Agreed about the application note. Some crucial information seemed missing. What sample rate was used with sima-delta converter, and what was the rise time of the test signal generator, for a couple of questions?
If one were trying to perform an accurate measurement of a system with energy present right around the the A/D Nyquist frequency, then of course a sharp filter could reasonably be expected to ring. And if a linear phase filter, there could be pre-ringing.
But if one were sampling at 24/192, and knew there was not significant energy around 96 kHz, then there shouldn't be much excitation to produce significant problems with ringing. And if one did see tale-tale ringing of significance, that would suggest looking at the system with another wider bandwidth instrument to investigate.
Spent a few minutes this morning and lo and behold someone has compiled the PortAudio (ASIO too claimed) .dll with Python bindings as well as a Python library for using it in scripts. I will get onto it. Claims simply dropping everything in the right directories just works, we'll see about that.
Hi Jan,
They start at $16,872.00 With the 1.5 MHz bandwidth ($3,894.00) and digital audio ($5,818.00), which would be approximately $26.584 Canadian. A quick call to Keysight got me a rough quote at $20,067 as I configured it. That's a pretty decent price for something that goes from DC to light.
You might have configured it differently. They come with Bluetooth, 4 and 8 channel models and software programs for analysis. Man, I would love to have one of these units. I can't remember what the frequency response is for the standard frequency response, but it's only a license, which means the top end response would have much error at all. The 1.5 MHz option likely has hardware that goes considerably higher if I know Keysight at all.
-Chris
They start at $16,872.00 With the 1.5 MHz bandwidth ($3,894.00) and digital audio ($5,818.00), which would be approximately $26.584 Canadian. A quick call to Keysight got me a rough quote at $20,067 as I configured it. That's a pretty decent price for something that goes from DC to light.
You might have configured it differently. They come with Bluetooth, 4 and 8 channel models and software programs for analysis. Man, I would love to have one of these units. I can't remember what the frequency response is for the standard frequency response, but it's only a license, which means the top end response would have much error at all. The 1.5 MHz option likely has hardware that goes considerably higher if I know Keysight at all.
-Chris
You can compile the dll library yourself as well. It's wirtten in C. Just download the ASIO SDK.
Go to the Portaudio web site. The instructions for building the lib are there and easy to follow. The example I mentioned is useful to see what the asio and other win drivers will do.
I suspect some of the problems you ran into are because the driver are limited.
Scott i found the problem. You need one of these.
"This is because their compilers link against msvcrt80.dll, msvcrt90.dll, msvcrt100.dll, msvcrt110.dll, msvcrt120.dll and msvcrt140.dll respectively, which are not shipped with windows.".
C++ redistributable. You can download if of the web and include in the .exe directory where the PortAudio app is located.
"This is because their compilers link against msvcrt80.dll, msvcrt90.dll, msvcrt100.dll, msvcrt110.dll, msvcrt120.dll and msvcrt140.dll respectively, which are not shipped with windows.".
C++ redistributable. You can download if of the web and include in the .exe directory where the PortAudio app is located.
"Sigh", this becomes like work again, besides PortAudio programming has it's own learning curve and it only gets better
. You can download all the manuals for the U8903B. It has two digitizers and a DAC based source module. The 1.5 MHz ADC is not "audio performance". It doesn't have an analog source or analog notch filters at least from the info in the manual. Its more like a good DAC + ADC with local processing.
It looks like it is really focused on high volume production of things like Bluetooth headsets and cell phones. It does not push the envelope on performance.
If you are outfitting a production line you would get several of these and they are cheap compared to stuff like the Bluetooth production test system which really does very little audio. They don't have speaker measurement stuff in it but its only software.
It's not like that Chris. It's the responsibility of the vendor to package required files.
Not that I'm a vendor. Generally an installer would do this.
Hi Demian,
I may not have optioned it out properly then. I'm pretty sure it has an audio source, or they might call it an "tracking generator" or some such thing. Now that I have them all excited after giving me a price, I expect they will be contacting me and I can ask them directly. I know it has filters that is implemented in DSP as they are license driven. Isn't everything these days?
I remember looking at the specs for the 8903A/B and not being impressed. Later (here) I find out that the box is 20 dB better than the spec would have you to believe. Had I known that, I would have bought an 8903B instead of the 339A that I have. Live and learn.
My recent experience with Keysite would lead me to believe that the equipment performs much better than the specs might suggest. Once I'm set up again, I'll request a demo of this new thing and really see where it's performance lies. For me, audio extends to 1 MHz, and I want to know what is up there. Older HP equipment would "see" that high and I have no doubt the current product does as well.
I won't be quick to judge this instrument until I see it in the flesh.
-Chris
I may not have optioned it out properly then. I'm pretty sure it has an audio source, or they might call it an "tracking generator" or some such thing. Now that I have them all excited after giving me a price, I expect they will be contacting me and I can ask them directly. I know it has filters that is implemented in DSP as they are license driven. Isn't everything these days?
I remember looking at the specs for the 8903A/B and not being impressed. Later (here) I find out that the box is 20 dB better than the spec would have you to believe. Had I known that, I would have bought an 8903B instead of the 339A that I have. Live and learn.
My recent experience with Keysite would lead me to believe that the equipment performs much better than the specs might suggest. Once I'm set up again, I'll request a demo of this new thing and really see where it's performance lies. For me, audio extends to 1 MHz, and I want to know what is up there. Older HP equipment would "see" that high and I have no doubt the current product does as well.
I won't be quick to judge this instrument until I see it in the flesh.
-Chris
Thanks that's what it asked for, as I've learned PortAudio is its own programming environment. It's not like you can drop it into your path and just say play() or record(). You can not avoid lots of work.
Just tried it on my laptop with the very simple built-in sound card. STDOUT produced 215 lines of output, and STDERR produced 686 lines of output. Seems like a lot of reading to figure out which drivers are there.
Well yes. It's not app, It's an API. We have to write our own code from it.
Lots of work ahead.
The hardest part about coding is trying to understand someone else's code.
I found a neat too called Dependency Walker. Load the exe in question and it shows all the dll calls.
Last edited:
I double checked tonight changing nothing and putting a 5kHz 5mV squarewave into both channels of my sound device only ARTA got the right answer. Ocenaudio is out of consideration. Octave gave the MME answer too ask for 24 bits you get garbage.
For the time being I'm playing dumb, I download the latest version of a package's release and execute the commands. Octave was particularly amusing ask for 16 bit samples at 96k you get a nice 22.05K filtered square wave at the right level. Ask for 24 bits and you get a 2V square wave at some random frequency (garbage).
ARTA is the only package that gave a 48kHz filtered square wave at 24 bit samples.
Chris, Live and learn AND live and learn.
Don't feel bad, because our departed friend Richard Moore did an analysis
and found the 8903 had some issues and wasn't too impressed with it's results.
I cannot say what the real issue was, it might be in his Moorepages or it might
be regarding the 8903 versus the 339A in the shorter Hp3389a thread.
It had something to do with the Microprocessor being right in the
middle of things which not only made it not so simple to make better
but also compromised its measurements.
RichEEM was much more enthusiastic about the HP339a.
Was it hardware versus software bias? I don't know.
Cheers,