A "Ground Loop Isolator" From Amazon Turned Out to Be Plain Wire, So the Speaker Built a Real One
イチケン / ICHIKEN
The speaker bought an audio "ground loop isolator" on Amazon, intending to test how well this kind of noise filter works. The product turned out to contain nothing but wire. The video explains what a ground loop isolator is supposed to do, shows how the purchased unit failed, and then measures a homemade transformer-based version to see how much ground loop noise it actually removes and where its limits are.
What a Ground Loop Isolator Is Supposed to Be
The speaker first corrects the framing used at the start of the video. Although it was introduced as a noise filter, a ground loop isolator is really a signal isolator. Noise gets onto a signal because of a ground loop. The isolator uses a transformer to electrically separate the signal path, which removes that effect.
In an audio ground loop isolator, the transformer converts the voltage of the audio AC signal. The goal is to keep the signal amplitude the same at input and output, so the turns ratio is often 1:1. Because the transformer sits in the signal path, the ground loop is broken at that point.
How a Ground Loop Produces Noise
To explain the mechanism, the speaker first draws the circuit without the transformer. Equipment A sends a signal and equipment B receives it, and the two are wired together. Both pieces of equipment are grounded, but for simplicity the speaker treats their grounds as separate, GND-A and GND-B. Suppose a noise source of unknown origin sits between these two grounds. The wiring then forms a loop, and the noise component is transmitted into B.
The speaker adds a qualification: having a ground loop does not guarantee that the loop will cause noise. It only means that it can happen.
Demonstrating Ground Loop Noise
The speaker set up a simple experiment without a transformer. Equipment A was a signal source and equipment B was a speaker. Their grounds, GND-A and GND-B, were connected by wiring, and a separate noise source was inserted between the two grounds. The connection was only a few tens of centimeters long. The speaker notes that in real setups this wiring could run several meters.
With only the signal applied, a 440 Hz sine wave was heard. When a 2 kHz noise signal was injected into the ground path, the noise became audible along with the tone.
The speaker explains the cause this way. When a voltage is applied across the loop, current flows through it. Real wiring has resistance that does not appear on the schematic, so that current produces a voltage drop. The drop changes the voltage applied to the speaker, and that change is heard as noise. With a ground loop isolator in place, the transformer provides electrical insulation and the loop that carried this current no longer exists.
The Purchased Isolator Did Nothing
The purchased isolator was then inserted into the same setup. The 440 Hz tone played as before. When the 2 kHz ground noise was added, it should have been inaudible if the device worked. It was still clearly audible.
The product page indicated that the device used a transformer. If so, the input and output terminals should be basically insulated from each other. The speaker measured resistance between terminals and found them connected. The other terminals were connected as well.
When the speaker opened the unit, there was no transformer. Copper wire simply ran from terminal to terminal. The speaker concluded that the device was just a terminal adapter box with no ground loop isolation at all. They called it a complete scam product and said they had been taken in by it.
Building a Homemade Isolator
The speaker still wanted to know how effective a real ground loop isolator is, so they designed their own circuit. The printed circuit boards were made by JLCPCB, and the capacitors, resistors, and transformers were bought from Akizuki and DigiKey.
Besides the isolation function, the design includes a high-pass filter meant to cut signals below a few tens of hertz. For its capacitors the speaker used Rubycon's PMLCAP, a type of film capacitor. They described it as small with large capacitance, and noted that its capacitance hardly changes even when the applied voltage changes. The speaker said a Rubycon representative told them it is often used in audio applications.
The finished board has the input on the left and the output on the right, with the ground split between the two sides. It carries two channels and uses one transformer per channel. The transformers are Sansui parts bought from Akizuki. The assembly went into an aluminum case. The XLR connectors were taken from the fake product, and the speaker remarked that the result looks quite good.
Listening Test
Without the isolator, injecting ground loop noise made the noise audible as before. With the homemade isolator, the signal was still heard, although the speaker noted that adding the isolator made the sound slightly different, a little "tinny." When ground loop noise was added, the speaker could no longer hear the noise component by ear.
Waveforms: Effective at 50 Hz, Leaky at 40 kHz
Next the speaker looked at the output waveform on an oscilloscope. The circuit had been modified so the isolator could be switched in or out for clarity.
Without the isolator, injecting 50 Hz ground loop noise superimposed a 50 Hz component on the original signal waveform. The speaker said this shows the ground loop noise appearing directly in the output. With the isolator, injecting the same 50 Hz noise left the waveform almost unaffected, with no visible change. The speaker called the isolator impressive at this frequency.
The speaker then cautioned that it is not a cure-all. When roughly 40 kHz ground loop noise was injected, high-frequency content appeared on top of the signal waveform even with the isolator. The speaker attributed this to the coupling capacitance between the transformer windings, which lets that component leak through.
How Much Noise Is Removed Across Frequency
To quantify the rejection, the speaker injected ground loop noise and measured its amplitude on channel 1 and the amplitude at the output on channel 2. Comparing the two channels' frequency responses shows how much of the noise the isolator removes.
At low frequencies the rejection was around −100 dB. The speaker described this as removing the ground loop noise quite well, with a high CMRR. As frequency rose, ground loop noise began to appear at the output starting around the 100 Hz mark. The speaker again attributed this to the interwinding capacitance: as frequency increases, more of the noise component gradually passes to the output side of the transformer.
At 100 kHz the rejection was about −43 dB. The speaker's interpretation is that 1 V of ground loop noise would put roughly 0.01 V onto the output signal. They added that this behavior differs greatly depending on which transformer is used.
Audio Frequency Response of the Homemade Unit
Since they had built the unit themselves, the speaker also measured how well it passes the audio signal. They said the frequency response came out roughly as expected.
Because of the added high-pass filter, low frequencies are not meant to pass. The measured cutoff was about 120 Hz. The speaker felt this might be a bit high and suggested lowering the cutoff somewhat. Above that the response stays flat for a while. Toward the upper end the phase begins to lag and the gain starts to drop slightly, which the speaker attributed to the transformer's characteristics. The high-side cutoff was 47 kHz, which the speaker thinks will not be a problem for audio use. Gain continues to fall beyond that, to about −6 dB at 100 kHz, which the speaker considered acceptable.
The speaker's assessment is that the homemade isolator is usable for audio signals from about 120 Hz to about 47 kHz. They said the high-pass cutoff only needs small adjustments to suit one's preference.
Conclusions and What Remains Unknown
To summarize, the speaker bought a fake isolator on Amazon, built their own, and verified its effect. The real isolator reduced ground loop noise enough to make it inaudible. That effectiveness, however, applies to low frequencies up to a few hundred hertz. Above that, some noise leaks through the capacitance between the transformer windings.
The speaker also admitted they have never actually run into noise caused by a ground loop, so they do not know how effective the isolator would be in a real environment. They plan to keep the unit safe in case they need it in the future.
This video is brought to you courtesy of JLCPCB. Hello. Today we will examine a noise filter I got from Amazon. This is a noise filter for audio applications. Specifically, it is called a ground loop isolator, a filter to reduce the effect of noise caused by a ground loop. What kind of filter is this? We will verify how effective it is.
But first, here is today's sponsor. Go ahead. If you want to make printed circuit boards at a lower cost, JLCPCB is quite recommended. When it comes to printed circuit boards, I imagine a hard green substrate like this one. JLCPCB can also make soft flexible boards. For wearable devices and more flexible component placement, flexible boards are almost mandatory.
At the moment, they want a lot of people to use flexible circuit boards, so they manufacture them at a much lower price: 100 × 100 mm, starting at $2.00 for five sheets. Quite inexpensive. For reinforcing flexible substrates, you can choose from a variety of reinforcement materials: polyimide, FR4, stainless steel, and 3M tape. In addition, EMI shielding film is also available as an option.
Flexible substrate coupons are also available at the link in the description, so we hope you will try making a variety of them, too! JLCPCB will also exhibit at NEPCON Japan, held from January 21, 2026. At the booth, we will distribute coupons that can be used when ordering from JLCPCB, along with other novelties. Please come!
First, we will see how the ground loop isolator is structured. I mentioned noise filters at the beginning of the video. It's more of a signal isolator. The idea is to eliminate the effect of noise caused by a ground loop by using this transformer to electrically isolate the signal.
An audio ground loop isolator has this structure. This part is the transformer. It transforms the voltage of the audio AC signal. Although it transforms the voltage, the transformation ratio is often 1:1 to keep the signal amplitude unchanged at the input and output. This transformer is designed to break the ground loop here.
Next, we will see how the ground loop isolator works. But first, I'm going to remove the transformer. I've removed the transformer and connected the wiring directly. For example, suppose we have equipment A that sends out a signal and equipment B that receives that signal. A and B are wired together. Both equipment A and B are grounded this time, but for simplicity, the grounding points are shown separately as GND-A and GND-B.
Suppose we have a noise source of unknown origin here. It is connected to this part of the ground. In that case, we can actually form a loop for the noise in this section. Then this noise component will be transmitted to B. Just because there is a ground loop does not mean that noise will always be caused by this loop. It means this can happen.
Let's try a little experiment here. The circuit configuration is without a transformer. This time, to make the experiment easier to understand, equipment A is the signal source and equipment B is a speaker. The signal source and speaker are connected by wiring. We put the signal that will be the noise source between GND-A and GND-B here.
This time the load is a speaker, and we have a fairly short connection of only a few dozen centimeters. It is conceivable that this wiring could extend several meters. We also have a separate noise source, which we'll connect between the two grounds here.
I'll try to see how much ground loop noise we can actually hear. First, just the signal. We hear a 440 Hz sine wave like this. Then inject the noise signal into the ground portion. Here we go. The noise component is set to 2 kHz this time. I hope you can hear it.
What is actually happening in this circuit is a loop in this section. In the actual circuit, this is the loop in this section. Current flows when voltage is applied here. It's not on the schematic, but actual wiring has a resistive component. The current flowing through it causes a voltage drop. It changes the voltage applied to the speaker. That's why you can hear the resulting noise.
What happens if we put a ground loop isolator here? The transformer provides electrical isolation, so the loop through which the current flows here is gone.
Now let's actually put in a ground loop isolator and see how well it works. First, only the 440 Hz sine wave is produced from A. It sounds like this. Next, I'll add some ground loop noise. Since I've put in a loop isolator, this noise component shouldn't be audible. Here we go.
I can hear it. I don't know. This 2 kHz noise component really shouldn't be audible, but I can hear it. Something is wrong with this ground loop isolator.
Let's take a look at the product page. There are indications of using a transformer. If it uses a transformer, these terminals should basically be insulated from each other. I'll actually measure the resistance. Here and here—there's some kind of connection. Shall we look at other holes? Here and here are also connected. Finally, here and here. It's all connected. Something's wrong with this one.
Since there's something wrong, I'll try to disassemble it. This way. What the hell is this? Do you see this? Oh, man. This isn't a ground loop isolator. It's just wire inside. This is terrible. This just connects the terminals with copper wire. It's just a terminal converter box. It has no function as a ground loop isolator at all. It is a complete scam product.
We've been stuck with this meaningless scam product, but I need to find out how effective a ground loop isolator is. I'll build my own circuit like this. I designed this circuit and contracted JLCPCB to manufacture the printed circuit boards. The other capacitors, resistors, and transformers on the circuit were purchased at Akizuki and DigiKey. All of them are listed in the description. Check it out if you're interested.
This time, apart from the ground loop isolator function, a high-pass filter is included to cut off signals below a few tens of Hz. In particular, for the capacitor section, we use Rubycon's PMLCAP. This is a kind of film capacitor. It has a small size and large capacitance. Even if the magnitude of the applied voltage changes, the capacitance hardly changes at all. Just hearing about it, it's a capacitor with amazing features. A person from Rubycon said it is often used for audio. So we're going to assemble the circuit.
Here is the finished product. The left is the input side, and the right is the output side. The ground is divided into the input and output sides. The signal is for two channels, and we use one transformer for each. I got the transformers at Akizuki. I'm using Sansui transformers.
Now let's listen to how much it changes with and without the ground loop isolator. First, without the loop isolator, I'll put in the signal. Then inject ground loop noise. It sounds like this.
Next is the case with the ground loop isolator. First, put in the signal. Although the sound is heard like this, it's a little tinny due to adding the ground loop isolator. In addition, ground loop noise is added. I can't hear the ground loop noise component by ear.
Let's look at the actual waveform and see how well it works. I'm looking at the one without an isolator first. I've modified this circuit so it can be enabled or disabled to make the comparison clear. The waveform displayed here now is the signal waveform at the output. No ground loop noise is being injected right now.
Inject 50 Hz ground loop noise here. Then the original signal waveform will look like a 50 Hz signal is superimposed on it. This means the ground loop noise component is coming out in the output signal.
Next is the one with the ground loop isolator. We will inject ground loop noise here. When we do that, it is almost unaffected. Nothing has changed in the waveform. This is the effectiveness of the ground loop isolator.
But this isn't a panacea, either. We'll see what happens when high-frequency ground loop noise comes in. I'm now injecting ground loop noise at about 40 kHz. High-frequency signals are appearing, superimposed on the original signal waveform. This is a component that passes through the coupling capacitance between the transformer windings.
So let's check how much ground loop noise is actually removed. In this setup, we're injecting loop noise. On channel 1, we're looking at the amplitude of the ground loop noise. On channel 2, we're looking at the amplitude on the output side to see how much ground loop noise is removed. We're comparing the frequency responses of channel 1 and channel 2. I'm going to wait for a while at this point.
When the frequency is low, it's about -100 dB, so the ground loop noise is removed fairly well. The CMRR is high. As the frequency increases, usually around the 100 Hz mark, ground loop noise starts appearing at the output. As we discussed earlier, a signal passes through the capacitance between the transformer windings. As the frequency increases, a ground loop noise component gradually passes through to the output side of the transformer.
Since it is about -43 dB at 100 kHz, for example, if there is 1 V of ground loop noise, about 0.01 V will be on the output signal. That's the way I see it. These characteristics vary greatly depending on the transformer used.
Since I went to the trouble of making it myself, we will also look at the ground loop isolator's frequency response. What I just measured was how much ground loop noise can be eliminated. What we're looking at now is how well the audio signal is transmitted. Leave it for a while until the measurement is complete.
The measurement is complete. This is the result. The frequency response is relatively as expected. This time I added a high-pass filter, so the low-frequency components aren't supposed to pass through. If we look at the cutoff frequency, it is approximately 120 Hz. Does that seem a little too high? You might want to bring the cutoff frequency a little lower.
After that, the characteristics remain flat for some time. The phase starts to lag a bit in this area. This is due to the transformer characteristics. At the same time, the gain starts to drop a little. The cutoff frequency on the high-frequency side is 47 kHz. I have a feeling it won't be a problem for audio use. After that, the gain gets smaller and smaller, about -6 dB at 100 kHz. I guess it's okay.
Therefore, the ground loop isolator we made this time will be usable for audio signals from 120 Hz to about 47 kHz. For the high-pass filter section, a few adjustments to the cutoff frequency to suit your preference are all that is needed.
I'm thinking about using it in the future if something goes wrong, so it is housed in an aluminum case like this one. The contents are also connected in this way. As for the XLR terminals, they're from the scam product. It's pretty good-looking when you look at it this way.
Today I verified the structure and role of a ground loop isolator, and how effective it actually is. We had some trouble getting stuck with a fake loop isolator from Amazon, but I managed to make my own and verify it. As for the key benefits, it was effective enough to make the noise component inaudible. But that is for low frequencies up to a few hundred Hz. When the frequency is higher than that, some components pass through the capacitance between the transformer windings.
In fact, I've never experienced noise caused by a ground loop, so how effective it will be in a real environment is unknown. I may end up using it in the future, so I'll keep this one in a safe place.
If you found this video helpful, please give us a high rating and subscribe to our channel! If you have any concerns, you can also leave a comment. Thank you for watching the video until the end.
Article published · Updated