Hi,
There must be DIYers out there who have used a spectrum analsyer for audio measurements, so for you guys i have a question.
How does one overcome the interface troubles of a node in an amplifier, could be of a vastly different impedance, driving the extremely vulnerable analsyer 50R input? Max 30dBm.
I am thinking either an audio transformer, or less appealing to me, a low distortion op-amp with ac coupling.
Thanks!
Craig
There must be DIYers out there who have used a spectrum analsyer for audio measurements, so for you guys i have a question.
How does one overcome the interface troubles of a node in an amplifier, could be of a vastly different impedance, driving the extremely vulnerable analsyer 50R input? Max 30dBm.
I am thinking either an audio transformer, or less appealing to me, a low distortion op-amp with ac coupling.
Thanks!
Craig
Last edited:
Resistor L pad, easy, linear and does not need power.
Actually, for audio work a PC with a good soundcard often makes a better SA then a real one which often lacks perforance at such a low frequency range, plus the soundcard approach can do things like plot group delay and phase shift which a scalar analyser cannot.
Regards, Dan.
Actually, for audio work a PC with a good soundcard often makes a better SA then a real one which often lacks perforance at such a low frequency range, plus the soundcard approach can do things like plot group delay and phase shift which a scalar analyser cannot.
Regards, Dan.
Hi dan, thanks very much.
A quick google points to this being a resistive divider basically, is that right?
If i wanted to measure a high impedance node for distortion, i guess this is still tricky with a 50R input.
I have never actually made a distortion measurement before so maybe people always select low impedance points.
Cheers
Craig
A quick google points to this being a resistive divider basically, is that right?
If i wanted to measure a high impedance node for distortion, i guess this is still tricky with a 50R input.
I have never actually made a distortion measurement before so maybe people always select low impedance points.
Cheers
Craig
What analyser do you have?
Most of the ones that go down low enough for audio work have a high Z input option.
You will find that an RF spectrum analyser is the wrong tool for audio distortion, dynamic range is usually not that great, and LO phase noise is often awful due to the very wide tuning range required.
Do note that you only need about -20dBm at that 50 ohm input, so the input resistor can be quite large sometimes, also note that some spectrum analysers will suffer damage if you put DC on the input.
Regards, Dan.
Most of the ones that go down low enough for audio work have a high Z input option.
You will find that an RF spectrum analyser is the wrong tool for audio distortion, dynamic range is usually not that great, and LO phase noise is often awful due to the very wide tuning range required.
Do note that you only need about -20dBm at that 50 ohm input, so the input resistor can be quite large sometimes, also note that some spectrum analysers will suffer damage if you put DC on the input.
Regards, Dan.
Hi,
I am using a R&S analyser, 20Hz to something like 30GHz. I dont have the model number to hand at the moment.
Point taken on AC coupling and amplitude limitation, if i blow it I bought it, and that would bankrupt me!
The input will tolerate a power of -30dBm with zero DC component and is only an RF 50R terminated input unfortunately, no high Z input.
I recall reading dynamic range was something ridiculous like 160dB, if I am correct then this is plenty for audio, although I dont recall the phase noise of the LO.
Is there a good way to measure from highZ nodes?
This may be a silly question, as it may turn out that all circuits I measure have a lowish impedance output due to feedback.
Cheers.
I am using a R&S analyser, 20Hz to something like 30GHz. I dont have the model number to hand at the moment.
Point taken on AC coupling and amplitude limitation, if i blow it I bought it, and that would bankrupt me!
The input will tolerate a power of -30dBm with zero DC component and is only an RF 50R terminated input unfortunately, no high Z input.
I recall reading dynamic range was something ridiculous like 160dB, if I am correct then this is plenty for audio, although I dont recall the phase noise of the LO.
Is there a good way to measure from highZ nodes?
This may be a silly question, as it may turn out that all circuits I measure have a lowish impedance output due to feedback.
Cheers.
My colleagues who frequently measure all sorts of things with spectrum analysers usually use a differential active probe for high-impedance nodes. Of course as soon as you use a differential active probe, you have to start worrying about its distortion. As a check, you could measure distortion, screw an attenuator on the active probe, measure again, see if you find the same answer.
You can measure low distortion levels on a spectrum analyser by using the smallest resolution bandwidth available (often 1 Hz) and increasing the attenuation until the distortion of the analyser (and probe, if any) is below the distortion you want to measure. Still, I don't believe you will get anywhere near -160 dB. By the way, on some types you explicitly need to switch on the DC coupled-option to get a reasonable sensitivity at low frequencies.
Craig405
I used to use a R&S specA...20Hz to 26.5GHz...($186,000 CDN) back in early 90's.
It had a tracking generator that went to 5GHz.
If yours has one:
Use a resistive divider and (terminate the gen in 50ohms) sweep the response to make sure it is dividing properly at the frequencies you are interested in.
Yes they are expensive to repair.
I used to use a R&S specA...20Hz to 26.5GHz...($186,000 CDN) back in early 90's.
It had a tracking generator that went to 5GHz.
If yours has one:
Use a resistive divider and (terminate the gen in 50ohms) sweep the response to make sure it is dividing properly at the frequencies you are interested in.
Yes they are expensive to repair.
As others have noted, a RF spectrum analyzer likely does not have sufficient resolution for any serious audio distortion measurements. Run the test--connect the generator you intend to use directly to the analyzer (with appropriate attenuation etc.) and see where the harmonics are.
The simplest form of medium-Z input would be simply a series resistor (plus perhaps a series capacitor for AC coupling). Take a 4950 Ohm resistor, and you have 5 kOhm input Z with 40 dB attenuation. When working with line- or even speaker-level signals, this should do the trick.
Samuel
The simplest form of medium-Z input would be simply a series resistor (plus perhaps a series capacitor for AC coupling). Take a 4950 Ohm resistor, and you have 5 kOhm input Z with 40 dB attenuation. When working with line- or even speaker-level signals, this should do the trick.
Samuel
Great responses, thanks.
@ Dangus, I think this will be the way to go for open loop measurements in the input stages of amps. I'll have to knock up a low distortion op-amp buffer to do it.
@ MarcelvDG, it looks like the frequency resolution is in the data sheet as 0.01Hz with a display resolution of 0.1Hz, I cant seem to get the display to do better than 1Hz/div with a 10Hz span about 20kHz though. I dont have a low distortion tone generator at the moment.
@ DUG, no tracking generator installed, doh.
@Samuel Groner, I will probably give this a go, I have a Hifi Power amp that has a data sheet distortion spec I would be interested to test out, should be able to see if i can get measurements of 0.000x% ballpark.
Cheers
Craig
@ Dangus, I think this will be the way to go for open loop measurements in the input stages of amps. I'll have to knock up a low distortion op-amp buffer to do it.
@ MarcelvDG, it looks like the frequency resolution is in the data sheet as 0.01Hz with a display resolution of 0.1Hz, I cant seem to get the display to do better than 1Hz/div with a 10Hz span about 20kHz though. I dont have a low distortion tone generator at the moment.
@ DUG, no tracking generator installed, doh.
@Samuel Groner, I will probably give this a go, I have a Hifi Power amp that has a data sheet distortion spec I would be interested to test out, should be able to see if i can get measurements of 0.000x% ballpark.
Cheers
Craig
A normal X10 passive oscilloscope probe has a 9 Mohm series resistor built in, which gives you 180001 times attenuation (105.1 dB) with a 50 ohm input, at least up to about 2 kHz. Above 2 kHz the parallel capacitor will determine the attenuation. Doesn't sound good to me.
Hi Graig,
Have you got an ultra low distortion sine wave generator to do the Samuel Groner test? A normal RF signal generator will be totally useless as will a cheap function generator, they are not designed for low distortion.
Regarding frequency resolution, I wonder if that refers to the minimum resolution bandwidth (IF bandwidth of the spectrum analyser, determines together with the noise density how far you can get the noise floor down) or just the number of digits in the frequencies annotated when you use the cursors (interesting when you want to compare crystal oscillators but not very relevant for audio measurements).
Best regards,
Marcel
MarcelvdG, no i havent but I will borrow one from work when the time comes, I'm just wondering how to interface for now, that is a future challenge!
Sounds like:
- for power amp outputs a big coupling cap and resistive divider will work
- for high impedance signal nodes then active intervention is sadly required.
..I can deal with these bombshells.
Thanks to all for sharing your knowledge!
Cheers
Craig
Sounds like:
- for power amp outputs a big coupling cap and resistive divider will work
- for high impedance signal nodes then active intervention is sadly required.
..I can deal with these bombshells
.Thanks to all for sharing your knowledge!
Cheers
Craig
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.