Boris is coming...

Thanks the response Paul - it has encouraged me to read for the 10th time my notes on grounding and having question to answer certainly helps learn something new. And I've just found an updated version of my main reference - seems clearer.

Even though this is an all DHT power amp, I'd like to use best practices where possible.

I will ground the input, using the noise blocking network. I grounded my last build at the input, connected to chassis close to the input without the network, and it sounded awful: lean, edgy. I've seen it recommended that the input be grounded by running a wire from the (isolated) input socket and connecting to the chassis close to the safety earth - is this what you would suggest?

I will earth the power transformer shields at the transformers - that is what I did with my last build.

Regarding the input and interstage transformer screens, I'm a bit at a loss with my limited understanding of how shields work and I need to read up on the screening behaviour of a chassis. A few thoughts and questions:
  • Are screens susceptible to picking up noise from ground or chassis; does it even matter?
  • Grounding the screens at the first PS capacitor seems wrong to me: there are strong current pulses (and noise) present in that location that I suspect could(?) contaminate the screens.
  • I probably would not want to ground them to their local stage 0V ground either: although the current "pulsing" would be minimal, would I want even a small amount of noise from the screen in my 0V ground - probably not.
  • So that leaves chassis, which I'm not entirely confident about either: can chassis noise contaminate the screens? My guess is to connect them to chassis at the same point as the input stage ground.
🤔
 
Last edited:
I've seen it recommended that the input be grounded by running a wire from the (isolated) input socket and connecting to the chassis close to the safety earth - is this what you would suggest?
That will send your input signal current all the way back to the chassis earth, then back through your power supply and up to the first stage. It wouldn't surprise me if this made a lot of extra noise.
Regarding the input and interstage transformer screens, I'm a bit at a loss with my limited understanding of how shields work and I need to read up on the screening behaviour of a chassis. A few thoughts and questions:
  • Are screens susceptible to picking up noise from ground or chassis; does it even matter?
  • Grounding the screens at the first PS capacitor seems wrong to me: there are strong current pulses (and noise) present in that location that I suspect could(?) contaminate the screens.
  • I probably would not want to ground them to their local stage 0V ground either: although the current "pulsing" would be minimal, would I want even a small amount of noise from the screen in my 0V ground - probably not.
  • So that leaves chassis, which I'm not entirely confident about either: can chassis noise contaminate the screens? My guess is to connect them to chassis at the same point as the input stage ground.
Remember that these noise currents need to flow in a loop, and that you're dumping this into the chassis plate and earth ground. If noise is going to come up through the chassis plate and into the shields, how does it complete the loop? The same goes for connecting the screens to the power supply.

You can always hook up a sensitive FFT and try the different configurations.
 
I am clearly no expert but I look at stuff like this as relating to a Faraday cage. Shields and screens should be hooked up to the chassis in order to dump any noise/stray currents into the house ground and not into anything carrying the signal (audio ground in other words). Do not listen to me for expert advise though..
 
Last edited:
I am clearly no expert but I look at stuff like this as relating to a Faraday cage. Shields and screens should be hooked up to the chassis in order to dump any noise/stray currents into the house ground and not into anything carrying the signal (audio ground in other words). Do not listen to me for expert advise though..
That is what I thought too... and that house mains and signal form a loop, through multiple components... meaning that the 0V signal ground point and chassis connection point also matter...

Our friend Derek has pointed me in the direction of a solid resource recommending for the screens to be effective, suggesting they should be connected to 0V signal ground, not directly to chassis.

Despite reading what I thought was some decent info on this topic, I still have much to learn, especially when it comes to the various current loops that are possible, noise, screens and the role of chassis etc. Or I leave a few options open in the design/layout and just try 'em.
 
Last edited:
That will send your input signal current all the way back to the chassis earth, then back through your power supply and up to the first stage. It wouldn't surprise me if this made a lot of extra noise.
I thought that would happen irrespective of where the 0V signal ground is connected to chassis? If connected near the input, wouldn't the current travel to chassis safety earth via chassis? If not, what path would the current likely take?

I'd like to learn more about more about where the currents want to return. I understand that okay for signal... but my understanding is not where it could be when incorporating chassis, system etc.
 
Last edited:
I think you may want to draw out
Our friend Derek has pointed me in the direction of a solid resource recommending for the screens to be effective, suggesting they should be connected to 0V signal ground, not directly to chassis.
Maybe you could post this resource here and we could all learn?
 
I thought that would happen irrespective of where the 0V signal ground is connected to chassis? If connected near the input, wouldn't the current travel to chassis safety earth via chassis? If not, what path would the current likely take?

I'd like to learn more about more about where the currents want to return. I understand that okay for signal... but my understanding is not where it could be when incorporating chassis, system etc.
I think it would help to draw out what you're thinking about. If audio ground and the chassis only touch in one place, signal current shouldn't flow through the chassis. You had mentioned running a wire from an isolated RCA jack back to the mains earth location, but not what the rest of the grounding scheme would look like.

Some of the variation in how shields are connected can depend a bit on the specific transformer design in question, and on whether earth and signal ground are the same, and how far apart they may be (both physically and electrically). When signal levels are super duper tiny, like a microphone or phono transformer, it's really easy to just put the earthing post by the input transformers and first audio stage, then earth and audio ground are very much the same thing at the ground post, and you can tie everything there.

In a power amp where you have 1+ volt of input signal, teasing out the differences is going to be a bit more difficult.
 
Last edited:
Here's an example where a different connection is specified by a manufacturer:
shields and screens.jpg
We can see two separate shields connected together and to the chassis, then an electrostatic shield going to load ground. If you put these MC step-up transformers into a phono preamp, those two points would be the same connection, while in an external box they would not be.

Having thought about this more, for your interstage transformer, the electrostatic screens would be better off connected to the audio ground. If you're using a ground breaker between audio and chassis ground up front, this may work against the electrostatic shield if said shield is tied to the chassis. It would be worth experimenting with this a bit. I don't find the ground isolators are generally helpful on a power amp, so I could still rationalize connecting signal ground to the chassis right up at the input and connecting all the shields in the input transformer to that point.
 
Last edited:
Here is the reference material that Shoshin references above.

From Henry W. Ott, Electromagnet Compatibility Engineering (Wiley: 2009) at pp. 204-06:

“… Actual transformers, not being ideal, have capacitance between the primary and secondary windings, as shown in Fig. 5-11, this allows noise coupling from primary to secondary.

This coupling can be eliminated by providing an electrostatic, or Faraday, shield (a grounded conductor placed between the two windings), as shown in Fig. 5-12. If properly designed, this shield does not affect the magnetic coupling, but it eliminates the capacitive coupling provided the shield is grounded. The shield must be grounded at point B in Fig. 5-12. If it is grounded to point A, the shield is at a potential of Vg and still couples noise through the capacitor C2 to the load. Therefore, the transformer should be located near the load in order to simplify the connection between the shield and point B. As a general rule, the shield should be connected to a point that is the other end of the noise source.”

Diagrams:

Ott, Capacitive coupling and trafo screen grounding_diagrams.jpg
 
Sorry for the late response - work has left me a bit depleted... still am, but your efforts to help me deserve a response. I just hope it makes sense. If bolded my questions below so they don't get lost among my ravings.

I think it would help to draw out what you're thinking about. If audio ground and the chassis only touch in one place, signal current shouldn't flow through the chassis. You had mentioned running a wire from an isolated RCA jack back to the mains earth location, but not what the rest of the grounding scheme would look like.
The grounding scheme will be much like that suggested in the Valve Wizard's grounding chapter: Grounding. Essentially a refined buss system, with each amplification stage's returns connecting at that stage's PS capacitor -ve (rather than progressively along the buss); it looks like multiple star grounds, daisy-chained, with only one connection from audio ground to chassis.

I understand that "audio ground and the chassis only touch in one place, signal current shouldn't flow through the chassis", but the audio ground is connected to source, the chassis is safety earthed to mains, and the mains is connected source... which seems like another loop via mains earth and source, partially running through the audio ground... How "partial", and how much runs through a decent conductor (if that matters), depends on where the audio ground is connected to chassis: where on the audio ground and where on the chassis. On my last amp, which has a large, nonmagnetic stainless steel chassis, audio ground connects in a single location, and no ground lift or noise blocking network is employed, I can measure the noise on the output changing as I vary the pressure of my hand against chassis (more pressure, up to 2mV less noise)... and when it was grounded at input, sounded dreadful (counterintuitively, sounds much better grounded from PS input capacitor to chassis close to safety earth). A ground lift and noise blocking network can make difference in some builds with a single audio ground connection to chassis, or does it? What do I make of this - what do I need to consider? Maybe I'm putting too much emphasis on a dubious loop?

Having thought about this more, for your interstage transformer, the electrostatic screens would be better off connected to the audio ground. If you're using a ground breaker between audio and chassis ground up front, this may work against the electrostatic shield if said shield is tied to the chassis. It would be worth experimenting with this a bit. I don't find the ground isolators are generally helpful on a power amp, so I could still rationalize connecting signal ground to the chassis right up at the input and connecting all the shields in the input transformer to that point.
Thanks Paul for thinking about this more and trying to help me out. I appreciate it. I will design the amplifier so that I can experiment with where I tie the shield connections, where I ground, and whether I use ground isolation. The first approach I try will connect input ground from the RCA to chassis close to the RCA, and I will connect the input transformer shield to that point also; I won't use the ground isolation. Should I ground the interstage electrostatic shield to that point (chassis) or connect to output stage audio ground at the PS capacitor's -ve (not chassis)? Power transformer shields will connect to chassis via a bolt on the the transformer. If that does not work out, I can experiment from that "known good" approach, using a ground isolator and moving the input transformer chassis connection to input audio ground. If that still does not work well, I can experiment the audio ground connection to chassis. Does that seem like a workable approach?

Here is the reference material that Shoshin references above.
Thanks Derek for your help also. I think we are getting close.

Pfft, who puts shields on transformers anyway? Madness. Now I need sleep.
 
Last edited:
I'm sure you saw that TVW recommends grounding the transformer shields to the chassis in 15.12: Miscellaneous Ground Connections.
 
I'm sure you saw that TVW recommends grounding the transformer shields to the chassis in 15.12: Miscellaneous Ground Connections.
Actually, I've not had the chance to read his updated version and that was not covered in the version I have. So no, I hadn't. A quick scan of it indicates it is much clearer than the version I am familiar with. The suggestion is that screens usually can be connected to any point on the chassis... otherwise to the -ve of the 1st PS capacitor.

There are a lot of different recommended approaches to grounding and synthesising them into "best practice" is a challenge: perspective is what I need, I think.

I will read the later version of TVW this weekend and think more, and less, on this. Section 15.9 just might clear up some confusion regarding the loop via mains earth too.
 
Maybe you could build the amp but temporarily clip lead the screen ground connections. And then scope the screens relative to signal ground to identify the connections that have the least noise.

cheers, Derek
 
Maybe you could build the amp but temporarily clip lead the screen ground connections. And then scope the screens relative to signal ground to identify the connections that have the least noise.

cheers, Derek
Good idea. The time I put into in learning and executing builds suggests that I will continue to build... a 'scope would probably be a sound investment. My building approach for this amp would certainly support what you suggest.

Thinking is no substitute for experiencing. 😉
 
Last edited:
  • Like
Reactions: Old
Well, it's been a while. I've been working on the physical design, more off than on, over the past few months. Lots of 'round in circles, exploring variations on sets of options over and over. I think my attachment, expectations, whatever for this project have turned it into a chore, one that I don't enjoy at the end of a demanding work day, week, quarter... I've rethought what I'd like to achieve, put some unhelpful thinking aside, am feeling in a better place, and just want to get on with it. Expect more frequent posts in future.

These might just become my best friends: 10mm M3 mounting cubes. I'd rather M4, but hey, I could not find any so M3 it is - will just need to use a few more.
 

Attachments

  • 10mm_M3_mounting_cube_resized.jpg
    10mm_M3_mounting_cube_resized.jpg
    452.6 KB · Views: 7
I'm parking Boris and much of my forum activity for a while. Perhaps I will feel like engaging with this project after some rest.

Wishing you all the festive season you'd like it to be. Take care.
 
I'm parking Boris and much of my forum activity for a while. Perhaps I will feel like engaging with this project after some rest.

Wishing you all the festive season you'd like it to be. Take care.
Take care of yourself bud. Stay in touch even if it’s just to shoot the breeze.
 
Back
Top