converter
>> into the
>> RTL-SDR.  I don't know if the OPs upconverter has gain, but let's
>> assume
>> not.
>>
>>
>> So we need -120dBm/Hz from the noise source.  A typical HP noise
>> source
>> has 13 dB ENR, or -174+13 = -161 dBm/Hz so you'd need 40dB of
>> gain.  The
>> ARRL handbook noise source is about 22 dB ENR, so you'd need
>> about 30 dB
>> gain.
>>
>> The total power for 50 MHz BW would be 77dB greater than the "per
>> Hertz"
>> level, or about -43 dBm, so not a big amplifier.   Then you have
>> all the
>> issues of knowing whether it's flat or not.
>>
>>
>> Upshot - if you happen to have a broadband noise source with some
>> amplifiers, then yeah, that's a quick way to see the bandwidth of a
>> converter.  If you don't, using the NanoVNA as a source, using some
>> scripting, is probably the easier way.
>>
>>
>>
>>
>> > Dave - WØLEV
>> >
>> > On Sat, Feb 19, 2022 at 10:06 PM Dave W6OQ via groups.io
>> <http://groups.io>
>> > <http://groups.io> <david.hostetler=ieee.org@groups.io> wrote:
>> >
>> >     When I made my suggestion, I was expecting a manual
>> operation with
>> >     CW from the nanoSNA and tuning the SDR to this frequency. To
>> >     evaluate the upper and lower frequency limits, you could take
>> >     bigish steps toward the expected limit and then take
>> decreasingly
>> >     smaller steps to narrow in on the 3 dB point.
>> >
>> >
>> >
>> > --
>> > *Dave - WØLEV*
>> > /*Just Let Darwin Work*/
>> >
>> >
>>
>>
>>
>>
>>
>>
>>
>>
>>
>> --
>> *Dave - WØLEV*
>> /*Just Let Darwin Work*/
>>
>

Thank you ALL very much!

I will try:
1. Using a noise source (I have one with Zener diode built into the Nooelec Ham It Up+ [that's not the DUT!]):
https://www.nooelec.com/store/downloads/dl/file/id/82/product/285/ham_it_up_plus_schematic.pdf
2. For the lower edge (possibly In the ELF..VLF range) I could e.g. use audio spectrum analyzers with the soundcard (?)
3. Using the NanoVNA CW mode
4. Using Satsagen program instead of SDR#

Will last a bit, but I'll report here (if I manage).

My 3rd try:

1. Nooelec Ham It Up Plus Noise Source connected to Upconverter IN (Att: - 30 dB)
2. Upconverter OUT connected to Airspy HF+ Discovery IN
3. Program SDR# displays the spectrum 100 MHz an up

Results:
1. Quite nice! I can see the lower edge of the spectrum.
2. Because of the low bandwidth of the Airspy I cannot display the upper edge of the spectrum in form of a typical bandwidth curve, but I can measure the decreasing gain:
100,5 MHz -70,7 dB
110,5 MHz -70,7 dB
120,5 MHz -77,8 dB
130,5 MHz -84,8 dB
140,5 MHz -88,1 dB
150,5 MHz -91,8 dB
160,5 MHz -97,2 dB
170,5 MHz -98,1 dB
> 170,5 MHz -98,1 dB (floor?)

I'm happy so far! Thanks!
What else to improve: Satsagen instead of SDR#, RTL-SDR stick instead of Airspy HF+ to display a wider spectrum for the upper edge.
If I look at the -3 dB rolloff, I cannot believe, that the upper rolloff point is at about 115 MHz (should be higher). So the question is to interprete this result correctly ...

Cascaded poorly designed or matched amplifiers can also make a
"respectable" noise source.

A "hefty" noise producer can be made with a noisy Zener diode followed by a
MMIC amplifier (or two). The MMIC not only supplies gain to yield a pretty
robust noise output, but also acts as an impedance stabilizer (in lieu of
the attenuator with just the Zener).

Dave - WØLEV

On Sun, Feb 20, 2022 at 9:34 AM Dirk <dottensm@gmail.com> wrote:

> Thank you ALL very much!
>
> I will try:
> 1. Using a noise source (I have one with Zener diode built into the
> Nooelec Ham It Up+ [that's not the DUT!]):
>
> https://www.nooelec.com/store/downloads/dl/file/id/82/product/285/ham_it_up_plus_schematic.pdf
> 2. For the lower edge (possibly In the ELF..VLF range) I could e.g. use
> audio spectrum analyzers with the soundcard (?)
> 3. Using the NanoVNA CW mode
> 4. Using Satsagen program instead of SDR#
>
> Will last a bit, but I'll report here (if I manage).
>
>
>

--
*Dave - WØLEV*
*Just Let Darwin Work*

My last try:

1. Nooelec Ham It Up Plus Noise Source connected to Airspy HF+ Discovery IN (Att: - 6 dB)
2. Program SDR# displays the spectrum 0..100 MHz
3. The peak values of the noise from 30 Hz to 90 MHz are manually read and input into an Excel sheet
4. Nooelec Ham It Up Plus Noise Source connected to upconverter IN (Att: - 6 dB)
5. Upconverter OUT connected to Airspy HF+ Discovery IN
6. Program SDR# displays the spectrum 100..190 MHz
7. The peak values of 3 upconverters from 100,00003 to 190 MHz are manually read and input into the abowe Excel sheet
8. The peak noise curve was subtracted from the 3 upconverter curves

The 3 upconverters are:
HU210 -> Chinese 100MHz upconverter board (eBay, etc.)
ELF-HF -> ELF-HF Up Converter V.1.0 (Janielectronics Hungary)
HItUp+ -> Ham It Up Plus (Nooelec)

Result: see picture.

ToDo:
- The used noise source seems to oscillate above ~62 kHz (also without connected upconverter). So the upper edge isn't shown correctly.
- The ELF area (very low edge) could perhaps be displayedshown with audio spectrum analyzers.
- The Airspy HF+ Discovery contains 2 receivers (HF/VHF) with different sensitivity (noise source was received with the HF part, the upconverted spectrum with the VHF part).

Because the solution is without NanoVNA, this is now a bit off topic, so I'll stop posting here.

You might consider that some of the VNAs are not pure sinusoidal drives
and the harmonic distortion may cause issues when driving in to active
electronics (IC's or transistor)
This is usually not an issue with passive networks (R-L-C or antennas or
couplers or diplexers).

There is a youtube video showing this somewhere,

On Thu, 17 Feb 2022 at 00:58, Dirk <dottensm@gmail.com> wrote:

> I'm using two RF upconverters (for SDR). They convert about 0..60 MHz up
> to 100..160 MHz.
> I would like to measure the real bandwidth of these upconverters. How good
> do they perform with low frequencies (ELF), where is the upper limit of the
> bandwidth and how large is the loss?
> Is there a "simple way" to do this job with a NanoVNA?
>
>
>

--
Joey +61 412 750 765