A Grounding challenge , please help..

[Elbert]

Hi,

Once again, I turn to the wisdom of the forum to seek advice..

I have buildt a pair of amplifier chassis with 3 separate amplifier channels each for stereo with 3-way active x-over.

My challenge is how to manage grounding and avoid ground loops when I hook the three amps up to an internal x-over card..

As you can see, I have enclosed a simple schematic.

The Supply consists of one Toroid with dual secondary windings, feeding three separate pairs of rectifiers.

Each of these recctifier pairs feed a separate bank of capacitors, one for each amplifier channel.

The thought behind this was to isolate the power supply for each amp channel, so that the current draw in one channel would have a minimum effect on another channel.

I used some random op-amp symbols to illustrate the amplifiers.

On the schematic, you will allso see a integrated circuit, which is supposed to illustrate the 3-way active x-over. This x-over is fed from a separate low-voltage winding frrom the toroid through a regulated supply (only rectifier shown)

The signal ground for both input and outputs for the x-over is common and connected to the 0V of the X-over power supply.

The input RCA ground is connected to the chassis.
The only other thing connected to the chassis is the mains safety ground.

So here's the problem:

When hooking up only one amplifier to an output of the x-over, everything is fine.

Hooking up two amplifiers to the x-over creates a lot of hum on the output of both those amplifier channels.
Obviously some sort of ground loop.

The solution that seems most likely, is to connect the 0V points in the amplifier power-supply together as a star point as they are now effectively 3 "floating" 0V points.

I checked the voltage between two of these 0v points, and there was a potential of about 100 mV across.

Is it safe to connect the 3 0V points in the power-supply?
Can it give rise to any other issues?
Will it solve my problem?
Are there any other workable solutions??

I hope I have described the problem at hand in a way that makes sense, and that some of you find my problem worthy your time and consideration, and any input will of course be greatly appreciated.

Best regards,

Elbert

[wintermute]

Personally I think I single bridge per channel with the Zero Volts coming from the transformer windings would be safest. That is how I did the powersupply for my Gainclone which has a single traffo, and separate rectifiction and cap bank for each channel. I don't know whether it will be safe to connect zero volts points together with your setup, as you might run into problems with differences in the diodes, not all conducting/rectifiying equally. (note that is just a suspicion, not based on any practical or theoretical knowledge, someone else would be better to answer with respect to your actual implementation I'm just relating what worked for me ).



I don't think adding a third channel would matter. The above is absolutely dead quiet.

Mine was all done on a single PCB as it was a low capacitance supply, zero volts point is a bolt in the middle of the PCB, Bottom lead goes back to the traffo terminal block, each channels speaker ground , followed by each channels power grounds and then finally signal grounds.



Tony.

[wintermute]

Hi thinking about this some more (your schematic makes my head hurt ) I'm a bit confused as to how your input wiring is organised.... The choice of symbols in your schematic could be better I think

if there is more than one path from your main input to the crossover through to the speaker output of the amps then you may have a problem there.. How have you got your speaker outputs 0V wired, are they completely separate to each "floating" point in the power supply?

Tony.

[Elbert]

I guess I could have done well enough with the conventional single rectifier per channel.

I just figured that a double set of rectifiers would be a more conservative design in terms of halving rectifier load, and according to some texts I've read, such a dual rectifier approach allso keeps the transformer wires completely separated from what's going on on the DC side.

Your point about diode diferences etc, is interresting, this could perhaps explain why I measured a difference of 0,1V between two of the power supply floating grounds.

I apologise for the schematic, Windows 7 doesn't like my old scanner, I could not do a sketch and post it, so I had to improvise with a layout program in stead.


The speakers are wired to 0V from their respective amplifier cards.


In my initial post, I mentioned ground-loop, but as far as I can see from my schematic, I have no ground loop.

On the other hand, whenever there is a voltage potential between any of the power supply floating 0V points, current will flow between these 0V points, and as the layout is now, the only way is through the signal ground connections to the X-over.

The more I think about it, the more convinced I get that I 'll have to tie together the three 0V points in the power supply as a star point, run a separate ground-wire from that point to the X-over, and make sure signal wires screens are only grounded in one end to avoid loops.

Hmm.. Am I on the right track here, or have I overlooked something??

[wintermute]

Hi Elbert,

I drew up what I think it your schematic so that I could try and trace the ground paths.. If I've drawn it correctly all of the zero volts points on the amps should be connected (via the signal grounds) back to the 0V on your crossover power supply.. You haven't given details as to how that power-supply's zero volts is derived..

The only path I can see for the ground loop is via the rectifiers between amp channels. I think if you tied the zero volts together at the amps it would actually make the situation worse (unless as you say you only connect one end of the signal shields).

You could try running wires from your zero volt points on the amps back to your star point (0V point on your crossover PSU) and only run the signal wire (no shield) to each amp. Although as you say perhaps locally connecting them all and then running a single wire back to your crossovers zero volts star point would be better as there will not be a chance for resistance differences between the paths back to the true zero volts point.

attached is the diagram I drew up.. note that I didn't draw the transformer properly (only one winding) to simplifiy the idea, it was more a block diagram to work out potential paths.. there would of course be two windings connection per rectifier block.

looking at this again, I would say that the zero volts from each rectifier should be connected to the zero volts point of the XO PS. and separate connections should be made from there to the XO and each AMP. (alternatively have a local star for the amp psu's which is connected back to the xover ps zero volts and each amp only connected to that local psu star with no other zero volts path back to the XO. Try to keep all connections as short as possible and the same length.

I hope someone else will chime in as I'm really not 100% certain about this.

Tony.

[Elbert]

Tony,

Your interpretation of my schematics seem to be spot on, well done! (must buy a working scanner soon..)

Regarding the 0 v of the x-over supply, its a bit of an odd one. I only had one low voltage winding available, but needed a +/-15V supply. One AC line is connected to the centre of a capacitor pair and the other AC line feeds two diodes of oposite direction which then alternately feeds the plus or the minus side of the capacitor pair. Following this "single-to dual supply converter" there is a pair of voltage regulators (LM series) and some more capacitors. And yes, it works very well!

So the ground of the X-over power supply is allso floating with no other connection to chassis, transformer center taps etc..

Yesterday, I jumped the gun and tied together all three 0V points in the power supply (on the capacitor banks). No smoke or other signs of impending doom!

I tried to hook up the X-over card in various ways. and with various ground-configurations, some according to your desccription. I was able to avoid any greal ground loops, but I did have some general hum and noise. I attribute this to the prototype connection of the cards which involves the odd un-shielded lab-wire etc..

When the X-over card is finished and installed in the chassis. I'll aim for one single ground connection to the X-over from the PSU star point, either through a separate ground cable or through the shield of just one of the signal connections betweenamp and X-over. I'll leave the shielding connected in one end only on the other cables to avoid loops.

[wintermute]

Hi Elbert, good to hear that you are making some progress

One suggestion I would try (and it goes against some of what you will read on grounding) is to not make the chassis connection from your signal ground, but to make it instead from your star point.

Another thing you can try is moving your transformer.... My integrated amp picks a LOT of noise from the toroid. Moving it outside the chassis pretty much eliminates a lot of hum and other undesirable noise.. it isn't excessive but it is noticable within about 1M of the speakers when nothing is playing. If moving it reduces the hum, then you can try rotating it or changing it's orientation.

Another thing, was this new hum with no inputs connected or with inputs connected? If with inputs connected do they also have 3 pin plugs (ie a safefty ground)? There could be a ground loop between equipment.

Tony.

[Elbert]

Tony,

Sorry for my much delayed response here!

I've been quite absorbed lately with active x-over design, so the grounding problem has sort of come a bit in the bacground..

Anyway, your suggestion is very interresting, and when the time comes to install the x-over cards properly in the chassis with screened cables all the way, It is one I'll certainly give a try!

Regarding the transformer, I don't have much leeway in terms of placement, nor do I wish to remove it from the chassis alltogether..

Having said that, I don't think the transformers are a big problem with regards to hum.

Earlier on, whern I first tested the amplifier cards, I connected a loudspeaker directly to the card, thereby bypassing the output relays of the protection cards.
With the card signal input shorted to ground, I could hear a faint buz if i put my ear just over the speaker.

I then switched off the amp, but the amp kept running for some seconds due to the rather large power-supply capacitors, and the hum was still there!

From that, I concluded that the hum came from an external source and that it would hopefully be reduced when I installed the cover and front panel on the amp.

In the enclosed image, you can see the placement of the Toroid transformer.
Allthough my design is fairly compact, the transformer sits in a separate compartment.
In addition to that, I placed a sheet of permalloy mu-metal on the compartment wall behind the transformer. I allso wrapped a strip of the same material around the circumference of the Toroid. How much this actually helps, I do not know, I just decided to do it as a precaution.

Annyway, When the dau\y comes to hook everything up properly, I'll make sure to post the findings here!

[magnoman]

Elbert,
If I'm not mistaken your bridge arrangement puts all the return side diodes (for each of the secondary windings) in parallel (they share a connection to the tx on one side and they share a common ground on the other). Though this will work the diodes wont share current evenly so make sure each is rated for the sum. This may also help in guiding you through your grounding paths.

Hope this helps
-Antonio

[wintermute]

Originally Posted by Elbert

Earlier on, whern I first tested the amplifier cards, I connected a loudspeaker directly to the card, thereby bypassing the output relays of the protection cards.
With the card signal input shorted to ground, I could hear a faint buz if i put my ear just over the speaker.

I then switched off the amp, but the amp kept running for some seconds due to the rather large power-supply capacitors, and the hum was still there!

From that, I concluded that the hum came from an external source and that it would hopefully be reduced when I installed the cover and front panel on the amp.

I think that is a very valid conclusion! so you may well be fine once all mounted in the proper chassis with all shielding in place! Probably a wise choice to be not chasing it any further at this stage.

Tony.