Choice of new Distortion Analyzer

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
1701 from Sound Technology was my tool at work.

In RF electronics, a lot of developments from individuals or littles companies conduct to a rich offer of low cost testing apparatus. Beside bigs (agilent...), enlightened enthusiasts can make precision testing.

In audio domain, I may be biased but close to bigs I only find use of souncards or assimilated.

Is Computer noisy environnement not altering measures, or limit precision floor ?
 
The measurement instrument is outside of the PC. QA401 is a USB soundcard too, only using a different communication protocol (which does not make it better suited for measurements) and equipped with a frontend aimed at measurement needs (which makes it the justified choice of many).
 
The measurement instrument is outside of the PC.

Noise is radiated all around computer :(

QA401 is a USB soundcard too, only using a different communication protocol (which does not make it better suited for measurements) and equipped with a frontend aimed at measurement needs (which makes it the justified choice of many).
.
soundcard are not dedicated for testing but without low cost alternative... it may help.
Pro: QA401 have attenuators/protection integrated comparing to standard soundcard.
Cons: if different protocol, may not accept others softwares ?
 
Is Computer noisy environnement not altering measures, or limit precision floor ?

The Audio Precision APx-555 is controlled by a user supplied Windows PC using a USB cable. So, at the very least, the "ground cooties" of a general purpose computer attached to a display and a USB link are not fatal to high precision analog.

Still, yes - an APx analyzer is not a PCI card in a generic PC backplane, but still, it is not all that far from that either. I think the important part is to try to separate the analog and signal related circuitry from the computer part of the system. If there's any way for the system to have electronics that use balanced signaling, probably powered by some sort of external box, then this might be a great advantage in trying to escape random computer hash.
 
IME that random computer hash is in most cases no noise radiation, but a plain ground loop - huge CPU/chipset currents (<1V => tens of amps) running somewhere through the motherboard ground along the path between the point from which PC output ground is taken (USB port ground, PCI(e) ground, internal soundcard ground for low-spec requirements) and the PC PSU to which the PE wire is connected. Both desktops and notebooks (mostly isolation class I these days) are affected by that. As you say - isolation or balanced connections fix that.

IMO the "great-sound-ready" motherboards with better quality onboard codecs have this issue taken care of in their ground layout design.
 
My DUT location on a wooden bench is 2ft from a PC and 1ft from multi-LED desk lamp and a monitor. The only spectrum spikes I notice up to 100kHz are from the LED lamp, so I turn that off when I want to capture a relatively clean spectrum plot.

I use aluminium foil BBQ trays to sit my external USB soundcard and battery in when I need to suppress anything and everything, and can use a 'slow' USB isolator for extra reduction in mains earth loop with mains powered DUT, although I could bring a laptop in if needed. Although my noise floor ain't AP level, it can be averaged down pretty far if I need to expose certain harmonics.

But it is worth saying that a spectrum plot can include a noticeable amount of spurious signals but not interfere with the DUT assessment, as the brain acts as a pretty good filter. Some people seem to overly obsess about requiring a 'clean sheet'. In an RF analogy its like doing OATS testing and living with FM radio stations.
 
But it is worth saying that a spectrum plot can include a noticeable amount of spurious signals but not interfere with the DUT assessment, as the brain acts as a pretty good filter.

Marking the harmonics by the measurement software helps a lot in this case, especially if the spurious signals are close to their frequencies. In addition it helps to choose a fundamental frequency a bit apart from the spurious (typically 1kHz multiples for USB-based instruments).
 

Attachments

  • VD-OP1656.png
    VD-OP1656.png
    120.2 KB · Views: 245
AX tech editor
Joined 2002
Paid Member
The Audio Precision APx-555 is controlled by a user supplied Windows PC using a USB cable. So, at the very least, the "ground cooties" of a general purpose computer attached to a display and a USB link are not fatal to high precision analog.

My 2722 also is USB controlled. I tested it with a USB isolator, which cuts any galvanic connection between the PC and the AP. I was not able to detect any differences even at the highest sensitivity.

I've had more noise and spuriae from well-regarded BNC test cables than from the system.

Jan
 
My 2722 also is USB controlled. I tested it with a USB isolator, which cuts any galvanic connection between the PC and the AP. I was not able to detect any differences even at the highest sensitivity.

I've had more noise and spuriae from well-regarded BNC test cables than from the system.

Jan

For what it is worth, although the 2722 betters the UPV-B1 by ~ 2dB THD+N @ 1KHz, @ 100 HZ our UPV beats the 2722. The reason I mention this is because the UPV is a standalone unit, with integral PC. If I send the oscillator from the 2722 to the UPV, it gives the same reading - suggesting its analyser is as good as the 2722 ( I haven't had a chance to compare if this is the case at all frequencies, but it definitely is at 100 / 1KHz)
 
That shows that involving a computer (UVP contains an x86 motherboard with custom PCI connectors for the peripherals and an ATX PSU) does not mean the performance will be flawed. I would dare to say it is the opposite - a proper combination of hardware and software can offer performance either exceeding that of an analog measurement device, or on par but with lower cost.
 
Remarks on noise was generics : when measuring very low level signals, intrinsec or environnement noise can limit floor level.
If you like a stand-alone solution a used Amber 3501a could fill the bill. Search on this forum and the net for lots of info. E
Thanks for recalling me this model. I've already heard good feedbacks about it. Seems that problems identified today are recap and... find model with options ;)
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.