KSTR said:
Color me blue, but I still don't get it. The AES48 document seems to address balanced configurations only. For single ended configurations, reference ground = shield = equipment ground = signal ground = ... all star'ed in a single point? If so, then I don't see anything new or special here.
FYI, HPS 3.1 uses such a scheme, with the amendment of using a pair of antiparallel diodes to separate the signal ground to the shield/case = equipment ground. Using a steel case, I got 50uV of harmonic noise at the output, mostly 60Hz and some 180Hz (3rd harmonic). Given the LF gain of 10k, this maps to 5nV harmonic noise at the input, which I think is very good. 2 x 25W Amveco Magnetics toroidal transformers are in the same case!
look @ guitar amps ....
definite chassis ground @ input
(ring of phone jack direct to chassis)
(also they have switchable (polarity)) on the mains in ...
definite chassis ground @ input
(ring of phone jack direct to chassis)
(also they have switchable (polarity)) on the mains in ...
hitsware said:
Who repairs them (at least in USA) is responsible for changing of the power cord and grounding the "white-green" wire.
That's right, nothing new if you just look at the galvanic connection as a concept. The salt is in the fine details how and where the individual bonds to that "net" are made (Star GND is a DC/LF-concept), most of all this detail: no isolated jacks (XLR or RCA, no matter) with long pigtails from the jack's shell to the chassis, or worse yet, wires/cables to the PCB. This makes the metal box effectively useless as a RFI-shield. Key pint is: we need to force any HF currents to flow only and only along the outside of the metal box.syn08 said:
some Info (not the best resource but the first I could remember o-the-fly)
http://www.audiosystemsgroup.com/publish.htm
Example from the AES-Papers section there:
http://www.audiosystemsgroup.com/AESPaperNYPin1-ASGWeb.pdf
- Klaus
BS is what obscured the mind of that fool who decided to design a supply concept which makes assumptions about potentials of the incoming mains. Exept for Safety-GND not assumptions of any kind should ever be made -- and that's why also a dual pole mains switch is a must for any equipment.hitsware said:
Funny thing though, here in Germany, where we have unpolarized mains plugs, single pole switches are legal (and this kills people, children mostly, day after day), which again is beyond anyone half-educated.
Strange world indeed.
- Klaus
Wavebourn said:
I didn't get the impression that Bear was asking about the safety ground. I'll rephrase my response relative to what I thought he was asking.
Sonically and stability-wise my experience places the reference at the power transformer center-tap. It's been a curiosity of mine for years. 🙂
Mike
Addendum, @Syn08:
http://www.audiosystemsgroup.com/AES-RFI-SF08.pdf
http://www.audiosystemsgroup.com/RFI-Ham.pdf
(The two go together, more ore less)
And the relevant schematic for us, see attached.
- Klaus
http://www.audiosystemsgroup.com/AES-RFI-SF08.pdf
http://www.audiosystemsgroup.com/RFI-Ham.pdf
(The two go together, more ore less)
And the relevant schematic for us, see attached.
- Klaus
Attachments
KSTR said:
Yep, I've seen that. But isn't this a trade? Connecting all grounds (I/O jacks, etc...) straight to the case/equipment ground is certainly good for RFI, but isn't it bad for LF noise, not to have a star'ed ground connection? Certainly, having LF currents potentially flowing through the case doesn't look like a good idea. Secondly, very few cases can be built without any cooling holes. If these holes are present (and of course this is an open gate for RFI) does it make any sense to still connect all grounds straight to the case?
I think we have a conundrum... if one designs for optimal RFI rejection, perhaps one is less than optimal at DC---> ultrasonic audio frequencies in terms of noise and hum?
The RF proof shell is inviting from the rejection point of view. But is that sufficient to keep things low noise in the audio region as well??
And, is the xfmr center tap really the right place or best place to "refer" to ground? What if one decoupled the supply ground via a theoretically large choke, and did the same with the rails, making the center of the caps (assuming + & - ) the virtual ground? Is that different?
And, how far "above ground" are the inputs and the output jack(s) then... does the length of wire matter?
Again, we're talking here about the Blowtorch design, which includes the input for phono level signals. But would this matter in a large power amp where the distances and currents are proportionally larger??
_-_-bear
PS. jneutron, where are you now? 🙄 (he seemed to know a bunch about this groundie thingie...)
PPS. Anatoly, there are ways of providing cooling up through gHz using perfs and weaves of metal... look at the front of a microwave oven or the mesh of a satellite dish (C band for example)?
The RF proof shell is inviting from the rejection point of view. But is that sufficient to keep things low noise in the audio region as well??
And, is the xfmr center tap really the right place or best place to "refer" to ground? What if one decoupled the supply ground via a theoretically large choke, and did the same with the rails, making the center of the caps (assuming + & - ) the virtual ground? Is that different?
And, how far "above ground" are the inputs and the output jack(s) then... does the length of wire matter?
Again, we're talking here about the Blowtorch design, which includes the input for phono level signals. But would this matter in a large power amp where the distances and currents are proportionally larger??
_-_-bear
PS. jneutron, where are you now? 🙄 (he seemed to know a bunch about this groundie thingie...)
PPS. Anatoly, there are ways of providing cooling up through gHz using perfs and weaves of metal... look at the front of a microwave oven or the mesh of a satellite dish (C band for example)?
No (but there surely are such statistics)... and I've got my share of this when I was a youngster: how could that damn mobile lamp zap me although I switched it off before changing the bulb. The inner contact looked a bit worn out so I thought I'd bend it back into shape a little.... DANG!! That wouldn't have happened with a dual pole switch.hitsware said:
- Klaus
KSTR,
The schematic you provided shows that there is no RF ground reference for the circuitry to the chassis at all because of the choke connecting it to the chassis.
So, I guess a bit depends on the value of the coil there?
What are they calling for?
If all the gear connected in the daisy chain is built like that, then the circuit is floating on whatever RF is on the chassis/shields??
😕
The schematic you provided shows that there is no RF ground reference for the circuitry to the chassis at all because of the choke connecting it to the chassis.
So, I guess a bit depends on the value of the coil there?
What are they calling for?
If all the gear connected in the daisy chain is built like that, then the circuit is floating on whatever RF is on the chassis/shields??
😕
Exactly. That's why hybrid schemes are in charge, optimizing for both.bear said:
Personally I always try to avoid thinking and drawing in GND terms... and leave it at that, for today (it's 5AM and alrewady dawning here and I'm still not in my bed!), no matter how interesting these things are... sorry. I'll come back to this tomorrow.
One last thing for today, those coils etc in the drawing are representing unwanted parasitics, not purposely installed elements.
- Klaus
There are techniques for HF-proof ventilation, you need a honeycomb structure, rather long but with small apertures.syn08 said:
EDIT: Link http://acousticsystems.com/page/all.cfm?i=iWaveGuideAirVents
So, but now finally good night to you all.
- Klaus
MikeBettinger said:
Mike, when more than one device is interconnected a safety ground becomes a signal ground, so it is wise to assume that the user will use an amp and some sources of signals that will be "safety grounded".
bear said:
I know Bear; and not only theoretically: my diploma project was design of Doppler-based microwave alarm system, it worked on 10.7 GHz. And I know that you know: I met you first at the radio forum.
The classic practice has been to isolate the connectors from the chassis and shunt them back to the chassis with small low inductance caps. I try to isolate the electrical reference grounds from the chassis and tie them back with a 10 to 100 Ohm resistor.
On a line level system where the supply is external I have bonded all of the connectors through a common panel (aluminum or stainless) and used it at the "star" but I have always had hum and noise problems using that for a phono level input.
The dual diode trick looks promising BUT in a high RF environment it may actually do its job of RF detector and make things worse.
Its extravagant but I try to have an isolated supply for each amplifier stage and no connection between the transformer and ground except the internal shield where present. The easiest way to keep the rectifier transients away from the audio is to give them no path back to the circuitry. The "ground reference" is actually defined by some active circuitry and caps well away from the raw DC.
On a line level system where the supply is external I have bonded all of the connectors through a common panel (aluminum or stainless) and used it at the "star" but I have always had hum and noise problems using that for a phono level input.
The dual diode trick looks promising BUT in a high RF environment it may actually do its job of RF detector and make things worse.
Its extravagant but I try to have an isolated supply for each amplifier stage and no connection between the transformer and ground except the internal shield where present. The easiest way to keep the rectifier transients away from the audio is to give them no path back to the circuitry. The "ground reference" is actually defined by some active circuitry and caps well away from the raw DC.
I'm all ears 🙂
Take everything from photos, schematics with grounding-scheme, whatever is not much bother for you to show!
Plan to do something similar, a fully packed preamp with phono+PSU inside and I have no intentions to mess up its noise.
Have fun, Hannes
Wavebourn said:
I understand this. When I feel them necessary I connect safety grounds to the chassis where the power comes in.
My comments on grounding have always been from the performance only side. I'm not building equipment to be sold and therfore have the freedom to experiment and build from a different perspective.
I thought the discussion might be heading in that direction, hence my comments.
Mike.
1audio said:
Hi Demian,
Here's a bit of food for thought as they say. My approach is similar to yours but different in a couple of ways. I would guess that any testing you might do on my grounding would find it a competent approach.
Since the line level ins and outs of a preamp (and amps) are all exposed to the world around them anyway, I bond them to a .025" copper plate and then define the path back to my power supply reference, which is bonded to the chassis. If properly executed this shunts the effects of any noise circulating between components 60/120 Hz along with any RFI, etc. circulating on the shields back to the supply reference. This sounds similar to your approach.
The regulator return loops have a similar defined loop and return to the same. If I used any bypassing I would approach it in a similar fasion, not allowing their return currents into any stage signal grounds, which are really what the term reference applies to.
For years I actually used the chassis as the return, but now I define and build a copper ground bus that incorporates all of the paths. I found aluminum to be a less than perfect conductor...
I've learned to deal with the addition of phono gain stages by treating them as an extension of the line stage loops with the stage signal ground coming from the same node as the line stage and keeping it as an isolated branch there. Works well.
I see parallels to what you are doing with your independent supplies and the actively defined reference; for my approach to work well I have gravitated to isolated supplies between channels, but not between stages (it's actually more involved than this). I differ from your approach in that I use a physical reference, but carefully defined. It seems to work.
One other comment, I have found the reference back to the center-tap to be critical and would caution against adding any chokes as Bear postulated if one is experimenting here.
Like I said, food for thought.
Mike.
- Status
- Not open for further replies.