Ferrite Cores: Are They Ruining Your Ethernet Speed?

[@WECB640]

Hi Hayden. I've done the same testing here years ago, and found the same results as you. Ethernet is twisted pair, which is the same as "balanced line" for RF. So by placing a CM choke on the line, it simply forces a current balance between each wire in the pair (IF there was any unbalance to begin with). I can see a case where the user is passing power over ethernet and is doing so using a noisy switch mode supply, then there may very well be RFI generated from the line and yet it should maintain good integrity of data speed transmission on the line. This is a case where a stout CM choke on each end could quiet down the QRM emitted by the ethernet cable. That's been my experience anyway. Thanks for debunking they myth. Great presentation and experiment! 73 OM

[@erlendse]

Honestly, do a image search for "ethernet magnetics". Whoever designed the standard isn't messing around. Isolation transformer per pair, and common mode choke per pair in addition! A core on the cable may help aginst RFI, but they have also done a nice job on the devices.

[@HamRadioDX]

Thanks mate!

[@JeffCowan]

I believe Ethernet cables use twisted pairs and differential mode signalling and thus are somewhat immune to common mode currents and RFI. I wonder if using shielded Cat 6 cable (assuming your terminating devices support shielded terminations) would be better at reducing potential interference.

[@LeeMcc_KI5YPR]

Yes. This type Ethernet uses differential signaling and the cable is twisted pair. This is 1000Base-T and T means twisted. Any Ethernet using RJ45 type connectors is twisted pair.

[@GoonyMclinux]

I've never seen an issue unless it was budget gear and cheap cable. massive RF feedback might do something but thats just skill issues.

[@HamRadioDX]

The location we're having problems has lots of runs of shielded cable running to AP's/links up a tower. There is some evidence there is noise being coupled into these cables - perhaps even on the shield. The next step is to identify all the unshielded cable, fix that up - use a probe to narrow down the noise sources a bit better and also look at potential issues with power supply/POE.

[@GordonHudson]

@@HamRadioDX Definitely on the shield. I have this in the house on the phone network side of the ADSL router. It's CAT 5. I swapped it for CAT6 and fitted clip on ferrites. I thought the shielding would split if I put it round a toroid.

[@don_n5skt]

STP and FTP cable is better and what I actually did when I was getting RFI on my long runs (over 30 meters running through the ceiling) was to switch to fiber optic. It was surprisingly cheap to do so. I will continue to use fiber for my long runs.

[@japanham5973]

I think you hit it.....the cable. Without knowing what cable the person used, this is not for sure, but cheap components very often result in problems. Thanks much for this video.

[@HamRadioDX]

Thanks! Yes using high quality shielded cable would be the first step I think in mitigating any noise, plus the addition of ferrites if needed - or swap out the whole cable if there are speed issues.

[@Super_Sad_Ham]

I suspected whoever it was that reported slowdown probably damaged their ethernet cables when wrapping them around toroids.

[@todddavid3695]

I use a common mode choke on my feed line, and have no toroids on my network, as I don't need them.

[@ThinKkBIGG]

1. Different Types of Ferrites and Their Effectiveness: Each ferrite material has a frequency range where it is primarily effective. It's important to understand the distinction between the full operating range and the range where attenuation is most significant. Ferrites can attenuate RFI without necessarily affecting Ethernet signal if they are not effective in the main band used by GbE. - Common Ferrites Used in Amateur Radio and Analyzed in Relation to GbE Ethernet Signal: Ferrite #31 (NiZn): Effective over a broad range, with significant attenuation between 1 MHz and 300 MHz. Although this range overlaps with the GbE signal (125 MHz for 1 Gb Ethernet), attenuation at 125 MHz is moderate. This material is commonly used for power cables and network cables, as it provides good attenuation of RFI at frequencies below 100 MHz, where much interference is generated. This can effectively reduce RFI without severely impacting Ethernet signal. - Ferrite #43 (NiZn): Effective mainly between 10 MHz and 300 MHz. Although this ferrite may also overlap with GbE frequencies, its primary effectiveness is below 100 MHz. This means it is very effective at attenuating low-frequency interference, often emitted by power supplies and electronic devices, but has minimal impact on GbE signal beyond 100 MHz. - Ferrite #61 (NiZn): Less effective at low frequencies and works better between 200 MHz and 2 GHz, which might coincide with some harmonics generated by GbE systems. It is less useful for attenuating interference below 100 MHz but might impact GbE signals. This means that ferrites of this type are more likely to block useful signal components, especially if poorly applied. It is often used for high-frequency applications but is generally not recommended for Ethernet cables as it does not benefit against RFI in frequencies below 100 MHz. - Ferrites #75 and #77: Ferrite #75: Formulated to be effective in very low frequencies (from a few hundred kHz to about 10 MHz). These materials are primarily used to eliminate low-frequency interference, such as that generated by power supplies or power equipment. RFI/EMI Effectiveness: Very effective for low frequencies, from 150 kHz to about 10 MHz. These ferrites are commonly used to block low-frequency interference on power supplies and power cables. Impact on GbE Signal: No impact on GbE signal. These ferrites do not overlap with the GbE signal frequency and thus will not affect data transmission. Applications: Ideal for power supplies of routers, switches, and other network equipment. However, they are not recommended for Ethernet cables as they do not offer protection against interference in the GbE frequency range. Ferrite #77: Similar to ferrite #75, with slightly better efficiency in the 1 MHz to 10 MHz range. RFI/EMI Effectiveness: Very effective for frequencies from 0.1 MHz to 10 MHz. No impact on GbE signal. Applications: Suitable for power supplies only. Summary of Potential Effects: #31: Low impact on GbE as attenuation at 125 MHz is moderate. #43: Minimal attenuation at 125 MHz, low risk of affecting GbE. #61: Higher potential risk of affecting GbE due to its higher frequency range. #75: Effective for frequencies between 0.1 MHz and 10 MHz - No impact on GbE - Power supplies only. #77: Similar to ferrite #75, with slightly better efficiency in the 1 MHz to 10 MHz range - No impact on GbE - Power supplies only. 2. Impact of Ferrites on GbE Signal: The difference between RFI/EMI interference and GbE signal is crucial. Although some ferrites cover a frequency band that includes GbE, their effectiveness (attenuation) in this range may be very low, limiting the impact on the signal. Ferrites #31 and #43, while covering the GbE frequency, have reduced attenuation in this band, making their impact on GbE performance minimal. #61 may potentially have a more significant impact but is less useful for attenuating RFI below 100 MHz, making it less recommended for Ethernet cables. 3. Placement of Ferrites: Ideally: Ferrites should be placed at the ends of cables to capture RFI emanating from devices (such as routers and switches). However, a critical point is the proximity to the RJ45 connector. If the ferrite is too close to the connector, it might possibly interact with the internal electronics of the case, potentially affecting Ethernet signal performance. Recommended Placement: Upstream of the connector, but far enough to avoid any interaction with electronics. Typically, placing the ferrite a few centimeters from the case is advisable. 4. Ferrites with One Turn vs. Multiple Turns: One Turn (Clipped Ferrites): Provides moderate attenuation, simpler to install. They are generally sufficient for reducing minor interference without affecting GbE performance. Multiple Turns (Cable Wrapping): Offers higher attenuation and is effective against stronger interference. However, this could induce signal loss if misused, especially with ferrites like #61. 5. RFI Generated by Routers and Switches: Most interference generated by routers and switches is below 10 MHz. This makes ferrites #31 and #43 particularly effective for these interferences without negatively impacting the Ethernet signal. 6. Choosing Ferrites for Power Supplies (Routers/Switches): Optimal Placement: Use ferrites on power cables near the routers and switches to limit the propagation of interference from the power circuits. You can also add a ferrite at the end of the power cable connecting to the mains. Recommended Ferrites: Ferrites #31, #43, #75, and #77 are all suitable for power cables. Ferrites #75 and #77 are particularly effective at low frequencies. For power cables, ferrites can be used without risking disruption to Ethernet transmission signals. Ferrite #31: Highly recommended for filtering interference from power supplies, particularly effective for frequencies below 100 MHz, thus reducing interference that could propagate to other network components. 7. Shielding and Grounding: Effective shielding of cables (such as those used in Cat6a and Cat8) is often employed to block external electromagnetic interference. This shielding is grounded to dissipate interference without disturbing the signal. Shielding of router and switch enclosures is also crucial to limit EMI emissions. Conclusion: Using ferrites such as #31 or #43 on Ethernet cables is generally safe for the GbE signal, as they primarily attenuate lower frequencies where RFI interference occurs, without significantly affecting GbE frequencies. Ferrites #31 and #43 are common choices for targeting typical EMI interference without much impact on Ethernet signal. Ferrites #75 and #77 are less likely to interfere with GbE signals but are also not very effective at high-frequency EMI attenuation. Ferrites #61, while less effective against low-frequency RFI, may affect the GbE signal, so their use on Ethernet cables is not recommended. Placement is crucial to avoid interaction with internal electronics of the case. For power supplies, #31 remains an excellent choice for limiting interference without risk to GbE signal. Choosing the ferrite based on the frequencies to be attenuated and positioning it ideally away from sensitive circuit points is important to avoid undesirable interaction.

[@egillis214]

Thanks!

[@hygri]

Going to shielded F/UTP Cat5 at our repeater site made a noticeable difference to the noise floor on 2m - I tried scattering ferrites around before changing all the cables and didn't notice any difference. Swapping the Ubiquiti 24V PoE switch out for a Netonix switch to run all our 5GHz PtP links made the biggest difference and took out all the sproggies.

[@HamRadioDX]

The repeater site we're having problems at (my 2m RFI problem) does use all shielded cable and a POE rail. There is some UTP too between that and the switch. The challenge lies in fixing the problem without actually disconnecting anything (i.e. no outages)

[@hygri]

@@HamRadioDX The joys of shared sites I suppose, argh. Good luck on your quest, watched your other vid & looks like you nailed it with the two-turn clip on ferrite solution. I'm inspired to have a play with my VNA and see what I can find in my junk box that performs well on 2m... cheers! All the best from G-land

[@aandpman]

Old fashioned Cat 5, which you mentioned you were using, I find won't reliably carry gigabit. A lot of times, the network will sync up at gig, then eventually drop back to 100 Mbps. In my line of work, we have been trying to eliminate all the cat5 (patch cables) we find and replace with cat5e (enhanced) or cat6. Cat 5e has a tighter twist and most are stamped "350 MHz", and cat6 still tighter and is stamped to "550 MHz". The higher speeds of gigabit and beyond need the tighter twist to handle the frequencies used in carrying gigabit (reliably anyway). Just my $0.02 worth. Love the videos.

[@HamRadioDX]

I should've clarified. The cable I had was actually Cat5e

[@KD0RC]

Nicely done Hayden! I have also heard that ferrites on Ethernet cables can hurt performance. I was having massive problems with RF getting into my switch and computer via the Ethernet and USB cables, so I loaded them up with ferrites which solved my problems. I am running a Flex 6400, so Ethernet performance is critical to me. I did not do the comprehensive testing that you did, but I also did not perceive any performance difference once I added them. Thanks for confirming my suspicions that the ferrites are fine in this application. 73, Len, KD0RC

[@HamRadioDX]

Thanks Len! Yeah I was intrigued too, so I wanted to test. I'd never heard of any issues with ferrites over ethernet cables until I found those posts in my research.

[@Nick_G7IZR]

Too tight a bend can affect an ethernet cable, but as long as the bend radius is being adhered to, ferrites shouldn't make any difference on speed.

[@robertmeyer4744]

Nice test. All my cables are CAT 6 and have snap on ferrite on them . I never noticed a speed drop at all. I choke everything . Reduces RFI radio noise and before I choked everything my Baofeng on UHF when I key up would take me off line if i was close to the switch. No longer does that with chokes . And the noise on AM radio is much less . I have had bad cables reduce speed and drop outs. 73

[@HamRadioDX]

Thanks Robert

[@joelaut12]

Thanks Hayden for the test. Very helpful diagnosing internet cable issues.

[@richardsparks9904]

Where it causes problem, I wonder if possibly the addition weight on the cable causes problems with the cable termination connector and switch socket.

[@KL5TY-Bill-Alaska]

I recommend cat 6e. If the switch is getting old replace it also check the firmware on the switch. Call your ISP and check with them on upgrading your router. That is where I would recommend starting.

[@manyshnooks]

Differential signaling is pretty resistant to RFI. You really need to be in a very noisy environment, and in that instance you'd be using STP or SFTP

[@HamRadioDX]

Yep. The noisy environment has STP. But many, many runs of it. Could be poor shielding, noisy POE/power supplies or a few of the UTP cables. The next part of the journey!

[@nickpeterson2042]

Shielded cable is the way to go!

[@godarklight]

This might be more applicable to HF and while it sounds silly, one way to avoid RFI from ethernet is to use wifi. This is obviously not going to help at the repeater site though.

[@ChrisRobinsonKF6NFW]

while using wifi, is technically faster then wired. it also adds to the overall noise floor as well in the vicinity of your radio station. This may not be a desired thing. Consider that you may be adding noise at the power source level, the wired router input level, and then any harmonic as well of the wifi band. So in many ways, this is not a desired product at a radio site. If you absolutely need it, then it is something you must deal with. Most of us are not able to separate our shacks from the internet or our homes in such a manner to isolate. However, we can do things to help our situation besides just adding to the noise floor. CHOKE IT!

[@michaelscott8388]

@@ChrisRobinsonKF6NFW wired is faster all around than wireless. Fastest wired is 100 Gb, while wifi is around 9.8 Gb at its fastest.

[@ChrisRobinsonKF6NFW]

@@michaelscott8388 UM no. Physics my friend. Wire has a much greater resistance than air, thus Wifi will always be faster. Your slow downs will be in what you decide to convolute your data stream with!

[@RobB_VK6ES]

If your Gb cable has a broken wire it will often fall back to 100 base using only 4 of the eight wires. Broken wires may be a result of too tight a bend radius particularly if the cable is tensioned in that area. Similarly badly or incorrectly terminated plugs can have the same downgrade effect. Cat cable typically bought in bulk reels will have solid core wires which is intended for fixed installations and not ideal for patch leads which are handled on a regular basis. Terminating with an RJ45 is seemingly a simple process but you would be surprised to learn even experienced cablers will have failures. In my experience the type of RJ45 that allows the wires to pass completely through offers the best chance to check the wiring order before crimping down but they also introduce some potential problems of their own. Finally the elephant in the room, the T568A and T568B connection standards. Do not mix them on the same network. Pick one and stick to it or match the standard that is already being used. It has taken some tight cable speciations and extreme engineering to raise the speeds of ethernet over the years which can be easily brought down with sloppy practices

[@klejb]

When I put a mix 31 toroid on the network cable to my shack computer, It dropped the connection completely when I transmitted on 80m with my Hermes lite 2 and a 300W amp. It was quite frustrating since I then lost the connection to the Hermes. It took me a while to figure out what it was. When I removed the toroid that I had put on a few days earlier, everything worked as normal again.

[@simongordon8182]

Of course there can be two different speed issues 1) doing something messes up speed negotiation so you drop to a lower speed 2) Ethernet packet loss, the way this causes speed loss is especially for TCP/IP as when tcp sees missed packets it slows down then speeds up gradually. This can cause a big drop in throughput

[@WelbyMcRoberts]

I've found that the noise issue is less prominent with "pure copper" cables, rather than the majority of patch cables that folks use which are CCS/CCA (coper clad steel/aluminium). I have a number of cat6 (and cat6a) runs both inside and outside, i have noticed that a clamp on ferrite, on each end, can make a small impact to the noise, but no statistically significant impact to the cable even when running at 10Gbit.

[@housesOTR]

Canon copiers/printers come with a jig with a ferrite core pre-installed. We found it doesn't really hurt unless you are approaching the 100 meter limit. On the other hand the cores totally blocked Netware traffic. The IPX/SPX traffic was too noisy and was blocked. Funny.

[@galaxy0088]

Love these videos. Keep up the great work. NF3R

[@HamRadioDX]

Thanks, will do!

[@akiddoo8113]

I tried several ferrites on my cable and my cat started shedding more and my coffee tasted funny. It did get rid of noise though.

[@HamRadioDX]

Noise is the most important thing. Don't worry about the cat

[@davidportch8837]

useful experiment - thanks

[@DK5ONV]

Good Job Hayden. Love your You Tube Channel...always useful Content shared with us. Tnx for all you do for HAM Radio. 73 de Uncle Günter across the Pond 💯👍🙋‍♂

[@HamRadioDX]

Thanks Gunter! 👍

[@kennethmcooke]

I notice you left the cable coiled up on the floor. Does that not also act as a filter/inductor?

[@ChrisRobinsonKF6NFW]

I use these all over the house, because almost everything is causing RFI. Our house is wired on every device if possible as we try to keep our RF devices to a minimum to keep the RF noise floor also minimum here, but that means RFI noise can go up in other ways. So we choke the noise and place a toroid on EVERY power cable and I/O cable for devices. I think total invested is around $125-$150 in ferrite material for the house, not including the shack. We have to keep in mind as well that the US for the most part does not have the same high speed internet as the rest of the world, and most folks are not using high quality cables. In the US, the noise floor from Ethernet wired devices is still fairly high because we have a lot of CAT5E cable in use. Now to figure out how to muffle the noise from the family during operations!

[@iowahank1]

Closing the door helps muffle the QRM.

[@ChrisRobinsonKF6NFW]

@@iowahank1 You all assume there is a door to close, thats funny stuff! I wish I had that again.

[@simccaffrey]

I tried 1G ethernet over 4x XLR cables recently...worked great lol...it's just crazy robust...a lot of that is to do with the quality of switches and network cards these days...they weren't that good 20 years ago.

[@Ressy66]

Yep, the cable, I've seen a number of issues over the years with cat5e and cat6, off... you guessed it, amazon, aliexpress and ebay, that are so flimsy, so lack of twists, they would fail or only narrowly pass a cat3 test for voice.

[@asknight]

The bend radii you are placing in the cable to install the coils can make more of a difference than the coils themselves.

[@bassangler73]

Thanks for the test!

[@HamRadioDX]

Cheers!

[@and0806]

perhaps you mix type A and B wired ethernet cable, that will give you speed issus. thanks for videos.

[@mrfreddyfudpucker2185]

An ethernet cable is more or less just 4 balanced antenna feeders. Notwithstanding, throughput with iperf isn't the be all end all (you should be getting consistently higher with unintended hosts on an unintended network), but the test could be more scientific by looking at frame errors/CRC as well as frame counts sent vs received at either end. In my experience, PoE devices like cameras are the noisiest, followed by junky devices which induce Switch-Mode noise into everything they come into contact with. The best solution being to use shielded CAT6 rather than UTP. Adding chokes on marginal runs or crummy cable can make them drop frames. But as you have demonstrated short good runs make minimal difference.

[@elvinhaak]

Well... problem is probably that with very tight spots and putting force on the wires, you make problems. So yes, they might be getting problems with the speed after the toroids have been placed. Even with good cables, just too much forces on them. We had speed problems many times in installations when people were using zipties very strongly on the cables. Sometimes working fine again after cutting the zipties away but sometimes just had to replace the cables. They are not made to get big forces on them...

[@FEPLabsRadio]

Great job, mate! Another FUD myth busted!

[@HamRadioDX]

Thanks JIM!

[@StianEklund]

RFI wise, shielded CAT6 does make a pretty big difference if the runs are as short as possible, made a big difference for me at least. Especially as far as RF getting into the modem and noise on HF. Wrapping this kind of cable around toroids is a game of chance, can easily mess up the cable.

[@HamRadioDX]

How does it mess up the cable?

[@StianEklund]

@@HamRadioDX I don't know exactly but in my experience I ended up with a reduced link speed after having wrapped it around a FT-240 43. I think it's likely it damaged one or several of the twisted pairs but I honestly didn't inspect it, I just cut the "bad" part off and put on a new connector. 😅

[@HamRadioDX]

@@StianEklund yeah the cable I used wouldn't be suitable to wrap around a ring toroid, that's why it was very loose in the video so I didn't damage it. Clamps would be better

[@jimrecord6505]

Do the ferrites keep RF in the shack from affecting the Ethernet?

[@Maker_Mikey]

not meaningfully, unless your antenna is inside the shack. My main problem is the noise from the ethernet getting into the shack :( the noise and harmonics of 250Mhz is horrid... which is the primary reason to get shielded ethernet. Even at the 100m mark, I have more problem with noise from the cable than noise into the cable.

[@jms019]

Maybe if somebody uses PoE and they've grounded one end of the supply so saturating the core or something.

[@y_x2]

You may have problem when you use TX close to the Ethernet cable. I saw it.

[@KI7JOM]

Can your switch do PoE? Maybe test that for the next video?

[@HamRadioDX]

Yep it can! Would be interesting to see if PoE devices drop out

[@KI7JOM]

@@HamRadioDX agreed, and perhaps how much noise you can choke out of the Ethernet cable

[@Maker_Mikey]

As others have noted though, you would need to create an alternate ground path. An isolated PoE connection would still create a differential pair back to the power injector.

[@KI7JOM]

​@@Maker_Mikey I've experienced issues with PoE making horrendous HF noise, and I've had to choke the heck out of the cables to reduce the noise. I also had drop outs on the same Ethernet cable. Perhaps they are related? Perhaps the noise was coming from the switch?

[@Shocker99]

What's the construction of your cable? Cat 5 isn't specific enough

[@muzzpvt7164]

CAT5e 7x strands per conductor flexible patch

[@Shocker99]

@@muzzpvt7164 Still isn't specific enough. There are more configurations.

[@Maker_Mikey]

Here is an unintelligent test... Try this on a PoE connection (AT or AF), or if you're brave, BT... (802.3BT that is) and then remove the isolation and ground the PoE powered device to earth! I think you will find very quickly the unpleasantries that a choke will have. Even if you're using CAT6A shielded (aerial or direct burial). As to the principle reason why it should have no impact is because Ethernet is already transformer isolated on both sides, and are by design: differential pairs.

[@LeeMcc_KI5YPR]

Good job. So far. 😉 1. (Easy) Attach an antenna or rf probe to a spectrum analyzer, and see if ferrites reduce emissions. 2. (Expensive) Repeat with 10G-Base-T adapters. I do not own one PC, let alone 2, that could load 20Gb/s, and none with fast enough PCIE slots for the adapters. Hence, expensive.

[@pauls8456]

Quality Shielded cable WITH shielded plugs will make the most difference…….

[@angrydachshund]

Ethernet cable has a minimum bend radius. Violating that might be the cause of the problems that were reported to you.

[@stephenwade8093]

I get a message from Telstra saying I have noise on my connection, did your tests reduce noise or interference at all, I realize you may not have looked at this side of the story,

[@LordGryllwotth]

But it is helping you on the 2M?

[@xephael3485]

IPerf3 is a bad program to test cable signal level. Cable length is also short and you could try 10gbe that's more sensitive to cable imperfections

[@godarklight]

It is an extremely good speed testing program though. I use it when messing around with wifi.

[@HamRadioDX]

I wasn’t testing cable signal level. I was testing throughout. Iperf3 is perfectly fine for such a test. Cable length was 20m, I don’t have anything longer. 10Gbe wouldn’t be in most hams shacks

[@antoineroquentin2297]

@@HamRadioDX How about ferrites on fiber optic cables ;)

[@xephael3485]

@@HamRadioDX 10gbe is becoming more popular and 2.5gbe is also being put on devices. Another thing you might want to look at with your test is that different ferrites utilize different materials that only affect certain frequencies. I believe the ferrites you're using are generally made to remove line noise from standard power connections. So they would be designed for the HZ range not MHZ.

[@xephael3485]

@@antoineroquentin2297 gravimetric ferrites because those can bend light 😜