Need help on RC filter for power supply

[Nounours18200]

Hi friends,

First of all I am not an expert in electronic design, so please be patient with me....

I have built a few months ago a new and upgraded power supply for my preamp, replacing the original P.S.
I have been using it for a few months and the improvement is really huge. The skematics of this (version 1) P.S. is herebelow:


The improvement in musicality compared to the stocked power supply is so huge, that I would like to improve further the power supply and I have been advised to add a RC network at the end of the P.S. flow. So the diagram of this version 2 would be as follow (the diagram is inspired by a dedicated article):


Compared with the version 1, (C3, C7 and C11) in the version 2 play the same role than C3 in the version 1 (they are the big capacitors decoupled by small ones).
But on the right of the diagram, we see a RC network (C15, C17 and R1/C19) that is supposed to get rid of the capacitor residual inductance (this is what is explained in the source article).
The values proposed in the article are:
C15=C17=100nF 100V
C19=680nF 100V
R1=1ohm 17W (this value is given for a power amplifier P.S., so should probably reduced to a few watts for a preamp).

My knowledge in electronic is enough to make me confortable with the version 1 (I have built it and it is a success) but I am not sure that the proposed values for the version 2 are appropriate for a preamp power supply. Could you tell me if it is OK ?
If you have any additional recommendation for improving the version 1, I will be very interested !

Thanks friends,

[wakibaki]

Some kind of problem when trying to load your images...

w

[Nounours18200]

Sorry, the links should be ok herebelow (I cannot edit my first post):

Version 1:


and version 2:


Tks

[wintermute]

I don't see any Resistors in there. for an example of how much effect on the ripple having a CRCRC circuit makes have a look at the below pics. edit: sorry just realised I was missing the point, and the RC was referring to the last resistor/cap in the circuit, oh well, I'll leave my irrelevant info below anyway

the attached image (at the end) is the circuit these measurements were made against (which is the filter stage before my reg circuit). I think from memory the load was around 160mA for these shots. The R's in the circuit are 3.3ohms and the input voltage was around 18V ac RMS.

The picture below is the ripple across the 10,000uF cap, scope set on 20mV/ div


Next is the ripple across the second cap scope set at 5mV / division



last is the ripple accross the third cap, scope set at 2mV / division.


Note that there is a penalty, The higher the current draw the greater the voltage drop will be across the resistors, smaller resistors mean a smaller voltage drop and for higher currents either higher voltage (inefficient), or smaller resistors (not as effective ripple reduction) will be required. You have to choose your tradeoff. However I think the above shows the difference that adding a few resistors can make Also you need to take into account the dissipation in the resistors and size them appropriately (ie high enough wattage).

Tony.

[Nounours18200]

Hi wintermute,

Thanks for your time...

I understand what you say, but not sure it addresses my question... it seems that your diagram with RC filters is different than the one I posted in version 2: does it mean that you replace the right part of my diagram (i.e. C15 to 20 and R1, R2) by your RC filter ?

Your diagram looks clear and efficient, but what is your opinion on the v2 diagram (I have designed the v1 not the v2) ?

[wintermute]

Yes sorry I realised that after I posted (hence the edit). As far as the second diagram is concerned I don't see a lot of point of all the extra parallel caps, IMO you will get far better value (assuming you have the voltage to spare) but also putting in small series resistance (if you are going to add extra caps), even something as low as 1/2 ohm will make a difference.

Is the output of this circuit feeding a voltage regulator? I've just looked at the values of the caps in your first diagram, I'd originally assumed that they were 22000 uF but now I'm wondering if that is 22uF Only reason is that it says they are nichicon FG's and they don't come in sizes that big (biggest is 10,000uF at 6.3V only).

With respect to the 2nd diagram, By paralleling multiple smaller caps the ESR will be reduced, but at the same the inductance will be increased. I'm not sure why each big cap would be paralleled by a small film cap, I can understand one being placed at the output to the supply, but this is something that is often debated as to whether it does more harm or good. Having one accross each big cap in the supply seems a bit strange.

The final small cap with resistor is something that caused a huge debate here on diyaudio a few years back, it was coined the snubberised supply, (which I think is a mis use of the term snubber). It was claimed to make all sorts of amazing transformations to the sound of the amplifier used on. I've not experimented with it, and don't have enough electronics knowledge to comment on it from a theoretical point of view. Certainly it wont require any significant expense to try it and you could do so on your existing v1 without adding all the extra caps of V2.

My circuit that I attached is overkill compared to what people normally feed a regualtor with, but my reasoning was that the lower the ripple on the input to the regulator, the easier its job is, and the lower the ripple should be on the output... this certainly seems to be the case in the sim. I need take out the 2nd and third caps and measure on the real circuit to see if there is a measurable difference in real life.

Tony.

[wintermute]

I just checked the difference on the output with only 10,000uF filter and then again with only 4700uF filter cap compared to the CRCRC before the reg. Whilst the filtering makes a big difference to the pre-reg ripple, it is either not making a difference post reg, or the resolution of my scope (and the noise pickup problems I have) is not good enough for me to easily see a difference (any difference seems to be in possibly lower spikes visible when the trace intensity is increased significantly (rectifier pulses?) with the additional filtering, but I'm not sure about that.

So the circuit I posted (when followed by a regulator) is definitely overkill, but that was the original idea (go with gross overkill and see if it actually made any difference) and it was intended to be a learning experience for me

Tony.

[Nounours18200]

Is the output of this circuit feeding a voltage regulator? I've just looked at the values of the caps in your first diagram, I'd originally assumed that they were 22000 uF but now I'm wondering if that is 22uF Only reason is that it says they are nichicon FG's and they don't come in sizes that big (biggest is 10,000uF at 6.3V only).

You are right: in my original design (version 1) they were 22kµF.

With respect to the 2nd diagram, By paralleling multiple smaller caps the ESR will be reduced, but at the same the inductance will be increased.

True: for my new design, I have chosen to reduce the ESR: I will see (ear) if it is better ...

I'm not sure why each big cap would be paralleled by a small film cap, I can understand one being placed at the output to the supply, but this is something that is often debated as to whether it does more harm or good. Having one accross each big cap in the supply seems a bit strange.

Not false, a unique small capacitor after the big ones is enough.

The final small cap with resistor is something that caused a huge debate here on diyaudio a few years back, it was coined the snubberised supply, (which I think is a mis use of the term snubber). It was claimed to make all sorts of amazing transformations to the sound of the amplifier used on. I've not experimented with it, and don't have enough electronics knowledge to comment on it from a theoretical point of view.

This is THE question I have... some people says it brings a benefit, some other says "no" like the debate you talk about. Like you, I do not have enough electronic knowledge to have a firm position on this point...
I am looking for somebody having a lot of expertise in P.S. building and designing, but it seems not easy to find !...

If somebody has some ideas on how to improve the v2 design without this additional right part, I am interested...

I send you PM to continue this discussion ...

[wintermute]

You could ask a mod to move this thread to the power supply forum, it has been quite active lately, and you might be more likely to get someone with the expertise you are looking for to chime in.

Tony.

[Nounours18200]

I fully agree with you Wintermute ! I also thought to ask the question again in the Power Supply forum....

I will ask if the post is not moved by a moderator

Tks