Is it wise?

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
For home theater amp, usually it is needed 5 or 6 amplifier module in 1 casing, with common power supply for all of the amps.

Is it wise to :

1. make 5 or 6 (2 is bridged for subwoofer) self-oscilating amp for this purpose? With 6 selfoscilating modules, there will be 6 working frequencies, while all is connected to common rail supply.
Will they all work properly (especially the sub woofer amp, 2 in bridged)? Or is it better to make triangle forced type, so all 6 modules is operating at the same clock? I noticed classD switching artifacts also injects to supply, making them has "crosstalk". Plus the problem of "rail pumping" for half bridge type?

2. Using selfoscilating classD for small loads in bridge mode, when the subwoofer load is 2 or 1 ohm, in bridge mode, each amp will experience 1 or 0.5ohm load. Is it wise to use selfoscilating type, or is it better to use triangle forced clock for this bridged+low impedance load?
 
I know people on here think that triangle/ carrier type amps are somewhat inferior but in your case i would reccomend that if you intend to use a high number of amps in one case then they are a good choice as you can use the same triangle or clock to synchronise them so you won't get clashing frequencys. I won't say any more on this subject because i stand alone in thinking that carrier type amps are good regardless of the application. People on here will prob read this and shoot me down but hey, it's a question of subjective taste!:D

Regards
Mad.P
 
lumanauw said:
Well, I'm seduced by the good quality of selfoscilating classD.
Is there a way to use 4 or 5 modules using the same 1 common power supply without any problem/crosstalk between each other?

It is impossible to do multichannel with a single supply with out multiple ground loops unless all inputs are fed separate balanced signals. I don’t personally know of a source for these signals unless you get into astronomically expensive units. One more practical solution would be an input transformer for each channel. This also would be expensive, about the same cost as the amp modules for reasonably high quality transformers. The less elegant solution would be a power supply for each module and all grounds floating. It would be a gamble to do it any other way and take a chance on the grounds. There have been others that have done multi channel in a single box but I don’t recall any reports on them.
The nicest solution would be a mono amp per channel. This would be the most complex in sheer number of parts.
Roger
 
It is impossible to do multichannel with a single supply with out multiple ground loops unless all inputs are fed separate balanced signals. I don’t personally know of a source for these signals unless you get into astronomically expensive units. One more practical solution would be an input transformer for each channel. This also would be expensive, about the same cost as the amp modules for reasonably high quality transformers. The less elegant solution would be a power supply for each module and all grounds floating. It would be a gamble to do it any other way and take a chance on the grounds. There have been others that have done multi channel in a single box but I don’t recall any reports on them.
The nicest solution would be a mono amp per channel. This would be the most complex in sheer number of parts.
Roger

Could you write down the specification for how much crosstalks are allowed? I have made a 5-channel UcD amplifier and could not heard any idle tones in the loudspeakers.

Jan-Peter
 
Hi, Jan Peter,

I have made a 5-channel UcD amplifier and could not heard any idle tones in the loudspeakers.

Once I made 1 power supply for feeding 2ch of classAB + 1ch of selfoscilating classD.

The artifacts of classD switching can be seen in the classAB output via scope. When I turned off the classD, those artifacts disappear.

I also read in IRFaudamp paper, that half bridge topology can do "bus pumping".

But if your experience that putting 5 modules with common supply (1 transformer, 1 set bridge diode, 1 set cap bank) is OK, I will try it :D

Oh, another thing, Jan Peter. Is selfoscilating classD suitable for driving low impedances load (like 2 or 1 or less ohm)? They won't change operating point (like oscilating frequency) when headed to low loads or bridge mode?
 
Crosstalk?

Jan-Peter said:


Could you write down the specification for how much crosstalks are allowed? I have made a 5-channel UcD amplifier and could not heard any idle tones in the loudspeakers.

Jan-Peter


Jan-Peter,
I wasn’t even thinking of crosstalk as the ground impedance is far lower than the input impedance so shared signal currents shouldn’t generate any significant crosstalk signal. Depending on the inductance of the cables used this could be a problem with higher frequencies and longer cables would be far more of a problem all around. The problem is that with all the grounds connected together at the source and amp, line frequency magnetically induced noise due to transformers, power cords or the refrigerator motor, can induce significant currents into the ground causing noise and hum. If the amps power supply is done with 2 bridge rectifiers to keep ripple current out of the grounds this end of the ground connections should not be that much of a problem. If the interconnects are all twisted together a couple of turns every foot or so the noise pickup can be minimized. In any audio system line noise is one degrading factor that we usually can do something about as DIY’ers. The closer we can keep grounds to being references only with no currents circulating the lower the noise floor will be and the better the sound staging will be too.
Reducing a line hum or buzz from something you can only hear when you are right at the speakers to something that can only be heard if you put your ear within an inch of the speaker is the difference between hearing the back wall of the sound stage or not! I have found every reduction of noise has a significant effect on sound quality and is worth pursuing even to extremes, like my beating on people to twist their wire pairs. Human hearing has an amazing ability to pick out coherent signals from the noise and this low level information retrieval is a significant clue to us to be able to put it all back together in our heads as real music. Any line related noise will distract from this ability.
Roger
 
lumanauw said:
For home theater amp, usually it is needed 5 or 6 amplifier module in 1 casing, with common power supply for all of the amps.

Is it wise to :

1. make 5 or 6 (2 is bridged for subwoofer) self-oscilating amp for this purpose? With 6 selfoscilating modules, there will be 6 working frequencies, while all is connected to common rail supply.
Will they all work properly (especially the sub woofer amp, 2 in bridged)? Or is it better to make triangle forced type, so all 6 modules is operating at the same clock? I noticed classD switching artifacts also injects to supply, making them has "crosstalk". Plus the problem of "rail pumping" for half bridge type?

2. Using selfoscilating classD for small loads in bridge mode, when the subwoofer load is 2 or 1 ohm, in bridge mode, each amp will experience 1 or 0.5ohm load. Is it wise to use selfoscilating type, or is it better to use triangle forced clock for this bridged+low impedance load?

For an unbalanced system: take 1 big transformer, give every module its own supply, bridge+2caps+fuses, ground at transformer/cap.bank 0V, take enough space between modules and twist all the wires!!!!!. Than take my modified input cable :D and feed them to the unbalanced inputs, ground free. These recommendations apply to the UCD's. This will have the best crosstalk possible hence Dolby/Cinema experience. But only if you use good caps(lowest ESR).
 
Bert's modified input cable

I have been giving this problem some thought and I do like the idea of connecting the power ground up to the signal ground at the input jack through a 10 ohm resistor and keeping a direct connection with the - signal input as Burt suggested. This gives the front end equipment or preamp a ground reference but breaks up the ground connection for induced noise very nicely. The only drawbacks I can see are twofold, there is the safety issue which can be solved by connecting parallel reversed 3 amp rectifier diodes across the 10ohm resistor and the RF issue that would be best taken care of with a .1uf film cap from the input jack - to the case at the closest possible point. This is basically a nice Band-Aid for the problem but I think it will do fine considering successful units have been done without this fix.
This is also a good ground solution for stereo units using a common power supply.
Roger
 
A copy of the post from Bruno, in an other thread;

The difficulty is obviously that there is never a truly "correct" way of wiring an unbalanced system, or certainly not a surefire recipe to get it right every time in all circumstances. Otherwise put, for every recipe there will always be someone who manages to set up a system that defeats it. Unbalanced audio should be banned by law... coax cables are for RF, not audio. The aim of using balanced connections is to relieve the ground connection of its secondary job of providing a signal reference and allowing it to do what it's intended for i.e. to roughly equalise the ground potentials either side of the connection. A correctly designed balanced connection is immune to ground loops. In fact, correctly designed balanced connections (see AES48) are almost guaranteed to produce ground loops but they do so without incurring any performance loss.

The problem in Bgt's setup is that part of the ground return current to the power supply circulates round the shield of his RCA cables, through his preamp. In an unbalanced system this is annoying for two reasons. One - this current creates a voltage drop on the audio cable shield which unfortunately is also the signal reference. Two - depending on the design of the preamp this current may well start flowing inside the preamp as well...
A trick which might be used in this situation is to short the two RCA grounds together. In this way, any ground current will be restricted to the shields of the input cables inside the cabinet. However, the input cables are *differential* which means that the inputs will cancel any common mode voltage appearing on the inputs as a result of this current.

Ideally this would be done by using uninsulated RCA connectors screwed on the back panel. Of course, the unbalanced demon pops up again there because this allows only one box in your audio system to be connected to mains earth and requires that all others be connected to ground only through their RCA connections. This is why consumer gear typically has no earth prong (and those that do often have hum problems) - in order to allow direct chassis connection of all RCA's via the chassis. This is the closest anyone can get to a surefire unbalanced recipe (I will add this note somewhere in the documentation) but keep in mind that it too is easily defeated by having two earthed boxes in one system of by using a power supply with excessive primary-to-secondary capacitance. Only balanced connections per AES48 are always guaranteed to work (caveat: only if both ends of the connection are per AES48 is the connection guaranteed to work).

As an aside (OT for those who abhor XLR but very relevant otherwise) I'd like to recap what AES48 prescribes: connect pin 1 and the outer part of XLR inputs and outputs straight to the chassis and only then to the PCB. This requires the use of silver-coloured XLR chassis parts (not the black kind) that will make positive electrical contact with the chassis. Connect pin 1 to the chassis lug on the XLR part (parts with pin 1 directly connected to chassis are also available) and connect the internal cable shield there (if at all).
This insures that:
*Your box cannot induce currents from one pin-1 to the other.
*No amount of current externally fed from one pin-1 to another can cause interference, including severe RFI.

-------------------------------------------------------------------------
Our recommondation is;

Use an XLR connector with Pin#1 to chassis ground, and make a XLR --> CINCH cable where at the Cinch side the shield is connected to -symmetrical wire, and these two together to the GND of the CINCH connector.

In a few weeks we will show on our website a demo model with 5-channel UcD180, how we recommend to get best sonic performance.

Regards,

Jan-Peter
 
Re: Bert's modified input cable

sx881663 said:
The only drawbacks I can see are twofold, there is the safety issue which can be solved by connecting parallel reversed 3 amp rectifier diodes across the 10ohm resistorRoger

Roger, what safety issue do you mean? Even if you blow the resistor nothing would happen, just hum at startup. I've used this kind of input connection for many years on the Crescendo amp and sometimes I touched one of the speaker connections on the ground of the cinch part. Nothing really happend. Just put a new resistor there.
 
Re: Re: Bert's modified input cable

Bgt said:


Roger, what safety issue do you mean? Even if you blow the resistor nothing would happen, just hum at startup. I've used this kind of input connection for many years on the Crescendo amp and sometimes I touched one of the speaker connections on the ground of the cinch part. Nothing really happend. Just put a new resistor there.

Bert,
The input chip can only handle voltages a little above its rail voltage before it blows up. This is one issue, the other is if you have 2 points that are at considerably different voltages and complete a connection between them with your body it can be quite serious! The diodes would limit this voltage to very harmless levels till a fuse blows if there is one.
Roger
 
Re: Re: Re: Bert's modified input cable

sx881663 said:


Bert,
The input chip can only handle voltages a little above its rail voltage before it blows up. This is one issue, the other is if you have 2 points that are at considerably different voltages and complete a connection between them with your body it can be quite serious! The diodes would limit this voltage to very harmless levels till a fuse blows if there is one.
Roger
Roger, this can happen to any opamp input also if they are in balanced mode. But you're talking about extremes. So I don't see the connection with the way I have the input organised and the sensitivity for a breakdown?
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.