USB3 to USBC and USB2 to USBC are in the box, as well as a standard USB3 cable. So any Thunderbolt3/4 or USB3 connection on the Mac would support the unit.

What would be the reason going for USB-C when it's not needed?

Sorry, but I don't get this whining over a connector when a fully sufficient and working standard is used. There's plenty of bandwidth on USB 3, It's compatible with all modern USB connectors, and I'd actually much rather have the sturdy and reliable USB 3 type B connector on the unit, compared to a smaller USB-C connector that I've seen turning into loose ones in frequent use.
Also, having a standard that would support higher bandwidths will not magically change the unit's internals or its bandwidth requirements, so I see more value in using a reliable connector with wider (backwards) compatibility than upgrading a connector just for the sake of having the latest of connector types.

What comes to differences between UFX III and UFX+, at least:

-New analog board with better SNR, up to 7dB THD+N, better converters.
-SteadyClock FS vs. Steadyclock on the UFX+,
-USB 3 Class compliant mode with full 94 I/O channel support (Linux, Mac or iPadOS)
-AES Multi-Mode (up to 3 AES/SPDIF I/O's)
-Lack of thunderbolt due to chip shortage and chips no longer being manufactured

"-New analog board with better SNR, up to 7dB THD+N, better converters. " - This sounds interesting, I'd love some elaborated details on this.

Regarding your response to the port, they should move at least to the USB-C connector as it's the standard at this point and a much better built connector vs this 15 year old port which is of objectively worse quality, the older port is much less durable.

On the point of providing PCIe (USB4 / TB3+4), there is no arguing RME are wizards when it comes to squeezing all they can out of USB.

However, USB drivers are less efficient in general due to the way USB indirectly interfaces with the CPU, USB itself involves a fixed bus latency. TB has the lowest impact possible on processing power, USB does use more CPU power over Thunderbolt, PCIe NICs generally offload much of the compute task from the CPU onto the on-board controller, giving your system some headroom during bandwidth intensive tasks. Compare this to USB, much of the compute task is still on the CPU side. USB is pretty CPU intensive at high throughputs. Thunderbolt should allow for lower CPU overhead to move data, even without DMA. Thunderbolt has DMA (Direct memory access) which bypasses the CPU for more functions. Thunderbolt has less latency, since it "talks" to the CPU directly using PCIe lanes, instead of going through the chipset (USB). TB can support Pitch (for those few who need it).

The fact that RME have the best drivers in the market for USB, that produce latency numbers which are close with Thunderbolt PCIe interfaces doesn't mean that they should stick with legacy USB and stop innovating. RME USB drivers are close in latency (but still not beating) Thunderbolt interfaces. They have already reached the limit of USB in terms of latency. If they managed to combine Thunderbolt's ability to use PCIe lanes with whatever clever techniques they are using on their USB drivers to achieve that low latency, the results would be amazing. Thunderbolt due to PCIe is a superior connection protocol, RME are complacent in having the best USB drivers in the present. Competing Thunderbolt solutions already have lower latency than RME and will keep getting lower with time, while I am sure RME have reached the limit with USB 3.0 by now. Thunderbolt  will always have the ability to go lower, since it talks directly to the CPU, bypassing the platform's chipset. It's not about bandwidth like you will see repeated over and over. It's about how fast the interface can "talk" to the CPU. RME will HAVE to switch to Thunderbolt (or whatever standard of like) one day, if they want to stay competitive/relevant.

At the end of the day we all know why they have released the UFXIII and discontinued the UFX+ and haven't even made a minimal change to use a USB type C port vs a 15 year old. Cost, profit, expense and work required. There's a reason they've made a marketing push both internally (their own YouTube channel and posts) and sponsored content (look to RME sponsored YouTube videos from different channels, Audio University, etc), explaining to people why USB is fine vs Thunderbolt. It's because they aren't willing to make the investment both in time and finance to innovate further which we should all expect from them.

USB-C connector is IMHO more tiny/fragile, I am glad that RME sticks to the current connector.
Higher bandwidth is not needed, doesn't bring any advantage latency wise.
Be glad that you can connect it to every plain USB3 port and, if needed, take an adapter cable.

Seems to be very close to my UFX+...

USB-C connector does not use pcie when used as USB. Thunderbolt share the same connector and it uses "pcie".
One big advantage with the USBC connector in this case it is power capability. Would have been possible to have made this
"bus-powered" with usbc. (Im not sure it would be a good idea though, since it uses more power that a Macbooc pro)

Sorry, but also this makes no sense to me.

#1 There are enough computers in use without USB-C and can't deliver power this way because they do not have USB-C.

#2 An interface in that quality range should get a dedicated PSU to be able to exclude any potential problems with power delivery from the PC side.

#3 The plugs are just too tiny. If you also send power through there, I would be afraid that with a lot of bad luck you might even destroy the electronics (*).

(*)Something like that happened with FW400 when you used power or FW. It was actually designed for hot-swap, but if you canted the plug when plugging something, then you grilled the electronics of the recording interface. Therefore, I would really not consider such a tiny plug and power over it for a device in that price range.

I trust in RME, that they made some good thoughts on what is the best solution for such a unit.

I completely agree that the old USB 3 connector is the best option here, paired with adapter cables, for the mentioned reasons:
- backward compatibility, especially the USB 2 mode works really well for long (and cheap) cable runs when MADI is not needed
- mechanical stability (USB-C bend and unplugs much more quickly)
And including thunderbolt (and maintaining the drivers) would probably raise the price even more.

The few technical improvements I don't find interesting at all and they certainly won't justify an upgrade from UFX+ for us.
I would rather have wished for more DSP features for room correction. Especially a delay and FIR based eq (or at least more than the 3 bands) on the outputs.
Maybe the DSP chip has enough power, so they could be added through a software update...?

#1 USBC will be mandatory for charging laptops in europe from 2026.
#2 A quality device clean up the power from what ever source it get it.
#3 USBC connector handle 240 Watts.

Congratulations on the new UFX III !!

+1 !

UFX III comes with USB 3 type B to USB-C cable, so no adapters required. Works directly with ipad too when set to class compliant mode.

MC could you post this explanation on TB in the RME FAQ maybe?

The topic comes up all the time in fora and it seems the facts are little known.

I’m always rational in gear purchases. I really don’t see what the problem is.

Wow...good luck, buddy.

There is still the UFX II between UCX II and UFX III.

Hope RME understands the amount of UFX interfaces it has.

We have the standard UFX 2, then the UFX+, and now the UFX3, and lets not forget the UFX, i think RME is a little bit UFX crazy.

So we understand that the UFX+ is no more development but that unit is going to be around for many many years just like the UFX is still going strong.

and in production RME has the UFX2 and the UFX3, one good thing is about this is RME no longer needs to rely on 3rd party intel

So independence is confidence, i guess lessen learned, hope it goes well from hear any one can see that this is a master hardware

Congrats

Hi

This post was quoted over on GS and is being discussed there. I wanted to add and/or correct a few details that I think are important for folks stumbling upon this post.

1. Intel has EOL'd their TB2 chips and one earlier set of Titan Ridge, from what I understand, but you can still get the other 7000 series Titan Ridge TB3 device controllers today. They aren't EOL'd on the Intel site, and Mouser has 700+ of each in stock, also not marked as EOL.

I assume you're talking about the computer-side controllers. Mouser shows that as EOL, but interestingly, the Intel site does not.

The 7440 is for peripherals. 7340 and 7540 are for computers.

https://www.mouser.com/c/?marcom=100578564

Intel doesn't show the 7340 or 7540 as EOL, but Mouser does, so that's a bit confusing. This one is the computer-side controller IC.

https://ark.intel.com/content/www/us/en … llers.html

But in any case, TB4 is backwards compatible with TB3 and TB3 peripheral controller chips are still supported and available. I wouldn't see that as a blocker for device manufacturers for TB3. It doesn't kill Thunderbolt peripherals, but it does make TB2 devices potentially obsolete. You don't need to adopt TB4 in your device to stay in the game.

So:

• Creating a TB3 device today: everything is available. TB3 works on TB3 and TB4 ports

• Creating a TB2 device today: nothing is really available to help

But maybe when you wrote "- Intel declared the TB3 chips EOL last year. All TB interfaces that one can still buy are made from stock of these chips. One can not buy these chips anymore, so sooner or later all existing TB audio interfaces will be gone." You actually meant TB2? Many audio interfaces out there today are still TB2. Nothing blocking companies from TB3 devices.

--

2. Microsoft never officially supported TB1 or TB2, even through adapters. So we never were in a position to "cease support" or "no longer support" Thunderbolt 1 and 2 protocols. We started Thunderbolt support with TB3 in Windows 10 around 2015, including support for Thunderbolt 3 audio devices. TB3 gave us the ability to have protections against exploits enabled by DMA over TB. Maybe you didn't mean to imply that Microsoft dropped support for something we previously supported, but it read that way. Backwards compatibility is very important to us.

That said, the Windows PC is an open ecosystem, so many folks have been able to get Thunderbolt 1 and Thunderbolt 2 devices working on Thunderbolt 3 ports, but it's really up to the device in use and the adapter used. It's finicky at best. Others have also gotten Thunderbolt 2 devices working with Thunderbolt 4 ports also using adapters the same way. TB2 backwards compatibility is not a requirement in TB4, but seems some had done so.

If your device is true Thunderbolt 3, then it would be supported on Windows, and would also work with TB4 without any weird adapters. If the device is Thunderbolt 2, regardless of connector or adapter, it is not officially supported by by Microsoft or the TB4 specification according to Intel, and may or may not work at all. https://www.thunderbolttechnology.net/tech/faq (note that there is wildly conflicting information about TB4 -> TB2 backwards compatibility, which is annoying to try to sort through)

TB4 users are in about the same situation as they were with TB3. Some adapters work, some don't. None are officially supported on Windows. The difference is the TB4 FAQ calls out TB2 compatibility is not supported, vs just Windows not officially supporting TB2. But, like I said, people are able to make it work like they could before.

So:

• If you create a device with TB4: all good. I get the complexity argument, however. Controllers are available.

• If you create a device with TB3: You're good. Peripheral controllers are available and not EOL. All the major OS companies support it. On the computer side, the TB4 controllers support this and all the major OS companies have TB4 support as well.

• If you are creating a device with TB2: On Windows this was never officially supported, but would often work. On macOS, this is officially supported with the Apple adapter.

For peripheral developers on Windows, the approach for TB3 audio is almost identical to creating a PCIe driver, but with an additional interface to support hot plug/unplug and rebalancing. There's no TB Audio Class, so all drivers across the operating systems are completely custom (or variations of one of the driver company drivers).

--

3. This is more my opinion than correction or clarification. In my experience, most protocol and connector designers don't really consider pro audio use cases, unfortunately. Firewire was designed for video transfer as I recall, USB for other peripherals with audio classes added later. What some may see as roadblocks or obstacles are often just things the original designers (in the case of TB, that would be Intel and Apple) never really considered. It's the ingenuity of peripheral designers at companies like RME who bend these protocols into shape for pro audio work and enable pro-audio gear to be used. This is not at all unusual given the niche (in terms of numbers, not influence) of pro-audio. So saying that Intel "never wanted audio" seems like an odd statement. My take is they just didn't really consider it a separate priority in the spec, similar to all the other protocols.

--

Microsoft still relies on Intel for Thunderbolt support, so our first-party devices have USB 4 and Thunderbolt 4 on the Intel-based Surface Pro 9, but have USB-C 3.2 on the SQ3 Arm64-based Surface Pro 9. Apple supports Thunderbolt 4 on their Pro and Studio, but it's Thunderbolt 3 on the non-pro Macs, last I checked. As co-inventor of the original TB spec, they may have an agreement in place to allow them to implement Thunderbolt 4 independently from Intel as part of the M2 Pro. That's speculation on my part, though, not fact. But TB3 is part of the USB-IF body of standards, despite still requiring Intel certification. TB4 has some Intel tech in it that maybe they license. I'm not at all involved in that, so again, speculation.

Aside: Apple sells a TB3 to TB2 adapter, but they haven't updated that to say TB4. They're usually really intentional about stuff like that. But they do list compatibility with their M2 pro, so maybe it makes the port run in TB3 mode? I'm not sure what to make of that.

Hope that helps.

Pete Brown
Microsoft

Just confirming we can still use ADAT 1 and 2 output on UFX III? The website has different info for II and III.

UFX II ADAT output spec has:
2 x TOSLINK
Standard: 2 x 8 channels

UFX III ADAT output spec has:
1 x TOSLINK, format according to Alesis specification
Standard: 8 channels 24 bit, up to 48 kHz

Of course, the UFX II is a flagship interface. It is the UFX+, the same revised analog section with the same technical data, but without MADI.

A modular interface would have made both interfaces only pricier and ultimately not given customers the price advantage of the current UFX II with USB2 / without MADI.

That's why I have always pointed out in my product consultations here in the forum that one should carefully consider whether it is not better to invest in a UFX+. Besides, for a while (before Corona, fire at AKM, delivery chain issues) the price difference was small.

Great thread.
My only take so far is that the new unit should have been called the UFX+II and not the UFX III.

I see your point, never the less all this steady clock, jitter free enhancement are now becoming standard all be it RME seems to be the where it all was birthed and a lot of developers have been taking notes,

but like Brainworks from Plugin Alliance rising up again from a flood out, i do admire RME for bouncing back from that fire,

the Fireface UFX 3 has a lot to admire because it allows the user to do very advanced applications with that unit, because of its connections, i guess just like the FF UFX 2 +

Good question! Hope for this too. Was looking for a TB4 laptop (Intel only) but if I understand it all correctly, a Ryzen with USB-only will do for our UFX+ since TB functionality seems to be EOL as well . So in the future (new laptops don't have tb3) we can't (correct me if I am wrong) use the full 188 channels via USB. Major bummer for us.

The studio machine has TB3 with a startech adapter to TB2 (ryzen) and it works super fast. USB on the other hand seems to have more load on my machine. So did this improve as well with the UFXIII?

And lastly will we be able to use the UFX+ with USB3 with full channel count in the future as the UFXIII seems capable...

Everyone is talking about the TB/USB connections, but I am asking: why not AVB or Dante? THIS would've made it a UFX Gen 3 in my opinion.
Right now it's just a quick fix for the discontinued UFX+.
I thought about buying, but what I really need is a future proof unit to work in modular environments.

If a UFX AVB/Dante (with MADI) drops, I'll buy it the day it comes out and add another 12micD on top. But with the current UFX III I feel like I would regret the purchase in the near future.

I tend to agree. I would really love to see a one-box-wonder that I could use for all different needs and applications. Something that would compete e.g. with Avid's MTRX Studio, because as of right now, that one seems to be the closest option.

UFX III plus Dante connectivity would be a real deal, and like you, I would also be shopping for 12micD or two on top of that.