M3 MBA won't support two NVMe Thunderbolt externals....not enough power

[Mike Boreham]

Rather surprised and disappointed to discover that if I connect two NVMe TB drives (two different Orico, also tried Envoy) into my M3 MBA, the second one to be plugged in gives:



Doesn't matter which port is used first or second. I can plug a Samsung T7 into second port and it is OK.

I normally connect the two NVMe TB drives through an OWC TB powered hub with no problems, but today I tried both direct into the MBA.

I can't see any info about port power in System Report. I thought there had been some in the past, but can't see any now.

 

[bzgnyc2]

I don't recall seeing the port power in System Report for Thunderbolt but it does show power usage/extra power for USB. Though it's a little confusing that macOS reports information on Thunderbolt/USB4 buses/devices completely separate from USB [implicitly 3.x and below] buses/devices with a fairly different sets of attributes.

By specification, the minimum power per port in the Thunderbolt 3/4 standard is 15W. However, if you pull the full 15W from each port = 30W, that's not going to leave a lot for the rest of the laptop. Are you using the bundled 30/35W power supply? I wonder if it would support both ports at 15W if you used a 70W charger?

In any case, it would appear that Apple's implementation of Thunderbolt 3/4 is not quite compliant with standard even if the reasons are practical...

 

[Mike Boreham]

I wonder if it would support both ports at 15W if you used a 70W charger?

Thanks and sorry for late response. That does seem to be the issue.

I was using the standard 30W charger. I don't have the Apple 70W charger but my normal usage is with an OWC powered Thunderbolt dock which supplies 60W to the MBA, and I can connect all three of my NVMe Thunderbolt enclosures to the OWC without getting the insufficient power message. I only discovered this limitation when trying to simplify by not using the OWC. So you are right.

 

[bzgnyc2]

Thanks and sorry for late response. That does seem to be the issue.

I was using the standard 30W charger. I don't have the Apple 70W charger but my normal usage is with an OWC powered Thunderbolt dock which supplies 60W to the MBA, and I can connect all three of my NVMe Thunderbolt enclosures to the OWC without getting the insufficient power message. I only discovered this limitation when trying to simplify by not using the OWC. So you are right.

I was actually thinking of 70W to the MacBook and didn't realize you were powering the Mac through the OWC dock. So it would be intereresting to know if the MacBook Air would support two NVMe enclosures directly when on the 70W charger though I doubt it given a) it didn't work with 60W from the OWC and b) Mac Mini and Mac Studio owners have the same issue just with 3 enclosures connected directly despite both systems having much beafier power supplies.

Then the only workaround at the moment appears to be the one you found -- either connect the drives through TB hubs (with their own power supply like you've done with the OWC) or use self-powered (not bus-powered) NVMe enclosures (of which I am not sure there are any TB5 out there yet).

I think Apple should document this better. I appreciate the engineering challenge of powering all TB ports @ 15W simultaneously but to the extent Apple can't do it/can't do it without too many other tradeoffs (cost, size, weight, etc), it should include these limits on the spec page (the same way they do as far as monitor resolutions).

 

[Mike Boreham]

I was actually thinking of 70W to the MacBook and didn't realize you were powering the Mac through the OWC dock. So it would be intereresting to know if the MacBook Air would support two NVMe enclosures directly when on the 70W charger though I doubt it given a) it didn't work with 60W from the OWC and b) Mac Mini and Mac Studio owners have the same issue just with 3 enclosures connected directly despite both systems having much beafier power supplies.

Then the only workaround at the moment appears to be the one you found -- either connect the drives through TB hubs (with their own power supply like you've done with the OWC) or use self-powered (not bus-powered) NVMe enclosures (of which I am not sure there are any TB5 out there yet).

I think Apple should document this better. I appreciate the engineering challenge of powering all TB ports @ 15W simultaneously but to the extent Apple can't do it/can't do it without too many other tradeoffs (cost, size, weight, etc), it should include these limits on the spec page (the same way they do as far as monitor resolutions).

Thanks. Not sure I have been completely clear. What I found was:

-Two NVMe TB drives were not supported when I connected them directly to the MBA with the MBA powered by just the Apple 30W charger.

-Three NVME TB drives connected to the OWC dock are supported, with the dock supplying 60W to the MBA.

-I did not try connecting the two NVMe TB drives direct to the MBA with the OWC supplying 60W to the MBA.

But I agree, I suspect two drives connected directly to MBA is probably not going to work with either the 60W OWC or 70W Apple charger powering the MBA. I think the power needs to go directly to the drives from the OWC dock, and that even with a 60/70W charger powering the MBA, the drives will not receive enough power.

I could actually try this when I get home, and I have an Apple 87W charger to try as well.

Will report back.

 

[Mike Boreham]

I could actually try this when I get home, and I have an Apple 87W charger to try as well.

Will report back.

We were right. Overall conclusions:

1. Two NVMe TB drives plugged directly into the MBA will not work even when the MBA is powered directly by the 87W Apple charger (and obviously not with the 60W OWC dock charger)

2. One NVMe TB drive will work even with the MBA on battery power.

3. Three NVMe TB drives work if all three are plugged directly into the OWC dock, with 60W being supplied to the MBA.

(In case the thought occurs to you, it is not physically possible to test whether the three drives plugged into the dock works without the 60W being suppled to the MBA, because the same lead makes all connections between the dock and the MBA.)

EDIT. A wild afterthought. The normal rules are that a charger with higher rated output will not supply any more power than the device requires. The M3 MBA is an exception, being supplied normally with a 30W charger but with option for 70W. My thought was whether there is a special characteristic of the 70W MBA charger itself which enables the MBA to accept 70W. If there is, then my older 87W charger might only be supplying 30W to the MBA because it would not have this special feature. The same would be true of the OWC dock 60W supply.

My hunch is that two TB drives will still need to be plugged directly into the OWC dock.

 

[bzgnyc2]

We were right. Overall conclusions:

1. Two NVMe TB drives plugged directly into the MBA will not work even when the MBA is powered directly by the 87W Apple charger (and obviously not with the 60W OWC dock charger)

2. One NVMe TB drive will work even with the MBA on battery power.

3. Three NVMe TB drives work if all three are plugged directly into the OWC dock, with 60W being supplied to the MBA.

(In case the thought occurs to you, it is not physically possible to test whether the three drives plugged into the dock works without the 60W being suppled to the MBA, because the same lead makes all connections between the dock and the MBA.)

EDIT. A wild afterthought. The normal rules are that a charger with higher rated output will not supply any more power than the device requires. The M3 MBA is an exception, being supplied normally with a 30W charger but with option for 70W. My thought was whether there is a special characteristic of the 70W MBA charger itself which enables the MBA to accept 70W. If there is, then my older 87W charger might only be supplying 30W to the MBA because it would not have this special feature. The same would be true of the OWC dock 60W supply.

Chargers these days basically have onboard computers running their own microOS so they can get complicated. In particular there is a whole negotiation of voltage and current between the charger and the computer/etc under the USB protocols. However, did find a site that reported that the 87W charger can provide all the power that the MacBook Air M4 (and presumably M3) can pull -- basically ~ 3.5A @ 20V = 70W.

Here are the details which I have to highlight because somebody really went to town documenting all this:

Charging Review of MacBook Air 13-inch M4

Now, the 2025 13-inch MacBook Air with the M4 chip further solidifies its reputation as the world’s best thin-and-light laptop, offering significant performance gains, smarter user experiences, and a more sustainable design. The new 13-inch MacBook Air with M4 features Apple’s cutting-edge 3nm chip,

www.chargerlab.com

My hunch is that two TB drives will still need to be plugged directly into the OWC dock.

Yes, I believe that is unavoidable with MacBook Air M3/M4 unless self-powered Thunderbolt/NVMe enclosures are used (which are relatively uncommon and I don't think I've seen any self-powered TB5 enclosures at all yet...and in any case a lot less clean). The issue is similar on Mac Mini M4 (base and Pro) and Mac Studio except the limit is two bus-powered TB NVMe enclosures rather than one.

Note there are two requirements for TB NVMe enclosures: 1) bandwidth and 2) power. The issue doesn't seem to be bandwidth or simultaneous devices just power. That's why the issue goes away once you insert a dock with its own power supply in the chain.

The TB spec allows devices to draw up to 15W and TB NVMe enclosures typically want all that. So two TB NVMe enclosures connected simultaneously will need 30W or 6A @ 5V. I'm guessing routing that around the system board would require Apple to make things bulkier than they wanted. Once you insert the TB hub into the chain, it assumes responsibility and your MacBook is relieved for providing all that power.

Last, using USB 3.x enclosures should avoid the issue as they are typically limited to 7.5W. Others' tests with the Mac Mini M4 have shown no issues simultaneously supporting 2xTB NVMe enclosures + 1xUSB SSD. I'm guessing your MacBook Air could support 1xTB NVMe + 1xUSB SSD simultaneously as well. Obviously the latter will run at a fraction of the speed of your NVMe enclosures.

 

[Mike Boreham]

Last, using USB 3.x enclosures should avoid the issue as they are typically limited to 7.5W. Others' tests with the Mac Mini M4 have shown no issues simultaneously supporting 2xTB NVMe enclosures + 1xUSB SSD. I'm guessing your MacBook Air could support 1xTB NVMe + 1xUSB SSD simultaneously as well. Obviously the latter will run at a fraction of the speed of your NVMe enclosures.

Thanks for all that! Yes, as I said in the OP, one NVMe TB and a USB Samsung T7 in the other port worked fine.

The detail in that Chargerlab link is staggering and completely negates my "afterthought" about whether the MBA 70W charger had some special feature.

 

[splifingate]

Just managed to get to seeing how my new 13" M4 Air BTO w/70W charger handles these things.

Interestingly, I can attach and use both my Acasis TBU405, and my Maiwo 1695, with/without the charger.

 

[bzgnyc2]

Just managed to get to seeing how my new 13" M4 Air BTO w/70W charger handles these things.

Interestingly, I can attach and use both my Acasis TBU405, and my Maiwo 1695, with/without the charger.

Didn't expect those results and interesting so thank you for sharing...just when you think you understand...

Brainstorming explanations:
a) MacBook Air M4 has different limits than MacBook Air M3
b) Apple treats USB4 different than TB3
c) The Maiwo 1695 actually draws significantly less power than the Acasis TBU405 and/or implements power management differently

I'd be surprised if a) but of course Apple could have tweaked this without making a big deal of it.

I would also be surprised by b) especially since USB4 was based on TB3 and Apple even puts them in the same System Information section.

I've heard USB4 only guarantees 7.5W compared to TB3's 15W and so maybe the ASM2464PD chip used in the Maiwo is more flexible about working in environments with less available power? I wouldn't have expected this as the ASM2464PD chip is reported to run hot and I assume the Maiwo needs to support NVMe SSD that draw >7.5W at peak not counting the ASMedia chip. Or maybe it is able to use USB-PD to draw power at higher voltage/lower current compared to TB3? That would be an interesting turn of events on this bus-powered NVMe drives issue.

Of course there's a small team at Apple who would know the actual hardware/engineering limits of each system offhand and another team in the OS division who knows how they programmed the limits but not sure how we would get questions around this to and from them...

 

[splifingate]

just when you think you understand...

USB4 on Silicon Macs regularly confuses me.

Pressed for time atm, but I have some interesting additions to add to your ideas when I get time..

 

[splifingate]

So, "USB4" on Apple Silicon.

A little backstory to the fore-posted screenies.

When I finally got-around to testing these dual externals (at the same time), one was listed as populated at 40Gb/s on the Thunderbolt/USB4 Bus(x), and one was listed as being populated at 10Gb/s on the USB 3.1 Bus.

I disconnected each, in succession, and randomly, yet the channel-partitioning status stayed the same (even with multiple access of About This Mac > More info...).

Only when I went to AmorphousDiskMark did the USB Bus external find-ground. The "10 Gb/s USB" Maiwo started pumping-out 2Tb/s readings via ADM! And, it had switched to populating the Thunderbolt/USB4 Bus(x)!

As with all ASM2464PD devices (that I have found (I've only used two variants, so what makes me an expert?)), the internal switching seems to initially (?) default to Lowest Common Denominator switching.

It usually takes a bit of tickling/nudging to get these multi-format devices to negotiate their Crown Speeds

It may be that the USB4 devices are better-able to negotiate the ebb-flow of the power limitations/requirements in the AAPL Silicon USB4/TB busses (?)

I finally have a semi-decent Windows setup that I can use to update the FW on my TB/USB4 Units (I find it ironic that the devices I own--most, if not all--which are Marketed as items I can use with my Mac Comps only have FW Updaters which can work on win64) *grr*

 

[splifingate]

the ASM2464PD chip is reported to run hot

Both my Units hold 42-43C in daily station.

Each can broach short 60C periods with intensive multi-transfer tasks, but I don't regularly copy 200MB+ in a normal setting.