Buying 0~10MHz Spectrum analyzer, solutions and advice

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Jump to Latest
Frex

Frex

Member
Joined 2009
Hello,

For all my audio measurements, i use my external ADC unit. Like a soundcard, it's an audio ADC chip, supporting sampling rate only up to 192kHz.
With a good software tool (like Spectraplus or others), make accurate measurements of many parameters is very easy.

But, unfortunately, as soon frequency is above some tens of kHz, the Nyquist limit is reached and sometime a part of the signal is not displayed...
(for example all the out of band noise of oversampling DAC).
Sometime, it can be very useful to know the spectrum shape above the 100kHz limit of soundcard.

Therefore, i look since many weeks what would be the better choice to satisfy my need.
First, i list below my preliminary need for this spectrum analyzer ;

0 to about 10MHz bandwidth (or more)
High sensitivity (µV range)
User friendly interface
Small from factor (i don't have much space !)


With just these parameters, i have selected few devices after many long time using our preferred web search engine,
and the result is listed below.

SignalHoud USB SA44 (750€)
RIGOL DSA1020 (3000€)
Picotech Picoscope 4224 (750€)
Handyscope HS3 (€)
Cleverscope CS320 (€ )
Bitscope BS325 (560€+shipping)
Agilent HP8568B 100Hz-1.5GHz Spectrum analyser USED (1500€+shipping)

Unless otherwise noted, all prices include VAT and shipping to my country (France).
(For more info about each device, you can click to go on it's main web page.)





SignalHoud USB SA44 (800€)
The SA44 is a new device, i'ts a small box connected to a PC, with a specific software to control it.
It can perform spectral analysis in the range of 1Hz to 4GHz.

It seem to be a very interesting device, above all with that price.
Unfortunately, we have no background for tis device and i find only official info about this unit.
Many real example of use of this unit is missing on the mfg web site, and it's hard to really know if the SA44 work well in the bandwidth of 0-10MHz.

pro:
wide bandwidth.
reasonable price.
small.

cons:
no background of how it work and it LF performance.
50 Ohms inputs not convenient for LF use.
computer needed.

RIGOL DSA1020 (3000€)
The Rigol DSA1020 is also a new product only available since some months.
Unlike the SA44, it's a real stand-alone tool, similar to an digital Oscilloscope.
The DSA1020 is a 9kHz to 2GHz spectrum analyzer, with many advanced functions. It's also the more costly solution...

pro:
performance
stand-alone , easy to use.
high dynamic range
very versatile
Large color screen,USB, Ethernet

cons:
Frequency start to 9kHz
Resolution bandwidth limited to 100Hz.
50 Ohms inputs not convenient for LF use.
Price


Picotech Picoscope 4224 (750€)

The picoscope is not mainly a spectrum analyzer, but the 4224 is a 12 bits 40Ms/s USB DSO. The software supplied with the 4224 is very powerful and allow to display spectrum (using FFT) between 0 to Fs/2. The 4224 offer all advanced 2 channels DSO functions. An averaging mode allow to push the ENOB from 12 to up to 16 bits, improving significantly the noise floor of the spectral analysis.
Demo software is downloadable on picotech web site.

pro:
Good performance
DSO channels functions 80Ms/s.
1Mpts FFT
versatile
reasonable price.
1M input impedance

cons:
Not a real spectrum analyzer
computer needed.


Handyscope HS3-25 (900€)

Like the Picoscope, the Handyscope HS3 is a USB DSO.It also allow spectrum analysis using FFT.
Price depend on the sampling rate you choose, but for 10MHz bandwidth, the HS3-25 (25Ms/s) is sufficient.
It's a 12bits ADC, but effective resolution depend on the sampling rate.
The biggest issue with this device, is the software...(only my opinion).
Demo software can be downloaded on the mfg web site.

pro:
Several auto measurement
Small
Reasonnable cost.

cons:
Very poor software interface (only my own opinion)
Resolution is dependent of sampling rate (8bits at 100Ms/s)
FFT only 32kpts


Cleverscope CS320 (1500€)
Another USB DSO, with FFT spectrum analysis.
High sampling rate, 12 bits.

pro:
Poor software

cons:
computer needed.


BitScope BS325 (€)

It seem to be a great product with good support and evolute software.
The demo software is convincing, easy to use.
It include a signal generator, 2 channels DSO, spectrum analyzer, logic analyzer
(8 digital inputs) and more.
It's a 40Ms/s ADC with
The interface can be USB or Ethernet.

pro:
very good software
generator and logic analyzer
signal generator, digital inputs
small

cons:
Not a real SA
low resolution (8 bits)
computer needed.

Agilent HP8568B 100Hz-1.5GHz Spectrum analyzer USED (1500€+shipping)

My first way to search a spectrum analyzer had to buy a HP8568, because it seems to be a logic choice.
Unfortunately, the selling price of this older unit is high (1500€ or more aften without any warranty!),
and it's very high weight (75kg i think) to make shipping extremely costly (often more than the price of the unit !).
The all these reason, i prefer to buy other thing. :)





It would be interesting if anybody have an experience with these units.
At this time, the DSA1020 have my favor, but i'm not yet decided i think that i'm probably not the only one which look for
this type of instrument.
Thank you.


Frex.
 
J

jcx

Member
Joined 2003
packaged products are nice but lag behind available monolithic ADC chips, 18 bit @ 2MHz, 16 bit @ >80 MHz are available with ~US$200-400 demo boards

may need fpga or dsp companion hardware to collect/buffer at full speed so system cost will be higher
 
Last edited:
B

bocka

Member
Joined 2003
synertronixx.de
Hi Frex,

sorry to say but it's better to save your money than buying anything of this crap. A spectrum analyzer ist still very costly. Although the Rigol looks still decent on paper (but this is a RF spectrum analyzer as well as the used HP item and the SignalHoud) you have to check out it on your lab. You'll see that any of them have some inacceptable weaknesses. A spec is still a $$$$$ part.

I wouldn't buy a used spec until it comes directly from Agilent, R&S, Tek or some other first class Mfr. If a used one fails you can throw it away in most cases.
 
Pano

Pano

Administrator
Joined 2004
Paid Member
I've used the picoscope (several years ago) and it's not bad. I know that a lot of folks don't find it "serious" enough. But it did what it was advertised to do and was easy to use. I found the spectrum analysis function very useful.
 
Frex

Frex

Member
Joined 2009
Hello,

Jcx,
do you know and had you try on of this EVM ?
I know some, but the provided software don't allow many option, and is it slower...
(i have try the evm software of TI and AD).


bocka,

I understand that 1000€~3000€ device can't be so good as a Tektronix or Agilent devices ten times more costly. Nevertheless, it can be also a very good compromise.
Like my DS1022CD Rigol DSO, 5x lower priced than a Tektronix DSO and fitting very well to my hobbyist need !

panomaniac,
Thank you for your opinion, but did you have yet a Picoscope ?



Frex
 
B

bocka

Member
Joined 2003
synertronixx.de
Frex said:
Hello,
I understand that 1000€~3000€ device can't be so good as a Tektronix or
Click to expand...
I ordered a Hameg HMS3000 some month ago as the price (about EUR 3,000) seems very promising for 3GHz spectrum analyzer. But when tested that part what a disappointment. I thought this was a bargin for the money but that part was that faulty, noisy, slow and so on that only after 2 days I sent the part back. Better to buy a used R&S (or similar) for about EUR 6,000 or just forget it. Not worth the money.
 
Last edited:
C

ChrisA

Member
Joined 2008
Frex said:
...
But, unfortunately, as soon frequency is above some tens of kHz, the Nyquist limit is reached and sometime a part of the signal is not displayed...
(for example all the out of band noise of oversampling DAC).
Sometime, it can be very useful to know the spectrum shape above the 100kHz limit of soundcard.
Click to expand...


You CAN see any frequency you want using only the typical 192K sample per second audio interface. The trick is to "heterodyne" the signal down to the audio range. Using this you can measure signals well up into the MHz range but you are limited to the same 100Khz bandwidth by Nyquist. But by tuning the frequency of a local oscillator you get to select which 100K bit of bandwidth you can see at one time. Basically you place a primitive kind of AM radio receiver in front of the audio interface.

Actually you can see up to 192Khz of bandwidth at a time by taking advantage of the fact that your audio interface has two channels and quadrature sampling but that is not so easy and requires different software
 
1

1audio

Member
Joined 2004
Paid Member
I have used the Picoscope ADC-200 a lot. It makes a pretty decent spectrum analyzer if you understand its limitations and the nature of what you are looking at. Their implementation and software was the best available when I got it. I tried all of them and returned all the others. The dynamic range is the biggest limitation compared to a self contained spectrum analyzer.
I also have a Tek 7L5 and a Tek 7L12, both of which are useful as well. High performance analyzers get very expensive fast and offer many features you may never use. For audio the extended range is useful for finding radiation and noise that you would not be aware of otherwise. You could also measure phase noise BUT if you can measure it with a general purpose spectrum analyzer it's way too high.
 
Frex

Frex

Member
Joined 2009
completed research

Hello,


Finally, and after many time to search, i have find and bought a used spectrum analyzer HP8568B.
I have buy it at a very good price, and it is fully working.
I have buy it at low price because the display was very faint.
Fortunatly, i was find in internet an easy way to restore old CRT.
After that, it display now a very bright spectrum.
:)) I'm very happy with it !

The 8568B have a 100Hz to 1.5GHZ bandwidth, and allow very narrow resolution to 10Hz.
It's a very convenient tool. I need time now to learn about it's use and to build some tools for it.
Thank to all DIYers who had gave their opinion.

Frex.
 
Frex

Frex

Member
Joined 2009
DIY spectrum analyser.

Hello,



Despite this buy, and because i'm very interested by that, i will start to build a DIY spectrum analyzer.
It already exist some on the internet , but only for RF use (starting at tens MHz) and using old components.

My objective is to made an simple device, with a bandwidth from some Hz to about 10~20MHz and with a high dynamic range.
For started, i wand use a AD831 as input mixer, a AD8307 as log detector and a DDS IC for the sweep generator.
As you have understand, it's a heterodyne SA (spectrum analyser), but to simplify the design it use a direct convertion (no
bandpass filter at mixer output). Only an active low noise, sharp rolloff analog lowpass filter will be used.

I wait now parts to starts this new fun project.
Of course i will post as the progress.

Frex.
 
1

1audio

Member
Joined 2004
Paid Member
Its an interesting project but I think you will discover that its not so simple. For direct conversion you need a tracking RF filter or the aliases will realy confuse what you are looking at. The filtering needs to be in front of the first mixer in direct conversion, or distributed in a double or triple conversion scheme. There are schemes with high speed ADC's but they still need filters in front. Typical design has a first conversion to an IF above the usable range, a fixed bandpass filter and second conversion to the variable IF (or to baseband). Going to 30 MHz makes this more of a problem. Usually you start with the math and look at the sum and difference products to figure out what you need.

There are plenty of downloadable manuals for HP, Tek and other spectrum/wave analyzers that you can use for guidance.

There is a project using a soundcard that does most of this on the web. I don't have a link but I think its related to "software defined radio" or sdr.

Good luck. I may want to build one once you have made progress.
 
Frex

Frex

Member
Joined 2009
Hello 1audio ,

I don't need a tracking filter. I use only a passive sharp rolloff low-pass filter at input at Fmax. At the mixer output, i use a passive or active LPF at very low cutoff frequency.
If i want a resolution bandwidth of 100Hz, i must use a 50Hz LPF(because frequency symetrie). Images frequency is always far from wanted signal.
I have now start prototyping on copper board. :)


Frex.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Top Bottom