Greetings and measurements from Maastricht

[Gerrit Boers]

Please allow me to introduce myself....

My name is Gerrit Boers. I'm a 46 year old pharmacist/IT architect. Music and audio have been a long time hobby of mine. I do this for fun and learning, but I repeat myself.
To get a better understanding of the subject at hand I perform a lot of measurements with my Clio system, here are some examples:

Mains and rectifier noise
AC supply noise
Capacitor comparison

Just recently I finished my first single ended amp:

SiTuIT

The amp itself is very simple, but the power supply is a 'tour de force'.


I joined this forum in the hope of getting some inspiration for further travels in the land of music and audio.

Regards,

Gerrit

[SY]

Gerrit, welcome to the forums!

I took a quick look at your power supply measurements. Could you say what the measurement parameters were (especially apodization, sample rate, and data length)? The lineshapes and baseline tilt suggest that those settings could be further optimized.

[Zoran]

hi,
i just peek at Your SE project...
rely nice example of good topology...
I hope that OT folows the top end requirements of whole design
thanks for the good photos...

[Gerrit Boers]

SY,

After a quick google on apodization (so that's what it's called),

The most important parameters are listed below the graphs:
CHB dBV 48KHz 65536 Rectangular
meaning: measured on channel B (unbalanced) with a sample rate of 48KHz, FFT window size is 64K (65536) and window type is rectangular.

I since realised that a rectangular window is not the best to use if there is a dominant peak in the spectrum. More recent measurements were made using a Hanning window, this seems to be a generally useful type.
I welcome any suggestions you might have on the subject as I have no formal training in this matter.

[Gerrit Boers]

Zoran,

Thanks for the compliment. The idea behind this design is to get a very small footprint, the amp is 12cm wide and 30cm deep. This topology also maximizes the distance between the 'dirty' part of the power supply (at the very bottom) and the amplifier circuit (at the top).
The placement of the components is such that the wire length is minimized.
The output transformer is an amorphous core Lundahl 1623/120mA, the sound is very, very sweet.

[SY]

Try the same measurement with a Blackman-Harris window. You'll get a much better idea of the relative sizes of the peaks. Also, for a measurement of lower frequency stuff, your resolution will be better if you drop the sample rate while making the number of data points as large as you can.

[Gerrit Boers]

SY,

Thanks for the advice, I will try this. I think I will put up a page to show the differences between various FFT window types so that other people do not need to repeat my mistakes.

I performed some cumulative spectral decay measurements on amplifiers. As you seem to have a thourough understanding of the subject, could you explain what's going on there?

Cumulative spectral decay

[SY]

The reality is that unless an amp has violent low frequency instability, the time resolution of a CSD is far too coarse to see anything. Think about what the time domain response of a step function is- it drops within a few microseconds to the new value. Even if there's any ringing, the period will be at least a couple of orders of magnitude smaller than the interval between waterfall curves. So all that's left is the artifacts of the measurement, mostly the bandwidth limitations caused by the length of the sample record. That's why you don't see the low frequencies drop. Also, please remember that the popular notion of what a CSD really is is wrong...

If you don't have Joe d'Appolito's book on loudspeaker measurement, I'd certainly recommend it. Very readable chapter on FFT/MLS analysis. And he does a good job of explaining what CSD actually is. My doctoral thesis was on the use of FFT methods in certain acousto-optical measurements, so I'm sensitized to nonsense, and d'Appolito didn't even make me wince one time. It will save you a lot of flailing around and make your measurement capability much, much more powerful.

[Cal Weldon]

Welcome to the forums Gerrit.

[Gerrit Boers]

Aha,

I

The reality is that unless an amp has violent low frequency instability, the time resolution of a CSD is far too coarse to see anything....

It seems I have some reading to do. I will follow up on your recommendation.

For the meantime I have removed the page in question