"Snubber" impedance graphs

[Joseph K]

Dear all,

It's been a loong time ago that my analyzer was fired up last.
This, and the fact that I got named in some recent thread flames, gave me the idea that I might as well serve some experimental basis to the
discussions.. ))

So I assambled, as usual, several experimental bypass configurations, and measured their impedance developing across a given frequency range. Basically, I would like to compare two configurations:
The latest [according to my knowledge] "snnuberisierts" bypass arrangement a' la Carlosfm, as it is described in
in this post
The other one is the simple [classic] 10000 uF//220uF// 100nF bypass arrangement applied in the MY_ref project of Mauro Penasa.

Though I could formulate this last statement also like this: "as it was suggested by Carlosfm half a year ago."

As it can be seen from the previous, I would rather leave alone for a moment the ubiquitous "snubber" problem, that is, the reservoir caps & snubber part. About that a lot was told and measured earlier on. And, as I 've already stated, it does not have a measurable influence on the local bypass impedance [on the amplifier pins]. Which statement does not imply that it could not have an effect in other ways.
So, what I would only test & compare here is this statement of Carlos:

Use 2200 uF +100nF (*) on the chip's Psu pins. 3,3uF (*) from V+ to V- pins. 0.1R/1W +3,3nF(*) snubber just before these caps.
(*) Use small 100V MKT caps

Let us see if I was able to reconstruct correctly:
The snubber cap is 2,4nF, because it is that what I had small.
The 100 nF is the smallest possible 100V mylar cap. The resistor is
as described, 0.1W/1R. The 3,3uF I had left out for a moment.
There is a 4.700 uF reservoir cap connected, in some distance, to simulate reality.

[Joseph K]

Here it is the classic setup:

4700 uF // 100 uF // 100nF

Here I used 100 uF instead of the 220 uF, about this later on.
The 100nF is the same as before.

[Jocko Homo]

Fire that network analyzer up, bub! At least one of us is interested.

Jocko

[Nordic]

Go go go, got a bag full of capacitors and a brand new transformer ... will be watching this corner..!

[Joseph K]

Now, I would like to take a pass backwards, and show a comparison between two setup:
The [highlighted] Trace 1 shows : 4700uF//2200uF(local)//100nF
The dimmer Trace 2 shows: 4700uF // 100uF (local)// 100nF

The second setup is the classic. It provides a reasonably balanced and low impedance reaching up to ~2 MHz, then peaks at 3.5 MHz, then we can see the beneficial effect of the small 100nF bypass cap extending up to ~ 40-60 MHz.
As it can be seen, changing the 100uF for 2200 has lowered the ESR value below 1MHz, and has emphatised the impedance peaking at 3.5 MHz, as it was expectable. [10dB difference]

[Joseph K]

Hey, thanks & nice to see You

So, now I would insert the "snubber"

Trace 1 is highlighted & fully "snubberisiert": 4700uF // 2200uF // 100nF // .1ohm+2.4nF

Trace 2 is there for comparison, and is the classic setup.

Well, as I had suspected, we introduced some more ringing..

[Joseph K]

Ok, but there is one more thing to check: This was not yet the originally suggested, full setup: that one includes the 3,3uF across the floor.. that is, the supply pins.
So, I built the whole thing, two times 2200 uF //100nF, one piece of an MKT 1.5 uF /63 V across [I had this, reasonably small] and I am peeking into this assymmetrically, from the point of view of the, let's say, positive supply pin: [versus ground]

[Joseph K]

And here it is the corresponding response:

[Mick_F]

Just making a marker to get email notification of new posts

Good work!!

Mick

[agent.5]

Is a snubber circuit designed for each chip or each channel? Let say I am using the calculation by Carlos and I am using the PSU for the XGC (which uses 2 chip amps per channel), do I need to have separate snubber circuits for each chip? If so, how will the calculation (capacitor values, etc) be affected?