Diagramm units
and the signal is +20dBm wich is +40dB above 0, so in total, the noise is 180dB below signal.
I think, that's the resolution of the analyzer.
Add: output resitance is 15R, so a level of +20dBm should be 100mW wich equals 2.25Vrms.
EDIT: The THD in fig.8 is 70dB (in dbms) wich equals 140db (in db) wich makes 0.00001 %, and that at 100kHz !
Have a look on the unit, it's dBm (10*log relative to mW), so in dB (20*log) it's 140dB below 0
and the signal is +20dBm wich is +40dB above 0, so in total, the noise is 180dB below signal.
I think, that's the resolution of the analyzer.
Add: output resitance is 15R, so a level of +20dBm should be 100mW wich equals 2.25Vrms.
EDIT: The THD in fig.8 is 70dB (in dbms) wich equals 140db (in db) wich makes 0.00001 %, and that at 100kHz !
Last edited:
Ooog.
I failed to comprehend the original graph. Thanks for laying that out.Changes the conclusion rather much.
You could also take a look at Mr. Groner's huge report.
http://www.nanovolt.ch/resources/ic_opamps/pdf/opamp_distortion.pdf
http://www.nanovolt.ch/resources/ic_opamps/pdf/opamp_distortion.pdf
page 3 -> AC Performance:
The analyzer could only deliver a maximum output of 0dBm at 50 (ohms), corresponding to a voltage of 223mVrms
seems to be dBm so mW and 10*log.
using dBm is quite normal for HF analyzer.
This fits (roughly) the 2.25V at 15 ohms from 20dBm. Datasheet says 5.2Vrms wich fits, if taken the 50R output resistor into account and the estimation of 20dBm (could also be 23dBm ?)
Voegelchen, can you go into a bit more detail why the distortion goes up on the lower output level on some of the opamps?
(Thanks for sharing this by the way)
THD + N (Emphasis on the latter!)
I mean that's the most common reason, otherwise it'd likely be due to under-bias of the output stage/crossover distortion. But, yes, thanks for sharing this survey and look forward to hearing what you have to say, Voegelchen.
I can only estimate: If you leave the class A region only a bit in a class AB amplifier, the amount of ditortion generated by the region "switching" is worse than staying in class A or having a large signal in B (and a small region in A only).
But this depends on the internal structure like quiscent current (and voltage, see D.Self) and transistor behaviour and the amount of feedback available.
Please consider also, that the soundcard output DAC isn't the best available, getting worser with falling level.
Yes, with long FFT traces you will pull the noise down tremendously, so that would minimize that effect. Otherwise, yes we're saying the same thing about the output stage.
But, yes, linearity in your ADC will definitely smear any low-level results, too. So probably difficult to say the exact mechanism, albeit all these factors are at play.
Edit, this is more head scratching than initially thought: 1Vrms is by no means a low-level signal.
But, yes, linearity in your ADC will definitely smear any low-level results, too. So probably difficult to say the exact mechanism, albeit all these factors are at play.
Edit, this is more head scratching than initially thought: 1Vrms is by no means a low-level signal.
But it could be on the edge of the class A region, because +/- 1.4mApk seem to be more than the typical 0.5..2.0mA of an opamps quiscent current.
Also, some opamps have a NPN-NPN (quasi complementary) output stage making things worser, especially with no feedback left (e.g. MC33078)
Still does not compute.
Go Convert dBm to dB conversion to volt mW voltage V dB dBu dBV dBm converter conversion and calculation analog audio milliwatt milliwatts kilowatt power level power watts convertor converter audio engineering sound recording - sengpielaudio Sengpiel Be, use the second calculator box, set 0dBm into 50R, calculate rms voltage (-->0.2236V), then use 20dBm, recalculate and you will see.... 2.2361V, 10x
Also sanity checking the plots in that linked BB pdf, Fig.3, generator spectrum, no harmonics above the noise floor which is 90dB down, a reasonable value for a RF analyser (pretty good, actually). 180dB++ sine purity at 20kHz is completely out of bounds, no way a RF analyser/generator from 30 yrs ago be that good. Even the best of the fixed freq audio sine generators (note the various threads here on DIYA) don't come close... let alone when the receiver doesn't have a precision notch filter (RF analyzers don't).
https://de.wikipedia.org/wiki/Leistungspegel
https://en.wikipedia.org/wiki/DBm
you are right KSTR, I intermixed the 10*log for POWER with the given scale VOLTAGE.
Beside that, I think the RF (HF is the german acronym for it) analyzer will work by mixing down (mixing diode) and then and then filtering (XTAL filter), so noise (beside the one generated inside) shouldn't be a problem and also the harmonics could be easieler measured by that mode (wich is unfortunately a bit uncomfortable for audio frequencies).
So, RF analyzers don't need an additional notch, it's always built in
https://en.wikipedia.org/wiki/DBm
you are right KSTR, I intermixed the 10*log for POWER with the given scale VOLTAGE.
Beside that, I think the RF (HF is the german acronym for it) analyzer will work by mixing down (mixing diode) and then and then filtering (XTAL filter), so noise (beside the one generated inside) shouldn't be a problem and also the harmonics could be easieler measured by that mode (wich is unfortunately a bit uncomfortable for audio frequencies
).So, RF analyzers don't need an additional notch, it's always built in
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.