Shielding boost converter for RF transceiver

[radiolistener]

Ok, this is much improved -- stating it as fact stands to mislead readers -- but stating it as opinion and experience leaves open the possibility that you just haven't seen every example, etc.  I apologize for the tone I took before.

Well, do you have a practical example of such SMPS whose noise is not detectable with a receiver that has an MDS sensitivity of -145 dBm @ 500 Hz? (about 0.012 µVrms @ 500 Hz)

To be clear, I do not exclude the possibility that such SMPS with so low noise can be created with modern cutting-edge technology, but I do not believe that they exist and are practically used.

And if you try to implement a DC/DC converter with noise below at least -120 dBm, I'm sure you'll end up using a linear regulator instead of an SMPS. Even if you add a good enough filter, you will still need to deal with leakage issues, such as magnetic coupling with inductor. That's the reality.

[Andy Chee]

To be clear, I do not exclude the possibility that such SMPS with so low noise can be created with modern cutting-edge technology, but I do not believe that they exist and are practically used.

What do you define as "cutting edge" technology?
What do you define as "old traditional" technology?

Modern power supplies have moved away from hard-switched PWM, towards soft-switching ZVS & LLC techniques.  This movement has been happening for many years now, over 20 years.  Much "old traditional" technology are probably in e-waste by now.

[radiolistener]

What do you define as "cutting edge" technology?
What do you define as "old traditional" technology?

With modern technology, you can use various ready-to-use low-noise solutions integrated into chips. Additionally, electromagnetic simulation tools are available to optimize PCB wiring and component layouts to reduce noise leakage.

In contrast, with older technology, such tools and ready-to-use components were not available, requiring significantly more time and money for the development process.

For example, using an old ST1S10 chip with good filters result in noticeable noise on the receiver. However, I don't exclude the possibility that with more modern chips, you might achieve better results. I also don't exclude the possibility that electromagnetic modeling of PCB layout could help to significantly reduce the level of noise leakage.

[T3sl4co1l]

Well, do you have a practical example of such SMPS whose noise is not detectable with a receiver that has an MDS sensitivity of -145 dBm @ 500 Hz? (about 0.012 µVrms @ 500 Hz)

Well sure, we already have one in the thread: for https://www.eevblog.com/forum/projects/shielding-boost-converter-for-rf-transceiver/?action=dlattach;attach=2365825;image above 500kHz, harmonics are dropping about as -80dB/dec(!), so probably by about 3MHz, it's below there.  We can't confirm that from the data given, but it seems likely to be true.  (The peaks at 50 and 100MHz may be aberrations of acquisition, or interference in the setup -- commercial stations, perhaps, but they might also be faint oscillation, for which ferrite beads might suffice.  We'd need a metal box, LISN, coaxial connections, and probably LNA, to have a better idea.  The scope might also not have good dynamic range in those frequencies.)

If a transceiver is insensitive to these lower frequencies but not the higher ones, then that's that, there you have it.

If you contend that there exists a receiver that's sensitive to any frequencies... well yeah I gotta say there are some pretty shit designs out there, that doesn't mean they should be used.

But even if so, these lower frequencies are easy to filter; it's hard to make a simple (as calculated) lowpass filter with more than, say, 2 or 3 decades of asymptotic performance, but evidently this supply only needs two decades worth of filtering; we might put in a two or three staged filter anyway (i.e., cutoffs of say 1kHz, 1MHz, 100MHz), just to be sure it's clean beyond that.

A capacitor multiplier might also be used, with or instead of an LDO.  -100dB at 20kHz isn't a compact filter but it's also not wild, and a staged approach can achieve that easily.

Things are more difficult if such a clean isolation barrier is required, but that too can be mitigated mostly by stuffing either side of the isolation barrier inside a metal box to block its common mode voltage; the CM path might still emit, but what's inside the box(es) doesn't see it.

So too, if the transceiver is sensitive to ambient fields, not just conducted (supply) noise.  Stuff them both in boxes and no one's the wiser.

To be clear, I do not exclude the possibility that such SMPS with so low noise can be created with modern cutting-edge technology, but I do not believe that they exist and are practically used.

And if you try to implement a DC/DC converter with noise below at least -120 dBm, I'm sure you'll end up using a linear regulator instead of an SMPS. Even if you add a good enough filter, you will still need to deal with leakage issues, such as magnetic coupling with inductor. That's the reality.

Curious, why "cutting edge"?

Is a 50/60Hz thyristor supply not "SMPS"?  Is an iron-cored inverter (audio frequency, or maybe just above to avoid whine, as the above example) not an "SMPS"?

Is a low-harmonic class-C oscillator/amplifier, plus resonant tank and rectifier, not an "SMPS"?  (It might not be as efficient as a switcher proper, but it's definitely more efficient than an equivalent transformerless linear supply for a high enough dropping ratio -- 70% efficiency should be feasible.)

Is a magnetron + rectenna not an "SMPS"?...okay, that's kind of out there, one would almost certainly not have to go to such lengths.  But it leaves open a further possibility: why not crank "Fsw" up even further, use LEDs + solar panels?  Total efficiency isn't going to be great (20% if you're lucky?), but that's still better than a linear supply of high enough dropping ratio.

Is an SMPS with linear postreg not an "SMPS"?

In short, we can put the tones and spurs wherever we want, and there exists a filtering strategy to deal with any of them -- given enough space or budget.  The mains pulses of the thyristor rectifier, or say a saturable reactor-controlled rectifier, can be filtered with enough iron and copper, even if we don't allow a minor voltage drop for a linear postreg; the carrier and harmonics of a traditional hard-switched buck, or switching transients of the thyristor, can be filtered with iron or ferrite; and the upper harmonics of fast switchers or VHF+ oscillators can be filtered with ferrite or air-core inductors, or transmission line structures, absorbed in resistive materials, and blocked with shielding.

If what you were saying were completely true, that absolutely positively nothing can hide -- SIGINT would love to hear about it; however, there are such things as signals-approved equipment (various internal signals, adequate filtering and shielding to prevent their leakage), and even filters (mains filtering from some kHz up).  Granted, those filters will have some standard minimum attenuation, not necessarily 150dB or more at any frequency, but as part of a system, easily more than 150dB from privileged sources to potential exterior points can be achieved.

Tim

[shabaz]

"No, this.."

"The real nonsense is the belief"

"example is flawed and frankly, laughable.  "

"you are mistaken."

"Seriously?  "

"just a piece of junk.  "

Since you're such a polite guy, I've attached some traces. I'm sure it won't convince you.

For anyone else; I'm no power supply- (or power supply measuring-) expert, I used a 10 nF capacitor for DC removal (should be fine for measurements beyond a few hundred kHz).

The traces are for 2 MHz chunks, randomly chosen to be 10k-2MHz, 3-5MHz, and 10-12MHz.
By the time I got to the 10-12MHz chunk, the double-shielded coax was insufficient, I was seeing noise from outside, at about 11.9 MHz, so I swapped over to semi-rigid coax at that point. Also needed to drop to 10 Hz RBW by then (the other traces are at 30 Hz RBW), to get the noise floor lower.

I've got max-hold applied, but I only waited for about three sweeps or so to reduce the time spent on the exercise.

[radiolistener]

For anyone else; I'm no power supply- (or power supply measuring-) expert, I used a 10 nF capacitor for DC removal (should be fine for measurements beyond a few hundred kHz).

Just try pairing your DC/DC converter with the receiver and see if its input isn't flooded with noise above the receiver's noise floor, which is around -145 dBm. Then you can return to your measurement setup and test what's wrong with it.

For better results, place inductor of your DC/DC close to receiver input filters inductors and make sure that receiver preamp is powered from it...   

At a glance, I suspect you just selected a frequency with minimal noise and didn't show the frequencies with the actual peak DC/DC noise. Previously, you showed a picture with -90 dBm noise peak at around 50 MHz. Could you please find the exact frequency of this peak and show us the noise details at that frequency?

Also, your previous picture shows -26 dBm (crazy noise!) noise peak at 11 kHz, -30 dBm at 50 kHz, and many other spurs below 1 MHz. Where did those noise peaks go in your last measurement? Did you manage to eliminate them? Could you please show us detailed picture of these noise peaks? 

Such noise below 1 MHz is very-very critical for receivers, because every received frequency is shifted down to exactly this frequency band.

I just wonder why you hide the frequencies with the maximum noise peak and instead decided to show a frequency where the noise is minimal? Did you do this deliberately, or was it a mistake?

Overall, based on your initial measurements, your DC/DC converter is extremely noisy, like a steam engine, and is entirely incompatible with receivers. A noise level of -26 dBm at 11 kHz is a very poor result, and it's the primary issue if you're considering compatibility with receivers. This frequency will propagate through the mixer, leading to noise spurs every 11 kHz step on the receiver. This is why I wrote "Are you kidding?" when you show these results and talked that your DC/DC is not noisy...

[shabaz]

Such noise below 1 MHz is very-very critical for receivers, because every received frequency is shifted down to exactly this frequency band.

This is clutching at straws. Pretty much every architecture will include amplification and signal processing (analog and/or digital) before down-conversion. And you say this is all very poor, but let's put it in perspective. It's 160 uV RMS, according to the automated measurement on the 'scope. Seems pretty good considering I used such a cheap LDO on the output.

[shabaz]

And here's the measurement at 50 MHz as requested. With max-hold turned off since you don't like that.

[radiolistener]

This is clutching at straws. Pretty much every architecture will include amplification and signal processing (analog and/or digital) before down-conversion.

The main issue with low-frequency noise for the receiver is that it propagates through the mixer, resulting in noise spurs at every 11 kHz interval. Consequently, a cluster of extremely high spurs below 1 MHz on your DC/DC converter becomes a significant problem for the receiver. This leads to numerous additional noise frequencies, potentially covering the entire IF bandwidth.

For a receiver, it's better to have higher amplitude noise at 1 MHz than lower amplitude noise at 10 kHz. This is because a 1 MHz spur interval is preferable instead to polluting the entire bandwidth with spurs every 10 kHz.

Also note that it is very difficult (almost impossible) to shield a receiver from low-frequency near-field coupling with your DC/DC converter at short distances, especially from the magnetic pulses generated by the DC/DC inductor. It's very hard to shield magnetic field at near field region.

I have a device that uses a DC/DC module, and due to its small size, it is placed about 40 mm away from the LPF inductors. This arrangement leads to significant noise interference. A field probe detected very high field strength around the DC/DC module. I tried various methods to shield it and even changed the orientation of the inductors, but without success. The only viable solution was to replace the DC/DC module with a linear regulator.

And you say this is all very poor, but let's put it in perspective. It's 160 uV RMS, according to the automated measurement on the 'scope. Seems pretty good considering I used such a cheap LDO on the output.

Just compare your 160 uV noise with 0.012 uV amplitude of processed signal...
Your noise amplitude is 13000 times higher than the signal!
And you say that this is pretty good?

[shabaz]

Yes, it is low noise; many linear supply power bricks for radio receivers will have higher noise or ripple. Some linear supplies will have noise at MHz frequencies. I'm presuming you're not now into audiofool territory.

For the portions of circuitry that needs lower noise, it's a solved problem to reduce noise below 1 MHz; I deliberately chose a low-cost LDO.

[radiolistener]

Yes, it is low noise; many linear supply power bricks for radio receivers will have higher noise or ripple.

Yes, and this is why it's almost impossible to listen radio below 7 MHz in modern cities. The reason - a lot of SMPS power bricks around...

We have power outage here and when it happens the noise floor drops down to 60-80 dB and I can listen very weak DX transmitters at 0.5-7 MHz frequency with no issue. But when power is enabled again, the noise floor rise up and it's impossible to listen even strong stations...

Some linear supplies will have noise at MHz frequencies.

That's not a problem if it uses proper shielding and filters. MHz band noise is more easy to filter and to make shielding. And it has less effect on the mixers and less spurs pollution on IF bandwidth, so it's preferable for receivers. But once its noise is emitted on the air it also will be an issue.

I'm presuming you're not now into audiofool territory.

No. For radio receivers, it is common situation to listen -130 dBm signal in the presence of a +20 dBm signal from a nearby transmitter on a different frequency. Therefore, having a good dynamic range is essential for an effective receiver. You can easily see the difference between low dynamic range and high dynamic range receiver. And this is the difference with audio, where it's hard to detect difference between 80 dB and 140 dB dynamic range.

I deliberately chose a low-cost

Low cost usually means worse results. You get what you pay for.
I recommend to try ST1S10, just to compare results.

[shabaz]

You do ramble. Almost none of what you've written is relevant.

You may have observed issues with the SMPS's you've used, I get that. Those manufacturers might not have cared, or had the requirements that you had or think you have.

On the one hand you state that the noise is a big problem, and then when I point out you'd have the same issue with many linear supplies you state that can be filtered out.

Then you go off on a completely irrelevant tangent about overloading, which has nothing to do with the issue the OP observed. He's not in a military environment (as far as I can tell), it doesn't apply at all. Perhaps I'm heavily mistaken, but he's using some typical sub-GHz modules and ICs not really intended for a battlefield.

Then you tell me low-cost usually means worse results. Very insightful. I selected that LDO because that's what I want to use.

[radiolistener]

On the one hand you state that the noise is a big problem, and then when I point out you'd have the same issue with many linear supplies you state that can be filtered out.

The same problem with linear PSU? 
Could you please explain, what you're talking about?

This is something new... I've never heard about noise issues from linear PSU for receivers...

To be clear, I was talking about switching mode PSU running at 1 MHz... Yes, they are more friendly for receivers, especially when you put their noise to unused frequency band.

Then you go off on a completely irrelevant tangent about overloading, which has nothing to do with the issue the OP observed.

It's not about overloading, but about sensitivity. Sometimes the impact of SMPS noise is much worse due to the low dynamic range of the receiver. This occurs when SMPS noise is so strong that the receiver's dynamic range isn't sufficient to maintain high sensitivity. To avoid overload, the receiver reduces its gain, which in turn reduces sensitivity. This is a common issue in inexpensive Chinese radios like Baofeng and Quansheng.

Overall, all SMPS units can cause interference with radio receivers and sensitive equipment. Sometimes SMPS EMI can affect even digital circuits, for example see inductor orientation issue for RP2350 boards. Therefore, it's preferable to use linear regulators when possible. The only downside of linear regulators is their lower efficiency.

I don't like SMPS due to their interference for receivers.

[T3sl4co1l]

A linear supply might be misbehaving to do this, but squegging or oscillation can indeed happen, and the PSRR at high frequencies is generally poor (better for emitter follower than open-drain LDOs, but there are high performance kinds of the latter too).

Even a normally-behaving one isn't much quieter than shown: typical ref + error amp noise is some µV RMS (say from 10Hz up to the control bandwidth of the device), even for a "low noise" one.

The device shown above, with the harmonics filtered down modestly, and maybe a little cancellation just to suppress the tones, is more or less indistinguishable from a linear supply.

Tim

[radiolistener]

A linear supply might be misbehaving to do this, but squegging or oscillation can indeed happen, and the PSRR at high frequencies is generally poor (better for emitter follower than open-drain LDOs, but there are high performance kinds of the latter too).

self oscillation is not normal mode for linear psu, it's fault mode.
Even if some linear PSU has low PSRR it still don't produce noise, it just suppresses them poorly.

If you need good PSRR you can use ultra-low noise linear regulators like LT3045

The device shown above, with the harmonics filtered down modestly, and maybe a little cancellation just to suppress the tones, is more or less indistinguishable from a linear supply.

If you're talking about an SMPS with hundreds of noise peaks ranging from 10 kHz to 1 MHz at a -26 dBm (14.1 mV peak) level, it's absurd to claim that it's better than a linear power supply. Using such an SMPS with a receiver will contaminate its IF stage with thousands of noise spurs at fairly high levels and approximately 11 kHz intervals. This is disastrous for a receiver.

Even with a poorly designed linear PSU, you would never experience such severe noise issues.

If you look more closely at the picture around 500 kHz, you'll notice numerous noise spurs. So, if a receiver uses a typical 455 kHz IF, it will suffer even more, as receivers are particularly vulnerable to noise around the IF frequency.

Despite the fact that above 1 MHz the noise is suppressed to -90 dBm, below 1 MHz the noise from this SMPS is so bad that I wouldn't even use it to power an LED lamp. Just because it produces too many noise which can affect my radio. It's worse than cheap Chinese SMPS...

Cheap linear regulator will have almost flat line near noise floor with some small peaks near mains frequency and its harmonics.

[shabaz]

A linear supply might be misbehaving to do this, but squegging or oscillation can indeed happen, and the PSRR at high frequencies is generally poor (better for emitter follower than open-drain LDOs, but there are high performance kinds of the latter too).

self oscillation is not normal mode for linear psu, it's fault mode.
Even if some linear PSU has low PSRR it still don't produce noise, it just suppresses them poorly.

If you need good PSRR you can use ultra-low noise linear regulators like LT3045

The device shown above, with the harmonics filtered down modestly, and maybe a little cancellation just to suppress the tones, is more or less indistinguishable from a linear supply.

If you're talking about an SMPS with hundreds of noise peaks ranging from 10 kHz to 1 MHz at a -26 dBm level, it's absurd to claim that it's better than a linear power supply. Using such an SMPS with a receiver will contaminate its IF stage with thousands of noise spurs at fairly high levels and approximately 11 kHz intervals. This is disastrous for a receiver.

Even with a poorly designed linear PSU, you would never experience such severe noise issues.

If you look more closely at the picture around 500 kHz, you'll notice numerous noise spurs. So, if a receiver uses a typical 455 kHz IF, it will suffer even more, as receivers are particularly vulnerable to noise around the IF frequency.

Despite the fact that above 1 MHz the noise is suppressed to -90 dBm, below 1 MHz the noise from this SMPS is so bad that I wouldn't even use it to power an LED lamp. Just because it produces too many noise which can affect my radio. It's worse than cheap Chinese SMPS...

Cheap linear regulator will have almost flat line near noise floor with some small peaks near mains frequency and its harmonics.

Your entire post, comparisons, and conclusions are misleading and dishonest.

You've deliberately chosen to present the output and values pre-clean-up.

[radiolistener]

Your entire post, comparisons, and conclusions are misleading and dishonest.

What is "misleading and dishonest"?
You posted picture from spectrum analyzer which shows a bunch of high power noise spurs from 10 kHz to 1 MHz and noise spike around 50 MHz. The max noise peak is at about 11 kHz = -40 dBV, this is about 14 mV peak, which well coresponds with waveform which shows pulse amplitude about 20 mV.

You will never get so bad noisy output even from cheap linear regulator.

I don't want to offend you, but such a power supply is completely unsuitable for radio receivers due to high noise.
And talking that its noise is better than noise from linear PSU is just ridiculous, because linear PSU don't have noise at all.

Sorry, but it seems like you're joking by posting example with so noisy switching mode regulator and claiming it has lower noise than a linear one. 

[T3sl4co1l]

The device shown above, with the harmonics filtered down modestly, and maybe a little cancellation just to suppress the tones, is more or less indistinguishable from a linear supply.

If you're talking about an SMPS with hundreds of noise peaks ranging from 10 kHz to 1 MHz at a -26 dBm (14.1 mV peak) level

Heyyyy waiwaiwait just a second, reading comprehension here.  I said one thing and you're going on with something else.  To put it simply:

Your entire post, comparisons, and conclusions are misleading and dishonest.

If someone is proposing one example, and you're willfully taking something different as your example, that's not very honest is it, right?

Like... you know you can put capacitors on the thing, right?  You know what a lowpass filter does, right?  And how to apply it?

Tim

[Phil1977]

Your entire post, comparisons, and conclusions are misleading and dishonest.

You've deliberately chosen to present the output and values pre-clean-up.

Unfortunately I can only confirm this statement. The spectrum shown by radiolistener looks like he has connected his analyser to some unfiltered non-sine oscillator.

Now I will ignore him again. But it´s a pitty that really informative posts like from T3SL appear in this environment. We could all be here exchanging knowledge but that doesn't work if someone just wants to be bigheaded and writes as if his impressions and experiences are of more worth than others. I can just repeat myself, in 99% of the cases where you think other people are doing nonsense you just didn't understand it by yourself. I suppose radiolistener is the 1% because he starts to insult anyone he doesn't understand.

[shabaz]

Your entire post, comparisons, and conclusions are misleading and dishonest.

What is "misleading and dishonest"?
You posted picture from spectrum analyzer which shows a lot of high power noise spurs from 10 kHz to 1 MHz and noise spike around 50 MHz. The max noise peak is at about 11 kHz = -40 dBV, this is about 14 mV peak, which well coresponds with waveform which shows pulse amplitude about 20 mV.

You will never get so bad noisy output even from cheap linear regulator.

I don't want to offend you, but such a power supply is completely unsuitable for radio receivers due to high noise.
And talking that its noise is better than noise from linear PSU is just ridiculous, because linear PSU don't have noise at all.

..And that's dishonest too. Anyone is welcome to read the detail, and will easily spot what you've done. If they are unsure, I will gladly point it out to them.

[radiolistener]

Unfortunately I can only confirm this statement. The spectrum shown by radiolistener looks like he has connected his analyser to some unfiltered non-sine oscillator.

I just magnified picture taken by shabaz on his spectrum analyzer for his switching mode regulator and attached that magnified picture to my post...

This is not mine regulator, this is shabaz's switching mode regulator and he claiming that it has better noise than linear regulator...   

Here is his original post with pictures from spectrum analyzer: https://www.eevblog.com/forum/projects/shielding-boost-converter-for-rf-transceiver/msg5635649/#msg5635649

And you're right it's very noisy, and this is what I'm talking about... 

He tried to filter it with linear regulator, but his second picture clearly shows that DC/DC noise peaks are still there, although with a smaller amplitude, but sufficient to create interference for the receiver. And noise at 50 MHz is the same as before linear regulator.

And I even don't talk about noise leakage from DC/DC to receiver components through common mode currents and through near field coupling.

It's very hard to eliminate noise from DC/DC and adding linear regulator on the output cannot help much, because once you create noise in circuit will be almost impossible to eliminate it. It will leak through near field coupling, it will leak through wires, through ground. You even cannot imagine all path how noise can leak from DC/DC to receiver. And this is why it's better to not create the noise than to fight with it.

I suppose radiolistener is the 1% because he starts to insult anyone he doesn't understand.

This is false claim. I don't insult anyone.
I respect shabaz efforts in creating and measuring his power source, he has done nice job.

But sorry, this power source is not suitable for radio receivers.
I just support my thesis that SMPS is evil for radio receivers. Nothing personal.

[T3sl4co1l]

It will leak through near field coupling, it will leak through wires, through ground. You even cannot imagine all path how noise can leak from DC/DC to receiver. And this is why it's better to not create the noise than to fight with it.

Pray tell --

Suppose I take a solid 1m cube of steel, drill a hole deep down into the center of it, and place an SMPS and battery there.  Springs ground the unit inside the hole, and a single wire exits the block.

Suppose we power a receiver from this wire.

How many paths of noise leakage exist?
1. Is there a leakage path through the near field?  (For what frequencies? Also, suppose the SMPS is a broadband comb generator, easy to spot tones in the output.)
2. Is there a leakage path through wiring?
3. Is there a leakage path through ground?
4. Are there any others? You should be able to enumerate them fully in this scenario.

Now, suppose I take to the block with an, erm, particularly long-life grinding wheel, shall we say.  How far can I grind down this cube, until the above list changes?

If your conviction is as strong as you claim it to be -- you should have an answer ready to apply to such a scenario!

Tim

[radiolistener]

1. Is there a leakage path through the near field?  (For what frequencies? Also, suppose the SMPS is a broadband comb generator, easy to spot tones in the output.)

There is missing information about second conductor which will be involved in near field coupling. It will depends on the distance and position against cube. But I won't attempt to calculate exact numbers. At a glance it seems to me that it will require electromagnetic near field modeling.

But I think it can be estimated if you provide more details about distance between cube and conductor involved into coupling and more details about geometry inside cube, especially the distance between SMPS and metal cube inner surface. And the thickness of the cube walls.

Also, in order to estimate it, we need to know metal properties - conductivity and permeability.

2. Is there a leakage path through wiring?

Yes, but its level will depends on impedance at wire feeding point. And impedance will depends on how much the wire going outside of cube. At some length we will get classic GP antenna.

3. Is there a leakage path through ground?

You mean through metal of cube?
Yes, the same as for wire.

4. Are there any others? You should be able to enumerate them fully in this scenario.

Through the hole with wire. It will works like wave guide.
Through far field radiation.

Technically, we can classify the leakage path into three parts:
- through near field capacitive coupling
- through near field inductive coupling
- through far field radiation

Additionally, there is some energy loss due to heating of the metal cube. And some correction which will increase these losses due to wave reflections inside metal walls of cube (between inner and outer surfaces).

Now, suppose I take to the block with an, erm, particularly long-life grinding wheel, shall we say.  How far can I grind down this cube, until the above list changes?

The list will not be changed. But it will change leakage level and proportion between different path. Any change in cube geometry will lead to change.

[T3sl4co1l]

There is missing information about second conductor which will be involved in near field coupling. It will depends on the distance and position against cube. But I won't attempt to calculate exact numbers. At a glance it seems to me that it will require electromagnetic near field modeling.

But I think it can be estimated if you provide more details about distance between cube and conductor involved into coupling and more details about geometry inside cube, especially the distance between SMPS and metal cube inner surface. And the thickness of the cube walls.

Also, in order to estimate it, we need to know metal properties - conductivity and permeability.

Well, I did say steel. Ok fine, AISI 1020 steel, as sand-cast. What more do you need to know?

As for hole, let's say 12mm ID, and wire to make 50 ohm coax. Wire is copper, C110 alloy half-hard let's say.  The copper might be held centered with a solid, foam or spider structure thingy of whatever material; if you insist, let's say it's foamed polyethylene.

2. Is there a leakage path through wiring?

Yes, but its level will depends on impedance at wire feeding point. And impedance will depends on how much the wire going outside of cube. At some length we will get classic GP antenna.

Say we attach the receiver at the surface of the cube. Wire does not extend beyond the cube. Is there still a leakage path?

3. Is there a leakage path through ground?

You mean through metal of cube?
Yes, the same as for wire.

Ok, that's incorrect.  You require further study on this point.

4. Are there any others? You should be able to enumerate them fully in this scenario.

Through the hole with wire. It will works like wave guide.
Through far field radiation.

Technically, we can classify the leakage path into three parts:
- through near field capacitive coupling
- through near field inductive coupling
- through far field radiation

Additionally, there is some energy loss due to heating of the metal cube. And some correction which will increase these losses due to wave reflections inside metal walls of cube (between inner and outer surfaces).

Inner...?

A reminder:

a solid 1m cube of steel,

There are no reflections, the material is wholly absorptive (skin effect dominant) and fields decrease exponentially with depth into the material.

Attenuation will be exceptional, even at fractional Hz.  It seems further study in this direction is also warranted.

By extension, as the block is solid and conductive, near electric field is zero by definition.

The list will not be changed. But it will change leakage level and proportion between different path. Any change in cube geometry will lead to change.

Hmm interesting. If I carve off the corners to make a maximally sized octahedron let's say, what parameters change, and by how much?  Surely you have a quantitative value in mind...

Tim

[Phil1977]

This is false claim. I don't insult anyone.

I do not assume myself to teach or change anyone in its behaviour. But in my world - call me allergic or whiny-voiced or whatever - it´s neither friendly nor professional to call others ideas or stuff crap, especially not if it´s just because it´s not highly expensive or because it´s from China. I also don't assume my experiences are more valuable than others, especially if they clearly are professionals and I am not. I would never call someone stupid because *I* don't understand what he/she is saying.

I still think the example of the R&S spectrum analyser nicely shows that it is possible to efficiently filter out SMPS noise. If this thread would be about technical content and not about personal vanity we could try to figure out what R&S has engineered to suppress SMPS noise that far that it is possible to operate a highly sensitive RF detection instrument. Do they use very good filters, do they use some clever spread spectrum technique (maybe actively pushing the converter frequency away from the momentary analysed frequency ?), do they use some cascaded grounding scheme?

Beside a really annoying "is it possible" discussion it would be nice to have a "how is it possible" approach.