Bandwidth problem

[JAY X]

Hi!

I wrote mhz, but i would say MHz,...megahertz.
That is the problem writing on a mobile phone... sorry!!

Jay x

 

[abbey road d enfer]

Hi!

I wrote mhz, but i would say MHz,...megahertz.
That is the problem writing on a mobile phone... sorry!!

Jay x

That's interference from all the crap from switchers, cellphones, WiFi... whatever. Just discount it.

 

[JAY X]

Hi,

Below  is a link for the pdf with ripple and noise measurements. The attachements does not work...

I said MHZ, but really was khz....sorry...my fault!  I think i confused the bandwidth of the picoscope, with the measurement. :

Jay x


download link:

https://mikrosum.webnode.es/_files/200000019-4a1984b0fa/VOLTAGE%20DOUBLER%20PSU%20RIPPLE%20V4.pdf

 

[abbey road d enfer]

Hi,

Below  is a link for the pdf with ripple and noise measurements. The attachements does not work...

I said MHZ, but really was khz....sorry...my fault!  I think i confused the bandwidth of the picoscope, with the measurement. :

Jay x


download link:

https://mikrosum.webnode.es/_files/200000019-4a1984b0fa/VOLTAGE%20DOUBLER%20PSU%20RIPPLE%20V4.pdf

All I can see is some signal in the MHz range, almost certainly picked up by the test cords. Not hum nor ripple.
The horizontal scale in the 1st graph is in us (microseconds); in order to catch hum, it should be in ms (milliseconds).

 

[JAY X]

Hi,

Here are some measurements i did with another CRC filter: 470uf/22R/470uf. These measurements are in ms.

I really thought that my problem with the frequency response at both low and high end were due to the noise of psu...but perhaps the cause is other...

Jay x

https://mikrosum.webnode.es/_files/200000020-6e38c6f325/VOLTAGE%20DOUBLER%20PSU%20RIPPLE%20%20IN%20MS.pdf

 

[abbey road d enfer]

Here are some measurements i did with another CRC filter: 470uf/22R/470uf. These measurements are in ms.

All these measurements are meaningless to me. I don't even know what you're measuring; is it the supply rail noise or the device's output noise. Anyway the waveforms do not look like hum/ripple.

I really thought that my problem with the frequency response at both low and high end were due to the noise of psu...but perhaps the cause is other...

There is no way ripple on the supply rail can result in BW limitations; look elsewhere.

 

[JAY X]

Hi Abbey!

The measurements are made at +V supply rail.

In any case, It is becoming clear that i have to investigate more to find out the causes.

Jay x

 

[JAY X]

Hi,

This is getting....weird :

I performed two audio tests, to see if the frequency response has changed or not.

The first fact is that, in my design, the PSU ripple filter affects the frequency response.
Why?...no clue..by now.

In the first test with CRC filter 1000uf/22R/1000uf, i got a notch in the frequency response, at 200hz -29,96dbu.

With the second CRC filter 470uf/22R/470uf i got a somewhat flatter response between 80 - 570hz,  and from there, the frequency response climbs 16dbu up to 20khz.

In both cases, this is not normal.
A PSU filter should not affect the frequency response of an audio  equipment.

So, there should be something wrong in my layout.

Maybe the Single point Ground?
Maybe...what?

Jay x

 

[JAY X]

Hi!

I'm going to give up with the voltage doubler psu. I will try a DC-DC converter from murata, and/or maybe a dc rail splitter circuit...
I will post schematics in the next coming weeks.

Jay x

 

[JAY X]

Hi!

Well... not so fast...
I was reviewing the schematic from the presonus eq3b, that i found in among the thecnical documents, and from the schematic It seems there are two Ground areas so to speak:

1. At the AC inlet.
2. From the filter caps towards the DC output, a shared ground.

So, both grounds are separated. This is my interpretation from the schematic. But i don't know if this is correct.

Thank you for your advise.

Jay x

 

[abbey road d enfer]

Hi!

Well... not so fast...
I was reviewing the schematic from the presonus eq3b, that i found in among the thecnical documents, and from the schematic It seems there are two Ground areas so to speak:

1. At the AC inlet.
2. From the filter caps towards the DC output, a shared ground.

So, both grounds are separated. This is my interpretation from the schematic. But i don't know if this is correct.

Thank you for your advise.

Jay x

I don't really understand what you man, can you post this schemo?

 

[JAY X]

Hi,
Attached is the PDF of the eq3b.
Page 2 is the PSU.

 

[abbey road d enfer]

I was reviewing the schematic from the presonus eq3b, that i found in among the thecnical documents, and from the schematic It seems there are two Ground areas so to speak:

1. At the AC inlet.
2. From the filter caps towards the DC output, a shared ground.

So, both grounds are separated.

Wrong interpretation. The ground symbol is a simplification that indicates that all the points that carry this symbol are electrically connected together. In reality there is a small resistance and inductance between all these points, but one may assume they are negligible; at least, that's what the designer assumes.
The reason is there are many nodes that are connected to ground, using this symbol helps cleaning the drawing, that's all. It certainly does not indicate two different grounds.
At the beginning of radio, there was only the earth symbol, but over the years was developed the necessity for discriminating grounds, analog ground, digital ground, clean ground, dirty ground, chassis ground; they're all supposed to be galvanically connected. When circuits are connected to different grounds, this is represented by different symbols, either a different shape, or an explicit marking.

 

[JAY X]

Ok.

Another question.

Should i tie any part of the PSU Ground, maybe from filter caps or the DC output, to chassis?

Jay x

 

[abbey road d enfer]

Ok.

Another question.

Should i tie any part of the PSU Ground, maybe from filter caps or the DC output, to chassis?

Jay x

If you use a metallic enclosure, this is mandatory. Best choice is the 0V connection to the rest of the circuit.

 

[JAY X]

Hi!

Good news!. After many hours of work cheking every soldering, rebuilding the psu filter, and making the chassis connection, now the frequency response is way better than initially.  And also the gain mismatch between L&R channels is within +/-0.5 dbu. Still i have to do some mods, but generally it works well. And thanks to this great forum and its members!!.

Jay x

 

[JAY X]

Hi!

Now i'm taking audio measurements. And  i'm stuck with THD+noise measure. I run a 1khz sine wave, and measured the level of the harmonics in dbu. But, i'm not sure... those dbu must be converted into RMS voltages...¿right?

And, on top of the equation i should place the values of 5 harmonics, make the square root, and divide by the fundamental (1khz) also in RMS volts.. ???

Jay x

 

[JohnRoberts]

The square root of the sum of the individual harmonics squared. The that compared to the fundamental level.

JR

 

[JAY X]

Hi,

I have calculated the THD (5 harmonics): 0.015%.

Also i tested the board (1 layer pcb), with a picoscope with a test signal: 1.23v @ 1khz sinewave.

the results were:  THD +N : -42,82 dbc ( below carrier)
                                        SINAD:    -42,82 dbc
                                        SNR:            43,27dbc
                                        IMD:            0,07%

I suppose these results can be better with a 2 layer pcb, good vias placing, and keeping sensitive opamp nodes clean of parasitic traces, etc...

What do you think?

Jay x

 

[JAY X]

Hi!

Two steps forward, one back...
Those measures, i Will repeat them... because i improved the harmonics distortion a bit...

Channel gain mismatch is still within 0,5dbu. Good.

Jay x