Subscriber Discussion

Cat5 Vs Cat6 Vs Cat6a Vs Cat7?

HC
Henry Cannon
Oct 09, 2013

I am about to do an installation of power-over-ethernet (POE) analog cameras in a large apartment building. The cable runs are mostly indoors, with just a few feet of the whole installation to be outside, to reach a couple of outdoor mounting spots.

I foresee upgrading these cameras in 5-10 years, as camera quality and value are advancing rapidly every year. Presumably they will be upgraded with digital cameras. I would like to run ethernet cable now that will be relatively "future ready" and will serve well in 5-10 years, to avoid having to run cable again.

As you likely know, at this point Cat5e or Cat6 is basically standard issue cable, and fairly economical at roughly $75 to $200 per 1000 foot roll. Cat 6a goes up in price somewhat, to $250-$350 per roll. And Cat 7 is a major jump in price, to somewhere in the range of $700+ per roll.

What I'm wondering is which cable features are actually the most important, or not important at all, in choosing a more "future ready" cable? How important is the internal pair-separating spline (found in Cat6)? What about basic shielding: just-inside-the-outside-cable-jacket foil wrapping (found in some Cat6 cable and all Cat7) vs Unshielded Twisted Pair or UTP? Or foil wrapping of each conductor pair (as found in Cat7 cable)? Is one type of shielding material better than another (e.g. Polyolefin (PO) vs. Fluorinated Ethylene Propylene (FEP); FEP seems to be much more expensive.)

I assume that solid conductors are better than stranded, but is that true? Does 23 AWG (or even 22 awg) provide much benefit over 24 AWG? Do larger 23 AWG cause problems with terminations? Is there a benefit of reduced crosstalk in terminating in T568A vs T568B? How do EZ RJ45 (pass-through) connectors affect things vs standard RJ45 ends vs RJ45 connectors with an internal load bar?

See this article for an opinion on RJ45 connector types.

If I get 23 AWG cable (larger cable), which connectors are recommended?

I realize that much of performance degradation comes down to the fine details and quality of terminations. That doesn't even concern me that much, because, the cost of re-terminating, if needed in 5 or 10 years to get improved performance, will be quite cheap (compared to needing to re-pull all cables). So I'm most concerned about which basic construction elements of a cable are worth paying a premium for, and why?

And according to this article, more than 80 percent of cable labeled as Cat6 does not actually perform at Cat6 standard. So, which brands are best/most trusted? According to that article, Belden is one safe choice.

I hope someone can give some good advice on what seems like a topic of potential interest to many, but for which I have found no easy answers.

Thanks!

JH
John Honovich
Oct 09, 2013
IPVM

Background posts on this topic include:

I'll let others with greater knowledge here, comment on your particular questions.

DM
Duncan Miller
Oct 09, 2013

Cat6 is more then sufficient for future proofing you installation. Cat6 can support Gigabit speeds and at this point no cameras are pushing even close to that. A typical HD camera is sending 1-10Mbit of data depending on several factors such as frame rate, compression the codec it is using. In the future as cameras move to higher resolutions such as 4K the compressions used will improve such as the adoption of h.265.

Don't bother with shielded Cat cable. It is not required in most installations and if not done right it will cause more problems then it will solve.

The internal spline isn't that important in my option as long as the cable is rated Cat6. Different manufactures use different techniques to achieve the Cat6 standard.

Solid cable should only be ran inside walls or areas where the cable will not be disturbed. You would never run solid cable as a elevator riser cable or outside the wall as the patch cable between the wall jack and computer.

The conductor sizes make no difference. Just buy Cat6 keystones for Cat6 cable.

I have never hear of a benefit of reduced crosstalk in terminating T568A vs T568B. We always terminate T568A and our failure rate in certification is less then 5%.

Buy a good cable like Belden or General Cable. Don't buy Chinese cable as there is way to much fake cable such as copper clad aluminum Cat6 and non FT4 cable that is marked FT4.

MI
Matt Ion
Oct 09, 2013

Just a slight correction/addition to Duncan's post: Cat5e supports gigabit just fine as well. You don't NEED to get into Cat6 until you start planning for 10GbE. Probably 99.9% of IP cameras out there, though, have only 10/100 FE ports, and I don't expect you'll see much call in the next 5-10 years for GbE on cameras except maybe in the VERY high-end (or just as GbE hardware gets cheap, cameras may implement it, without using anywhere near that bandwidth).

As for the wire size, keep in mind that ethernet spec is still limited to 100m per segment... given how broad the PoE spec is, you'll hit the segment length wall long before PoE voltage loss becomes an issue. In other words: there's no benefit to the larger wire, unless MAYBE you run into cameras pushing the extreme high end of 802.3at.

Avatar
Brian Rhodes
Oct 09, 2013
IPVMU Certified

Fluorinated Ethylene Propylene (FEP) is used in 'Plenum Rated' cable, because it has a melting point of >450 degrees (F).

Other jacket materials burn, smolder, and ignite at lower temps. Given the oxygen-rich environment of plenum spaces, non-rated cabling can act like fuses and carry fire to a widespread area inside a building. The FEP jacket puts a stop to this risk.

Despite lazy specs, there is no need to 'run everything plenum rated' on a job, not even 'riser' runs. Like the OP noted, FEP isn't cheap, and it just adds needless cost.

Avatar
Brian Karas
Oct 09, 2013
Pelican Zero

"power-over-ethernet (POE) analog cameras"

Say what?

"How do EZ RJ45 (pass-through) connectors affect things vs standard RJ45 ends vs RJ45 connectors with an internal load bar?"

If you're installing a cable plant that you want to future-proof, things should really be punched down properly into Cat53 or better rated outlets on both ends, not terminated directly into RJ-45 "plugs".

HC
Henry Cannon
Oct 09, 2013

Thanks for all the great comments already. I am concluding thus far that I should be fine with Cat6, and should avoid any STP shielding that is "ground/drain" type shielding (typically an alumninum foil), because it can cause worse problems if not terminated/handled properly in all places. Since this to be installed in a regular apartment building, at first glance there should be no demanding electrical interference. However, the cable will be run in chases near power wiring mains, at least a dozen cables will be running next to each other for 20+ feet closest to the video monitoring/DVR location and (in a few places) the cables will run close to flourescent lights and inductive sources like HVAC equipment.

With that in mind, I'm wondering if I would be wise to opt for some "basic shielding" like
a 'cable screening' where a single layer or metallic foil or mesh covers the entire bundle of wires.

I'm also curious about the issue of terminating the runs with keystones vs. RJ45 plugs. I understand that a keystone enables a higher-quality termination, but using a keystone means having to add 3 connections (the keystone itself, plus a patch cable with plugs at both ends) rather than 1 connection (just the plug termination), which, logically, seems like introducing 3 times as many failure/problem points, which is generally to be avoided when possible.

Even if I can easily purchase short Cat6 patch cables, and replace them as a commodity, why would I want to introduce those additional potential problem links at both ends of every cable run? Especially when I read elsewhere that more than 80% of patch cables, from a wide spectrum of manufacturers and sources, fail to meet the Cat6 spec when tested, indicating, if nothing else, that patch cables vary substantially in quality and the average end user like myself (with no expensive testing equipment) has no way to determine the quality of a given patch cable.

Thanks again for all the great feedback.

Avatar
Brian Karas
Oct 09, 2013
Pelican Zero

"With that in mind, I'm wondering if I would be wise to opt for some "basic shielding" like
a 'cable screening' where a single layer or metallic foil or mesh covers the entire bundle of wires."

Normally not. Cat5 is a twisted-pair cable, using a balanced signalling method. This is what gives it immunity to noise sources (within reason). Improper shielding techniques *could* potentially cause more interference from reflected signals or things like that. Instead run your cables in the most logical manner. Attempt to have as much separation from noise sources as possible. Don't kink or bend the cables, use quality terminations and so forth.

"Especially when I read elsewhere that more than 80% of patch cables, from a wide spectrum of manufacturers and sources, fail to meet the Cat6 spec when tested"

So, your hand-made crimps are going to be more reliable than professionally made patch cables?

The basic idea of a cable plant with solid-core cabling is that you install it and it "never moves". I know there are some online debates about this, but most people that have done cabling for a living for a long time all come around to the same basic concept. Your connections would/should be secure and weatherized, so the additional connection points won't have any negative impact on the system. Over the lifetime of the cable plant they should have a *positive* impact because you won't ever be moving or flexing that solid-core cable, which can cause breaks and weakened connections over time.

Avatar
Marc Pichaud
Oct 09, 2013

Hi

I would recommand : CAT6/CAT6A which are naturally shielded to support high frequencies used to carry 10 Gb/sec and with a very competitive pricing versus Cat5E. Connectors also are more secured and isolated.

Cat7 doesn't used same RJ45 connectors..

So no more need for SSTP ... and full compatibility with current Ethernet RJ45 devices.

For example, Veracity recommand CAT6A for his long range 600 meters POE connections. (with Ethernet you are still limited to 100 meters)

HC
Henry Cannon
Oct 09, 2013

Normally not. Cat5 is a twisted-pair cable, using a balanced signalling method. This is what gives it immunity to noise sources (within reason). Improper shielding techniques *could* potentially cause more interference from reflected signals or things like that. Instead run your cables in the most logical manner. Attempt to have as much separation from noise sources as possible.

Is the balanced signaling design inherent to twisted-pair ethernet cable affected/degraded by using 2 of the conductors for power-over-ethernet?

And what about bonded pairs? Is that a feature with real value?

Avatar
Brian Karas
Oct 09, 2013
Pelican Zero

No, the benefit of balanced signalling over twisted pair is noise rejection. PoE is also DC instead of AC, which makes it less prone to inducing noise.

The noise immunity of a twisted pair cable, and by relation maximum transmission speed, is very closed tied to the number of twists and the integrity of the twists - maintaining each of the wires in the pair at the same length relative to each other. Bonded pairs in theory help maintain the consistency of the twists around bends and whatnot. However when it comes time to terminate the cables they're harder to work with.

Unless you have any plans for the cable plant to support speeds beyond Gig-E I wouldn't worry about it too much. Gig-E is also a little more forgiving than people realize, because the carrier frequency is not actually 10x what it is for 100 meg Ethernet... Gig-E uses a 4-level encoding scheme instead of a standard binary encoding scheme. So the base frequency is lower, but more data can be transmitted each cycle.

Avatar
Jon Dillabaugh
Oct 09, 2013
Pro Focus LLC

Unless you anticipate camera bandwidth jumping from around 10 Mbit to over 1 Gbit, it makes little sense to worry about cable types. We installed CAT5e outdoor cable for the apartment complex we finished this year. Our cables ran through the attic and had about 2 feet of exposed wiring outside. We could have ran conduit and jboxes, but the customer didn't want all of that visible on the outside of his building. He wanted minimal additions.

This is the cable we used was from Sewell Direct link

Avatar
Marc Pichaud
Oct 09, 2013

You are not right Jon. Shiled is always recommended.

Axis recommended shiledied CAT5E because crosstalk issues. Now that we have CAT6A we solve most crosstalk issues

JH
John Honovich
Oct 09, 2013
IPVM

Axis does recommend shielded, but not everyone agrees. Here is our coverage of shielded vs unshielded.

Avatar
Brian Karas
Oct 09, 2013
Pelican Zero

"You are not right Jon. Shiled is always recommended."

Sorry, I disagree with this. I'm sure there are people that recommend it because most people WON'T run STP, and then you have an excuse to say "it must be the cable plant".

But, talk to any cable supplier and UTP will outsell STP 100:1. If you're not pulling cable through a particularly noisy environment, STP is just going to be an added cost and pain in the ass for no benefit.

Avatar
Luis Carmona
Oct 09, 2013
Geutebruck USA • IPVMU Certified

Axis can take a flying leap. They can say they always recomemnd STP, but it's not always recommended or necessary.

Avatar
Marc Pichaud
Oct 09, 2013

You are right Luis, not mandatory in most cases, when you control the installation and you know exactly where you cables are...now if you add some other electrical deviced or lights in 5 years ....

but when you don't know who will install, where, and if installations will evoluate ... better don't take risk. Stp was sometimes worse than UTP due to shield to ground polarity issues I agree, but than's not the case of CAT6A

Avatar
Jon Dillabaugh
Oct 09, 2013
Pro Focus LLC

Not that I am some proponent of sheilding to begin with, but the link I gave above WAS SHEILDED. Looking closer, this actually isn't the exact wire we used, but very close. The wire we used DIDN'T have a sheild, but just the second jacket.

So I agree that sheilding isn't really needed IMO, but I am far from the expert on sheilding per se. I just have never had issues with UTP. I know that Ubiquiti really stresses the need for STP, but we have never used it for any wireless installs either.

DM
Duncan Miller
Oct 09, 2013

Ubiquity stresses shielded cabel on wireless devices for a tottaly different reason. If you don't drain the static off the antenna of a wireless device that can be caused by high winds or electric magnetic storms you will end up blowing the electronics eventually.

Avatar
Marc Pichaud
Oct 09, 2013

using IP CAT6A is much better than STP, because it does resist much more to high frequencies noise (radio, TV, and now 3G, 4G) and Stp was a shield on a UTP not a high frequencies copper

20 years ago we didn't had these new noises... Most cable manufacturers don't know anything about IP (my experience) dont' know we can do POE on coax or twisted pairs using 2,3 different technologies from EThernet

I do personnaly IPerf bandwitdth testings and measurements as IP trainer and can garantee that you reach comparable real throughputs as you had previsouly with UTP or STP. Now with CAT6A we have a "natural" SSTP

So my believe is : never trust in distributors or manufactures, do the testing with independant software and reproductible testing methods

My believe is also : how many bad installations due to subcontractor putting wires close from power or UPS or high frequencies lights ..

HL
Horace Lasell
Oct 10, 2013

While our applications aren't commercial, our most bandwidth intensive network uses are video of one strain or another.

With recent availability of 10GE over CAT7 copper, backwardly compatibile to 100/1000, we've chosen to do all our future work in CAT7.

For limited duration operations, system interconnects have been a real cost driver. Correctly connecting 20+ cables per rack across multiple racks, then troubleshooting emergent issues, is a real time killer.

Designs requiring only power and one or a few 10GE interconnects per rack simplify set up and break down, justifying the cost premium for both CAT7 and 10GE. Formerly 10GE was available only in fiber, but now copper backward compatibility enables standard applications to coexist without modification.

One decision space is the value of consolidating from a rats nest to a few cables. This may be less relevant in a facility installation, more-or-less star configured pushing network cables out to a few cameras each. If you can't imagine ever demanding anywhere near 10G of bandwidth across these sorts of networks, it could be difficult to justify the CAT7 cost premium. On the other hand, if you can imagine bandwidth growth at the end nodes, or if the architecture supports aggregation and interconnects among system elements become confusing, intermittent, or a constant source of troubleshooting, then those could be areas that could benefit from 10GE growth path or aggregation to a more limited number of interfaces and interconnections.

UM
Undisclosed Manufacturer #1
Oct 10, 2013

Just an FYI- if you do decide to use STP, you'll need the STP patch cables, patch panels, modules and plugs or you'll have wasted a pile of money.

Avatar
Marc Pichaud
Oct 10, 2013

Yes, but the idea here is to use a Cat6A because its natural strong shield and parasite isolation as a normal cable for "normal" video transmissions 100 or 1000 Kb/s or Agreggated links. 10Gb is only used on backbones for larger installations. The Cat6A doesn't impact the global costs, ..but a 10Gb core swicht does.

HL
Horace Lasell
Oct 10, 2013

I agree, the 10GE links require all elements to be STP to suppress crosstalk, while legacy 100/1000 CAT5E and CAT6 can interface at the 10GE switch, enabling aggregation across the higher capacity links.

HL
Horace Lasell
Oct 11, 2013

I'd like to attempt a brief summary of the merits of CAT7:

The general consensus seems to be that there is little justification for the cost driver of CAT7 in most installations.

My offering was, it might be helpful to be be aware of a few cases that may justify CAT7 costs. For our temporary and often-moved installations, we have found that troubleshooting connections among a rats nest of cables has been a major cost driver that has justified reducing our inter-rack connections to a single CAT7 10GE backbone. Another case might be, if you are installing wiring today and can envision the possibility of substantial future growth along a particular cable, CAT7 cabling is affordable insurance against future installation costs. Particularly during construction, once accesses are closed, it's very costly to expand installed capacity. You might want to provide the customer with price vs capability options for 1GE vs 10GE installed capacity at the physical and/or network levels. Some customers might appreciate the future cost avoidance benefits of this forward-thinking approach, while others may see no benefit in incurring those costs. Either way, you're the one who's thinking ahead and offering value-added choices to support their needs.

Users should appreciate that installing CAT7 cabling without proper terminations and connectors will not support 10GE capacity. However, future proofing does not require immediate purchase of the currently rather expensive CAT7 gear. To get a sense of cost, I'll quickly summarize ballpark pricing on 10GE switches as of Oct 2013:

Netgear M7100 24-port: on the order of $5K

Dell Powerconnect 8132 24-port: on the order of $10K

Cisco Nexus 3064-T 48-port: on the order of $40K

While these special circumstances may not apply in most applications, installers might want to consider if and when aggregation or future-proofing makes sense. There may be cases in which it makes sense to structure a proposal with appropriate price/capacity options and especially to make wise choices before laying in the cable.

PM
Peter Morrissey
Oct 13, 2013

It would not be a good idea to count on supporting 10Gig over Cat6. I believe it is supposed to work under 37 meters for 10 GbaseT, but most installations are based upon the 100 meter limit. There are also issues with alien crosstalk and mixing cables. What you really have to ask yourself is not so much about the properties of the cable, but what technology it will be plugged into. Cat5e supports Gigabit Ethernet perfectly. To get to the next speed of Ethernet, you have to go to Cat6A. Not only is Cat6A more expensive for both cable and terminations, but it is also more difficult (more expensive) to install due to the fact that it is thicker, it is stiffer, and there are very strict guidelines as to how tightly it can be bundled. Not only do the conduits to the wall plates have to be bigger, but the raceways have to be significantly bigger as well. I agree with the other posts that it is hard to imagine cameras needing more than a Gig of bandwitch within the next 5-10 years.

jw
jim warner
Oct 14, 2013

> The cable runs are mostly indoors, with just a few feet of the whole installation to be outside,

Independent of cat 5/5e/6/7, if you run a "few feet" of the cable exposed to sunlight, then a few feet of the cable will age from UV exposure. It will crack. Then rain will find its way into the cable and wick its way into your equipment. The fine points about cable performance will be lost once cable is wet inside.

JK
Jayant Kumar
Oct 14, 2013

"Probably 99.9% of IP cameras out there, though, have only 10/100 FE ports, and I don't expect you'll see much call in the next 5-10 years for GbE on cameras except maybe in the VERY high-end (or just as GbE hardware gets cheap, cameras may implement it, without using anywhere near that bandwidth).

As for the wire size, keep in mind that ethernet spec is still limited to 100m per segment"

Starting with the above statement, I would like to take this discussion down a different road:

If cable specs are far better than what is really needed, say CAT 6 in place of CAT5e , for connecting a 2MP camera (no upgrades for the next ten years) to a switch, why can I not run this length to more than 100m, say 120m? Since bandwidth/data rates in use will be much lower, the attenuation too would be lower in the high performance cable. This should allow a longer length to be used before hitting any performance barriers. I am not talking about back bone, just the camera to switch section. True the standard specifies the length at 100m but that specification is for convenience and for performance at 100 Mbps or higher. At lower data rates one should be able to use a longer length. Am I missing something?

Avatar
Marc Pichaud
Oct 14, 2013

Jayant, you can't play with network rules and technologies that Way. EThernet on Twisted pair is straight 100 Meters whatever it's quality is. (sometimes we can have 105 or 110 meters but it's just a lucky combination of switch, wires and plugs without any rules, and unreproducible. Ethernet Technologies are described in 802.3 IEEE rules which are the networking bible. (802.11 describes wireless ..and so and so )

If you want to extand the distance you need small autopowered POE Ethernet Repeaters for example or use an other technology like PLC (600 meters on TP) or VDSL2 ( up to 1,5 Km) or proprietary like Veracity (600meters POE on Twisted CAT6A) or OPtical Fiber or IP over coax ( all technologies are working up to 500 meters and more on good Kx6)

So Cat6A brings only more isolation and less risk with environment parasites, that's it.

Unfortunately, a lot of installers still beleive like you did that 110 meters are like 100 meters , and leave more cables on site, or untwist pairs to insert RJ45 plugs and lose pairs isolation.

.

JK
Jayant Kumar
Oct 14, 2013

While I agree 100% that leaving loose random lengths or untwisting of pairs or loose pairs are not permissible (the characteristic impedance would change and the susceptibility to pickup noise would increase) and I also agree that using longer than 100m lengths should not be attempted by those who are not comfortable with their electronics and maths, I would still like to understand where the technological barrier of 100m really comes from.

It has already been stated that the camera puts out signals at a relative low data rate, hence this should not create the distance limitation that we are discussing. Does the switch communicate with the camera at the full 100Mbps data rate and could that be what can impose the distance limitation when using lower grade cables? Would a better cable then not be able to partly overcome this barrier? Or are we trying to account for some time delay/latency related issue where a longer cable would mean that we may have unacceptable higher latency?

I shall be obliged if someone can go deeper into this and help resolve my doubts.

Avatar
Luis Carmona
Oct 14, 2013
Geutebruck USA • IPVMU Certified

Jayant, I don't think htis is the place for a detailed dissertation on the limitations of Cat5/6/7 cable, though anyone else is free to chime in. But in short, spec's that outline distance limits are like a guarantee. If everything is to spec, then it's a way of saying as long as you don't go past 100 meters, everyhting is pretty much guarenteed to work. If you go out of spec, for example past distance limits, it may work up to a certain point, but it's not gaurenteed.

I saw on another forum where an installer tested different makes and models of cameras connected with lenghts past 100 meters. He found some makes and models made it up to 160 meters, others only made it to 110 meters. Also keep in mind sometimes the device may appear to work when pushed passed limits, say a camera at 120 meters, but it only works intermittently. You cannot expect something working outside of spec will always work.

Please don't be that manufacturer's nigthmare customer where you're using a camera or any other device past distance limits or other specification, and having problems with it working, and you're calling them for support telling them their camera or device isn't working and don't tell them your using it outside specification. And don't blame them if they don't help you with troubleshooting it into you get all equipment back within specifications.

JK
Jayant Kumar
Oct 14, 2013

Luis,

Thank you very much for your detailed reply. Your point that this is not the right forum for a detailed discussion on the relative merits of the various cables and also not to blame the manufacturer or harass him with complaints of unexpected behaviour are both well taken.

However, you have answered my question by quoting others' experiences, in that, yes, the cable length can be longer, though one needs to take cognizance of the various factors involved, including source impedance of the camera, and the termination impedance of the switch, in addition to the characteristics of the cable.

Thanks once again!

Avatar
Luis Carmona
Oct 14, 2013
Geutebruck USA • IPVMU Certified

You are most welcome. Thank for you taking my comments as being constructive as they were intended.

JP
James Pralle
Oct 14, 2013

I see a lot of references to a 100M distance limitation. You must keep in mind that these 100M's include any patch cords necessary for connectivity from the patch panel (typical) to the switch, and in a VOIP or workstation environment, from the wall plate to the device. Per BICSI, EIA/TIA and the rest of the low voltage industry standands, the actual physical distance limitation of a Category cable in an Ethernet based system is only 90M- slightly less than 300 Ft.

Avatar
Marc Pichaud
Oct 14, 2013

I agree for the 90 meters instaed of 100 meters. It's more secured.

To end, instead of talking when you have any doubts , test with bandwidth IP real throughputs... and you wil see if your 0 and 1 go through it. (what we are doing before taking over any old or new twisted pair, telephones wires or coax )

On Cat5E or Cat6A 100 Mb you should get a real 94/95 Mbits everytime (up/down) . And round 900 Mbits on Gb then sudddendly "0" in case of errors because Ethernet can't stand errors and attenuations/latencies, where xDsl, Plc, Wifi can stand -45 dB attenuations...(analog can't go over -35 dB)

JH
Jerome Humery
Dec 23, 2013

Any recommendations for quality cable other than Belden & General Cable?

JP
James Pralle
Dec 23, 2013

CommScope, CommScope Systimax, Superior Essex, Mohawk, take your pick. Including the 2 you referenced, I'd suggest that these are the top 6 manufacturers of category cable in the US- these are your Tier 1 cables.

JH
Jerome Humery
Dec 23, 2013

Thanks so much James....much appreciated.

UM
Undisclosed Manufacturer #2
Dec 23, 2013

We've started using Windy City Wire and they seem to have a pretty good product and good service.

JH
Jerome Humery
Dec 29, 2013

Has anyone used or have an opinion on Black Box cable/products?

New discussion

Ask questions and get answers to your physical security questions from IPVM team members and fellow subscribers.

Newest discussions