Indeed, DiAna does it in a similar way, except that triggering is derived from the phase of the fundamental.
Cheers, E.
Hi Demian,
If the signal has been low pass filtered, these higher harmonics must be just noise. What else it could be?
Cheers, E.
Basically, that's correct.
It depends to which spectrum you are looking at.
The so called 'THD spectrum' is based on the residual, which in turn is phase locked on the fundamental. Since these residuals are summed together, noise and spurious signals are averaged out. The longer the measurement time, the more they are suppressed.
The same applies more or less to the so called 'coherent raw spectrum', because also in this case the averaging process takes the phase of the fundamental into account. However, not so in case of the 'incoherent raw spectrum'. Then DiAna might be fooled by spurious signals if the resolution of the FHT is too small. The resolution of the 'raw' spectra is the sampling rate divided by 2*record length.
Cheers, E.
BTW, a new version is coming soon that automatically detect the notch frequency and attenuation of a band-stop filter, as well as simulation of a standard passive twin T filter.
arms race / ultra low distortion measurements
Any ideas?
Regarding tunability (to compensate for frequency drift) I think a tunable oscillator is the way to go, that is, a digital one (because easy to tune). But they are not as clean as high-end analogue oscillators. That's why I need a band-pass filter in order to remove the last traces of harmonics.
Regarding FFT's with a variable test frequency? Preliminary trials showed that DiAna can handle that.
Any thoughts about above approach?
Cheers,
E.
Yes, a battle between the 'analogues' and the 'digitals', but the 'digitals' still need two (simple) analogue parts: a band-pass filter and a band-stop (notch) filter, both passive and not tunable. I don't trust tunable filters because the necessity of active components that will inevitably increase the distortion. LC passive band-pass filters without any magnetic core material are easy to build for frequencies of say 1kHz and higher, but how about 20Hz or so?. Then one need some magnetic core material, otherwise the dimensions of the inductors get way too large. But then things like saturation and hysteresis come into play and will (most likely) increase distortion.
Any ideas?
Regarding tunability (to compensate for frequency drift) I think a tunable oscillator is the way to go, that is, a digital one (because easy to tune). But they are not as clean as high-end analogue oscillators. That's why I need a band-pass filter in order to remove the last traces of harmonics.
Regarding FFT's with a variable test frequency? Preliminary trials showed that DiAna can handle that.
Any thoughts about above approach?
Cheers,
E.
Band-pass filters
Hi Gerhard,
I'm not sure that this kind of core is most appropriate for filters, as it is primary intended for high power applications. I was more thinking about a ferrite pot core because of its self-shielding geometry. Regarding low hysteresis losses, some MnZn composition seems the material of choice, right?
Cheers,
E.
Hi Gerhard,
I'm not sure that this kind of core is most appropriate for filters, as it is primary intended for high power applications. I was more thinking about a ferrite pot core because of its self-shielding geometry. Regarding low hysteresis losses, some MnZn composition seems the material of choice, right?
Cheers,
E.
I'm no specialist in ferrites, but I have used EE48 or EE54? cores once in
a generator for geophysical measurements. I think the material was N48
or a more modern successor.
Lepaisant e.a. use N48 material in pot core form for their ultra-low-voltage noise
amplifier:
< https://pdfs.semanticscholar.org/49b5/63335faea2836e65c6dee411d5c8b04d5a3e.pdf >
EE cores are probably cheaper and more common. I don't think that stray fields
are a problem because the air gap, if any, is in the middle of the core. If you want
maximum Q, stay away with the copper from the air gap by upholstering with some
isolation tape. (eddy currents)
a generator for geophysical measurements. I think the material was N48
or a more modern successor.
Lepaisant e.a. use N48 material in pot core form for their ultra-low-voltage noise
amplifier:
< https://pdfs.semanticscholar.org/49b5/63335faea2836e65c6dee411d5c8b04d5a3e.pdf >
EE cores are probably cheaper and more common. I don't think that stray fields
are a problem because the air gap, if any, is in the middle of the core. If you want
maximum Q, stay away with the copper from the air gap by upholstering with some
isolation tape. (eddy currents)
Last edited:
Hi Gerhard,
Thank you for additional information. As for stray fields, I'm more afraid of stray fields from the PSU of my RTX6001. They radiate hum in the filters (and cables) I'm currently using. In the meantime I've got a copy of "Modern Ferrite Technology" by Alex Goldman. Let's hope this book provides further insight.
Cheers,
E.
Thank you for additional information. As for stray fields, I'm more afraid of stray fields from the PSU of my RTX6001. They radiate hum in the filters (and cables) I'm currently using. In the meantime I've got a copy of "Modern Ferrite Technology" by Alex Goldman. Let's hope this book provides further insight.
Cheers,
E.