Excellent proof of the design points that you make. I had not thought about prioritizing inputs over outputs. That's a good point and I will add it to my bag of tricks.
Great demonstration. This video make me thinking of my DIY Class AB audio amplifier back in 1985. I solved the Hum noise issue by placing the PCB in vertical position near power transformer. The circuit loops is in parallel to the magnetic fluxes emitted from the power transformer.
Great fix! I never considered how orientation has an impact - if you set the circuit in just the right way, even if its close to the noise source, you can cancel the effect.
So, in summary, to make a circuit more immune to external noise: · Use shorter wires/traces and make loops smaller. · Use PCBs instead of wires. · Use surface-mounted components instead of through-hole components. · Shielding the circuit.
Very nice and practical explanation - this is much more intuitive compared to traditional teaching of loop areas and EMC in general. Good work and inspiring for my future electronics videos :)
I've got an idea of making circuit as 'sandwich' of PCBs can further reduce loop area, because the whole circuit becomes less 'spread'/flat and more compact in volume.
you can decrease more noise by choosing lower value for resistor too, and screening too , make trace guard at high input impedance will also decrease noise
Of course there are multiple things that can be done to make the circuit work better. I was just focusing on the specifics of circuit and component size. All of things you mentioned are valid of course and can be used together with smaller circuit loops to get an even better result.
Great video .. Thank you so much for such deep analysis and sharing with us . I noticed you twisted your power supply wires does it make any difference? Do they reject common mode noise 🤔
Indeed. Keeping the wires twisted will reduce the noise pick-up -> each loop picks up noise in opposing directions so it cancels out. At the same time I used coax cable on the output, again to keep it as shielded as possible. I wanted to keep the cables as immune as possible, so all the measured noise comes from the circuit rather than other sources.
An important issue in high gain (sensitive) amplifiers is maintaining proper order of the ground connections. You don't want to put a decoupling capacitor, or the output ground conection, in the same loop as the amp's input ground connections. It's ideal to keep the different sections grouped, and then "star grounding" (or at least, first group the sections (input, output, filters), then bring them together at a common ground point. In this way, the differences in ground impedances do not induce unwanted signals in the input/neg. feedback path
I have a lot of broken cores from SMPS pcbs. Can I crush them into a power and use very little epoxy, to tamp the mixture in to a mold and make new cores?
It will work, in the sense that the end result will have magnetic proprieties, but to get this to work well you will need to grind the old cores into a very fine powder so that the final shape has more ferrite and less glue. I remember a viewer once told me they did a similar thing but with black sand (ferrite oxide naturally found in rivers) and they had good results.
You are right, when having multiple layers, this principle can be applied by having an underlying ground plane. The principle can be further developed by saying that the smaller the distance between layers the better - typically with 2 layers it will be worse than with 4.
Of course! but I was trying to illustrate that circuit size (regardless of usage of other layers) is an important factor; by adding in a ground plane, this effect would be far more difficult to show.
You put a lot of work in these videos, they are really good. Have you tried the effect of the pcb material. The ratsnest one couples to your inductor through air, that gives a certain capacity. Probably to small to have much effect, but if you would glue it to a piece of FR4 in theory the very small capacity becomes a lot bigger. I never tried to measure this but I think it will perform worse as the ratsnest alone. You really could use a VNA for your experiments. I have done a lot of experiments like you but then with the SDRkits VNWA, back then the only affordable VNA. I learned so much from experimenting with it. (made some tutorials about using a VNA, you can find them on my website ( pa4tim.nl) )
Hello Fred! If capacitive coupling is intended, then yes, replacing the air with FR4 will yield better coupling trough higher capacity. On the other hand, that is not really the situation I was trying to replicate - I was trying to show how the geometry of the circuit will affect noise pickup from a distance - so the thin FR4 already in the board does not have that much contribution. On the other hand if the both the noise source and the sensitive circuit are on the same board - in close proximity, then of course the proprieties of the PCB will have a big impact. If I remember correctly, FR4 has a relative dielectric constant of around 4.
I was trying to say that its not connected to a signal source. Of course, if I would have not added the 100k resistor the output would end up swinging to one extreme, so it would not work.
Iron oxide is not conductive, it only has magnetic permeability, so it will help with magnetic fields, but it should not impact the electric fields. You can get the highest shielding benefits by using a conductive box and then applying the magnetic film onto it. This way you get the best of both worlds.
6:35 Is this a Hall Effect sensor? Can you also induce a current in the inductor by passing a current through the loop (instead of a magnetic field)? Or am i talking about the same thing: magnetic field generated by a current? Is it correct that it also works with DC? Isn't that similar to a Rogowski AC Current Sensor? Why must a Rogowski be a coil instead of a simple loop? Why does the Hall Effect work with both AC and DC, but the Rogowski only AC? Thx! circuitdigest.com/article/how-to-measure-current-in-a-circuit-with-different-current-sensing-techniques
One important concern is that the output also contains the amplified input offset - some op-amps have input offset as big as 10-20mV (NE5532 for example has maximum of 5mV); this input offset is multiplied by the gain and it can be observed as a voltage shift at the output; since the offset can be either positive or negative, your useable output voltage swing is reduced by 2xoffsetx gain (for 5mV and 100 gain, we get 1V) Another important concern is the bandwidth of the amplifier; The actual BW you will get will be the GBW (gain bandwidth product) divided by the gain - so a 1MHz op-amp configured to amplify by 100 can only amplify a 10KHz signal. The GBW in the datasheet is of course a typical value, so a lower number should be used to keep a safe margin. Depending on the application high gain stages are used but this needs to take into account the actual circuit needs, and special op-amps may be necessary (ex with low offset). By using 2 amplifiers with 10x gain, the effects are reduced.
Its there to provide a fixed voltage level on the input - the input is pulled to the reference voltage (Vcc/2). Also this resistor sets the input impedance of the amplifier since its much smaller than the op-amps internal impedance.
@@sumit_kashyap.. Most probably no, it's output probably already has a decoupling capacitor so it doesn't get affected by DC, and if it doesn't have it, the 100k ohm is high impedance so very low current will flow into your music player.
Even if the output of the signal source has a series capacitor, if its a different piece of equipment than the next block its connected to, its good practice to also have a capacitor on the amplifier input - just to make sure no unexpected DC voltages are interconnected.
*empties coffee machine into prevailing wind* GROUND LOOP!!!! meh. the first stage is the most critical, and as everything after that is amplified, targeting SNR here has the largest effect. one also must consider the mains earth grounding and the amplifier grounding... the circuit should reference to the chassis at one point only, and that point should be the input jack. rebuilding old guitar tube amps, i find so many ground loops and complete disregard of basics that it makes one want to scream... a big one is all plugs/jacks and pots being "grounded"... a really bad habit, does more harm than good. unfortunately the old school metal jacks are already grounded and need isolating bushes... except that ONE input jack. a recent one... someone had added a "ground lift" switch... by switching the main earth... tch tch! thats dangerous... deadly dangerous... if you must have a "ground lift"... its to isolate the CIRCUIT from the chassis... never, EVER remove the mains grounding! the "ground lift" is for when you have unavoidable ground loops due to chaining things like foot pedals and PAs together... theyre all hooked up to the mains earth as well as the "signal ground"... ONE is for safety. you dont mess with it. EVER. im not even going to start on general grounding basics...
This experiment cannot be relied upon to arrive at any conclusions about the relationship between a system’s susceptibility to RFI/EMI without further testing. Measuring at 2kHz only fails to tell us anything about how the induced noise in the circuits will vary as the frequency is swept through various ranges. Being that each circuit is basically acting as a tuned antenna, My guess is that each circuit and particular footprint has a resonant frequency where it will become most susceptible to particular frequencies. So, drawing any conclusions about size vs immunity to noise based on this experiment alone seems to be either ignorant of the truth, simplistic, or outright disingenuous. Maybe I am the ignorant one. That is possible, as I am not an expert in this filed, but this is how I see it.
@@FesZElectronics yes, I agree, detuningthe coupling would work, in fact it could work both ways I imagine. Making a larger than necessary circuit would work, for example in some special application (think space travel), but it would certainly create other behaviors, that may not be wanted. The trick would be to incorporate those secondary effects into an overall design. For example, maybe a circuit could be designed to actually serve as an antenna, and decoupled from the primary circuit for use in another part of a larger system, thus eliminating the need for a separate antenna. In a way, repurposing the circuit as is done with devices that use AC Mains wiring circuits to pass Ethernet signals.
Very clear explanations of concepts with practical demonstrations. No possible doubts. I like your videos very much.
Excellent proof of the design points that you make. I had not thought about prioritizing inputs over outputs. That's a good point and I will add it to my bag of tricks.
This is better then many videos I've seen on EEVblog or even w2aew!!! This is fundamental knowledge!
Thank you for explaining and proving those important principle so clearly
very useful and relevant to so many circuits - thanks for all the thought, wisdom and work you are putting into everything you do.
Great demonstration. This video make me thinking of my DIY Class AB audio amplifier back in 1985. I solved the Hum noise issue by placing the PCB in vertical position near power transformer. The circuit loops is in parallel to the magnetic fluxes emitted from the power transformer.
Great fix! I never considered how orientation has an impact - if you set the circuit in just the right way, even if its close to the noise source, you can cancel the effect.
dude this channel needs WAY more subs. Ive been doing this stuff for over a decade and im always learning something here.
Thanks for posting. Production values are great in this video 📷.
Happy to discover your tutorials. They are of very high quality.
So, in summary, to make a circuit more immune to external noise:
· Use shorter wires/traces and make loops smaller.
· Use PCBs instead of wires.
· Use surface-mounted components instead of through-hole components.
· Shielding the circuit.
Except for electrolytic capacitors. Through-hole capacitors have smaller footprint. You can see it in 15:20
Wow very well explained
Great content, as usual. Thanks !
Very informative! Thanks
Very nice and practical explanation - this is much more intuitive compared to traditional teaching of loop areas and EMC in general. Good work and inspiring for my future electronics videos :)
I've got an idea of making circuit as 'sandwich' of PCBs can further reduce loop area, because the whole circuit becomes less 'spread'/flat and more compact in volume.
you can decrease more noise by choosing lower value for resistor too, and screening too , make trace guard at high input impedance will also decrease noise
Of course there are multiple things that can be done to make the circuit work better. I was just focusing on the specifics of circuit and component size. All of things you mentioned are valid of course and can be used together with smaller circuit loops to get an even better result.
Great video
Great video ..
Thank you so much for such deep analysis and sharing with us .
I noticed you twisted your power supply wires does it make any difference?
Do they reject common mode noise 🤔
Indeed. Keeping the wires twisted will reduce the noise pick-up -> each loop picks up noise in opposing directions so it cancels out. At the same time I used coax cable on the output, again to keep it as shielded as possible. I wanted to keep the cables as immune as possible, so all the measured noise comes from the circuit rather than other sources.
@@FesZElectronics Got it. Thank you ☺️
I would have loved to see the effect of a gnd plane, even if it was on a single layer.
Great content, as always!
Really interesting, thanks.
An important issue in high gain (sensitive) amplifiers is maintaining proper order of the ground connections. You don't want to put a decoupling capacitor, or the output ground conection, in the same loop as the amp's input ground connections. It's ideal to keep the different sections grouped, and then "star grounding" (or at least, first group the sections (input, output, filters), then bring them together at a common ground point. In this way, the differences in ground impedances do not induce unwanted signals in the input/neg. feedback path
Your deep analog knowledge shines. Can you do a video about track impedance matching?
Thank u Sir ...
thank you!
Hi great video, did this circuit work with a MCP6021?
I have a lot of broken cores from SMPS pcbs.
Can I crush them into a power and use very little epoxy, to tamp the mixture in to a mold and make new cores?
It will work, in the sense that the end result will have magnetic proprieties, but to get this to work well you will need to grind the old cores into a very fine powder so that the final shape has more ferrite and less glue. I remember a viewer once told me they did a similar thing but with black sand (ferrite oxide naturally found in rivers) and they had good results.
@@FesZElectronics Makes sense. Thanks.
If you have a 2 layer board, have the return path overlapping the forward path on adjacent layers would make for absolute min loop area.
You are right, when having multiple layers, this principle can be applied by having an underlying ground plane. The principle can be further developed by saying that the smaller the distance between layers the better - typically with 2 layers it will be worse than with 4.
What about the induced signal on your cables? I see they are twisted pairs but is it enough?
All the supply cables where twisted in the same way, so even if there was any noise induced that way, it was the same amount in every experiment.
Why am only learning about this great channel now, from Hackaday?
Do better, TH-cam :-(
The use of both sides of the pcb and a ground plane would gain a bit more attenuation.
Of course! but I was trying to illustrate that circuit size (regardless of usage of other layers) is an important factor; by adding in a ground plane, this effect would be far more difficult to show.
You put a lot of work in these videos, they are really good. Have you tried the effect of the pcb material. The ratsnest one couples to your inductor through air, that gives a certain capacity. Probably to small to have much effect, but if you would glue it to a piece of FR4 in theory the very small capacity becomes a lot bigger. I never tried to measure this but I think it will perform worse as the ratsnest alone.
You really could use a VNA for your experiments. I have done a lot of experiments like you but then with the SDRkits VNWA, back then the only affordable VNA. I learned so much from experimenting with it. (made some tutorials about using a VNA, you can find them on my website ( pa4tim.nl) )
Hello Fred! If capacitive coupling is intended, then yes, replacing the air with FR4 will yield better coupling trough higher capacity. On the other hand, that is not really the situation I was trying to replicate - I was trying to show how the geometry of the circuit will affect noise pickup from a distance - so the thin FR4 already in the board does not have that much contribution. On the other hand if the both the noise source and the sensitive circuit are on the same board - in close proximity, then of course the proprieties of the PCB will have a big impact.
If I remember correctly, FR4 has a relative dielectric constant of around 4.
How is the input floating, when its connected via a 100K resistor to 1/2 the input voltage?
I was trying to say that its not connected to a signal source. Of course, if I would have not added the 100k resistor the output would end up swinging to one extreme, so it would not work.
@@FesZElectronics Ah right, that makes sense, thanks.
Can a box made of iron-oxide film provide shielding against electrical and magnetic fields?
Iron oxide is not conductive, it only has magnetic permeability, so it will help with magnetic fields, but it should not impact the electric fields. You can get the highest shielding benefits by using a conductive box and then applying the magnetic film onto it. This way you get the best of both worlds.
6:35 Is this a Hall Effect sensor? Can you also induce a current in the inductor by passing a current through the loop (instead of a magnetic field)? Or am i talking about the same thing: magnetic field generated by a current? Is it correct that it also works with DC?
Isn't that similar to a Rogowski AC Current Sensor? Why must a Rogowski be a coil instead of a simple loop? Why does the Hall Effect work with both AC and DC, but the Rogowski only AC? Thx!
circuitdigest.com/article/how-to-measure-current-in-a-circuit-with-different-current-sensing-techniques
At 1:07 "a single stage amplifier with 101 gain is not that practical". Why is that?
One important concern is that the output also contains the amplified input offset - some op-amps have input offset as big as 10-20mV (NE5532 for example has maximum of 5mV); this input offset is multiplied by the gain and it can be observed as a voltage shift at the output; since the offset can be either positive or negative, your useable output voltage swing is reduced by 2xoffsetx gain (for 5mV and 100 gain, we get 1V)
Another important concern is the bandwidth of the amplifier; The actual BW you will get will be the GBW (gain bandwidth product) divided by the gain - so a 1MHz op-amp configured to amplify by 100 can only amplify a 10KHz signal. The GBW in the datasheet is of course a typical value, so a lower number should be used to keep a safe margin.
Depending on the application high gain stages are used but this needs to take into account the actual circuit needs, and special op-amps may be necessary (ex with low offset). By using 2 amplifiers with 10x gain, the effects are reduced.
@@FesZElectronics Thank you for the explanation. The first point could be nice video idea!
What is use of R1 ?
Its there to provide a fixed voltage level on the input - the input is pulled to the reference voltage (Vcc/2). Also this resistor sets the input impedance of the amplifier since its much smaller than the op-amps internal impedance.
What if your small music player runs on 3.3v . Pulling up the output of music play can destroy output of the music player ?
@@sumit_kashyap.. Most probably no, it's output probably already has a decoupling capacitor so it doesn't get affected by DC, and if it doesn't have it, the 100k ohm is high impedance so very low current will flow into your music player.
Even if the output of the signal source has a series capacitor, if its a different piece of equipment than the next block its connected to, its good practice to also have a capacitor on the amplifier input - just to make sure no unexpected DC voltages are interconnected.
Thanks for your great explanation 😊.
are you trapped inside a football? (referring to your background) xD
Other viewers called it "space tiles" :D
*empties coffee machine into prevailing wind*
GROUND LOOP!!!!
meh. the first stage is the most critical, and as everything after that is amplified, targeting SNR here has the largest effect.
one also must consider the mains earth grounding and the amplifier grounding... the circuit should reference to the chassis at one point only, and that point should be the input jack.
rebuilding old guitar tube amps, i find so many ground loops and complete disregard of basics that it makes one want to scream... a big one is all plugs/jacks and pots being "grounded"... a really bad habit, does more harm than good. unfortunately the old school metal jacks are already grounded and need isolating bushes... except that ONE input jack.
a recent one... someone had added a "ground lift" switch... by switching the main earth... tch tch! thats dangerous... deadly dangerous... if you must have a "ground lift"... its to isolate the CIRCUIT from the chassis... never, EVER remove the mains grounding! the "ground lift" is for when you have unavoidable ground loops due to chaining things like foot pedals and PAs together... theyre all hooked up to the mains earth as well as the "signal ground"...
ONE is for safety. you dont mess with it. EVER.
im not even going to start on general grounding basics...
This experiment cannot be relied upon to arrive at any conclusions about the relationship between a system’s susceptibility to RFI/EMI without further testing. Measuring at 2kHz only fails to tell us anything about how the induced noise in the circuits will vary as the frequency is swept through various ranges. Being that each circuit is basically acting as a tuned antenna, My guess is that each circuit and particular footprint has a resonant frequency where it will become most susceptible to particular frequencies. So, drawing any conclusions about size vs immunity to noise based on this experiment alone seems to be either ignorant of the truth, simplistic, or outright disingenuous. Maybe I am the ignorant one. That is possible, as I am not an expert in this filed, but this is how I see it.
You are right that circuit size will lead to a specific resonance frequency but a single turn loop of this size (diameter
@@FesZElectronics yes, I agree, detuningthe coupling would work, in fact it could work both ways I imagine. Making a larger than necessary circuit would work, for example in some special application (think space travel), but it would certainly create other behaviors, that may not be wanted. The trick would be to incorporate those secondary effects into an overall design. For example, maybe a circuit could be designed to actually serve as an antenna, and decoupled from the primary circuit for use in another part of a larger system, thus eliminating the need for a separate antenna. In a way, repurposing the circuit as is done with devices that use AC Mains wiring circuits to pass Ethernet signals.