Paradise Builders

My sim has not included any additional passive filtering. In case you have some RC's in a final design, it will help a lot.

Regarding metal enclosure, I support you totally. I almost cry when I see the diy-builders put the pcb boards into wooden boxes, or free in the air without any shielding. Such high gain and sensitive circuits. I cannot understand it, for the reason that all the efforts may be lost when using unshielded case or wrong wiring... Such is DIY :(
 
And floating DC at the output is also gone?

Chris

I wish :) It's a separate issue altogether.

So far it seems possible to significantly reduce it by keeping the input transistors' temperature very steady. Thin copper foil thermally glued to all of them may do the job. It's just a pity they are not facing each other in a more compact formation on the board. Not sure if using MAT02/03 will be possible but may be worth investigating.

As for the suggestions to encase the boards airtight, it would be great if it wasn't for all the heat dissipated by the shunts. Ventilation is a must.

If it was just a question of slowly drifting offset i would just ignore it, but this is more akin to infra low frequency noise. It just cannot be good for sonics.
 
Is anything gained by putting prereg and xformers in a seperate space within same box with only aluminum plate dividing the two spaces?

Yes, use a box in the box for the prereg. Do not ground it (the prereg box) (not needed) just keep it isolated from the rest of the electronics and mechanical parts, mount it using Teflon bolts/nuts. It will be a cage-of-Faraday and it will stop most HF leaking from the prereg to the rest of the circuits.

Also, you should put the transformer in its own box, and if possible this box should be placed in the prereg box, so now we have a box-in-a-box-in-a-box :) This box also should be isolated from all other circuits and/or mechanical parts.

The mains connector should be mounted directly on the transformer box and the cord should have a metal shield that may be attached to the transformer box and sticks out of the main enclosure not touching any other parts of the construction.

Any way, you can go as far as you like, implementing a box-in-the-box-in-the-box schema goes pretty far :) I have never seen it done (commercially).
 
Member
Joined 2005
Paid Member
Any way, you can go as far as you like, implementing a box-in-the-box-in-the-box schema goes pretty far :) I have never seen it done (commercially).

That being said, shielded transformer in a power supply having its own enclosure is a common sight. More so for they noise they are generating than picking up, though.....

Comments about the arrangements are very good. Generally, it is good practice to put an EMI filter right at the power cable input of the power supply, and these usually come in metal packages so nothing unshielded can propagate towards the inside of the case. A shielded pre-regulator in a separate box is your best bet for high-gain, low-noise amplifiers (PMA your comments are very true....)

In my pre-reg, I built my own EMI filter, it is two-stage (higher damping for RF than one-stage), component values adjusted for the characteristic impedance to have good impedance matching and no ringing, and not too big components to have high self-resonance frequency. In other words, there is more to EMI filter design than meets the eye :D

Building your own EMI filter also allows you to separate the grounds right at the input, I use two back-to-back power diodes (and Nelson Pass and a few others do similar things), to reduce / avoid contamination through the ground - that way, you can control the ground loops rather than trying to fight them.
 
Yes, use a box in the box for the prereg. Do not ground it (the prereg box) (not needed) just keep it isolated from the rest of the electronics and mechanical parts, mount it using Teflon bolts/nuts. It will be a cage-of-Faraday and it will stop most HF leaking from the prereg to the rest of the circuits.

Are you sure .....
Normally the shielding box should be connected in one point with board ground, reference 0V or something like this. In case you leave metal box floating, it will NOT serve as a Faraday cage. It will act as an antenna and it will couple the EMI noise received on its surface thru coupling capacitance (just geometrically defined) into the circuit board (do not forget you have the wires going IN and OUT of the box). If you connect the box with board reference ground, then it WILL act as a Faraday shield. :)
 
Get used with it, or AC couple the pre.

Sorry, not sure what your link points to.

AC coupling is a possibility, maybe even a must in case this is followed with a dc coupled line stage. It does feel a bit of a waste to have a servo though. In order to minimise coupling cap size i may consider providing a bias resistor ahead of the output buffer and including the cap between the riaa and buffer. That is, unless someone comes up with a smarter suggestion. In any case, even without the servo and going for ac coupling it will be useful to achieve good thermal coupling between the input devices.
 
Are you sure .....
Normally the shielding box should be connected in one point with board ground, reference 0V or something like this. In case you leave metal box floating, it will NOT serve as a Faraday cage. It will act as an antenna and it will couple the EMI noise received on its surface thru coupling capacitance (just geometrically defined) into the circuit board (do not forget you have the wires going IN and OUT of the box). If you connect the box with board reference ground, then it WILL act as a Faraday shield. :)

I'm sure, a metal box, grounded or not, will alwais work as a cage of Faraday. No radiation can escape (a true) cage of Faraday.
Faraday cage - Wikipedia, the free encyclopedia
 
Are you sure .....
Normally the shielding box should be connected in one point with board ground, reference 0V or something like this. In case you leave metal box floating, it will NOT serve as a Faraday cage. It will act as an antenna and it will couple the EMI noise received on its surface thru coupling capacitance (just geometrically defined) into the circuit board (do not forget you have the wires going IN and OUT of the box). If you connect the box with board reference ground, then it WILL act as a Faraday shield. :)

In the box-in-a-box-in-a-box case the outer box should be chassis ground (and (at the star point) be connected (using a DDRC I would suggest) connected to the signal ground).
 
Are you sure .....
Normally the shielding box should be connected in one point with board ground, reference 0V or something like this. In case you leave metal box floating, it will NOT serve as a Faraday cage. It will act as an antenna and it will couple the EMI noise received on its surface thru coupling capacitance (just geometrically defined) into the circuit board (do not forget you have the wires going IN and OUT of the box). If you connect the box with board reference ground, then it WILL act as a Faraday shield. :)

And (last point) if it is an inner box there is nothing to receive (for the antenna, from the outside any way). On the other hand, an gage of Faraday could be looked upon as a shorted antenna. In the proposed case, the antenna would 'receive' all radiation that is in the box, and then short it into oblivion (actually heat :)) as long as it is not bouncing around (Hermann von Helmholtz http://en.wikipedia.org/wiki/Helmholtz).
 
Last edited:
Also, it is a good indication for the need to put the prereg in a separate metal enclosure :) The question is, how far are you (the builder) prepared to go :).

Good question but I now can say that MyParadise R2 is prety quiet
Good (realy good) board design prety good Power suply and regulators
It does not oscillate as far as I can tell and the litle bit of ofsset on the output do not seem to bother mine or another pair of speakers I have eard

So tick box for trafo and box Phonos tick main filter supressor and tick for buss of star on earth.

And use Steinless Stell 304 and a further jacket of cooper but is it realy necessary as it performs quite well just sitting on a carpet

Wuld one design new board for separate power suply and one for phono addressing thermals?
 
Your posts like this are completely useless, you are bringing no insight here. Similar as in case of your not understanding how the DC servo works.

Man, you are one nervous guy! And I though I clarified the servo problem (and also the input bias current red herring) :).

They are not Bode plots, at least not in the sense required for a complete circuit stability analysis. A simple plot of the crcuit transfer function vs. frequency is pretty much useless for analyzing a circuit stability.

But perhaps I am wrong, so please specify the Y units on the loop gain plot and the method used to determine the gain and phase plots, also the feedback loop(s) you are analyzing. Otherwise, everything is mirrors and smoke.
 
Sorry, not sure what your link points to.

AC coupling is a possibility, maybe even a must in case this is followed with a dc coupled line stage. It does feel a bit of a waste to have a servo though. In order to minimise coupling cap size i may consider providing a bias resistor ahead of the output buffer and including the cap between the riaa and buffer. That is, unless someone comes up with a smarter suggestion. In any case, even without the servo and going for ac coupling it will be useful to achieve good thermal coupling between the input devices.

You almost got it. In a very high gain servo controlled, DC coupled amp, noise fluctuations are visible at the output, that's unavoidable, happens even in proffessional metrology equipment. Input noise has a quickly increasing power with decreasing frequency, so .1Hz fluctuations have a much larger amplitude than, say, 10Hz. If you get 200mV, that's a lot though, could be another additional cause as well (like a bad contact, or a joint thermoelectric contact pile).

Somewhere in the audio chain, there is a high pass filter removing those fluctuations. If you want to control these, just add a 5Hz (or even less) pole somewhere.