~$400 oscilloscope FFT for spectrum analysis

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
I'm getting an oscilloscope for general audio work, and noticed that some of the newer digital storage oscilloscopes now include FFT function. I have heard conflicting reports about whether these are useful for audio spectrum analysis work; some say the 8 bit accuracy of most DSOs makes them not very useful, but others say it's a good tool to have.

My main use for the FFT would be to characterize the frequency response of components (e.g. amps, crossovers, custom circuits). I'd also like to use it with a microphone for getting a sense of room response of speakers.

My leading contender is the $400 Rigol DS1052E because of many positive user reviews, solid build quality, and 50Mhz bandwidth hackable to 100Mhz. But I'm also considering the $270 OWON 5022S which has a bigger screen. The $300 ATTEN ADS1022C has a unique split FFT/channel display, and the $350 Instek GDS-1022 has a very limited built in signal generator.

Here is a screen shot that I think shows the Rigol 1052 operating in FFT mode. The frequency scale seems to be linear rather than log, and it's not clear that the whole audio 20-20k range would fit on one screen.

I was hoping to replace my TrueRTA pc-based RTA with the FFT mode of a DSO, but trueRTA generates a much more readable graph of frequency response.

Any thoughts on whether its feasible to use these FFT DSOs as RTAs or basic audio spectrum analyzers are welcome. And which of the models above are best for FFT?
 
I use a Tek TDS series, and it functions just fine. You will be able to control the window width of the FFT display, so 20-20k will be possible.

I have found the ability to measure about 55dB down from the fundamental, and that's about it. So it is a useful troubleshooting tool, but does not have the ability to provide those wonderful -110dB measurements you see posted in these forums to properly analyze preamp/amp performance.

I have an external firewire sound card that I plan to use for more serious measurements, but haven't decided on the right OSX software package.

That being said, if you are serious about getting into audio, you NEED a scope, and having the additional FFT function seems the right path to go.
 
You don't need an FFT scope to measure frequency (amplitude vs. frequency) response. Instead, you should use a function generator that can produce a sine wave at between (at least) 20 Hz and 20 kHz and at a constant amplitude. Then just measure the amplitude of the output of the device under test at various frequencies and plot manually. You can also get a really nice plot using a sweep generator (log frequency) and a digital scope that can do a very slow timebase, like 1 second/division. Set the scope to 1s/div and 'envelope' mode if available, set the sweep gen to 1 octave per second. trigger the scope off the sweep generator's trigger output. The scope display will effectively plot the frequency response as one octave per division and linear amplitude.

If you try to use FFT (by inserting a noise signal then taking FFT of the output) you will not get satisfactory results. An FFT of random white noise is not the nice flat line that you might expect; it can look just as spurious and 'noisy' as the noise signal in the time domain, unless you average many many FFT results together, which a typical scope w/FFT can't do. As you discovered, all or most scopes will plot the FFT result with linear frequency axis, since that is the real format of the FFT data. Audio software tools artificially distort the frequency axis to give a more familar log axis for easier visualization by us humans. Most will however do a log amplitude at least.

I use the FFT on my scope to examine the distortion residual signal from my distortion analyzer, which is basically the output of the analyzer's notch filter. For that, it is very useful, and exceeds the capability of a sound card due to the much higher sampling rate than what a sound card can do.

I agree with zigzagflux, you should have a "real" scope, not just a PC sound card based wannabe. But for frequency response type measurements, the PC solution is excellent.
 
Last edited:
Looking through the instruction manuals online for the FFT modes on these DSOs, it seems like they are all pretty much the same functionality. There are a few minor differences about how many samples are used (1024 in one 2048 in another IIRC), but nothing that looks important.

Macboy, that's a good point about the DSO FFT not doing time-based averaging/smoothing like the PC product would. I hadn't considered that issue, and I can see that making it quite unusable for RTA purposes.

The idea of capturing a sweep to do a frequency plot would probably work better... great idea! I don't have a device that would generate it, but I do have test CDs with sweeps that I could use.
 
Just to follow up, I am about to pull the trigger on an Instek 1062a oscilloscope, which seems to be a bit newer and better than the rigol for about the same price.

I also got an instek function generator GFG-8255a on ebay that has sweep function, so I can try out the approach macboy suggested of making my own frequency response graphs inside the DSO. That sounds like a very simple and stable way to get the data, as long as my function generator response is flat (we'll see).
 
Member
Joined 2009
Paid Member
If you want to do speaker and/or room measurements, get a USB or Firewire sound card and a calibrated Dayton EMM-6 from Cross Spectrum Labs. Then use the free but very capable Holm Impulse software to perform the frequency sweep.

With software like TrueRTA and others, you can use the same sound card for FFT measurements.

I find the sound card based "FFT analyzers" are not as convenient to use as stand-alone gear, but also not as expensive. Tradeoffs, tradeoffs.

~Tom
 
There was just a bunch (about 40 of them) of Agilent DSO-3062A's that went on Ebay. They went for 200.00 plus shipping. Many of the guys including myself on the HP-Agilent Yahoo Group bought them. One can NEVER have too many scopes after all! All but a few of them turned out to be a steal, the few that were not just had slight BNC damage that was easily repaired. Mine had only been powered on 58 times when I got it. I upgraded the firmware to 4.0.16 and grabbed the free "Connect" software that Agilent has for these scopes. Connect allows one to display as well as operate the scope on his PC computer or laptop. The Agilent DSO series is made for them by Rigol and after seeing the spectaculuar performance of these scopes I would say stick with the Rigol. Many of the Yahoo Group guys have measured their -3db bandwidth to be >/=160 mhz!!! Thats pretty amazing for what is advertised and marked as a 60 mhz scope! Rigol really rocks!

Mark
 
I wrote up a piece on a XMOS USB board eval with a QA100 scope. The scope has a super deep buffer, and so it can do fft with fractional Hz resolution. The scope can also produce dual tone (audio range) arbitrary waveforms. Sine is most interesting here, because you can do 1 KHz and 900 Hz and look for mixing products.



New to the Lab: QuantAsylum QA100 Oscilloscope & XMOS USB Audio engineeringtopics
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.