Bob Cordell Interview: Power Supplies

[forr]

Were'nt two diode bridges used in some Mark Levinson amplifiers ?

MIKEB
--- no advantage in using 2 bridge rectifiers, in the contrary, additionally to the doubled diode drop you also get double the diode ringing...---

To me, this seems a good argument against the twin bridge approach.
However, it could imply that for circuits with low dissipated wattage like preamplifiers, it would be an advantage, in the aim to minimise diode HF garbage, to use a twin diode rectifier and a transformer without a center tap instead of the ubiquitous bridge scheme.

 

[pooge]

Re: Re: Re: Re: Re: Actual example

I have never seen rational grounds for this either.

Good thing you brought it up; perhaps someone will come up with a good explanation.

Here

See also post #s 2, 16 and 48 in this thread.

 

[Nelson Pass]

Re: Re: Re: Re: Actual example

There are two popular ways of connecting the transformer-rectifier-reservoir capacitors to provide the pos and neg rails for a power amplifier. In the first, a center-tapped transformer has its center tap grounded, and its two arms go to the a.c. inputs of a single bridge rectifier, whose + and - outputs go to a pair of reservoir capacitors whose common node goes to ground. This is the most common and popular setup.

In the second, a transformer with two isolated secondaries is used (no center tap). Each secondary goes to a bridge rectifier. The pos and neg outputs of each bridge connect to a reservoir capacitor. The two reservoir capacitors then share a common node connected to ground.

Some claim that the second approach, which uses two bridge rectifiers, is superior. However, I have never seen a clear explanation of why it is superior. Indeed, the second arrangement appears to suffer an extra diode drop in the rectification process.

Is the second approach superior?

If so, why?

What are the tradeoffs?

Most customers want total silence from the amplifier, including
mechanical noise. If there is not complete matching between the
secondary coils and only 1 rectifier bridge, any net DC imbalance
between the current of the + supply and the - will tend to
saturate the core of the transformer and create noise. This is
seen for quite low current differences and can also show up with
low frequency output. Using two bridges eliminates the problem.

In the eyes of a manufacturer, any other subtle differences
pale in comparison to the cost of having to replace a transformer
in the field due to mechanical noise.

 

[jacco vermeulen]

the "blip" is coupled through the magnetic field.

Option No 3:
1 transformer + 1 rect. bridge for 1 capacitor bank ?

(i'll save the usual amp listing for the history channel)

 

[Bob Cordell]

Re: Re: Re: Re: Re: Actual example

Most customers want total silence from the amplifier, including
mechanical noise. If there is not complete matching between the
secondary coils and only 1 rectifier bridge, any net DC imbalance
between the current of the + supply and the - will tend to
saturate the core of the transformer and create noise. This is
seen for quite low current differences and can also show up with
low frequency output. Using two bridges eliminates the problem.

In the eyes of a manufacturer, any other subtle differences
pale in comparison to the cost of having to replace a transformer
in the field due to mechanical noise.

Thanks, Nelson! Great insight.

Bob

 

[Ultima Thule]

Re: Re: Re: Re: Re: Actual example

Most customers want total silence from the amplifier, including
mechanical noise. If there is not complete matching between the
secondary coils and only 1 rectifier bridge, any net DC imbalance
between the current of the + supply and the - will tend to
saturate the core of the transformer and create noise. This is
seen for quite low current differences and can also show up with
low frequency output. Using two bridges eliminates the problem.

In the eyes of a manufacturer, any other subtle differences
pale in comparison to the cost of having to replace a transformer
in the field due to mechanical noise.

Yes, that may be very well a truth I admittedly was not aware of though I know the importance of pure AC balanced voltage supply on the primary where small DC imbalance can easily saturate the transformer.
I understand very well the equation that must be fullfilled as stated by NP for the saturation to occure, eg. BOTH the secondary windings AND + and - supply must be in imbalance. Only either of the imbalances will not create a saturation as far as I see then.

Thanks NP!

Cheers Michael

 

[Ultima Thule]

Option No 3:

the "Blip" sensation?

 

[pooge]

Re: Re: Re: Re: Re: Actual example

Most customers want total silence from the amplifier, including
mechanical noise. If there is not complete matching between the
secondary coils and only 1 rectifier bridge, any net DC imbalance
between the current of the + supply and the - will tend to
saturate the core of the transformer and create noise. This is
seen for quite low current differences and can also show up with
low frequency output. Using two bridges eliminates the problem.

This must be what I meant by friendly flux!

 

[jacco vermeulen]

Oh Dear,

They released Mr Dynamic,....again.

 

[pinkmouse]

... which is the best is obvious...

I quite often find the things that people say are obvious are not. Could you ellucidate please?

 

[lumanauw]

Oh Dear,

They released Mr Dynamic,....again.

Who was it that mentioned "NFB Dynamics" not long ago?

 

[Babowana]

Re: Re: Re: Re: Re: Actual example

Most customers want total silence from the amplifier, including
mechanical noise. If there is not complete matching between the
secondary coils and only 1 rectifier bridge, any net DC imbalance
between the current of the + supply and the - will tend to
saturate the core of the transformer and create noise. This is
seen for quite low current differences and can also show up with
low frequency output. Using two bridges eliminates the problem.

In the eyes of a manufacturer, any other subtle differences
pale in comparison to the cost of having to replace a transformer
in the field due to mechanical noise.

Already many quotes; I just want to add another.
As you might know, every night I’m enjoying your great Zen V5, which having two bridge rectifiers. By the way, I read somewhat different +/- rail voltages of ZV5. Now I could (probably) understand that if I use only one bridge, the secondary coil might see this difference as DC input.

Thanks, Nelson, for all.

 

[Nelson Pass]

As I said, this is only an issue when there is an imbalance
of windings and net DC current, which is not a typical
situation. It only occurs if there is DC offset at the output
of the amp which flows to ground, or possibly when very low
frequencies are present in the signal.

 

[Babowana]

As I said, this is only an issue when there is an imbalance
of windings and net DC current, which is not a typical
situation. It only occurs if there is DC offset at the output
of the amp which flows to ground, or possibly when very low
frequencies are present in the signal.

I have no problem with Zen V5 at all.

By the way, when I turn on some power amps, their transformers start crying and calm down after certain minutes. The above (DC offset) explains these? (on the assumption that the transformers have approved specifications)

 

[john_ellis]

Babowana

Transformers work on the basis that the magnetising field in the core is constant - that is, there is a continuously changing voltage which the inductance generates a continuously opposing emf. Immediately at the point of switch-on, the "continuous" part of the current does not apply. Instead the magnetic field cycles through a couple of loops or so before the steady-state condition is reached. It is possible that if you turn the transformer on at the point where the mains voltage is highest, there will be a significant in-rush current which saturates the core. The current settles down once it has dropped below the saturation point.

Sometimes cores running close to saturation also tend to hum sooner on load even though in theory the magnetic field in the core shoudn't change.

cheers
John

 

[Babowana]

Babowana

Transformers work on the basis that the magnetising field in the core is constant - that is, there is a continuously changing voltage which the inductance generates a continuously opposing emf. Immediately at the point of switch-on, the "continuous" part of the current does not apply. Instead the magnetic field cycles through a couple of loops or so before the steady-state condition is reached. It is possible that if you turn the transformer on at the point where the mains voltage is highest, there will be a significant in-rush current which saturates the core. The current settles down once it has dropped below the saturation point.

Sometimes cores running close to saturation also tend to hum sooner on load even though in theory the magnetic field in the core shoudn't change.

cheers
John

Thanks, John.

Your explanation makes me understand how the steady-state condition is reached and what are happening before and after the steady-state condition.

Regards,

 

[john_ellis]

Hi darkfenriz

I agree that another couple of diodes is a good idea. Actually, I use these precisely to make sure that the earlier stages have at least nearly the same voltage as the main output stages.

Especially if you run the doubler diodes between one rail and ground (so that if you have a 25-0-25 transformer to give 35-0-35V the doubled outputs are 70-0-70 not 105-0-105!

cheers
John

 

[MikeBettinger]

Mike,

I still don't really understand what you have been trying to say and you constantly ignore the question being repeated over and over again.

What happened from my point of view was that you made a claim about the center tap. Your argument for this claim was the problem, since it seemed to go against basic electricity theory.

Several people before me pointed this out and asked for a clarification of what you meant. However, you just reiterated things you had said before and seemed to either not understand the question or even avoiding it.
Since you do seem to understand EE well, your behvaiour puzzled me, since I would then presume that also you believe Kirchoffs laws hold for real circuits.

Something you said in one of you more recent posts gave me a feeling that you are using words and terminology in a way that seems very strange to me. Perhaps there are others who have understood what you mean.

Interesting assessment. My point was simple and I was trying to present it in plain English for the people asking how to ground their amplifiers and I might add based on many practical applications.

I've found the grounding approach presented most often to be less than optimal, affecting the stability and noise floor of the circuit it's applied to.

The concept that was returned repeatedly to me in the discussion was that no current returned through the centertap and that the return from the load followed a path through the filter caps to the opposite end of the secondaries. The load should be returned to the filter grounds.

This is correct as far as it goes but once the amp starts delivering current to the load, the loop that creates the motion in the speaker flows through either half of the output stage, through the load back into the centertap. Making this a very important connection in the circuit; the point I was trying to make through it all (conceptually, without relying on textbook terminology).

The image posted in post 137 shows what I was attempting to say. Kirchoff's law applied.

I apologize for the prior blunt response, but think about how your comments might have been perceived.

Regards, Mike.

 

[AndrewT]

Hi Mike,
I agree one point that you made and seemingly missed by most others.
The centre tap ONLY carries current when the two power rails are supplying DIFFERENT current values.

Lets take an example.
If the dual polarity power rails each carry exactly 100mA when the amplifier is idling then the centre tap OUTSIDE the transformer carries ZERO current.
The Centre tap legs INSIDE the transformer carry the charging pulses.

If the amp draws different rails currents when idling then the centre tap carries the difference. eg. +ve rail 100mA & -ve rail 90mA then Centre tap carries 10mA. But most importantly this is a DC current when the amp is idling. There is NO MODULATION of voltage due to current flow when the amp is idling. i.e. this point is quiet.

There is a local loop here that benefits from being coupled to "enclose" it's modulating effect. Namely the charging circuit.
If the centre tap and smoothing capacitor 0v (common) are connected then all the charging pulses are confined inside the local loop formed by the transformer/rectifier/smoothing cap pair.

Now consider what happens when the amplifier draws asymetrical currents from it's supply rails.
When the amp sends current to the load the centre tap connection carries the difference in current in the two supply rails. Again this is overlooked by many.
The load current falls into two classes, below the two Times Iq (of the output stage) and above the two times Iq.
When the output current plus the difference in quiescent current is less than two time Iq the current variation in both supply rails follows the audio signal in the load. This is a smoothly changing current and is similar to the operation of a ClassA amplifier, (that's where the name ClassAB comes from).
The current in the Centre tap connection also changes with the audio signal.
The voltage modulation on the centre tap connection also changes smoothly in line with the smoothly changing ClassA output current.

The big change comes when the current to the load plus the difference in quiescent current exceeds two times Iq (the ClassB mode).
Now the current in one of the rails changes abruptly as it approaches zero current. (at this moment the other rail continues to increase smoothly in response to extra load current :- the rail current follows the audio frequency load current). This abrupt change in current will cause an abrupt change in centre tap connection current.
This can be mitigated by connecting the speaker return cable to the centre tap connection in a local loop. This consigns the abrupt current changes to within the local loop and those connections outside the local loop never feel the modulation that happens inside that local loop.

After this I think our views diverge.
Are we agreed so far?


I have just described two local loops a). smoothing cap 0v and centre tap, b). centre tap connection and speaker return.

 

[Bob Cordell]

Hi Mike,
I agree one point that you made and seemingly missed by most others.
The centre tap ONLY carries current when the two power rails are supplying DIFFERENT current values.

Lets take an example.
If the dual polarity power rails each carry exactly 100mA when the amplifier is idling then the centre tap OUTSIDE the transformer carries ZERO current.
The Centre tap legs INSIDE the transformer carry the charging pulses.

If the amp draws different rails currents when idling then the centre tap carries the difference. eg. +ve rail 100mA & -ve rail 90mA then Centre tap carries 10mA. But most importantly this is a DC current when the amp is idling. There is NO MODULATION of voltage due to current flow when the amp is idling. i.e. this point is quiet.

There is a local loop here that benefits from being coupled to "enclose" it's modulating effect. Namely the charging circuit.
If the centre tap and smoothing capacitor 0v (common) are connected then all the charging pulses are confined inside the local loop formed by the transformer/rectifier/smoothing cap pair.

Now consider what happens when the amplifier draws asymetrical currents from it's supply rails.
When the amp sends current to the load the centre tap connection carries the difference in current in the two supply rails. Again this is overlooked by many.
The load current falls into two classes, below the two Times Iq (of the output stage) and above the two times Iq.
When the output current plus the difference in quiescent current is less than two time Iq the current variation in both supply rails follows the audio signal in the load. This is a smoothly changing current and is similar to the operation of a ClassA amplifier, (that's where the name ClassAB comes from).
The current in the Centre tap connection also changes with the audio signal.
The voltage modulation on the centre tap connection also changes smoothly in line with the smoothly changing ClassA output current.

The big change comes when the current to the load plus the difference in quiescent current exceeds two times Iq (the ClassB mode).
Now the current in one of the rails changes abruptly as it approaches zero current. (at this moment the other rail continues to increase smoothly in response to extra load current :- the rail current follows the audio frequency load current). This abrupt change in current will cause an abrupt change in centre tap connection current.
This can be mitigated by connecting the speaker return cable to the centre tap connection in a local loop. This consigns the abrupt current changes to within the local loop and those connections outside the local loop never feel the modulation that happens inside that local loop.

After this I think our views diverge.
Are we agreed so far?


I have just described two local loops a). smoothing cap 0v and centre tap, b). centre tap connection and speaker return.

You and Mike are both right in a sense.

First, the center tap is very important to the extent that the filter caps are not infinitely huge. They stabilize the rail voltages with respect to ground during the asymmetrical currents that do in fact flow at low frequencies. But the center tap current only flows in pulses at twice the ac line frequency.

Secondly, it is very imporatnt to recognize that there are two loops at work. One loop involves the center tap and the reservoir capacitors. It involves current pulses at twice the a.c. line rate. The second loop involves the speaker return and the reservoir capacitors. It involves current pulses from the Class-AB output stage.

These two loops can be largely separated at ac by devoting two sets of reservoir capacitors, one to rectification and located near the transformer and rectifiers, and the second devoted to the amplifier output and located near the output transistors. The junctions of these two sets of capacitots each form a star where the relevant a.c. pulses are at least partially resolved locally. These two star junctions are connected together as part of a star-on-star topology. The rail lines connecting the two sets of capacitors can have some small resistance in them to force the current pulses to circulate locally.

Bob