Read the data sheets on various transformers. You will then understand the leakage and EMI measurements.
I have a hard time to find this kind of data, most is commercialy oriented so you'll never know what you buy. It looks hopeless to compare.
I found on a forum that toroïdals are not born equal: About 3 different brands of about the same power, one was badly noisy, shielding could not get rid off, second was pretty good, third was better.
Here:
Hum differences in toroidal power transformers...
To shield from magnetic field you need iron, soft steel, nickel.
The best is Mumetal a Nickel Iron alloy.
Mumetal Sheets & Foils – Permanent Magnets Ltd
Permalloy is another Ni Fe alloy like Mu Metal.
However, when a transformer does leak badly, it will be extremely hard to shield it real good. Here I am talking about microphone preamplifiers aiming at ultra low noise where hum must be knocked down extremly low, below the noise floor. Here you need a power supply with low leak to begin with....Or use batteries.
Hum can come from poor grounding too.
The best is Mumetal a Nickel Iron alloy.
Mumetal Sheets & Foils – Permanent Magnets Ltd
Permalloy is another Ni Fe alloy like Mu Metal.
However, when a transformer does leak badly, it will be extremely hard to shield it real good. Here I am talking about microphone preamplifiers aiming at ultra low noise where hum must be knocked down extremly low, below the noise floor. Here you need a power supply with low leak to begin with....Or use batteries.
Hum can come from poor grounding too.
Last edited:
Stray field in a transformer has different possible sources:
-Geometry: in principle, a well built toroidal is ideal; R-cores can also be quite good, but it is essential they have half the primary and half the secondary on each leg (normal construction)
-Construction: even a toroidal can be poor if some rules are not adhered to.
Both the primary and secondary need to cover the entirety of the core homogeneously, have an even number of layers with the direction of layers alternating.
For EI transformers, the "modern" construction of side-by-side primary and secondary is the worst possible. Older, superposed windings types are much better in this respect (they also have a larger P/S capacitance.
-Material quality: poor laminations have a low permeability, and some zones let field escape, because of saturation, imperfect orientation and mechanical stress.
-Loading of the secondary: the current in the secondary creates a reaction flux that tends to eject the magnetic field.
Thus, a good, small toroidal would be ideal for low-noise applications, but very small toroidal tend to be rare or non-existent.
An alternative is to operate a small EI transformer at a much reduced voltage.
It should also operate at a minimal current, which fortunately is compatible with a preamp supply.
This means that the supply should not be of the shunt variety, to minimize the secondary current.
Since it is difficult to find small transformers made for a primary voltage higher than 230V, a solution to reduce the operating voltage is to insert a dropping impedance.
If the power is really low, a resistance is acceptable, otherwise a capacitance keeps the dissipation within acceptable limits, but then, it cannot be used in the primary, because it will cause a ferro-magnetic resonance.
The solution is to reverse the primary and secondary. The output current will be minuscule, but for a low-power preamp, it is workable.
The capacitance value needs to be determined experimentally: use a 24V or 30V transformer with the supply circuit connected to the primary and test various capacitor values between the mains an the primary.
Use the lowest value that gives an acceptable regulation margin for the worst case supply current.
The capacitor could be in the 100nF to 2.2µF range and needs to be X-rated, of course.
Also include a safety shunt zener across the filter cap, in case the current falls to zero.
With such an arrangement, the transformer will operate at a 10th of its nominal voltage or less and at a low current, which means extremely low leakage
-Geometry: in principle, a well built toroidal is ideal; R-cores can also be quite good, but it is essential they have half the primary and half the secondary on each leg (normal construction)
-Construction: even a toroidal can be poor if some rules are not adhered to.
Both the primary and secondary need to cover the entirety of the core homogeneously, have an even number of layers with the direction of layers alternating.
For EI transformers, the "modern" construction of side-by-side primary and secondary is the worst possible. Older, superposed windings types are much better in this respect (they also have a larger P/S capacitance.
-Material quality: poor laminations have a low permeability, and some zones let field escape, because of saturation, imperfect orientation and mechanical stress.
-Loading of the secondary: the current in the secondary creates a reaction flux that tends to eject the magnetic field.
Thus, a good, small toroidal would be ideal for low-noise applications, but very small toroidal tend to be rare or non-existent.
An alternative is to operate a small EI transformer at a much reduced voltage.
It should also operate at a minimal current, which fortunately is compatible with a preamp supply.
This means that the supply should not be of the shunt variety, to minimize the secondary current.
Since it is difficult to find small transformers made for a primary voltage higher than 230V, a solution to reduce the operating voltage is to insert a dropping impedance.
If the power is really low, a resistance is acceptable, otherwise a capacitance keeps the dissipation within acceptable limits, but then, it cannot be used in the primary, because it will cause a ferro-magnetic resonance.
The solution is to reverse the primary and secondary. The output current will be minuscule, but for a low-power preamp, it is workable.
The capacitance value needs to be determined experimentally: use a 24V or 30V transformer with the supply circuit connected to the primary and test various capacitor values between the mains an the primary.
Use the lowest value that gives an acceptable regulation margin for the worst case supply current.
The capacitor could be in the 100nF to 2.2µF range and needs to be X-rated, of course.
Also include a safety shunt zener across the filter cap, in case the current falls to zero.
With such an arrangement, the transformer will operate at a 10th of its nominal voltage or less and at a low current, which means extremely low leakage
Thanks Elvee.
I do not understand the primary secondary reversal.
I understand underating the primary voltage should decrease stray field, but isn't it the same as underating the drawn power. In both cases the core is running at a flux well below saturation, no ?
One thing I will try is serie resistors at the secondaries in front of the reservoir caps, I suspect the current spikes to charge the capacitors when diodes turn on, do induce bad transformer flux spikes.
Another investigation: I heard of switching PSUs that can be quieter than linears.
Not that I think of designing one, here I would buy it, and eventually make it quieter with appropriate shielding.
I do not understand the primary secondary reversal.
I understand underating the primary voltage should decrease stray field, but isn't it the same as underating the drawn power. In both cases the core is running at a flux well below saturation, no ?
One thing I will try is serie resistors at the secondaries in front of the reservoir caps, I suspect the current spikes to charge the capacitors when diodes turn on, do induce bad transformer flux spikes.
Another investigation: I heard of switching PSUs that can be quieter than linears.
Not that I think of designing one, here I would buy it, and eventually make it quieter with appropriate shielding.
Toroidal transformers is the best type but you can also order them with a magnetic shield. Here is one with shield. AA-98050-AU Toroidtransformator 800 VA 2 x 42VAC Noratel
Produktdetaljer
Information om produktfamilj
Speciellt avsedd för audioslutsteg
Dubbla sekndärlindningar
Minimalt magnetiskt läckfält
Magnetisk skärm
Montering utan rondeller, försedd med fastgjuten innerdel
Tillverkningsnorm EN61558
It says "magnetisk skärm" => Magnetic shield but the datasheet doesn't say a thing about it but you see that it looks a bit special with the white ring. I have also seen this transformer in real life. It looks impressive.
Information om produktfamilj
Speciellt avsedd för audioslutsteg
Dubbla sekndärlindningar
Minimalt magnetiskt läckfält
Magnetisk skärm
Montering utan rondeller, försedd med fastgjuten innerdel
Tillverkningsnorm EN61558
It says "magnetisk skärm" => Magnetic shield but the datasheet doesn't say a thing about it but you see that it looks a bit special with the white ring. I have also seen this transformer in real life. It looks impressive.
Shield about transformers can also mean somthing that separate the primary from the secondary windings. I have one like that ( not a toroidal ), I think it is a one layer winding ( may be a foil ) with only one wire out, to connect to chassis.
I have seen shields like a round box for toroidals, but no serious data about the effectiveness.
Capot de Blindage Acier pour Transformateur Torique 90x40mm - Audiophonics
They say it is made of steel, it limits radiation and dress transformers.
( Made of steel... Too bad, it is not Mu Metal nor Permalloy ).
I have seen shields like a round box for toroidals, but no serious data about the effectiveness.
Capot de Blindage Acier pour Transformateur Torique 90x40mm - Audiophonics
They say it is made of steel, it limits radiation and dress transformers.
( Made of steel... Too bad, it is not Mu Metal nor Permalloy ).
In an ideal world, in order to decrease the induction generated by the primary (which is in fact the root cause of stray fields, even if eliminating it completely would be unwise, because the transformer would cease to function), the normal method would involve an increase of primary turns, ie. an increase in theoretical input voltage.
In the real world, increasing the primary turn count is fraught with difficulties: first, standard voltages include 400V for low to medium power transformers, but no more, and even if you have yours custom-wound, low-power, high voltage windings will require impossibly thin wire gauges, less than 0.02mm, something wind shops do not like very much.
The solution is to reduce the apparent primary voltage by using a dropping impedance, and a resistive one is perfectly OK, except it will dissipate a non-negligible amount of power.
Up to 1W might be acceptable, but beyond you have to find a more effective solution.
Capacitors should be alright, except that any useful value in series with the primary will cause a phenomenon called ferromagnetic resonance.
This can be useful, think of saturated iron regulators, but it goes completely against the idea of magnetic quietness: it will cause saturation, and a strong emission of 50Hz harmonics.
To get rid of this resonance, you need a large ratio between the reactances of the capacitive dropper and the magnetizing inductance.
The only practical way of achieving that is to use the secondary as a primary
No, it is not the same: I suggested both as a remedy: reducing the primary voltage, because it reduces the induction, which is the primary cause of stray fields (they are a percentage of the induction), and reducing the secondary current (~= drawn power), which impacts the percentage.
Note that reducing the output power does not reduce the induction: in fact, the opposite is true, because the induction is related only to the actual magnetic primary voltage, ie. the primary voltage minus the ohmic losses, which slightly increase under load.
This means that stray fields under load decrease because of this (marginal) effect, but they also increase because of the imperfect no-core coupling between primary and secondary.
The resulting effect is completely unpredictable without a complete and detailed knowledge of all parameters, but a sure bet is to reduce both the input voltage and secondary current
It certainly can help, and even if the magnetic fields do not originate from the transformer, it will improve matters
I have resorted to sinewave oscillators used as low power supplies, search my subjects, I have shown some examples.
The difficulty is elsewhere: to reduce HF perturbations, mostly common-mode ones (perfectly doable).
Ready made switching supplies are difficult to use, because they are built around EMC standards, which paradoxically makes them worse in a practical environment: in particular, they rely on class I and/or on Y caps to meet the regulations, which causes countless problems (make a forum search on the subject of noise with SMPS modules or bricks, it is edifying)
Magnetic and electrostatic shields are completely different: electrostatic shields are part of the transformer, and have a drain wire.
Magnetic shields are external, and could be left floating (but they are generally grounded).
Electrostatic shields have no effect on magnetic fields.
Some EI transformers have a "Gauss band": it is a magnetic shield, but effective mostly on stray fields caused by secondary current loading.
I understand that good magnetic shields are made of high permeability materials like Mu Metal or Permalloy.
So copper or aluminium should be no good. However I heard there is some effect with Cu or Al because of eddy currents. I think variable magnetic fields induce eddy currents and they counter interact the magnetic field, but I have no idea wether this gives a worthwhile shielding effect.
I had a very confusing experiment.
Very sensitive microphone préamplifiers in a ALUMINUM enclosure with a PSU inside using a toroidal for +15v -15V +48V.
All went fine testing with microphones, until, I closed the top of the box.....Hum came in. Then, with box lid off, hum went away. Moving the toroidal had definetly an effect but no position was giving no hum in all four preamps.
Top of the box open, about no hum ( amp gains at full max ), while closing the box hum increasing and worse when fully closed.
How can an aluminium plate scatter magnetic field all over inside the box ?
Good when open, bad when closed ?
One more oddity: More hum with microphone inputs shorted than with them left open.
Needless to say that the wiring does follow the rules. No ground loop, shielded cables grounded at only one of the two ends, PSU twisted wires, signal ground only connected at chassis at one point. All circuitry is balanced, XLR in, XLR out.
So copper or aluminium should be no good. However I heard there is some effect with Cu or Al because of eddy currents. I think variable magnetic fields induce eddy currents and they counter interact the magnetic field, but I have no idea wether this gives a worthwhile shielding effect.
I had a very confusing experiment.
Very sensitive microphone préamplifiers in a ALUMINUM enclosure with a PSU inside using a toroidal for +15v -15V +48V.
All went fine testing with microphones, until, I closed the top of the box.....Hum came in. Then, with box lid off, hum went away. Moving the toroidal had definetly an effect but no position was giving no hum in all four preamps.
Top of the box open, about no hum ( amp gains at full max ), while closing the box hum increasing and worse when fully closed.
How can an aluminium plate scatter magnetic field all over inside the box ?
Good when open, bad when closed ?
One more oddity: More hum with microphone inputs shorted than with them left open.
Needless to say that the wiring does follow the rules. No ground loop, shielded cables grounded at only one of the two ends, PSU twisted wires, signal ground only connected at chassis at one point. All circuitry is balanced, XLR in, XLR out.
Last edited:
For static or low-frequency magnetic fields, there is no substitute for magnetic materials, preferably high-perm ones like µ-metal, however conducting materials can also shield for alternating fields: think of IF cans made of brass or aluminum.
That said, for 50Hz it is an uphill struggle: it all depends on the penetration of the magnetic field in the conducting material, and for aluminum or copper it is a few mm (you can find tables or on-line calculators for skin effect on the net).
Your experience with the lid is not uncommon at all: the lid probably closes a loop subjected to a weak stray field, and the rest of the shield and circuit which was hitherto ~equipotential now sees various potentials along the loop thus created.
Idem with the inputs shorted.
You have basically two options: (and since they are not mutually exclusive, I recommend you use both)
-Reduce the ambient field by shielding, distance, orientation or similar means
-Minimize any possible loop area (you have to think creatively, first to locate the offending loop, second to eliminate or minimize it, keeping in mind that it is probably part of the root construction of your device, be it mechanical or in the wiring philosophy/architecture)
That said, for 50Hz it is an uphill struggle: it all depends on the penetration of the magnetic field in the conducting material, and for aluminum or copper it is a few mm (you can find tables or on-line calculators for skin effect on the net).
Your experience with the lid is not uncommon at all: the lid probably closes a loop subjected to a weak stray field, and the rest of the shield and circuit which was hitherto ~equipotential now sees various potentials along the loop thus created.
Idem with the inputs shorted.
You have basically two options: (and since they are not mutually exclusive, I recommend you use both)
-Reduce the ambient field by shielding, distance, orientation or similar means
-Minimize any possible loop area (you have to think creatively, first to locate the offending loop, second to eliminate or minimize it, keeping in mind that it is probably part of the root construction of your device, be it mechanical or in the wiring philosophy/architecture)
Thanks Elvee, I appreciate your help.
This stuff, is not at my place now, so I cannot look into it further now, but I am ready to go at it and definately, your advices are the way to go.
I might have missed a loop.
I am not sure the ground for the 48V phantom power is separate of the +15V -15V ground at the PSU, I must check on this. Phantom ground should be chassis ground XLR pin1 nothing else.
I am thinking of using a hum sniffer; A small wire loop, to locate humming fields.
I wonder whether there is stray fields radiated outside the box when closed. It seems not, but I must make sure of this too.
BTW, I am now at the design of the next generation of these preamps, to go at even lower noise level.
Ultra low noise and ultra hight CMR.
This stuff, is not at my place now, so I cannot look into it further now, but I am ready to go at it and definately, your advices are the way to go.
I might have missed a loop.
I am not sure the ground for the 48V phantom power is separate of the +15V -15V ground at the PSU, I must check on this. Phantom ground should be chassis ground XLR pin1 nothing else.
I am thinking of using a hum sniffer; A small wire loop, to locate humming fields.
I wonder whether there is stray fields radiated outside the box when closed. It seems not, but I must make sure of this too.
BTW, I am now at the design of the next generation of these preamps, to go at even lower noise level.
Ultra low noise and ultra hight CMR.
Last edited:
A requirement was an equipement in a 2U standard chassis powered from 240V mains power. To be used together with other equipements in a mobile bay.
Indeed I can make a version with a remote PSU for a fixed installation, that is the way to ultimately get rid of hums.
This 4 channel preamp has hum at a level audible only with all gains at max and silent microphones. At gains used for real audio there is no audible hum.
Indeed I can make a version with a remote PSU for a fixed installation, that is the way to ultimately get rid of hums.
This 4 channel preamp has hum at a level audible only with all gains at max and silent microphones. At gains used for real audio there is no audible hum.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.