EEVblog Electronics Community Forum

Products => Test Equipment => Topic started by: _Wim_ on October 21, 2017, 09:38:43 am

Title: Looking for a dynamic signal analyzer with extended bandwidth
Post by: _Wim_ on October 21, 2017, 09:38:43 am
I am looking for a “dynamic signal analyzer” with extended bandwidth (“DC” to approx. 10MHz). This will be used during experimentation with analog front ends for optical sensors (mainly interested in noise and FFT of the modulated light).  Signals of interest are typically small (1mV <>100mV), but sometimes also riding on “large” DC (+-15V). I would like to have input ranges that go down to around -40dBVrms (10mV), and still have a good noise floor (-80dbFs preferred). Input impedance 1M to prevent loading of the DUT.  Budget is around 1500€ but stretchable . I prefer more modern instruments, but off course considering my budget, I will need to make compromises somewhere...

This is my list of possible options I found so far:

1) Dynamic signal analyzer + spectrum analyzer
Most (affordable) spectrum analyzers start only somewhere around 9 KHz.  I could use one of these combined with an older dynamic signal analyzer like an HP 35665A, but these are huge, quite old and go still for quite some money. Also it is annoying to have to use 2 unit for the range I am interested in. Most spectrum analyzers will need some kind of active probe for the higher input impedance required (could be something homebrew because it only needs to go up to 10MHz). Not the most practical solution. I already have a CMU200, so the additional range offered by the spectrum analyzer is of no real interest.

2) USB  Spectrum Analyzer:
 Signal Hound sa44b (https://signalhound.com/download/usb-sa44b-datasheet) covers” the range I am interested in as it goes from 1 Hz to 4.4GHz, I am however unsure if no compromises were are made at the low end to go up this high. The datasheet does not specify a lot. Needs the (homebrew) active probe and a DC block. Unsure about the performance for my goals.

3) High Resolution Scope
An alternative could be the  picoscope 4262  (https://www.picotech.com/oscilloscope/4262/picoscope-4262-specifications). This one is a little low on bandwidth, but so far seems like a reasonable solution.  It does seem quite expensive for what it only offers.  If bandwidth would be a little higher, this was a winner. The picoscope 5000 series also claim 16 bit of resolution, but their noise specs are much worse.

4) HP 89410-89441A   (http://www.testequipmenthq.com/datasheets/Agilent-89441A-Datasheet.pdf)(with or without RF-downconverter)
This seems like a good solution for what I need. Disadvantages are they are very big and heavy, which makes importing them from the US  to Belgium very expensive . In Europe they do not seem to be common, and if they appear on Ebay, they are way overpriced.  They are typically quite old (1994-98), and there seems to be a common problem with the corners of the screen getting quite dim (there is a VGA port available however). Temped by this solution also as it also offers network analyzer capabilities if the second channel is installed.

Any advice, tips, other units to consider?

Edit: added hyperlinks to specification datasheets
Title: Re: Looking for a dynamic signal analyzer with extended bandwidth
Post by: _Wim_ on October 21, 2017, 09:56:33 am
Maybe also a quick question. Many spectrum analyzer specify DANL in dBm/Hz. If the spectrum analyzer has a RBW of 0.1Hz, can you subtract 20dbm additional from the specified specs (eg, in the case of the signalhoud, DANL at 10Hz is specified as -124dBm/hz, but as this unit has a possible RBW of 0.1 Hz, noise at 10Hz would be -144dBm, correct?)

Title: Re: Looking for a dynamic signal analyzer with extended bandwidth
Post by: nctnico on October 21, 2017, 10:23:05 am
Maybe you can get an Anritsu MS4630B for that budget. That is a network analyser from 10Hz to 300MHz and it does have 1M Ohm inputs.

Another option would be the GW Instek MDO2000E series:
http://www.gwinstek.com/en-global/products/Oscilloscopes/Digital_Storage_Oscilloscopes/MDO-2000E_Series (http://www.gwinstek.com/en-global/products/Oscilloscopes/Digital_Storage_Oscilloscopes/MDO-2000E_Series)
If the sensitivity is too low you could use an external amplifier. The advantage of using an FFT based oscilloscope is that you get the spectrum from one acquisition instead of a scan during a period of time. The MDO-2000E also has input filtering you can play a bit with the effects of filtering on your signal. I have used my GW Instek GDS2204E (which only has standard FFT) for a similar purpose; look at signals coming from a sensor to see what kind of noise is there and what kind of filtering/signal conditioning works well. I kinda bought it specifically for that project.
Title: Re: Looking for a dynamic signal analyzer with extended bandwidth
Post by: _Wim_ on October 21, 2017, 10:31:59 am
Thanks for your reply. The Anritsu does look like an interesting device. Do you want to sell yours? I willing to pay the double of what you paid  ;D

The MDO2000E still remains an 8-bit device. As I am interested in looking at my signal together with the noise at the same time,so I need a bigger dynamic range then the 8-bit can offer, and pre-amp does not help for this.
Title: Re: Looking for a dynamic signal analyzer with extended bandwidth
Post by: nctnico on October 21, 2017, 10:54:31 am
I think I want to hang on to my MS4630B for a bit longer.
Another option might be the 'analog discovery 2'. It ticks all your boxes but I don't know if the software works for your purpose.
Title: Re: Looking for a dynamic signal analyzer with extended bandwidth
Post by: _Wim_ on October 21, 2017, 11:30:25 am
Another option might be the 'analog discovery 2'. It ticks all your boxes but I don't know if the software works for your purpose.

This looks indeed like a possible candidate, although a bit toy-like. It will need a low noise pre-amp for sure, as the low input ranges use "digital zoom", resulting in a max resolution of 0.32mV on the input range of 0.5mV/div  :palm:, which is off course way too low for my needs. But still, for the price, it seems quite a versatile device.

No clear specs for the noise performance of the entire system however, and if I see the specs of the the ADC and buffers, I think it will not be so fantastic. But maybe worth a play around with...
Title: Re: Looking for a dynamic signal analyzer with extended bandwidth
Post by: _Wim_ on October 21, 2017, 11:50:34 am
For future reference for others: here is the manual for the  Anritsu MS4630B (https://isurplus.com.au/manuals/Anritsu%20MS4630B%20User%20Manual.pdf) (specs in section 1.5)

@nctnico, do I reed this correct that this unit has only 2 input ranges,one of 0dBm and one of 20dBm? For measuring small signals a pre-amp would still be necesarry (not really an issue, but worth mentioning).
Title: Re: Looking for a dynamic signal analyzer with extended bandwidth
Post by: David Hess on October 21, 2017, 03:45:03 pm
Something from Cleverscope (https://cleverscope.com/) might be suitable.
Title: Re: Looking for a dynamic signal analyzer with extended bandwidth
Post by: egonotto on October 21, 2017, 06:52:32 pm
Hello,

possibly a picoscope 5000 can do it.

The pictures are with 20MHz limit.
The first 3 are with input from the signalgenerator with 1MHz sine and an attenuator.
The last 2 are with an 50 Ohm termination to show the noise level.

If you have questions feel free to ask (too for picoscope 4262 and analog discovery).


Best regards
egonotto


Title: Re: Looking for a dynamic signal analyzer with extended bandwidth
Post by: _Wim_ on October 21, 2017, 07:28:15 pm
Hello,

possibly a picoscope 5000 can do it.

The pictures are with 20MHz limit.
The first 3 are with input from the signalgenerator with 1MHz sine and an attenuator.
The last 2 are with an 50 Ohm termination to show the noise level.

If you have questions feel free to ask (too for picoscope 4262 and analog discovery).


Best regards
egonotto

Thanks for the screenshot. The 5000 series indeed perform a bit better than I expected. They still need a preamp however. If you look at the plot with the 50 ohm termination, there is still around 40µV of noise and some DC offset (not a real problem if it is stable). The noise is better then the spec (100µV for 12 Bit, 70µV for 16 bit) , but when measuring signals of a couple of mV, this is still too high. The 4262 spec is around 8µV for 16 bit, which means it is at least 10 times better noise wise (all other specs are however worse however).

Title: Re: Looking for a dynamic signal analyzer with extended bandwidth
Post by: _Wim_ on October 21, 2017, 07:29:19 pm
Something from Cleverscope (https://cleverscope.com/) might be suitable.

Thanks, I will have a good look at their specs. Haven't come across one of these before.

Edit: just checked, no real noise specifications for these. Typiccally that means they are not fenomenal. Do you use one of these?
Title: Re: Looking for a dynamic signal analyzer with extended bandwidth
Post by: egonotto on October 21, 2017, 08:14:55 pm
Hello,

as the picoscope 5000 have 500MS/s (12bit) you can use more bit.

With 14 bit you have 31.25 MS/s if i understand it right (one channel).
Than you get lower noise in time view.
The noise in 10mV is not much lower than the noise in 50mV so the pictures are with 50mV.

You can zero the DC offset. I measure the drift and post the result later.

Yes the 4246 is better and the DC-drift is better.

Best regards
egonotto



Title: Re: Looking for a dynamic signal analyzer with extended bandwidth
Post by: _Wim_ on October 21, 2017, 08:33:13 pm
Hello,

as the picoscope 5000 have 500MS/s (12bit) you can use more bit.

With 14 bit you have 31.25 MS/s if i understand it right (one channel).
Than you get lower noise in time view.
The noise in 10mV is not much lower than the noise in 50mV so the pictures are with 50mV.

You can zero the DC offset. I measure the drift and post the result later.

Yes the 4246 is better and the DC-drift is better.

Best regards
egonotto

Thanks for these last pictures, very informative. The 5000 series perform way better then spec! 12µV RMS noise is indeed comparable to the 4262. I do not understand were the 70µV rms comes from in the spec for the 5000 series?


Edit: removed my confusion
Title: Re: Looking for a dynamic signal analyzer with extended bandwidth
Post by: hagster on October 21, 2017, 08:58:04 pm
How about an Airspy HF+?

http://airspy.com/airspy-hf-plus/ (http://airspy.com/airspy-hf-plus/)

Should be availiable soon and has 18bit dynamic range and DC to 30mhz coverage.


https://www.rtl-sdr.com/airspy-hf-upcoming-low-cost-yet-high-performance-hf-sdr/ (https://www.rtl-sdr.com/airspy-hf-upcoming-low-cost-yet-high-performance-hf-sdr/)
Title: Re: Looking for a dynamic signal analyzer with extended bandwidth
Post by: egonotto on October 21, 2017, 09:14:03 pm
Hello,

it is the min and max value for the 10 traces.

I don't know how picoscope 6 make the measure exakt.
Use they all samples or only a subset?
To update the screen fast, they take only a subset of the samples to show the trace.

I think my 4262 is in spec.

In the datasheet they say it is 8.5uV RMS in 10mV range.
My 4262 has in 10mV range near 8.5mV RMS noise.
And the noise spec from Pico Technology are only typical value.

When I remember right, say write me, that they tolerate 20%.

This measure is with a 50Ohm termination with a metal cap.
Without termination the noise is far bigger.

The peak-peak measures on my picoscopes are about 10 times greater than the RMS measures (most people say that peak-peak values are about 6 times greater than the RMS value)

The DC drift on the 4234A is about 100uV and on the 4262 about 25uV after 40 minutes

I can save a trace and than we can look was is real.


Best regards
egonotto

Title: Re: Looking for a dynamic signal analyzer with extended bandwidth
Post by: _Wim_ on October 21, 2017, 09:17:43 pm
How about an Airspy HF+?

http://airspy.com/airspy-hf-plus/ (http://airspy.com/airspy-hf-plus/)

Should be availiable soon and has 18bit dynamic range and DC to 30mhz coverage.


https://www.rtl-sdr.com/airspy-hf-upcoming-low-cost-yet-high-performance-hf-sdr/ (https://www.rtl-sdr.com/airspy-hf-upcoming-low-cost-yet-high-performance-hf-sdr/)

I find it hard to correlate these specs to what I want to do. The 18bit refers to the DDC (digital down converter), but I am unsure how this relates to the dynamic range. I have also no idea how to interpreted the minimum detectable signal specs (MDS). I can image this device being able to tune to a very narrow band, and by doing that, achieve a very low noise. But I do not see any specs about scan speed of the spectrum, so I am not sure if you can use these as a normal spectrum analyzer. I need to read a bit how these SDR devices work...
Title: Re: Looking for a dynamic signal analyzer with extended bandwidth
Post by: _Wim_ on October 21, 2017, 09:22:40 pm
Thanks for this, I already figured it out and updated my post. The 5000 series start to look nicer and nicer. I used a picoscope before (parallel port device, ADC-212 if I remember correctly) and really liked it. Software was very intuitive and worked flawlessly. I was just a bit thrown off by their 70µV RMS noise spec for the 16 bit mode.

Title: Re: Looking for a dynamic signal analyzer with extended bandwidth
Post by: David Hess on October 21, 2017, 09:47:10 pm
Something from Cleverscope (https://cleverscope.com/) might be suitable.

Thanks, I will have a good look at their specs. Haven't come across one of these before.

Edit: just checked, no real noise specifications for these.  Typically that means they are not phenomenal. Do you use one of these?

Cleverscope was brought to my attention a couple months ago when I expressed a similar interest in low frequency network analysis but I do not have one and I have not directly evaluated one yet.

As far as noise, oscilloscope 1 megohm inputs are all pretty poor in this regard compared to low impedance and audio inputs and they also suffer from high flicker noise; this is just because they use high frequency JFET or MOSFET followers to allow a high impedance input.

So 80dB below 10mVrms over a 10 MHz bandwidth is not going to happen; 40dB would be more realistic.  However if you are using a long FFT to measure noise over a 1 Hz bandwidth, better than 80dB under the same conditions should be easy to achieve.  This is the standard "noise marker" measurement made by good spectrum analyzers; the problem is finding an oscilloscope which supports it.

I was checking the Cleverscope documentation after I wrote the above and they give lots of design and performance details including noise performance here:

https://www.cleverscope.com/files/Cleverscope%20CS328A%20performance.pdf (https://www.cleverscope.com/files/Cleverscope%20CS328A%20performance.pdf)

However I was not able to figure out from their documentation if they can produce an FFT with calibrated equivalent noise bandwidth.  I did not see anything in the Picotech documentation about this ether.
Title: Re: Looking for a dynamic signal analyzer with extended bandwidth
Post by: egonotto on October 21, 2017, 09:47:33 pm
Hello

_Wim_ wrote:

"Thanks for these last pictures, very informative. The 5000 series perform way better then spec! 12µV RMS noise is indeed comparable to the 4262. I do not understand were the 70µV rms comes from in the spec for the 5000 series?"

Unfortunately my 5243A is worse in noise as the spec tells.
I had a forum with people from Pico Technology.
"https://www.picotech.com/support/post49411.html#p49411"
and with email.

But the spec tells the typical noise only without band width limiting and without average.
With band width limiting and with average you can get much lower noise.

In the picture (picoscope 4262) is the band width 200kHz and the most average (256 sample average (20bit)). Consider is only 40uV/div.


Best regards
egonotto



Title: Re: Looking for a dynamic signal analyzer with extended bandwidth
Post by: KE5FX on October 21, 2017, 09:54:14 pm
Thanks for this, I already figured it out and updated my post. The 5000 series start to look nicer and nicer. I used a picoscope before (parallel port device, ADC-212 if I remember correctly) and really liked it. Software was very intuitive and worked flawlessly. I was just a bit thrown off by their 70µV RMS noise spec for the 16 bit mode.

If you really need to go down to near DC, the white noise specification might not be enough for your needs.  Most ADCs have some excess 1/f noise below a kHz or so, and DC offset also comes into play since it can leak into the displayed spectrum if not removed by a (very laggy) HPF. 

A few parts, e.g. AD9253, are starting to offer a chopping feature to move the DC offset and 1/f noise to Nyquist, but I wouldn't expect this to be common in general-purpose scopes yet. 

Might not be a big deal, just something to keep in mind.
Title: Re: Looking for a dynamic signal analyzer with extended bandwidth
Post by: David Hess on October 21, 2017, 10:06:07 pm
If you really need to go down to near DC, the white noise specification might not be enough for your needs.  Most ADCs have some excess 1/f noise below a kHz or so, and DC offset also comes into play since it can leak into the displayed spectrum if not removed by a (very laggy) HPF. 

A few parts, e.g. AD9253, are starting to offer a chopping feature to move the DC offset and 1/f noise to Nyquist, but I wouldn't expect this to be common in general-purpose scopes yet. 

It is actually much worse than that.  A wide bandwidth (200 MHz) CMOS operational amplifier used for the high impedance input buffer might have a 5nV/sqrHz white noise specification but a 10 MHz corner frequency yielding 1000nV/sqrHz of noise at 10 Hz.  Discrete designs even from 40 years ago are *much* better than this.
Title: Re: Looking for a dynamic signal analyzer with extended bandwidth
Post by: precaud on October 21, 2017, 10:21:57 pm
If an MS4630B will work, you might also consider an Anritsu MS420K or B. 10Hz-30MHz VNA with spec an functionality. Older (90's-90's) unit but well built and quite serviceable. I like mine very much. They were the darling of the telecom industry prior to DSL, so there are quite few of them out there.
Title: Re: Looking for a dynamic signal analyzer with extended bandwidth
Post by: _Wim_ on October 22, 2017, 06:47:29 am
If an MS4630B will work, you might also consider an Anritsu MS420K or B. 10Hz-30MHz VNA with spec an functionality. Older (90's-90's) unit but well built and quite serviceable. I like mine very much. They were the darling of the telecom industry prior to DSL, so there are quite few of them out there.

I will have a look at these also. These older more purpose designed units for sure also have their advantages, but so has a small picoscope with many modern features and warranty.

2 years ago I repaired and HP4276A LCR meter (older unit), but since it has broken down twice. As I mainly experiment in my limited spare time, it can be frustrating having to repair my test gear first, hence I am a bit reluctant for the older gear.

I do like to buy broken gear on Ebay for cheap , and then repair them. But once repaired, they should keep working forever!  :) ?
Title: Re: Looking for a dynamic signal analyzer with extended bandwidth
Post by: _Wim_ on October 22, 2017, 06:52:16 am
If you really need to go down to near DC, the white noise specification might not be enough for your needs.  Most ADCs have some excess 1/f noise below a kHz or so, and DC offset also comes into play since it can leak into the displayed spectrum if not removed by a (very laggy) HPF. 

A few parts, e.g. AD9253, are starting to offer a chopping feature to move the DC offset and 1/f noise to Nyquist, but I wouldn't expect this to be common in general-purpose scopes yet. 

It is actually much worse than that.  A wide bandwidth (200 MHz) CMOS operational amplifier used for the high impedance input buffer might have a 5nV/sqrHz white noise specification but a 10 MHz corner frequency yielding 1000nV/sqrHz of noise at 10 Hz.  Discrete designs even from 40 years ago are *much* better than this.

Unfortunately, these kind of details are typically not published for the scopes (and other gear).

To answer the question of low frequencies, yes, I need them. Light sources used for testing are typically modulated with frequencies between 300Hz an 20Khz, so this range is very interesting for me. My initial thought was to just by a dynamic signal analyzer, but there just so limiting in higher frequencies. Still, noise wise, they are probably still by far the best for my application...
Title: Re: Looking for a dynamic signal analyzer with extended bandwidth
Post by: hagster on October 22, 2017, 06:55:55 am
I think you are right regarding the Airspy. It seems to work differently to many SDRs that I am used to. Most capture the entire bandwidth directly.

There are others such as the Perseus SDR that works as a 10khz to 40Mz real time spectrum analyser. You do have to add an external matching pad to increase the impedence of the 50ohm input, but thats not hard to build.

How about an Airspy HF+?

http://airspy.com/airspy-hf-plus/ (http://airspy.com/airspy-hf-plus/)

Should be availiable soon and has 18bit dynamic range and DC to 30mhz coverage.


https://www.rtl-sdr.com/airspy-hf-upcoming-low-cost-yet-high-performance-hf-sdr/ (https://www.rtl-sdr.com/airspy-hf-upcoming-low-cost-yet-high-performance-hf-sdr/)

I find it hard to correlate these specs to what I want to do. The 18bit refers to the DDC (digital down converter), but I am unsure how this relates to the dynamic range. I have also no idea how to interpreted the minimum detectable signal specs (MDS). I can image this device being able to tune to a very narrow band, and by doing that, achieve a very low noise. But I do not see any specs about scan speed of the spectrum, so I am not sure if you can use these as a normal spectrum analyzer. I need to read a bit how these SDR devices work...
Title: Re: Looking for a dynamic signal analyzer with extended bandwidth
Post by: nfmax on October 22, 2017, 07:48:11 am
Have you considered using a lock-in amplifier as an alternative? If your light source is modulated with a 'simple' waveform (e.g. Rather than a pulse compression coding) nothing else comes close in terms of signal recovery. But of course it is only a single frequency measurement, and requires a reference signal.
Title: Re: Looking for a dynamic signal analyzer with extended bandwidth
Post by: nctnico on October 22, 2017, 10:47:20 am
One note on the Pico scopes: you have to check very carefully if they use ADCs with the specified resolution or specificy resolution after doing oversampling. When using oversampling you may not get the extra bits if there isn't enough noise in the signal.
Title: Re: Looking for a dynamic signal analyzer with extended bandwidth
Post by: _Wim_ on October 22, 2017, 12:03:45 pm
Have you considered using a lock-in amplifier as an alternative? If your light source is modulated with a 'simple' waveform (e.g. Rather than a pulse compression coding) nothing else comes close in terms of signal recovery. But of course it is only a single frequency measurement, and requires a reference signal.

Hi, I have a couple of lock-in amplifiers. I agree to extract the signal nothing is better. But I am more trying to understand the impact of changing variables (loading resistors, feedback capacitors, light intensity, modulation frequency...). Currently I can see the degradation in the signal, but I do not always see clearly why. By visualizing the entire spectrum, I might get a more clear idea...

Title: Re: Looking for a dynamic signal analyzer with extended bandwidth
Post by: _Wim_ on October 22, 2017, 12:09:36 pm
One note on the Pico scopes: you have to check very carefully if they use ADCs with the specified resolution or specificy resolution after doing oversampling. When using oversampling you may not get the extra bits if there isn't enough noise in the signal.

Yes, this is correct. That is the biggest advantage over the 4262, it uses a real 16-bit AD, hence also the lower sampling rate only.
The 5000-series uses AD-converters of the other channels in parallel to get increased resolution.
On top of that, both have an enhanced resolution mode were oversampling is used (+4 bits max). This allows both to get up to 20 bits theoretically.
Title: Re: Looking for a dynamic signal analyzer with extended bandwidth
Post by: _Wim_ on October 22, 2017, 12:11:52 pm
No users here that use the HP 89441A or 89410A? I am still very temped by this solution also, but would like to read some user feedback (google turns up nothing, probably because these are quite old and not commonly used in the DIY community)

Edit: just noticed in the TEA thread user Berni snatched one of these recently (or posted recently about is)...
Title: Re: Looking for a dynamic signal analyzer with extended bandwidth
Post by: David Hess on October 22, 2017, 05:28:21 pm

It is actually much worse than that.  A wide bandwidth (200 MHz) CMOS operational amplifier used for the high impedance input buffer might have a 5nV/sqrHz white noise specification but a 10 MHz corner frequency yielding 1000nV/sqrHz of noise at 10 Hz.  Discrete designs even from 40 years ago are *much* better than this.

Unfortunately, these kind of details are typically not published for the scopes (and other gear).

To answer the question of low frequencies, yes, I need them. Light sources used for testing are typically modulated with frequencies between 300Hz an 20Khz, so this range is very interesting for me. My initial thought was to just by a dynamic signal analyzer, but there just so limiting in higher frequencies. Still, noise wise, they are probably still by far the best for my application...

I was not even being hypothetical; the Cleverscope design uses an OPA2355 as the input buffer.  This does not necessarily make it bad; who knows what the Picoscopes use?  And an integrated wide bandwidth low input current transimpedance amplifier used for wide bandwidth photodiode amplification will have a similarly high noise anyway.  Noise is a major problem with photodiode amplifiers which I assume is why you are interested in this.

Maybe they use a separate low frequency correction loop which became common in oscilloscope designs starting in the 1980s in place of expensive matched dual JFET to lower DC drift.  One way to find out would be to short the input and run an FFT on the resulting 20 MHz bandwidth limited noise; this should be done anyway to find out exactly what the lowest detectable noise level will be at different frequencies.

A sure way to solve this is to implement a simple external preamplifier.  It just has to have enough gain to overcome the noise of the test instrument and it removes the requirement for a 1 megohm input.  Also keep in mind that a 1 megohm input when used with a probe or cable will have considerable input capacitance.  A x10 probe solves this but increases the input noise by 10 times.
Title: Re: Looking for a dynamic signal analyzer with extended bandwidth
Post by: _Wim_ on October 22, 2017, 06:39:12 pm
A sure way to solve this is to implement a simple external preamplifier.  It just has to have enough gain to overcome the noise of the test instrument and it removes the requirement for a 1 megohm input.  Also keep in mind that a 1 megohm input when used with a probe or cable will have considerable input capacitance.  A x10 probe solves this but increases the input noise by 10 times.

A good pre-amp is indeed something that I need to look for. I know Standford Resaerch and EG&G make some very nice ones, typically also fitted with low and high pass filters to limit total noise. I have bid on a few in the past, but never managed the get one.

Noise is indeed a big problem with photo-amplifiers, but I find it a very interesting topic. Currently I am mostly looking at older commercial sensor systems, trying to understand how they were designed. I am quite a beginner at all of this (more of an automation guy than an electronics guy), but by understanding the design choices better will also allow me to purchase the correct sensor platforms during my day automation job. And maybe one day I can make a design of my own, but at the rate I am going, I will not get my hopes up...
Thanks by the way for the cleverscope doc about noise and SNR, I printed it for a good bedtime read  :)
Title: Re: Looking for a dynamic signal analyzer with extended bandwidth
Post by: David Hess on October 22, 2017, 07:37:03 pm
Noise is indeed a big problem with photo-amplifiers, but I find it a very interesting topic.

Currently I am mostly looking at older commercial sensor systems, trying to understand how they were designed. I am quite a beginner at all of this (more of an automation guy than an electronics guy), but by understanding the design choices better will also allow me to purchase the correct sensor platforms during my day automation job. And maybe one day I can make a design of my own, but at the rate I am going, I will not get my hopes up...

Burr-Brown published a number of great application notes on this subject which include detailed circuit analysis.  It is a difficult problem because the high capacitance of the photodiode greatly constrains the performance of the amplifier.

Quote
Thanks by the way for the cleverscope doc about noise and SNR, I printed it for a good bedtime read  :)

I was not aware of that document until I ran across it for this discussion.  That Cleverscope publishes so much detail leads me to trust their designs more.
Title: Re: Looking for a dynamic signal analyzer with extended bandwidth
Post by: _Wim_ on October 28, 2017, 09:58:56 am
Just a quick update, I just ordered a picoscope 5442B here: https://www.reichelt.de/USB-Oscilloscopes/PS-5442B/3/index.html?ACTION=3&GROUPID=5905&ARTICLE=146212&SHOW=0&START=0&OFFSET=16&SID=95Wb13EKwQAT4AADKebe825c2d261a5a91ed39db57cde2a9f417b&CTYPE=1&MWSTFREE=1 (https://www.reichelt.de/USB-Oscilloscopes/PS-5442B/3/index.html?ACTION=3&GROUPID=5905&ARTICLE=146212&SHOW=0&START=0&OFFSET=16&SID=95Wb13EKwQAT4AADKebe825c2d261a5a91ed39db57cde2a9f417b&CTYPE=1&MWSTFREE=1)

The cleverscope also seemed a very good option, but was much more expensive, and seeing the good results egonotto achieved with his 5000 series scope, and the fact that I used a picoscope before and reallly liked it, made me choice the pico.

Many thanks for all of your replies.

Title: Re: Looking for a dynamic signal analyzer with extended bandwidth
Post by: thanasisk on June 10, 2018, 11:10:09 am
Hello,

Did you get similar good results as the ones posted by egonotto with your new picoscope? How is the noise performance compared to the specs?
Title: Re: Looking for a dynamic signal analyzer with extended bandwidth
Post by: _Wim_ on June 10, 2018, 01:20:07 pm
Yes, quite satisfied with my purchase.

I will post some pictures with the result I get when I am in the lab. The scope fits my needs, but on the software side of things there could be some improvements (I am mainly missing a good way to add overlays in FFT window to compare results easily). But the great thing about picoscope is that you can download new software versions for free, so any new features implemented in the future you get also.
Title: Re: Looking for a dynamic signal analyzer with extended bandwidth
Post by: egonotto on June 11, 2018, 05:25:31 pm
Hello,

thanasisk wrote:
"Did you get similar good results as the ones posted by egonotto with your new picoscope?"

Please don't misinterpret my measures. There is a reduced band width and sometimes a kind of averaging. This is not written in the specs.

In the specs you only find typical noise with full band width.
(There is also no accurate description how to measure the noise).

And there is the PicoScope 5000 not so good. And my PicoScope 5243A is a little worse as in the spec.

From full band width noise consideration there seems better oscilloscopes. For instance the new Siglent or the new Rohde und Schwarz RTB2000

Best regards
egonotto
Title: Re: Looking for a dynamic signal analyzer with extended bandwidth
Post by: thanasisk on June 11, 2018, 05:48:27 pm
A specified 70uVrms at 2mV/div at full bandwidth delivers approx 5 bits ENOB according to theory for the 5000 pico.

Is the siglent or r&s delivering better enob despite being 8bit and 10bit scopes?

Title: Re: Looking for a dynamic signal analyzer with extended bandwidth
Post by: _Wim_ on June 11, 2018, 07:56:47 pm
Here are some quick tests I have done. If you want me to test anything else, please feel free to ask:

- FFT with shorted input for 10mV range an 1V range
- FFT 1Mhz sine from internal generator and Rigol DG1032Z
- noise with shorted input 16 bit and 20 bit (oversampling)
- noise when using EG&G premap with gain set to 100, both for 16 bit and 20 bit (remark: bandwidth pre-amp is only 1Mhz)

Remark: no bandwidth limits or averaging
Title: Re: Looking for a dynamic signal analyzer with extended bandwidth
Post by: _Wim_ on June 11, 2018, 08:16:05 pm
Did some more tests:

- in the above during the first tests for noise 2 power supplies were turned on next to the scope, hence the 50Hz noise
- also repeated the tests above at max sample rate (62.5MS/s for 16 bit)

Again no bandwidth limit or averaging...
Title: Re: Looking for a dynamic signal analyzer with extended bandwidth
Post by: thanasisk on June 11, 2018, 11:40:37 pm
Thank you Wim for the interesting measurements!

May I also ask for :

- +-10mV range, shorted input, BW limit ON, 16bit/62.5MS/s: measure AC RMS + P-P noise  [to see how effective is the BW limit on noise reduction]
- same with +-1V range [to see how the input frontend noise changes]
- FFT with 1kHz sine 2Vpp from internal generator: measure THD,THD+N,SFDR, (+AC coupling to remove DC offset?) [to see the performance of the gen]

Title: Re: Looking for a dynamic signal analyzer with extended bandwidth
Post by: thanasisk on June 12, 2018, 12:47:36 am
As a side note,  sampling at 62.5MS yields a low effective bandwidth of approx 12.5MHz. In that case not using the BW limit will cause aliasing (thus affecting the noise measurements). Perhaps decreasing the number of bits from 16 to e.g 10 to increase the sampling rate sufficiently would show a different picture?
Title: Re: Looking for a dynamic signal analyzer with extended bandwidth
Post by: KrudyZ on June 12, 2018, 03:32:40 am
Another oldie that would fit the bill:
HP3585A @0 Hz - 40 MHz with tracking generator, 1M Ohm input in addition to 75 and 50 Ohm.
Very heavy. Repairable with available schematics.
Title: Re: Looking for a dynamic signal analyzer with extended bandwidth
Post by: _Wim_ on June 12, 2018, 06:53:08 pm
Thank you Wim for the interesting measurements!

May I also ask for :

- +-10mV range, shorted input, BW limit ON, 16bit/62.5MS/s: measure AC RMS + P-P noise  [to see how effective is the BW limit on noise reduction]
- same with +-1V range [to see how the input frontend noise changes]
- FFT with 1kHz sine 2Vpp from internal generator: measure THD,THD+N,SFDR, (+AC coupling to remove DC offset?) [to see the performance of the gen]

Here you are...
Title: Re: Looking for a dynamic signal analyzer with extended bandwidth
Post by: _Wim_ on June 12, 2018, 07:03:25 pm
As a side note,  sampling at 62.5MS yields a low effective bandwidth of approx 12.5MHz. In that case not using the BW limit will cause aliasing (thus affecting the noise measurements). Perhaps decreasing the number of bits from 16 to e.g 10 to increase the sampling rate sufficiently would show a different picture?

Here is the measurement on different sample rates and resolutions. No BW limit configured for all of them. 14 bit seems to be not the ideal choice becasue of low sampling rate and higher nosie...
Title: Re: Looking for a dynamic signal analyzer with extended bandwidth
Post by: Marco on June 12, 2018, 09:12:25 pm
How about the Red Pitaya? Has 14 bit 125 MSPS.
Title: Re: Looking for a dynamic signal analyzer with extended bandwidth
Post by: egonotto on June 13, 2018, 12:12:58 am
Hello,

Red Pitaya works with digital zoom. You have +- 1V with 14 Bit. The more sensitive V/div settings are only zoom.

Best regards
egonotto


Title: Re: Looking for a dynamic signal analyzer with extended bandwidth
Post by: thanasisk on June 13, 2018, 10:24:30 am
I think the SNR of the 5000, because the design utilizes multiplexed ADCs, does not improve in higher ranges.  Did you get any significant  benefit from utilizing the preamp?

Perhaps the following test would be appropriate:
1. 20mV sine gen 1khz -> channel 1 of ps5000 (+-10mV range)
2. 20mV sine gen 1khz -> preamp 100x  -> channel 1 of ps5000 (+-1V range, custom 1:100 probe configuration in settings)

For both cases measure SNR. BW limit on at 16bits mode, no hires or averaging.
Title: Re: Looking for a dynamic signal analyzer with extended bandwidth
Post by: thanasisk on June 13, 2018, 11:50:58 pm
I calculated the ENOBs for your scope.

16bit@2mV/div@20MHzLIMIT yields 6.19bits ENOB
16bit@200mV/div@20MHzLIMIT yields 8.86bits ENOB

So if you were to use the preamp, a maximum of approx. 2.5bits ENOB could be gained depending on the noise of the preamp.
Title: Re: Looking for a dynamic signal analyzer with extended bandwidth
Post by: _Wim_ on June 14, 2018, 08:07:55 pm
I think the SNR of the 5000, because the design utilizes multiplexed ADCs, does not improve in higher ranges.  Did you get any significant  benefit from utilizing the preamp?

Perhaps the following test would be appropriate:
1. 20mV sine gen 1khz -> channel 1 of ps5000 (+-10mV range)
2. 20mV sine gen 1khz -> preamp 100x  -> channel 1 of ps5000 (+-1V range, custom 1:100 probe configuration in settings)

For both cases measure SNR. BW limit on at 16bits mode, no hires or averaging.

Yes, quite a significant benefit. Attached the comparison.

Setup:

DG1032Z 2Vpp sine => Attenuator /100 => Preamp x100 => Picoscope  (with custom 1:100 probe)
DG1032Z 2Vpp sine => Attenuator /100 => Picoscope


Title: Re: Looking for a dynamic signal analyzer with extended bandwidth
Post by: _Wim_ on June 14, 2018, 08:12:40 pm
I calculated the ENOBs for your scope.

16bit@2mV/div@20MHzLIMIT yields 6.19bits ENOB
16bit@200mV/div@20MHzLIMIT yields 8.86bits ENOB

So if you were to use the preamp, a maximum of approx. 2.5bits ENOB could be gained depending on the noise of the preamp.

2.5 bits ENOB is quite significant, not (16.8db). Off course the preamp add distorsion, so it depends on what is more important..
Title: Re: Looking for a dynamic signal analyzer with extended bandwidth
Post by: thanasisk on June 15, 2018, 12:57:30 pm
After seeings several measurements here (many thanx wim and egonotto!) and in the picoscope support forum, i can see that the 5000 series yield approx 7-10 bits of enob across ranges (1 bit of which is because of the bandwidth limit) which can be further improved  by 1.5 bits with res.enhancement and 1-1.5 bits by using a preamp. The dynamic range would be limited by distortion to around 67dB).

As for the 4262 I am wondering about the noise across ranges with and without bandwidth limit and with/without res enhancement. Perhaps egonotto you could kindly offer some further insight/measurements?
Title: Re: Looking for a dynamic signal analyzer with extended bandwidth
Post by: Marco on June 15, 2018, 01:32:38 pm
Eyeballing the FFT, I'd say the Red Pitaya has around the same dBFs noise on the 2Vpp input as the 5000, but is more linear.
Title: Re: Looking for a dynamic signal analyzer with extended bandwidth
Post by: _Wim_ on June 15, 2018, 03:09:42 pm
Eyeballing the FFT, I'd say the Red Pitaya has around the same dBFs noise on the 2Vpp input as the 5000, but is more linear.

Unfortunately the Red Pitaya lacks low input ranges, which makes it less usefull for my application. Combined with the preamp this could have been a potential candidate, but the preamp also introduces additional distorsion...

How is the software for the red pitaya? The hardware seems very capable for the price, but I am not sure how good the software is.
Title: Re: Looking for a dynamic signal analyzer with extended bandwidth
Post by: egonotto on June 15, 2018, 03:15:35 pm
Hello,

Marco writes: "but is more linear."
What do you mean with this.

I think Red Pitaya has good hardware, but the software is frugal.

The Red Pitaya has 16kB/channel. If one would get knowledge about the noise you should consider the whole 16kB.

Best regards
egonotto
Title: Re: Looking for a dynamic signal analyzer with extended bandwidth
Post by: Marco on June 15, 2018, 03:29:22 pm
What do you mean with this.
Their FFT (http://redpitaya.readthedocs.io/en/latest/developerGuide/125-14/fastIO.html#analog-inputs) of a 0dBm 2 MHz signal seems to show less harmonics than the one Wim showed for 1 MHz. Although on second thought, that might be the fault of the Rigol DG1032Z Wim used.
Title: Re: Looking for a dynamic signal analyzer with extended bandwidth
Post by: _Wim_ on June 15, 2018, 03:38:15 pm
Although on second thought, that might be the fault of the Rigol DG1032Z Wim used.

Unfortunately, that is the best (and only siggen I have that can produce a 1MHz signal. I do have somewhere a very low THD 10Khz audio oscillator, but I do not think this will prove much. Maybe egonotto has a very clean siggen to make the comparison?
Title: Re: Looking for a dynamic signal analyzer with extended bandwidth
Post by: egonotto on June 15, 2018, 04:06:58 pm
Hello.

_Wim_ wrote: "How is the software for the red pitaya? The hardware seems very capable for the price, but I am not sure how good the software is"

I think it is possible to try if you want.
You can use my Red Pitaya over web. My version now is 0.94

Best regards
egonotto
Title: Re: Looking for a dynamic signal analyzer with extended bandwidth
Post by: _Wim_ on June 15, 2018, 05:36:20 pm
Hello.

_Wim_ wrote: "How is the software for the red pitaya? The hardware seems very capable for the price, but I am not sure how good the software is"

I think it is possible to try if you want.
You can use my Red Pitaya over web. My version now is 0.94

Best regards
egonotto

Thanks for the offer, but for the moment it is not necesarry. I was more wondering how the software was, as devices like this (and also the pico) only have real value is the software is up to the task...
Title: Re: Looking for a dynamic signal analyzer with extended bandwidth
Post by: innkeeper on June 15, 2018, 06:31:59 pm
look into CMU-B41 for cmu200
Title: Re: Looking for a dynamic signal analyzer with extended bandwidth
Post by: egonotto on June 15, 2018, 06:51:10 pm
Hello,

here are some picture from PicoScope 4262. Both inputs have a 50 Ohm closure (without the noise in 10mV  would be far more)
Today the noise in 10mV RMS range is particular low most it is near 8.5 uV
The quotient p-p/RMS is very high. Much higher as other oszilloscope have. Look the excellent test from nctnico "https://www.eevblog.com/forum/testgear/rohde-schwarz-rtm3000-review/".

Best regards
egonotto


Title: Re: Looking for a dynamic signal analyzer with extended bandwidth
Post by: thanasisk on June 16, 2018, 12:24:53 am
So I can see that the 4262 delivers ENOB of 9..12 bits across ranges (BW limit does not appear to have a huge effect on the outcome), which can be increased to 10..14 bits using hi-res mode.

That is 2 extra bits compared to a BW limited 5000.  :clap:
Title: Re: Looking for a dynamic signal analyzer with extended bandwidth
Post by: _Wim_ on June 16, 2018, 07:12:22 am
look into CMU-B41 for cmu200

If you mean as a signal generator, this one is quite limited. It only goes up to 20Khz, and THD+N is around 80dB. A sound card on a pc performs better then this.
Title: Re: Looking for a dynamic signal analyzer with extended bandwidth
Post by: _Wim_ on June 16, 2018, 08:01:24 am
Hello,

here are some picture from PicoScope 4262. Both inputs have a 50 Ohm closure (without the noise in 10mV  would be far more)
Today the noise in 10mV RMS range is particular low most it is near 8.5 uV
The quotient p-p/RMS is very high. Much higher as other oszilloscope have. Look the excellent test from nctnico "https://www.eevblog.com/forum/testgear/rohde-schwarz-rtm3000-review/".

Best regards
egonotto

Thanks for the additional measurements. Looking at the tests of nctnico, I was a bit supprized the noise levels were much higher with the 50ohm input impedance. As expected before, the 4262 is indeed still much better then the 5000 series. If only it had a little more bandwidth.

The RTM3000 also looks like an excellent device, but is above the level I want to spent...
Title: Re: Looking for a dynamic signal analyzer with extended bandwidth
Post by: egonotto on June 16, 2018, 07:26:16 pm
Hello,

sometimes one can the improve the noise through addition the same signal.

Best regards
egonotto

Title: Re: Looking for a dynamic signal analyzer with extended bandwidth
Post by: thanasisk on June 16, 2018, 08:48:57 pm
Hello,

sometimes one can the improve the noise through addition the same signal.

Best regards
egonotto

This is a cool idea. You could even squeeze some more SNR  if you compensate for dc offset  and weight each channel according to the rms noise.  Not sure if the math channel of picoscope could support this though.
Title: Re: Looking for a dynamic signal analyzer with extended bandwidth
Post by: nctnico on June 16, 2018, 09:09:45 pm
Hello,

here are some picture from PicoScope 4262. Both inputs have a 50 Ohm closure (without the noise in 10mV  would be far more)
Today the noise in 10mV RMS range is particular low most it is near 8.5 uV
The quotient p-p/RMS is very high. Much higher as other oszilloscope have. Look the excellent test from nctnico "https://www.eevblog.com/forum/testgear/rohde-schwarz-rtm3000-review/".

Best regards
egonotto

Thanks for the additional measurements. Looking at the tests of nctnico, I was a bit supprized the noise levels were much higher with the 50ohm input impedance. As expected before, the 4262 is indeed still much better then the 5000 series. If only it had a little more bandwidth.

The RTM3000 also looks like an excellent device, but is above the level I want to spent...
That has to do with the lower bandwidth of the inputs in 1M Ohm mode With the 20MHz bandwidth limit on the 50 Ohm mode on the RTM3000 has a much lower noise floor.
Title: Re: Looking for a dynamic signal analyzer with extended bandwidth
Post by: _Wim_ on June 17, 2018, 05:20:30 am
That has to do with the lower bandwidth of the inputs in 1M Ohm mode With the 20MHz bandwidth limit on the 50 Ohm mode on the RTM3000 has a much lower noise floor.

I was looking at the bandwidth limited case! But I see now, the name of the pictures are mixed up, and I was unsure how to interprete the channel setup "DC" and "Ohm"...
Title: Re: Looking for a dynamic signal analyzer with extended bandwidth
Post by: _Wim_ on June 17, 2018, 05:43:04 am
But I see now, the name of the pictures are mixed up

In that case, with 20M BWLimit and 1M input, the RTM is noiser than the pico5000 (Vrms 33µV vs 63µV, Vpp 321µV vs 518µV). But the pico does not have 50ohm input impedance (not very usefull for me because of too much loading), so this comparison cannot be made.
Title: Re: Looking for a dynamic signal analyzer with extended bandwidth
Post by: David Hess on June 17, 2018, 12:51:12 pm
In that case, with 20M BWLimit and 1M input, the RTM is noiser than the pico5000 (Vrms 33µV vs 63µV, Vpp 321µV vs 518µV). But the pico does not have 50ohm input impedance (not very usefull for me because of too much loading), so this comparison cannot be made.

Is there any way to determine how much of that is input noise and how much is quantization noise?

33uVrms over 20MHz is high for a 1Mohm input but may be consistent with a CMOS implementation.
Title: Re: Looking for a dynamic signal analyzer with extended bandwidth
Post by: bozidarms on June 17, 2018, 05:26:27 pm
HP 89410A continue to be the most noiseless analyzer till this days - effective about only 8 nV/√Hz,
better that some dedicated LNA-s!
Title: Re: Looking for a dynamic signal analyzer with extended bandwidth
Post by: _Wim_ on June 17, 2018, 06:07:07 pm
HP 89410A continue to be the most noiseless analyzer till this days - effective about only 8 nV/√Hz,
better that some dedicated LNA-s!

 8 nV/√Hz equals to 35µVrms at 20MHz bandwidth, so that is quite simular than the pico. Is that for the 50 ohm input or for the 1Mohm input? But I agree it is a very nice analyzer, and if I lived in the states, I would probably have bought this one instead of the pico. But in Europe these are much more expensive, and shipping a beast like that aint cheap...
Title: Re: Looking for a dynamic signal analyzer with extended bandwidth
Post by: bozidarms on June 17, 2018, 06:41:24 pm
For 1Mega ohm.
Yes, still one exceptional instrument indeed, unfortunately expensive in EU.
Title: Re: Looking for a dynamic signal analyzer with extended bandwidth
Post by: Gerhard_dk4xp on June 17, 2018, 11:56:12 pm
HP 89410A continue to be the most noiseless analyzer till this days - effective about only 8 nV/√Hz,
better that some dedicated LNA-s!

Aaarghhh! - No!

Below 100 Hz its 1/f noise is breathtaking. For my low noise amplifiers,
an 80 dB preamp is not enough to mask it. Rising the gain above 80 dB
is impossible because of overflows. And the preamp has 200pV/rt Hz.

regards, Gerhard

(none the less, I kinda like it. Contact me if you want a remote control
C/Linux program to do FFT plots over 9 decades via network.
Requires Gnuplot.)
Title: Re: Looking for a dynamic signal analyzer with extended bandwidth
Post by: _Wim_ on June 18, 2018, 07:27:23 pm
For my low noise amplifiers,  an 80 dB preamp is not enough to mask it. Rising the gain above 80 dB
is impossible because of overflows. And the preamp has 200pV/rt Hz.

Nice pre-amp! That sure beats my 5113 in noise for low impedance sources, looks ideaal for power supply noise measurements!

For others,  see here: http://www.hoffmann-hochfrequenz.de/downloads/lono.pdf (http://www.hoffmann-hochfrequenz.de/downloads/lono.pdf)
Title: Re: Looking for a dynamic signal analyzer with extended bandwidth
Post by: thanasisk on June 19, 2018, 01:33:44 pm
Very interesting preamp design  :-+ i will learn a lot by studying it
Title: Re: Looking for a dynamic signal analyzer with extended bandwidth
Post by: thanasisk on June 23, 2018, 11:53:27 am
I found this short review of the picoscope 4262:

https://www.amplifier.cd/Test_Equipment/other/Picoscope4262/Picoscope_4262.html (https://www.amplifier.cd/Test_Equipment/other/Picoscope4262/Picoscope_4262.html)

From the pictures I can see that the LTC2202 ADC is used. 10MSps 16 bit..

http://www.analog.com/en/products/analog-to-digital-converters/standard-adc/high-speed-ad-10msps/ltc2202.html (http://www.analog.com/en/products/analog-to-digital-converters/standard-adc/high-speed-ad-10msps/ltc2202.html)

Its specs in terms of sample rate, SFDR, SNR are still quite competitive, whereas ADCs with higher sample rates have worse SFDR and list up to 14.5 ENOB (and are at at least double the price at mouser).

Cant tell from the picture which analog devices DAC is used for the generator though.