Need educated model/feature advice for oscilloscope for doing high-quality audio

[Franki]

hi folks. I'm new here.

I used to use an old Hameg 20MHz from the early 80s for years, but I need something better now.

I want to start building audio devices with much better quality: headphone amps(for headphones like Beyer DT770, AKG K701), power amps(active ones for home, studio monitors or multichannel for home theaters), active filters, mixer consoles etc... for audio sources like cd-audio/aac/vorbis up to uncompressed 32bit 192kHz uncompressed PCM that is supposed to sound like that.

AFAIAAO I need a DSO/DPO scope with at least 2 channels, indep, +, -, x, / operations, fft spectrum analyzer(at least 1024 band), continuous waveform and single shot mode. This scope should have a very low noise floor(I don't like most DSOs because of their noisy bucket brigade memory), very low distortions and have a high dynamic resolution(8bit won't be enough, if that's the actual sample dynamic resolution) so as to easily identify different small noises and their profiles(designing audio equipment of that grade requires minimizing very small noises and other non-linearities). Should be real-time sampling only. Important: To minimize noise, this scope should support hi-res multisampling in various modes. Of course a good HF-shielded box and connectors are valuable too. Also, it should be able to compute average and effective/rms values(per full view and single cycle). It should have a storage of at least some 100kSa. USB/RS232 connectivity is a must.

AFAIK I don't need excessively high bandwidth, 20MHz is more that I need. Input capacity, either of 10pF or 100pF is not an issue with audio circuits.

And it should be available for purchase in Germany. Absolute budget limit is 1000€ (~ US$ 1350).

I'm still undecided as to whether use a DSO or DPO(if there is such a thing whithin my price range) scope, as my experience is limited to analog scopes.

THX in advance

[EEVblog]

Check out the 12 bit Picoscope's:
http://www.picotech.com/oscilloscope-specifications.html

Dave.

[alm]

High resolution and measuring low distortion aren't typically strengths of a DSO, almost all of them have an 8-bit ADC, although some can get more resolution by averaging. No idea about the Picoscope, usually I think their products are overpriced and under performing, but this may be an exception. An analog scope with a fine trace and differential inputs is the usual suggestion in this case, but you won't find them new. Something like a dedicated audio spectrum analyzer or distortion meter may be a better tool for the job, I'm not sure if any DSO will satisfy your requirements. A scope typically has about 1-3% tolerance (not sure if they publish any distortion specs), so I'm not sure how useful this is for measuring .01% distortion. A differential high-gain pre-amplifier (like the Tek AM502) may be helpful for measuring noise, but it's not available new either. I think Lecroy and Preamble has/had something similar, but it's probably expensive.

[EEVblog]

Pretty much only a PC based DSO will get you the performance you need, one that sacrifices bandwidht for resolution and noise performance.
This Picoscope is just in the budget and is specced for low level signal measurement:
http://www.picotech.com/precision-oscilloscopes.html
1% accuracy, but no noise spec.

Dave.

[Franki]

Pretty much only a PC based DSO will get you the performance you need, one that sacrifices bandwidht for resolution and noise performance.

Why do you think that only a PC-based DSO may reach such requirements? Is there a technical reason for that, or are non-PC based ones not demanded enough for being considered to be built?

On the other hand, I would have no problem with a PC-based scope, really, any kind of interface/connector from RS-232, RS-485, USB up to version 2.0, ethernet up to gigabit would work, as long as it doesn't exercise poor signal quality and is usable under non-Windows OSes, like FreeBSD, Linux, L4...

I forgot some more requirements: It must be at least partially-portable, meaning it should be small enough to be carried in a small back-pack or regular suit-case, but it's okay if it always needs power connection, though having a rechargeable battery or 12V connector are nice-to-have. And it must have an externally triggerable.


The thing with PC-based scope is this: They don't just work alone, so I would have to carry my laptop besides the scope. Laptop weights 5kg by itself. Possible, but ugly.

I just had a glance on the PicoScope 4224 you suggested. Good specs, fulfills all requirements, except that it only runs on Windows and doesn't work on its own...FAIL

Here's another one with promising specs: AEMC OX7102-C unfortunately, I have no clue how reliable and trustworthy that company AEMC is...haven't heard of them yet, anyone an idea? And I don't know whether they ship to Germany, and besides, that device is a little too much overpriced

[NiHaoMike]

The base model of the Agilent Dave just reviewed is just within your budget and would probably offer the most features. It even has a high resolution mode that takes advantage of averaging to increase resolution to 12 bits at low speeds. (Forget about more than 8 bits at high speeds with your budget.)

20MHz is a little on the low side for some digital amplifiers and definitely way too low for digital audio. Go for at least 50MHz. Even that's a little low since I2S can reach about 12MHz.

You'll also need an isolated probe if you'll be working on the power supply or BTL outputs. (Using two channels and difference mode might work in a pinch on BTL outputs.) You might also need to monitor the drive signals to high side MOSFETs, another place where an isolated probe is the only proper way to do it.

[alm]

Why do you think that only a PC-based DSO may reach such requirements? Is there a technical reason for that, or are non-PC based ones not demanded enough for being considered to be built?

My guess is that because they suck compared to a stand-alone DSO for 99% of the work, they try to focus on niche markets.

On the other hand, I would have no problem with a PC-based scope, really, any kind of interface/connector from RS-232, RS-485, USB up to version 2.0, ethernet up to gigabit would work,

USB 2.0 is basically the only choice until you get to the $$$ stuff like lXI.

as long as it doesn't exercise poor signal quality

Sample rate and attenuator are usually worse. Not sure how critical this is for your use.

and is usable under non-Windows OSes, like FreeBSD, Linux, L4...

Forget this. If you're lucky, you may be able to figure out the protocol (either by asking or reverse engineering) and write your own.

[EEVblog]

Pretty much only a PC based DSO will get you the performance you need, one that sacrifices bandwidht for resolution and noise performance.

Why do you think that only a PC-based DSO may reach such requirements? Is there a technical reason for that, or are non-PC based ones not demanded enough for being considered to be built?

You tell me, I don't know what your exact requirements are.
You said that a regular scope may not do the job because of it's 8 bit converter?
To get greater than an 8bit converter your only choice is basically a PC based scope, or maybe one of the handheld scope might do greater than 8bit.
Almost every general purpose bench scope on the market uses an 8bit converter. Sure some like the new Agilent have a 12bit averaging mode, but it ain't the same as havign a real 12 or 16 bit converter with a low noise floor.
I presume you wanted to capture data and do decent THD analysis or something?

Dave.

[EEVblog]

The thing with PC-based scope is this: They don't just work alone, so I would have to carry my laptop besides the scope. Laptop weights 5kg by itself.

A laptop from the 1980's maybe.

Get a cheap Windows netbook to go with it then. Small and cheap, and battery lives of 10hrs are common. Nothing like 5kg.
You can likely get a good USB scope and netbook combined within your budget.

Dave.

[Zero999]

How about simply buying a good sound card and using that?

As you say, you don't need the bandwidth but the resolution is important so a good sound card and software aught do the job.

[NiHaoMike]

The best sound card does 192kHz sample rate at 24 bits with an analog bandwidth of about 20kHz. That would work fine for line level audio, but you won't be able to see the carrier transitions in a digital amplifier.

For a good analyzer software, try http://baudline.com/ .

[alm]

The best sound card does 192kHz sample rate at 24 bits with an analog bandwidth of about 20kHz. That would work fine for line level audio, but you won't be able to see the carrier transitions in a digital amplifier.

Keep in mind that although the ADC produces 24-bits, it usually has nowhere near 24 effective bits. I wouldn't expect any accuracy either (who cares about 3dB change in volume, it's barely perceptible and volume controls are closed loop anyway (too loud? just turn it down). The proper tool for this job would probably be something by companies like Audio Precision.

[Zero999]

I've never seen the point of a 192kHz sample rate for audio, it's just overkill, anything above is more than good enough and anything above 64kHz is probably imperceivable, even to someone with the best hearing.

[saturation]

Now and then, the Acer Aspire One netbook appears for $200 or less, fully configured.  I bought one some months ago as my preferred travel PC and use it to run lab stuff when I'm not on the road: for data collection or a USB instrument.

They come in different battery life lengths.  Weighs ~ 1.2 kg with 8 hr battery, gives practically 6 hour continuous battery life surfing the web [my timing, not the manufacturers rated 8 hours], but minimum 10 hours, up to 15 hours when working as a PC alone [off/on web access.]  

Smaller footprint fits limited bench space nicely.




Win 7 starter works well, I'm able to multithread some apps that support such very nicely, such that it runs faster than the same apps on Vista or XP, such as z.zip.

The thing with PC-based scope is this: They don't just work alone, so I would have to carry my laptop besides the scope. Laptop weights 5kg by itself.

A laptop from the 1980's maybe.

Get a cheap Windows netbook to go with it then. Small and cheap, and battery lives of 10hrs are common. Nothing like 5kg.
You can likely get a good USB scope and netbook combined within your budget.

Dave.

[Franki]

The base model of the Agilent Dave just reviewed is just within your budget and would probably offer the most features. It even has a high resolution mode that takes advantage of averaging to increase resolution to 12 bits at low speeds. (Forget about more than 8 bits at high speeds with your budget.)

I hope the preamp stage of the Agilent stands up to its 12 bits in hi-res mode in terms of noise and thd.

20MHz is a little on the low side for some digital amplifiers and definitely way too low for digital audio. Go for at least 50MHz. Even that's a little low since I2S can reach about 12MHz.

I'm not using that device on digital amplifiers, audio or other pwm stuff, since I have a dedicated 8 channel logic-analyzer.

On the other hand, a 20MHz scope may be more than enough for usual audio, but what if I'm using the hi-res mode that is averaging adjacend samples to increase the dynamic resolution and to decrease the noise, than 20MHz bandwidth may turn out to be just 1.25MHz or 78.125kHz of bandwidth which is definitely not enough.

So, if there is no scope with native 12bit dynamic resolution in my price range, than I end up of needing a scope with a hi-res mode and higher bandwidth.

You'll also need an isolated probe if you'll be working on the power supply or BTL outputs. (Using two channels and difference mode might work in a pinch on BTL outputs.) You might also need to monitor the drive signals to high side MOSFETs, another place where an isolated probe is the only proper way to do it.

of course

[Franki]

as long as it doesn't exercise poor signal quality

Sample rate and attenuator are usually worse. Not sure how critical this is for your use.

That's something I've heard in other places too, especially for the low/mid price range

and is usable under non-Windows OSes, like FreeBSD, Linux, L4...

Forget this. If you're lucky, you may be able to figure out the protocol (either by asking or reverse engineering) and write your own.

I can't forget about it, unless it would have all the required features on its own. I'm a little experienced in reverse engineering digital protocols, but I really don't have the nerves to do a full USB configuration reverse engineering. Really, doing this with USB really sucks.

[Franki]

Almost every general purpose bench scope on the market uses an 8bit converter. Sure some like the new Agilent have a 12bit averaging mode, but it ain't the same as havign a real 12 or 16 bit converter with a low noise floor.

This is a bit of a sad situation, as my intuition tells me that there are more people that want more than 8 bits of dynamic range. Hopefully the Agilent's 12 bit averarging mode stands up to its name and is not fully covered by its input stage noise or the like.

Also, as you already have an Agilent, can you tell me if this is the case?
I'm also interested to know whether the averaging mode of the Agilent averages adjacent samples or samples of different cycles (sometimes called multisampling)?

Thanks in advance

I presume you wanted to capture data and do decent THD analysis or something?

exactly. thd, frequency response and noise analysis on inputs vs outputs, power supply, adjacent channels and on individual components at different test waves (first 100Hz, 1kHz, 10kHz broad and then finer on sine, square waves), then making frequency response analyzes (sweeps, white - pink noise) this is why I need differential calculations, rms/effective value calculation and a fft spectrum analyzer is real handy for that too.

[Franki]

The best sound card does 192kHz sample rate at 24 bits with an analog bandwidth of about 20kHz. That would work fine for line level audio, but you won't be able to see the carrier transitions in a digital amplifier.

For a good analyzer software, try http://baudline.com/ .

of course, but finding a laptop with such a high quality sound card isn't easy. Plus I really doubt they deliver the full 24 bit of accuracy, but even 16bit would be enough.

I have such a sound card in my PC, but in a laptop you basically need an onboard sound card(since most laptops don't have expansion card slots), and from my experience, many onboard sound chips really suck when it comes to being immune to noise from digital signals especially on the supply lines, even the more expensive ones. You have all kinds of whirring, and you can't just filter it out with a ferrite choke or metal shielding since it is on the supply line, and you can't just easily put an auxillary power supply(that is supposed to be mobile) just for the analog amps on your onboard sound card.

So, in conclusion, unless you have a good quality sound card specifically immune to noise, foremost HF noise, on the supply line or with an additional power supply connector, which are most likely found on dedicated ATX style PC expansion slot sound cards, your sound card is useless for doing audio oscilloscopy and actually useless for doing any serious audio recording.

It's not that the internals of digital scopes don't produce HF noise, but a) the manufacturer knows exactly what's in the scope, so he may exactly know what kind of noise there is and can account for required shielding, HF-suppression etc... where he wants and how he wants, whereas in a PC, no manufacturer can really know this, and most components aren't shielded inside of the PC casing and b) you probably don't have as big HF power variation as in a PC (hard drives, disc drives, current hungry cpus and graphic cards with maybe the exceptions of fans), and even if, the manufacturer can account for that and put adequate chokes, capacitors and other HF voltage stabilizing in the power line.

[Franki]

I've never seen the point of a 192kHz sample rate for audio, it's just overkill, anything above is more than good enough and anything above 64kHz is probably imperceivable, even to someone with the best hearing.

Yeah, the 192kHz audio sample rate, you really just need it for intermediate processing, especially non-linear computations/effects like speed changes, convolvers and other fancy stuff to not degrade quality

[Franki]

You might also need to monitor the drive signals to high side MOSFETs, another place where an isolated probe is the only proper way to do it.

This is what I'm mostly concerned of. I still have active probes from my old hameg 20MHz from the early 80 ies, but I have no clue whether they would be compatible to my new scope.

If not, how comes that there aren't any active probes for the Agilent 2000 series on their site http://www.home.agilent.com/agilent/product.jspx?nid=-536902770.0.00 just 1000, 3000, 5000 etc... series

[deephaven]

You could use a USB external audio box like this one http://www.m-audio.com/products/en_us/MobilePre.html which has professional XLR inputs, has the advantage that it's external so won't pick up computer noise, and can be freely moved from one PC to another. Then use some analysing software to do your FFT, distortion analysis etc.


Nigel.

[jahonen]

I second the soundcard approach. A scope is an all-round-instrument, it measures pretty much anything but is not particularly accurate or sensitive on almost any frontier. What OP is trying to approach, can only be achieved using an audio analyzer (or good quality soundcard). Haven't actually checked how good results one can get with Agilent 6000-series, but I'd prefer the soundcard if one wants to make distortion measurements for much less than 1% THD.

Regards,
Janne

[Franki]

I second the soundcard approach. A scope is an all-round-instrument, it measures pretty much anything but is not particularly accurate or sensitive on almost any frontier. What OP is trying to approach, can only be achieved using an audio analyzer (or good quality soundcard). Haven't actually checked how good results one can get with Agilent 6000-series, but I'd prefer the soundcard if one wants to make distortion measurements for much less than 1% THD.

Regards,
Janne

You're right, I also thought of using an audio analyzer, but itend to not only use it as an input/output comparator, but also to test how output power or incoming HF-noise on the powerline affects audio equipment on the power line. And some minor other test.

Bottom line is, I not only want to know how much noise or THD something has, but what kind of distortion this is (which harmonics and what does the distortion look like) in order to identify where the problem is: bad capacitors/transistors, too much amplification, too much/few filtering of some frequency etc... And the problem with most audio analyzer in the price range up to 1500$ is that they don't have the visualization capabilites a good scope has.

Most usable Audio Analyzers are just too expensive, for example the Agilent 8903B, its successor U8903A or the Prism Sound dScope Series III, price is around 10.000-20.000US$ and much too high
.

[deephaven]

How about this: http://tinyurl.com/nxndzk

This shows what it can do: http://tinyurl.com/atcylj

[Franki]

How about this: http://tinyurl.com/nxndzk

This shows what it can do: http://tinyurl.com/atcylj

THX, good equipment, but out of budget by an order of magnitude (quick search reveals 10000-20000US$ new, maybe a tenth for a used one).

I'm thinking of buying a high-def external USB/IEEE1394 sound card with external power and known good HF-noise and power ripple shielding/filtering. I just need to search some more to find some models, if they do excist.