EEVblog Electronics Community Forum
Products => Test Equipment => Topic started by: Sylvi on September 12, 2018, 08:06:39 pm
-
Hi
What is a good DSO to use for looking at audio signals?
Not just for sine wave testing, but also to look at dynamic music signals.
Thanks
-
How about a sound card scope in this case?
-
If you keep a computer on your bench look into the Digilent Analog Discovery 1 or 2. They are well thought of here and particularly for audio. Kind of the Swiss Army knife of of T&M .
Analog Discovery 2
Two-channel USB digital oscilloscope (1MΩ, ±25V, differential, 14-bit, 100MS/s, 30MHz+ bandwidth - with the Analog Discovery BNC Adapter Board)
Two-channel arbitrary function generator (±5V, 14-bit, 100MS/s, 12MHz+ bandwidth - with the Analog Discovery BNC Adapter Board)
Stereo audio amplifier to drive external headphones or speakers with replicated AWG signals
16-channel digital logic analyzer (3.3V CMOS and 1.8V or 5V tolerant, 100MS/s)
16-channel pattern generator (3.3V CMOS, 100MS/s)
16-channel virtual digital I/O including buttons, switches, and LEDs – perfect for logic training applications
Two input/output digital trigger signals for linking multiple instruments (3.3V CMOS)
Single channel voltmeter (AC, DC, ±25V)
Network Analyzer – Bode, Nyquist, Nichols transfer diagrams of a circuit. Range: 1Hz to 10MHz
Spectrum Analyzer – power spectrum and spectral measurements (noise floor, SFDR, SNR, THD, etc.)
Digital Bus Analyzers (SPI, I²C, UART, Parallel)
Data Logger - Exportable data and plot functionality
Impedance Analyzer - Capacitive and Inductive Elements
Protocol Analyzer - SPI, I2C, and UART
Two programmable power supplies (0…+5V , 0…-5V). The maximum available output current and power depend on the Analog Discovery 2 powering choice:
500mW total when powered through USB. (Each supply can provide between 0mW and 500mW so long as the total does not exceed 500mW.)
2.1W max for each supply when powered by an auxiliary supply.
700mA maximum current for each supply.
-
What is is that you are looking to see in music signals?
-
Most oscilloscopes should meet your needs, but I think you should focus on the refresh rate.
-
How about a sound card scope in this case?
The main problem with a sound card scope is that it's not calibrated.. but relative measurements can be made if you have a generator of some sort with a known amplitude that can be used as a reference
The other problem is that most aren't DC coupled, it could be a problem for him
-
Thanks for the replies!
I looked into sound card options before which at first seem pretty good using a 24-bit card and something like the Dr. Jordan MLS software, which can do THD analysis, IM, and other things apart from showing the wave. But as posted above, the card itself is not calibrated and in the case of a THD measurement, you want the signal to be very close to the maximum level allowed by the card, both for the generator tone and for the returned signal. This maintains the best resolution of both signals.
Back when I looked at the sound cards computers were still full-size boxes. It might be more practical today with the mini-computers. I didn't really want to have a computer on my work bench although I would plug a monitor into a DSO to make it easier for me to see.
As for basic oscilloscope functions, I don't need anything too tricky but being able to see that the wave is smooth and the correct shape is helpful. The jagged 8-bit sine waves I see in the scope ads and brochures is not inspiring! The steadiness of the DSO's display IS very nice!
Low signal level resolution is also nice when assessing noise and I've read complaints on this forum about certain DSO front-ends, like OWON, which I am most interested in. They seem to have the highest bit-count for the lowest price. I rarely use 2-ch let alone 4, so a really good 2-ch scope would suit me fine - or a really really good 1-ch - :)
-
I don't see how common DSO can help in the audio field, unless you are fixing switching power supplies for audio equipment [chuckle]. How you are going to measure THD+N with device having 8 bit ADC, 3% gain and 1% offset accuracy? (specs of decent quite popular Rigol scope) I doubt that anyone plans to work on landline telephone audio quality tests today.
If someone say that there are 14bit or 16 bit DSO's - fine. They exist. For what price? Most likely such scopes are (much) more expensive than state of the art audio analyzers with calibration certificates. IMHO quality audio capture adapter with proper software is way to go for constrained budget. Other option: R&S UPV Audio Analyzer, if you can afford such.
-
Hi
I'm not looking at using the DSO to do THD measurements; rather, just to be able to see signals when I place the probe. Also to test circuit rise-times to step inputs and to help look for noise or oscillation. For the latter, the high band-width that DSOs all seem to have would be beneficial since many of the audio circuits have response out to 10MHz before their response is shped and BW-limited.
I don't need to decode digital audio protocols or any other digital signal, so none of that is of interest to me.
I do have to be able to see rise times of DC control voltages and relative timing of different DC points rising/falling at turn-on, say, for initialisation of the channel switching circuits (which are not computer-based but use logic circuits). DSO seem to be better for this than analogue scopes.
-
I do have to be able to see rise times of DC control voltages and relative timing of different DC points rising/falling at turn-on, say, for initialisation of the channel switching circuits
Well, then why we even have this discussion? - Get DSO!
-
I'm not looking at using the DSO to do THD measurements; rather, just to be able to see signals when I place the probe. Also to test circuit rise-times to step inputs and to help look for noise or oscillation. For the latter, the high band-width that DSOs all seem to have would be beneficial since many of the audio circuits have response out to 10MHz before their response is shped and BW-limited.
Any new DSO will do that, no problem at all.
-
I looked into sound card options before which at first seem pretty good using a 24-bit card and something like the Dr. Jordan MLS software, which can do THD analysis, IM, and other things apart from showing the wave. But as posted above, the card itself is not calibrated and in the case of a THD measurement, you want the signal to be very close to the maximum level allowed by the card, both for the generator tone and for the returned signal. This maintains the best resolution of both signals.
Back when I looked at the sound cards computers were still full-size boxes. It might be more practical today with the mini-computers. I didn't really want to have a computer on my work bench although I would plug a monitor into a DSO to make it easier for me to see.
As for basic oscilloscope functions, I don't need anything too tricky but being able to see that the wave is smooth and the correct shape is helpful. The jagged 8-bit sine waves I see in the scope ads and brochures is not inspiring! The steadiness of the DSO's display IS very nice!
I use a program called Signal Scope (https://www.faberacoustical.com/apps/mac/signalscope/) on my Mac for FFT and oscilloscope display using standard audio interfaces (I have an older FireWire MOTU 828 mk2). You can calibrate your audio interface so it can display 'scope traces in units of volts. It supports proper oscilloscope triggering, too. Obviously the measurements are only as good as your audio interface, but for most things you're probably fine.
-
Signal Scope looks nice. Is there an open source sound card based audio measurement project that is worth consideration? I haven't had much luck finding one, but then, it might be my googlefoo needs some tweaking. Maybe just a discussion of the different packages that are available would be interesting?
-
Signal Scope looks nice. Is there an open source sound card based audio measurement project that is worth consideration? I haven't had much luck finding one, but then, it might be my googlefoo needs some tweaking. Maybe just a discussion of the different packages that are available would be interesting?
I have not really looked to see what’s available. There are a few programs that do FFTs and such, and are targeting the acoustics-measurement crowd (Room EQ Wizard, for example). I haven’t seen an oscilloscope program that uses standard audio devices that’s better than Signal Scope. I don’t particularly care whether a program is open-source or not, I just want it to work.
-
Signal Scope looks nice. Is there an open source sound card based audio measurement project that is worth consideration? I haven't had much luck finding one, but then, it might be my googlefoo needs some tweaking. Maybe just a discussion of the different packages that are available would be interesting?
I have not really looked to see what’s available. There are a few programs that do FFTs and such, and are targeting the acoustics-measurement crowd (Room EQ Wizard, for example). I haven’t seen an oscilloscope program that uses standard audio devices that’s better than Signal Scope. I don’t particularly care whether a program is open-source or not, I just want it to work.
Sure, I was asking more generally since the topic is "DSO for audio." I'm aware of Room EQ Wizard and a couple of others, but I'd really like to have something that's open source so I could adapt it as necessary.
-
Signal Scope looks nice. Is there an open source sound card based audio measurement project that is worth consideration? I haven't had much luck finding one, but then, it might be my googlefoo needs some tweaking. Maybe just a discussion of the different packages that are available would be interesting?
there are few here... https://sourceforge.net/directory/os:windows/?q=oscilloscope+sound+card
-
Hi
Thanks for the advice and info!
After a lot of looking at manufacturers' sites and at stores on Aliexpress, and Mouser, I decided to get the OWON XDS2102A which is 12-bit 100MHz and has a VGA output. It can connect to a pictbridge printer, too, and LAN, although I don't think I'll need the latter except to update the firmware? Some vendors show pics of the scope with or without jacks or features contradicting the product description, and then either never answer emails or are really slow to answer. The particular one I bought was on sale at about 35% off the store's regular price. Whether that is list price who knows? :)
I also got some x100 probes for looking at tube circuits.
I thought the unitised oscilloscope would be better for me at the moment given the comments about scrolling through menus and latency of USB scopes. I know their are menus to deal with in the stand-alone DSOs, but I only have to get used to one machine. I won't really know if another brand is "totally superior and easy" compared to this until maybe a few years down the road if I get another one.
It's pretty exciting to be getting some new test equipment and to move into the 21st century ;D
-
When i first started looking at DSOs seriously in the recent weeks, the first ones I saw were the Hanteks and the rebranded Hanteks. These all had 5.7" displays and were 8-bits vertical resolution. Then I started seeing the units with 7" WVGA displays which were still mostly the same specs but with larger displays. There were some with nice clean layouts and simple panels, sort of like I'm used to, so those were appealing and were still only $250 to 350.
Then I saw units with 8" displays! 800 x 600 pixels. These looked really nice but again all 8-bit until I started finding OWON. Actually, the first 8" display might also have been OWON, but for my wonky eyes the larger display was a real winner. Then I saw models that offered a VGA output option, or as a standard feature. Then I found the 12-bit and 14-bit OWONs that also had VGA options and of course much higher sampling rates and more memory.
While I was trying to learn about the specs and the options the same model scope could have, and figure out some of the lingo used by each manufacture - which was convoluted by the stores - I looked upthe manufacturers's sites to see what they said about their own products. The Sigilent site was crap, with very little information about the scopes unless you clicked on each model in turn and looked at whatever data was available. There was no easy way to compare models or to see the differences. OWON's site was great in that respect, and they have little videos explaining aspects of their scope features, a lot of which applies to all DSOs. The Hantek site showed up the fact that they only make entry-level cheapo scopes, and then the reviews I read on Amazon pointed out fairly bad issues with their firmware. I had kept one Hantek model on my list simply because it was so inexpensive that it would allow me to do more stuff with the little bit of cash I had available at the time. Fortunately more $$ arrived and I could go with what I really wanted to get - the OWON XDS2102A.
I looked at the XDS3000s as well, but I did not need the multi-meter nor was i interested in data loggers. The arbitrary waveform generator (AWG) was enticing but not a deal breaker. The 3000s have twice the memoroy depth at 40M compared to 20M for the 2102, and a bit higher sampling rate at 75k versus 55k, again not a deal breaker for my needs. The specific models I was looking at were 2-channel, 100MHz and 12-bit, so the main characteristics of all the scopes are identical. The 2102A was on sale, so I went for that one and saved almost 300$ off the store's list price for the model.
You could easily argue that the external VGA display with its limited resolution will detract from the capability of the scope, and certainly would not look as sharp as the smaller display built into the scope. But I think it will benefit me quite a bit and is certainly easy to hook up. VGA displays are still plentiful and cheap and the ones i have here seem very robust. It's an adventure ;D
-
Signal Scope looks nice. Is there an open source sound card based audio measurement project that is worth consideration? I haven't had much luck finding one, but then, it might be my googlefoo needs some tweaking. Maybe just a discussion of the different packages that are available would be interesting?
there are few here... https://sourceforge.net/directory/os:windows/?q=oscilloscope+sound+card
Haha, yeah, thanks! My googlefoo is at least that good. I was really hoping that people had some experience and that there was some gem that I hadn't heard about. There are a log of oscope programs, most are not that interesting.
-
Congrats Sylvi
looks like you have more than enough for your initial needs!
-
This always happens to me when I buy something expensive at Aliexpress:
I got my OWON scope on a 35% off sale.
Today it is 40% off - haha
It's still a great deal even at the list price considering the capability of the machine.
-
Hi Guys
The OWON XDS2102A arrived and it is really impressive! It is much bigger in person than the pictures suggest even though I watched OWONs videos which have the XDS3102A which is the same size.
DSOs are certainly very good at locking in a signal but I somehow got into a mode where the behaviour was distinctly "analog-y". The image ws shifting back and forth just like an untriggered CRT scope might do. Then the image was overlayed by an inverted form, again like the triggering was gone. I turned off the scope and turned it on again and I guess it went back to the factory settings? Anyway, it was working again like the first time I turned it on.
On my bench is an audio preamp with SMPS supplies; an AC-DC master supporting two DC-DC secondary converters. If I plug the sine generator into the input of the preamp with the preamp power off, the sine is clean. Turning the power on there is converter hash on the input jack. With the CRT scope, I never saw this hash even though it is significant in amplitude compared to the sine wave, and listening tests up till then did exhibit hiss. I'm sure a major part of the hiss is this hash. Poking around the supply rails the hash is everywhere - it is highest on the DC-link voltage between the converters. Fortunately this is all easily fixed and I'm glad to have the OWON to see these things clearly.
I have barely dipped my to into what the XD2102A can do. It's really fantastic and I recommend to anyone working on audio to consider the 12-bit scopes from OWON. There is a lag between live audio at the probe and what appears on the display, but I'm sure this can be minimised more than it is now, since I am still learning about the controls.
-
my Oldie is also good for audio analyzing, the spectrum analyzer you see in slot B :)
-
@Sylvi
I have barely dipped my to into what the XD2102A can do. It's really fantastic and I recommend to anyone working on audio to consider the 12-bit scopes from OWON. There is a lag between live audio at the probe and what appears on the display, but I'm sure this can be minimised more than it is now, since I am still learning about the controls.
The 12 bit resolution is impressive but there is a limit.
When you look in the manual at the technical specification of this model in chapter 8
you will see that 12 bit is only available by 500 MS/s Sample-Rate.
If you choose 1 GS/s Sample-rate you will have only 8bit available.
Have a look at this video (the title is wrong it is a XDS2102A):
https://www.youtube.com/watch?v=-wGB49mSAis (https://www.youtube.com/watch?v=-wGB49mSAis)
Pay attention when he switched to 12 bit the sample rate changes immediately to 500 MS/S.
-
Hi
Yes, I saw the vertical-bit connection to sample rate in the data sheet before I bought the scope. I also watched all the videos on OWON's site and could see how the memory depth, sample rate and bit-resolution ALL work together to show detail.
Overall, I went with the XDS2102A instead of the XDS3102A partly because I did not think the extra 50% sample rate or doubling of memory would be of benefit to me right now. Plus, I couldn't afford the XDS3102A and could not find one configured as the XDS2102A was. At some point when I learn more how to use the scope to its limit, I might regret that choice, but in the mean time I am enjoying what I do know how to do with it ;D
As I said above, the price went down a bit after I bought mine, then dropped further. I was pestering a friend of mine to buy one and he did. Usually he gets these new toys before me - haha
I love this site!
-
Sound cards are the way to go as no scope will have the sort of dynamic range and distortion.
I built my own sound card out of the best audio ADC that TI had to offer. It gives 192KHz 24bit as its often the case for modern sound cards. But it has excellent spur free noise performance down to -155dB and a measured THD of about 0,0005% (Its hard to tell really as THD measurements become really hard this low). Lots of software out there to make sound cards into useful instruments such as scopes or spectrum analyzers.
Its so sensitive that you can measure the 1/f noise of even pretty low noise opamps by simply connecting it to the opamps output.
A piece of real test gear that does a similar thing might be the HP 89410A (Vector Signal Analyzer DC-10 Mhz)
-
you will see that 12 bit is only available by 500 MS/s Sample-Rate.
That's enough for audio...
-
Hi Berni,
I'd be interested in the circuit diagram/firmware/ W10driver of your sound card to fully reproduce it.
Because this would give me full control over the frequency response. I.e. really flat from DC to 100 kHz.
The M-Audio Audiophile 192 I am using for measurements with the now discontinued Audio Tester software
is not flat enough above 70 kHz or so.
Ideally I would like to go even higher using a sampling rate of 384 kHz for filter measurement purposes.
-
It was built using a PCM4220 as the audio ADC. Very nice high performance chip in terms of noise and distortion. This was run by a XMOS XS1 microcontroller with the manufacturers asynchronous USB audio firmware that was modified slightly for my use. The analog front end to the ADC also consisted of some fancy expensive opamps to get the noise down and local LDOs with fancy specs feeding everything to keep interference out. The whole thing ran pretty darn warm as a result. Oh and it was also galvanicaly isolated to avoid ground loops on the analog side.
Using these special async drivers means that they don't just auto install on Windows (But work just fine on Mac/Linux) but it has a benefit of letting the sound card dictate the datarate so there is no resampling needed to keep the PC and the ADC in sync so you get bit perfect samples from the ADC getting to the PC.
Most sigma delta DACs can already be seen creating quantization noise even with 100KHz of bandwidth. The noise floor on them starts rapidly going up once you get out of the audio band.
-
Sounds interesting Berni. I have been thinking about something similar around AK5394A (although it has just been announced reaching EOL early next year). It's just the USB part I am not sure how to implement, and wouldn't want to dive too deep into. I have been thinking about using a ready-made USB-I2S module. And of course the analog front-end is not trivial either but I'm sure it is relatively easy to get at least somewhat excellent performance out. Squeezing out the last dBs will require lots of effort, time, and experimenting if willing to do that.
I have APx to use at work so with the help of that I'd like to develop some measurement capability to home as well. Currently only have a scope, and although it is a good scope, it has very little use in audio other than functionality tests.
-
You can just buy a board that does the USB part like this for example:
http://jlsounds.com/i2soverusb_n.html (http://jlsounds.com/i2soverusb_n.html)
You can probably find cheaper ones too, but this one looks pretty solid. You do want to be using asyncronous USB Audio 2.0 drivers as that grantees you get bit perfect samples from it. All the Xmos chip based stuff uses it so that's an easy way to know.
The usual drawbacks in terms of accuracy still apply tho. Audio ADCs don't have external reference inputs and the internal references are optimized for noise rather than accuracy. Due to this you have to take amplitude measurements with a grain of salt. Two ADC chips might not measure quite the same amplitude and it might measure slightly different when hot versus cold. But all of this should be within about 5% so at least the first digit is pretty confident. Anything else however is very precise, ratios between two tones in a signal will be bang on due to the ruler flat frequency response. Frequency is driven from whatever oscillator you hook up so you can get it as accurate as you want in terms of time. And the usual warnings on the input apply, most nice input front ends will not handle large voltages sent into them so you have to make sure to be careful what you probe or add some input protection(That hopefully doesn't degrade the performance too much)
-
12 bit is only available by 500 MS/s Sample-Rate.
"Only" :). 12bit at such is very-very good for this sample rate. Even getting 10bits of actual resolution (ENOB) at such sample rate is a challenge, let alone 12 bits. I'd be happy even if it has full 8 bit enob at maximum sample rate and frequency.
I'd take resolution specifications with a grain of salt, unless proven otherwise.
-
Hi Guys
I forgot to mention that I only use one input most of the time and the (OWON XDS2102A) display indicates 12-bit resolution. There are very times I need two inputs, so all the 4-ch scopes were of no interest to me. This "trick" is mentioned right on the data sheet.
The discussion about sound cards is interesting. I thought for the best signal-to-noise and lowest-distortion you would want to use a differential circuit? Or is that just for DAC?
I also read recently about 32-bit ADCs but there was a question if there was any real benefit over 24-bit given all the considerations of file formats, communication, front-end noise etc.at least with respect to audio playback. Maybe for instrumentation use things are different?
How can you roll your own ADC these days with everything being surface-mount parts? My eyes have trouble with too-tight through-hole assemblies - I might go nuts if I had to try SM :o
Also, when you get to super-low-noise, grounding has to be perfect, ground planes and guards are needed, a super-quiet supply is required - easy to do linearly - and even cleaning the board is critical.
One of the reasons I wanted the DSO was to see the noise of the SMPS modules I'm using for a preamp project. My CRT scope did not show this but it is blazing with the DSO! I'm at the point where I'm going to make a separate PSU board to experiment with to keep the cost down (right now, everything is on one large PCB, so expensive to make changes). I would like to get the output of the switching supply cleaned up enough before it hits the discrete post regulators that the latter does not need a lot of bandaids like ferrite beads and high-value ceramic caps.
-
12-bit (if it is) is great for a scope and plenty for many applications, unfortunately audio is not one of those. You need 20 bits or more to do proper performance measurements. Less is enough for power amps.
Regarding the DIY "sound cards", they can be made differential. All top end ADC ICs are differential input anyway.
32-bits is more marketing thing. Even getting to 24-bit is very difficult with audio signal levels. Noise has physical limits, you cannot go below certain level as thermal noise sets the limits. Therefore, to get more effective bits and higher dynamic range, the only way is to increase the maximum level. But having even 24 bit dynamic range with 10V signal level means little if your practical signal level then is 1V. Of course, if all your equipment is DIY, you can choose to have higher signal levels throughout the system and may be able to keep higher dynamic range.
You get used to surface mount, and they are essential in getting the best performance, especially when you have digital stuff as well. If you don't choose the smallest possible, they are not difficult to work with. IMO 0805 size is quite good compromise. Some new ICs can be tricky as there may not be easy packages available anymore.
-
You need 20 bits or more to do proper performance measurements.
00
How much more? :) I'll be surprised if best equipment would be able to squeeze even 20 bit. You can build ADC with any resolution, but noise floor and non-linearity will make it useless. A 20bit 3.3V ADC would have a ~3uV resolution, that's virtually unmeasurable for a sound ADC.
I'd say anything better that 17 ENOB is really good equipment.
-
20 bits is 120 dB corresponding dynamic range. Nothing unseen in audio. Not that every piece of kit would be like that, but if we are talking about a measurement instrument to benchmark other kit, at least I would like the performance to be a step or two better.
Ok yes you can do with less but if you'd like to measure very high quality DACs for example, that's what you need.
-
Honestly, you shouldn't even be using a scope for precise measurements of audio level, noise or distortion; there are far more accurate tools available which are purpose-designed.
And as far as resolving DAC or even linear artifacts on a scope, 20 bits is way overkill. Even in a high quality signal from any real world analog source, the content of the bottom few bits of a 20-bit measuring device are going to be all noise and useless for benchmarking. The whole 20/24 bit discussion is more about the end user feeling good than showing any provable audible benefit.
-
Hi Sylvi,
yes the eevblog is a very great site.
Also Daves videos are very interesting.
I am also want to buy the xds2102a and searching now the web about reviews and tests.
At this time there existst only 2 of them yet. Unfortunately they shown only basics.
The model is currently very young (it first came on the market on Aug 2018).
So there are not very much profound tests available.
The most things i want to know on new models are bugs in software.
You said something about that.
The second thing is the low voltage handling.
Most of them are struggling when they must display signals in the mV-Area.
Some of them are very slow when the must display Lissajous (x-y Mode).
The most interesting thing is the FFT and the quality of them.
I know that it can handle only a bandwith with maximum of the nyqist frequency.
That must be on this model 50 Mhz.
Hope you can make some tests and report about it.
I know its a bit offTopic but maybe we can connect these two things together. ;-)
regards
-
Hi James38
I haven't found a problem triggering on low signals provided they are clean. I have two prototype preamps, one with a linear PSU and the other with SMPS. The linear unit is clean throughout. When I turn on the SMPS unit, I see noise impressed on the input signal! For this, triggering is unstable because the spikes are randomly distributed all over the sine wave input signal.
I bought my OWON XDS2102A from Hilitand Brand Store on Aliexpress. It was on sale and went on a further discount afterwards. If you work with tube stuff, get some x100 probes, too - the x10 gets you +/-100V where x100 allows 1kV points to be monitored ;D
I have not learned too much about the features of the scope yet but really enjoy it. It is such a pleasure to use and is not toy-like at all, which turned me away from Rigol, Hantek, Uni-T and others.