LF Spectrum Analyser

Do you have a question? Post it now! No Registration Necessary

Translate This Thread From English to

Threaded View
Any recommendations for a spectrum analyser, say <10%Hz to >=1MHz, maybe  
USB, hopefully inexpensive, definitely available now, in the UK?

Aaronia any good?  Not generally very good reviews, but people here may  
know better.

Cheers
--  
Clive

Re: LF Spectrum Analyser
Am 27.06.2017 um 22:25 schrieb Clive Arthur:
Quoted text here. Click to load it

A used Agilent 89441A. The IF unit would be enough probably.

Available cheaply because people think it is usable only
for checking 3G cell phones.

cheers, Gerhard

Re: LF Spectrum Analyser

Quoted text here. Click to load it

Just playing around with one. Performance about ok for my purpose
(generator magnet field EMC measurements), but the software feels
uncertain at least on Mac. As an example, I'm having hard time to
keep the unit as dBuV while changing span or some other parameters.

I'll compare with HP 3585A when I get back from the gig.

--
mikko


Re: LF Spectrum Analyser
On Tuesday, June 27, 2017 at 4:26:01 PM UTC-4, Clive Arthur wrote:
Quoted text here. Click to load it

What's low cost?  I've been lusting after a Rigol SA.  I see there is now a  
low end (100 kHz to 500 MHz) one ~ $700.00, but I would like a DSA815 with TG.  There seems to be a divide in the SA market at ~100 kHz.  
With 'audio' SA's from milli-Hertz to 100kHz, and then the RF stuff.  

I use the FFT on my 'scope as a low end SA, but it's not very good at  
zooming in on one limited frequency window.  What are you doing?  

George H.  

Re: LF Spectrum Analyser
On Wed, 28 Jun 2017 06:06:36 -0700 (PDT), George Herold

Quoted text here. Click to load it

Do these 100 kHz 'audio' SAs have proper S&H at input ?

In that case use a local oscillator (LO) to mix up a frequency segment
of interest to a convenient intermediate frequency say 45 MHz. Using a
100 kHz roofing filter to limit the spectrum to 45.0 to 45.1 MHz and
feed it to the 100 kHz audio SA.  

To select a different 100 kHz segment,just step he LO in 100 kHz
steps. Due to the big frequency step, even a PLL can be used as the
LO, without introducing too much phase noise.

Search for "waterfall display" and WebSDR. These often use an I/Q
(Zero IF)  front end with cheap "audio"  A/D converters.


Re: LF Spectrum Analyser
On Wednesday, June 28, 2017 at 10:32:57 AM UTC-4, snipped-for-privacy@downunder.com wrote:
Quoted text here. Click to load it
Sample and Hold?  I don't think so.  We've got an SRS770.  
I've done the mixing trick to look at somewhat higher frequencies.  
(a few MHz.)  But not the best filtering, so all sorts of  
aliasing (if that's the right term) of stuff into the 100 kHz window.  
(I'd figure out what is what by changing the LO and seeing how things
moved.. not the best.)  

George H.  
Quoted text here. Click to load it


Re: LF Spectrum Analyser
On Wed, 28 Jun 2017 07:49:56 -0700 (PDT), George Herold

Quoted text here. Click to load it

The reason that I was asking is that while in traditional ADCs with
S&H can easily handle _bandlimited_ high frequency signal by
decimation, the only critical parameter is  _transition_ time between
sample and hold state. Thus, if the transition time is  about 1 ns,
quite high frequencies can be sampled without too much aperture error
from say a band limited 50 MHz IF signal.

But how about delta/sigma converters, can they handle  bandlimited
high frequency signals and properly decimate or is a S&H stage needed
in between ?

Quoted text here. Click to load it

Using some common intermediate frequencies, such as 455 kHz, 10.7 MHz
or 45 MHz and there is a huge number of commercial ceramic or crystal
filters with a few kHz to at least 300 kHz bandwidth available. These
filters will help getting away with most of the aliasis.

For a more modern approach, take a look at www.websdr.org and try out
some of these receiving station. These typically digitize a whole
amateur radio band, producing a spectrum and waterfall display.  

In addition, several clients over the internet can connect to this
station, independently tune around the digitized band, perform
demodulation (CW, SSB, AM or FM) and transfer the audio over the
internet to individual clients. Of course SSB "demodulation" is just a
frequency transfer, so the end user could do any analysis to that
"audio" bandwidth.

Most of those WebSDR stations are simple PCs with a stereo 192 kHz
sampling rate sound card. Some use I/Q mixers driving the left and
right audio channel. The local oscillator can be a sin/cos DDS or
simply a crystal with 90 phase shifter.

All the rest is done in PC software.

Try one of those.

I still do not understand how somebody can sell 100 kHz SAs (SRS770)
starting from USD5000 :-)

  


Re: LF Spectrum Analyser
On Thursday, June 29, 2017 at 6:19:42 AM UTC-4, snipped-for-privacy@downunder.com wrote:
Quoted text here. Click to load it
OK Thanks... I don't know anything about SDR.  (except in theory)
I would mostly want a box that I could turn knobs and get a spectrum.
Quoted text here. Click to load it

Well that's their old (1980's) version.  It's got a decent low noise  
front end.  They make a newer one,
http://www.thinksrs.com/products/SR1.htm
~$10k.  Audio market. :^)  

George H.  

Re: LF Spectrum Analyser
On 06/29/2017 06:19 AM, snipped-for-privacy@downunder.com wrote:
Quoted text here. Click to load it

I have an HP 35665A that I love.  I paid $350 for it about 8 or 9 years
ago, and it works flawlessly.

They're very good for tuning analogue control loops, for instance, or
for measuring the transfer function of some slow plant such as a
TEC/thermistor or voice coil/LVDT.

It also has swept sine and Gaussian white noise sources and a bunch of
other useful functions.  Plus it runs good old Rocky Mountain Basic, so
you can use it to control other instruments. (Shades of grad school.)

I got the All Options ROM from GLK Instruments for $50, which unlocks
all the built-in goodies.  They have to customize it for your unit's
serial number.

Cheers

Phil Hobbs

--  
Dr Philip C D Hobbs
Principal Consultant
We've slightly trimmed the long signature. Click to see the full one.
Re: LF Spectrum Analyser
On 29/06/2017 15:03, Phil Hobbs wrote:
Quoted text here. Click to load it

I'd be loathed to pay anyone $50 for 30 seconds of their time  
programming a 50c part.

I upgraded mine thanks to the instructions shared by George.

Hi all,
This topic about the eeprom in the HP 35665A seems to be quiet for  
sometime. I
thought that I would share my recent findings about the 35665A options.  
A few
months ago, I purchased a like new 35665A with all documentation except the
service manual. It had only one option ANA. Unfortunately, I have not  
been able
to locate the service manual or options disks and there is no support from
Agilent. After some detective work and from the post #21350 and others,  
I was
able to active all of the options.

The bits to activate options 1D4, 1D3, 1D2, 1D1, 1D0, 1C2, and 1C1 are  
stored at
hex address locations 23h, 24h, 25h, 26h, 27h, 28h, 29h, respectively.  
To active
all options or any option, fill the address location(s) with 76h. Here  
is the
eeprom hex dump from my instrument:

0000: 00 13 20 5D 04 05 06 00 08 09 48 50 33 35 36 36 .. ]......HP3566
0010: 35 41 20 20 33 34 34 35 41 30 32 36 36 36 1E 1F 5A 3445A02666..
0020: 20 21 00 76 76 76 76 76 76 19 2A 2B 2C 2D 2E 2F !.vvvvvv.*+,-./
0030: 48 45 57 4C 45 54 54 2D 50 41 43 4B 41 52 44 20 HEWLETT-PACKARD
0040: 33 35 36 36 35 41 20 44 59 4E 41 4D 49 43 20 20 35665A DYNAMIC
0050: 53 49 47 4E 41 4C 20 41 4E 41 4C 59 5A 45 52 20 SIGNAL ANALYZER
0060: 60 61 62 63 64 65 66 67 68 69 6A 6B 6C 6D 6E 6F `abcdefghijklmno
0070: 70 71 72 73 74 75 41 77 78 79 7A 7B 7C 4D 48 99 pqrstuAwxyz{|MH.

I will also upload the binary file. With a hex editor, you can edit the  
binary
file and put in your instrument's serial number in place of 3445A02666  
although
this is not necessary. Remember that options 1C2 and ANA are mutually  
exclusive.

Also, the 85C72 eeproms are obsolete (and expensive if you order from IC  
dealer
who specializes in obselete ICs). Use Microchip's 24C01 eeprom - they  
are $0.29
from Mouser. They are equivalent and work fine in the 35665A. Also, you  
might
want to replace the lithium battery (B200 on the main board) while you  
have the
instrument open.

Regards, George

Re: LF Spectrum Analyser
On 06/29/2017 12:37 PM, JM wrote:
Quoted text here. Click to load it

You're just cheap, is all.  $50 to a public benefactor like that is
money very sell spent.

Thanks for sharing the instructions.

Cheers

Phil Hobbs
Quoted text here. Click to load it


--  
Dr Philip C D Hobbs
Principal Consultant
We've slightly trimmed the long signature. Click to see the full one.
Re: LF Spectrum Analyser
On 29/06/17 20:19, snipped-for-privacy@downunder.com wrote:
Quoted text here. Click to load it

Very few, in fact. There's a whole range of types, your review
was very limited.

Quoted text here. Click to load it

Mostly it's dynamic range, and image rejection.

Zero-IF converters get a lot of LO2 feedthrough.

Low-IF still has a number of image problems unless you have
very good front-end filters (and few SDRs do).

8-bit converters (like in RTL-SDR) might be adequate to
receive digital TV, but nearly useless as an SA.

Really wide-band front-ends have massive noise problems
that makes wide dynamic range a near impossibility.

In short, the better hobbyist-grade SDRs are good to
perhaps 40dB, whereas pro equipment goes beyond 70dB.

The hobbyist community seems blissfully unaware of many of
these problems, because they don't own and perhaps have never
even used pro-quality instruments, and don't understand why
they are so complicated, with cavity resonator filters and
banks of switchable or electronically-tuned filters.

Clifford Heath.

Re: LF Spectrum Analyser
On Fri, 30 Jun 2017 17:09:50 +1000, Clifford Heath

Quoted text here. Click to load it

Listen to http://websdr.ewi.utwente.nl:8901/
with 16 bit sampling of 0 to 29 MHz.  

Use "View Spectrum" to get the traditional spectrum display instead of
waterfall display. Left click on the frequency scale to change
frequency.

Unfortunately  the receiver is on a campus with high background noise
and of course there is a lot of band noise on HF and especially on
MF/LF, so the true potential of the ADC is hard to determine. The
strongest signal seems to be a nearby pager at 27 MHz at about 0 dBm.  

At upper HF the background noise is about -100 dBm in 3 kHz bandwidth.
To me, this works surprisingly well for its intended purpose, but of
course, this is not a calibrated test instrument.

Quoted text here. Click to load it

The problem is that in past decades quite complex structures were
needed for spectrum analyzers and general coverage receivers (such as
multiple PLLs). Today, when good components are available, such as
NCOs, DACs and ADCs the same performance can be achieved with less
complexity and hence less cost.

Quoted text here. Click to load it


Re: LF Spectrum Analyser
On 02/07/17 07:14, snipped-for-privacy@downunder.com wrote:
Quoted text here. Click to load it

Not the same performance. Adequate for many tasks, but no way the same.
The best wide-band SDRs have big banks of RF filters, and they're
still not as good as a proper communications receiver. Good enough
to make pretty waterfall displays, but no way the same.

Re: LF Spectrum Analyser
On Sunday, July 2, 2017 at 1:08:15 AM UTC+2, Clifford Heath wrote:
Quoted text here. Click to load it

For low frequency high res I have a National Instrument DAQ 24 bit aquisiti
on unit:

http://www.ni.com/datasheet/pdf/en/ds-324

And a cheap SW I got for 100 USD that does all sorts of tricks

I can dig up signals buried 100dB below a carrier

I actually have one unit for sale for highest bidder ?

Cheers

Klaus

Re: LF Spectrum Analyser
wrote:

Quoted text here. Click to load it

The weakest point in both SA as well as general coverage receivers and
scanners is the mixer.The situation can be somewhat helped by using
special mixer structures and/or using a lot of power (not possible in
a portable device).

In addition, if you are driving a mixer with a square wave, there will
be mixing products between the odd harmonics of the LO and input
spectrum at the IF distance of the odd harmonics of the LO. For this
reason, octave filters are used to keep frequencies close to the 3rd
harmonics (and higher) out of the mixer.  

For a SA with 1 kHz to 3 GHz range, you will need more than 20 octave
filters to cover the range. The good thing is that you can attenuate
each octave separately and fix the corresponding attenuation in the
display. In fact suboctave front end filters might be helpful to
filter out second order mixing products of two strong input signals at
the lower end  of the octave, which would fall  on the upper end of
the octave.    

However if the upper frequency requirement is more modest (say
HF/VHF), you could get away with the problematic mixer and run the
signal directly into the ADC.

It should be noted that a 16 bit ADC the theoretical weak signal
resolution is 97 dB, there are always going to be band and front end
noise, which work as dither, making it possible to resolve signals
well below 100 dB FSD.

While the problematic mixing has been used for decades, the
performance of ADCs will constantly  increase, moving the practical
dividing line to higher and higher frequencies.


Re: LF Spectrum Analyser
On 02/07/17 16:46, snipped-for-privacy@downunder.com wrote:
Quoted text here. Click to load it

Mixers are non-linear devices. Even a perfect mixer will still have the
same image problems that plague real receivers.

Quoted text here. Click to load it

Zero-IF receivers are also afflicted by 2LO, not just 3LO,
because the frequencies "fold around" zero.

There's quite a good review of available devices in the slides here
(though these still pass over some of the problems):
<http://microhams.blob.core.windows.net/content/2017/03/RTL-SDR-dongle.pdf

Clifford Heath.

Re: LF Spectrum Analyser
On 07/03/2017 09:15 PM, Clifford Heath wrote:
Quoted text here. Click to load it
              ^^^^^^^^^^

A small quibble, because this bit of Bad Info keeps popping up. A  
perfect mixer is a bilinear device: the IF output is linear in both the  
LO and RF signals.  If you put some DC in the LO port, the RF->IF path  
satisfies all the requirements for a linear, time invariant network.

 > Even a perfect mixer will still have the
 > same image problems that plague real receivers.

As you say, the image exists even with a perfectly linear multiplier,  
because of the identity

2 sin A sin B = cos(a-b)-cos(a+b).

Cheers

Phil Hobbs

--  
Dr Philip C D Hobbs
Principal Consultant
We've slightly trimmed the long signature. Click to see the full one.
Re: LF Spectrum Analyser
On 05/07/17 03:00, Phil Hobbs wrote:
Quoted text here. Click to load it

I don't think a physical realisation exists at VHF or higher?
Maybe for *very* small signals.

It's not really relevant what's possible under 1MHz, LM13700 etc.

Clifford Heath.

Re: LF Spectrum Analyser
On Tuesday, July 4, 2017 at 8:55:38 PM UTC-7, Clifford Heath wrote:
Quoted text here. Click to load it




It's not challenging, but for those of us who chose our modulations,
it's interesting.   It's as relevant as we choose to make it.

Site Timeline