Wes Stewart wrote:

I haven't really been following this thread but if you are looking for
noise diodes and associated stuff, you should take a look at:

http://www.noisecom.com/

Seems to me that at one time they would calibrate a homebrew source for
a modest fee. That may have been a ham that worked there paving the
way; not sure.

In connection with the QST article by Bill Sabin, W0IYH, NoiseCom had a
program of single-quantity sales to amateurs, and offered a low-cost
calibration service. I'm not sure whether that program still operates,
but some years ago NoiseCom were very helpful in supplying a free sample
outside of the USA.

If I were building a noise source again, I wouldn't put NoiseCom to any
trouble. Instead, I would use the base-emitter junction of a
small-signal UHF/microwave transistor. It seems that just about any
junction that goes into avalanche breakdown at a reverse voltage of 5-7V
will give an Excess Noise Ratio of about 35dB. With a modern SMD
package, the device parasitics will be lower than the wire-ended diodes
that NC were offering, so the ENR will be flatter with frequency into
the microwave region.

ENR flatness of the whole noise source depends on your construction. The
other important factor is that the output impedance must not change
significantly between the 'on' and 'off' states of the diode. To achieve
both of these, the best technique is to build the noise source using
SMD/microstrip technique including an attenuator of about 7dB made from
0805 SMD resistors; and then buy a high-class 20dB attenuator (N or SMA)
which need not be expensive at a hamfest. This attenuator then becomes a
permanent part of the noise source - not to be taken off and used for
something else!

This level of attenuation will give you an ENR of about 5-6dB, which is
what you need to measure typical modern low-noise amplifiers.

On the DC side, it is a very good idea to include a constant current
source for long-term stability of ENR. Most devices will give a peak of
ENR at a few mA, so you need to adjust the current to the top of this
peak where the variability is least. (There is much more noise at
currents of a few hundred uA, but you don't want to go there - the noise
output there is far too sensitive to the DC current, temperature, color
of carpet, phase of moon etc.)

It is VERY important to design for the industry-standard power supply of
+28V DC, so that your noise source is a simple plug-in replacement for
any professional source... because that is how you're going to get your
ENR calibration.

All of these ideas come from an article by DJ9BV in DUBUS magazine which
described a very high-class noise source, good up to 10GHz. Mine uses
simpler and much less precise construction, so the ENR begins to wobble
above a few GHz due to resonances.

The DJ9BV articles (in both English and German) are on the DUBUS
website, at:
http://www.marsport.demon.co.uk/archive.htm
Look around 1990, and there are a few follow-ups in later years.
(Unfortunately both archive sites are down right now, but do keep trying
- these articles are *exactly* what you've been looking for!)

As for calibration, the best way is to take your noise head to a
microwave meet where there is professional NF measuring equipment, and
take a cal from the HP346A there. Alternatively, get to know someone -
anywhere - who has access to these facilities, and can do it for you one
lunchtime.

Other useful background information is in an Agilent App Note, at:
http://literature.agilent.com/litweb/pdf/5952-3706E.pdf

If you can still find a copy of 'The VHF/UHF/DX Book' (out of print)
there's a lot of information in there too.


--
73 from Ian G3SEK 'In Practice' columnist for RadCom (RSGB)
http://www.ifwtech.co.uk/g3sek

"Ian White, G3SEK" wrote in message ...

Instead, I would use the base-emitter junction of a
small-signal UHF/microwave transistor. It seems that just about any
junction that goes into avalanche breakdown at a reverse voltage of 5-7V
will give an Excess Noise Ratio of about 35dB. With a modern SMD
package, the device parasitics will be lower than the wire-ended diodes
that NC were offering, so the ENR will be flatter with frequency into
the microwave region.

That's an interesting suggestion...but presumably the 35 dB ENR can't
be counted upon to the level needed for measurements without
calibration.

The DJ9BV articles (in both English and German) are on the DUBUS
website, at:
http://www.marsport.demon.co.uk/archive.htm
Look around 1990, and there are a few follow-ups in later years.
(Unfortunately both archive sites are down right now, but do keep trying
- these articles are *exactly* what you've been looking for!)

Thanks, Ian. I found the DUBUS articles...though I haven't read them
through yet. They look very interesting. Though perhaps not *exactly*
what I was after, as they use a real microwave noise diode, whereas my
thrust is more in the direction of what can be done with more common
parts and without any need for calibration of individual noise
standards. I recognize that this probably won't be possible in the
microwave range but it seems likely (to me, anyway) that reasonable
accuracy (not good enough for EME preamps !) can probably be attained
up to VHF, even with these constraints.

73,
Steve, VE3SMA

Steve Kavanagh wrote:
"Ian White, G3SEK" wrote in message
...

Instead, I would use the base-emitter junction of a
small-signal UHF/microwave transistor. It seems that just about any
junction that goes into avalanche breakdown at a reverse voltage of 5-7V
will give an Excess Noise Ratio of about 35dB. With a modern SMD
package, the device parasitics will be lower than the wire-ended diodes
that NC were offering, so the ENR will be flatter with frequency into
the microwave region.

That's an interesting suggestion...but presumably the 35 dB ENR can't
be counted upon to the level needed for measurements without
calibration.

That's correct; it is only a design guide to the amount of attenuation
that will be needed to give an ENR that's in the right ballpark. But
then you need to know what the actual value *is* - and for that, you
still need a calibration.


The DJ9BV articles (in both English and German) are on the DUBUS
website, at:
http://www.marsport.demon.co.uk/archive.htm
Look around 1990, and there are a few follow-ups in later years.
(Unfortunately both archive sites are down right now, but do keep trying
- these articles are *exactly* what you've been looking for!)

Thanks, Ian. I found the DUBUS articles...though I haven't read them
through yet. They look very interesting. Though perhaps not *exactly*
what I was after, as they use a real microwave noise diode, whereas my
thrust is more in the direction of what can be done with more common
parts and without any need for calibration of individual noise
standards.

My point was that you don't need a real microwave noise diode - any
small, cheap UHF/microwave bipolar transistor will give almost the same
performance up to several GHz.

I recognize that this probably won't be possible in the
microwave range but it seems likely (to me, anyway) that reasonable
accuracy (not good enough for EME preamps !) can probably be attained
up to VHF, even with these constraints.

You can optimize NF using an uncalibrated source, but with this kind of
source you cannot can make worthwhile quantitative measurements without
an external calibration.



--
73 from Ian G3SEK 'In Practice' columnist for RadCom (RSGB)
http://www.ifwtech.co.uk/g3sek

"Stewart Bryant" wrote in message
news:1097795826.35817@sj-nntpcache-3...
Can you measure the noise power of the noise source by comparing
the output (through the RX) with the power from a signal
generator (again through the RX) of known output power?
Presumably using some form of averaging of multiple readings
across the RX passband.

- Stewart G3YSX

Steve Kavanagh wrote:

Some assorted comments and follow-on questions on your suggestions:

- (from bviel) The Elecraft noise generator is not a calibrated unit.
They give a typical ENR but specifically state that it varies from
unit to unit. Their low-level signal generator is calibrated (though
perhaps not quite accurately enough) but this would bring in the issue
of measurement errors between noise and sine-wave powers. And it only
works at one frequency.

- (from bviel) Which MMIC did you find had flat noise figure to 1 GHz
? My experience is that MMIC NF specs are usually at a frequency
above the flat part of the spectrum, as this is more indicative of the
high frequency performance. Are there any with noise figure also
specified, or at least well characterized, at low frequencies (within
the spectral region where the NF is flat) ? But the other question
with MMICs is the unit-to-unit variation in noise figure which I don't
believe is ever specified. If all units of a given part have, say,
between 2 and 3 dB noise figure then a MMIC amp could be a fairly good
standard for amateur use, but if the variation is from 1 to 4 dB then
the usefulness is limited, in my opinion.

- (from bviel) I don't have the JT-44 software but I did look at the
(full) manual. On page 35 it describes the "Measure Sub-Mode" which
allows noise levels to be compared. However it does not appear to
support absolute noise figure measurements (unless you use a
calibrated noise source). Is there more in the online help ?

- (from Mike Andrews) Terry Ritter's stuff seems mostly concerned with
the degree of randomness rather than absolute output power, which
makes sense as he is concerned with cryptography. But I did find,
elsewhere on his web page,

http://www.ciphersbyritter.com/REALRAND/92102201.HTM

a low frequency noise generator based on a diode source followed by a
limiting amp. I imagine this concept could be extended to HF by
suitable choice of amplifier. I am not sure of the significance of
the very different output waveform (pulses of random width and fixed
amplitude) compared to the true random (Gaussian) noise from the diode
itself. I am not sure if I am willing to tackle the math needed to
understand this or to calculate the ENR - can anyone help ?

73,
Steve VE3SMA


The answer is: Sort of.

Another post I made mentions that if you know the gain of your DUT you can
calculate the noise figure by measuring the noise power output with the
input terminated properly. Likewise, I think you could determine the ENR of
a noise source by the same method.

Someone else mentioned calculating the noise power of the ENR from the
current used. This would give the total noise power, but what is needed is
the noise power at a very narrow band of frequencies. If you are trying to
measure a receiver, the receiver determines the bandwidth, but with an LNA
you need a narrow band detector.

When I was doing the afore mentioned software, I used a FFT power spectrum
and made four measurements (DUT with noise source, DUT with out noise
source, test receiver with noise source, test receiver without noise
source). The FFT provided the narrow band filtering.

Of course, watch out for double sideband fold over in the mixer!

Jim
N8EE

Mike Andrews wrote:

[...]

You might find Terry Ritter's work on getting a good noise source to
be of at least a bit (ahem!) of interest:

http://www.ciphersbyritter.com/NEWS5/FMRNG.HTM

There are lots of other hits in Google for a search on
'calibrated "pseudo random" noise'
(without the outer single quotes).

It's hard to generate good noise, and at least as hard to find it.

--
Mike Andrews

Tired old sysadmin

Also take a look at Terry's analysis of various other noise sources:

http://www.ciphersbyritter.com/NOISE/NOISRC.HTM

Good noise is hard to find

Mike Monett

I read Terry's work again, the comments and links.

White noise generated with the B-E junction of a high
Ft transistor, B-E junction is a zener.
If you downconvert above 100kHz, that's where white
noise starts, and display it with your soundcard you
"should" get a flat respons.
The methods where right, the spectrum used was too
low.
Add zener noise to your soundcard and you get pink
noise, because at low frequenties the noise behavior
is always pink + the 1/f semiconductor flicker noise.
You can justify it, but that's manmade white noise.
If I was interested in random numbers, I would use
real white noise, real random.
The keyword here is downconvert in KISS concept,
simple as possible.
Semiconductors produce also white noise.
If you can't display white noise, that does not mean
the noise generator produces non.
I would not trust my sound card at all, because with no
input the FFT shows pink noise, ok,at very low level,
but its added to your not anymore random signal.
I have tried Spectran FFT software.
Peaks from noise floor up to 20dB in the range of zero
to hundred hertz.
The computer is full of signals inside, that's not random
compared to white noise.
If you are measuring relative great signals, the little noise
does not improve the S/N ratio so much that you cannot
copy the signal anymore.
But random noise added with little pink noise is no longer
random, how small the error is, especially if the error is only
at a specifiek part of the spectum.
Maybe a professional A/D chip can do the job ?
Unfortunate a spectrum analyzer can show you the white noise,
but cannot make numbers out of it.
To make numbers out of it is the difficulty I think, not to get
white noise.
The analog world and digital world have a " love and hate"
relationship, sometimes they work fine together,
sometimes not.
I don't say the above statements are right, its just how I think
about the experiments at my point of view.