Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.

whodiini

macrumors regular
Original poster
Aug 16, 2021
157
63
I just purchased the OWC thunderbolt 4 hub that was on sale for $99. I am trying to use it to expand the TB ports on my M1 Mac mini (2020). I connect my M1 mac mini to the OWC TB hub. I test the hub by connecting a NVME external enclosure. It works fine. I now want to connect it to my Studio Max so I can run thunderbolt IP between the 2 macs. So I connect with a TB cable from the OWC TB hub to a Caldigit elements hub connected to my Studio Max. With the OWC TB hub, the connection with the studio max is not recognized. I go to System report, Thunderbolt/USB4 and I see the OWC as a hub, but the port shows no device active. I know it is not cables, as I tried different (good) cables. I can get the connection to be recognized with the same setup and cables if I 1) bypass the OWC and connect direct to the M1 Mac mini. System report shows a connection to the Studio Max. 2) replace the OWC hub with a different hub. The different hub shows a connection with the Studio Max via the caldigit elements. So I conclude that there must be a problem with the firmware or chip used by OWC not to support the complete TB3 protocols including IP over TB. Has anyone else tried to use the OWC thunderbolt hub to connect to another mac? BTW, since the other Studio Max is not recognized, I cant share internet, share files, use IP, so it is broader than just IP over thunderbolt.

Doesnt work:
M1 mac mini -> OWC TB hub <-> caldigits element TB hub <- Studio Max

Works:
M1 mac mini <-> caldigits element TB hub <- Studio Max
M1 mac mini -> another (noname) TB hub <-> caldigits element Tb hub <-Studio Max
 

whodiini

macrumors regular
Original poster
Aug 16, 2021
157
63
M1 mac mini -> OWC TB hub <- Studio Max works.
So the only one that doesnt work is M1 mac mini -> OWC TB hub <-> caldigits element TB hub <- Studio Max

Strange. According to reviews, both the OWC and Caldigit use the Goshen Ridge Chipset from Intel, so they should be compatible.
 

whodiini

macrumors regular
Original poster
Aug 16, 2021
157
63
Someone may ask why not just plug it straight into the M1 mac mini. The reason is to maximize the bandwidth of the TB ports. IP over thunderbolt is about 10Gbps. Effective bandwidth of the TB port on the M1 is 32Gbps. So I am leaving 22 Gbps on the table. That is why I want to hook it up to the hub and use TB port on the M1 mini to connect to an external NVMe drive, which I can pretty much saturate the bandwidth.

oh, and of course, I tried all 3 TB ports on the OWC and none work to make a connection to the caldigit.
 
Last edited:

whodiini

macrumors regular
Original poster
Aug 16, 2021
157
63
Well, I shouldnt believe everything I read. The calldigits element hub shows up in my system report as using the intel ash creek chipset, not the goshen ridge. There is no chipset reported for the OWC hub
 

joevt

Contributor
Jun 21, 2012
6,843
4,136
M1 mac mini -> OWC TB hub <- Studio Max works.
So the only one that doesnt work is M1 mac mini -> OWC TB hub <-> caldigits element TB hub <- Studio Max

Strange. According to reviews, both the OWC and Caldigit use the Goshen Ridge Chipset from Intel, so they should be compatible.
Can you chain Thunderbolt 4 hubs? Just hubs, no second Mac:
M1 mac mini -> caldigits element TB hub <-> OWC TB hub
M1 mac mini -> OWC TB hub <-> caldigits element TB hub

What if you change the order?
M1 mac mini -> caldigits element TB hub <-> OWC TB hub <- Studio Max

Someone may ask why not just plug it straight into the M1 mac mini. The reason is to maximize the bandwidth of the TB ports. IP over thunderbolt is about 10Gbps. Effective bandwidth of the TB port on the M1 is 32Gbps. So I am leaving 22 Gbps on the table. That is why I want to hook it up to the hub and use TB port on the M1 mini to connect to an external NVMe drive, which I can pretty much saturate the bandwidth.

oh, and of course, I tried all 3 TB ports on the OWC and none work to make a connection to the caldigit.
You mean TB IP doesn't use all the bandwidth so you don't want to waste an entire Thunderbolt port for just TB IP.

Well, I shouldnt believe everything I read. The calldigits element hub shows up in my system report as using the intel ash creek chipset, not the goshen ridge. There is no chipset reported for the OWC hub
Can you show screenshots of where you are seeing this intel ash creek chipset? I have no idea what that is.
 

whodiini

macrumors regular
Original poster
Aug 16, 2021
157
63
I just tried to daisy chain. I realized my first message was not clear how I plugged into the hubs. The hub has one connection to the host (master) and 3 connections to peripherials.

So in more detail, this doesnt work:
Studio Max-> (master port) Caldigit elements hub -> one of the 3 caldigit elements thunderbolt ports <-> one of the 3 OWC TB ports TB hub <- (master port OWC TB hub) <- M1 mac mini

These do work:
Studio Max -> one of the 3 OWC TB ports TB hub <- (master port OWC TB hub) <- M1 mac mini
Studio Max-> (master port) Caldigit elements hub -> one of the 3 caldigit elements thunderbolt ports <- M1 mac mini
Studio Max-> <- M1 mac mini

So daisy chaining this way:
Studio Max-> (master port) Caldigit elements hub -> one of the 3 caldigit elements thunderbolt ports <-> - (master port OWC TB hub) works. On the system report, both hubs are reported.

Tried this next:
Studio Max-> (master port) Caldigit elements hub -> one of the 3 caldigit elements thunderbolt ports <-> (master port OWC TB hub) <- one of the 3 OWC TB ports TB hub <- M1 mac min
This works. But the OWC ports act as a hub to the studio Max, not to the M1 mac mini which defeats the purpose of the hub

This doesnt work:
Studio Max-> (master port) Caldigit elements hub -> one of the 3 caldigit elements thunderbolt ports <-> one of the 3 OWC TB ports TB hub <- (master port OWC TB hub) <-nothing connected

System report on Studio Max doesnt show the OWC hub. So it isnt negotiating between the 2 hubs/

I will try swapping the hubs next and report back.
 

whodiini

macrumors regular
Original poster
Aug 16, 2021
157
63
I tried swapping the hubs. No luck. You asked for a screen shot - here is the system report with the swapped hub:
I also realized ask creek does not refer to the chip on the hub. It refers to hte device (external tb3 NVme enclosure) which has the intel ask creek chipset.
 

Attachments

  • Screenshot 2023-02-13 at 7.52.07 PM.png
    Screenshot 2023-02-13 at 7.52.07 PM.png
    256.2 KB · Views: 57

whodiini

macrumors regular
Original poster
Aug 16, 2021
157
63
I am wondering if the thunderbolt protocol cant go thru 2 hubs if each hub is not a master. When I go thru one hub and go directly to the other mac, then it connects to a master (host). When it goes from the hub into the master port of the other hub, it also works. When I tried the no-name hub that worked, it was a hub that had a pass thru TB3 connection (master to the computer), the other was a pass thru TB port - maybe that pass thru port maintained its master status??
 

joevt

Contributor
Jun 21, 2012
6,843
4,136
You want one TB4 hub to belong to one Mac and another TB4 hub to belong to the second Mac. Then you want to be able to connect the TB4 hubs together so that each Mac can TB IP the other.

So, while each Mac is on, connect a TB4 hub to each. They should appear to be working with 3 downstream ports with nothing connected to them.

Then you should be able to connect a cable between the TB4 hubs. If things work as expected, then each Mac should be able to see each other. But you haven't been able to get this part to work?

I don't think it should matter what length of devices (chain depth) is connected to each Mac. At least, this has never been a problem with Thunderbolt 1/2/3 devices. You should be able to connect a chain of 5 devices to one Mac and 5 to the other, and still be able to connect the two chains. Or maybe the total length needs to be less than 6. I've never tried connecting two chains that long.

I just tried to daisy chain. I realized my first message was not clear how I plugged into the hubs. The hub has one connection to the host (master) and 3 connections to peripherials.
Correct. One port is expected to be upstream (in the direction of the host). The other 3 ports are expected to be downstream (in the direction of peripherals).

So in more detail, this doesnt work:
Studio Max-> (master port) Caldigit elements hub -> one of the 3 caldigit elements thunderbolt ports <-> one of the 3 OWC TB ports TB hub <- (master port OWC TB hub) <- M1 Mac mini
The order of connection matters. With power on, connect each hub to each host. Only if they are confirmed to be working connect downstream port of one hub to the other hub. I suppose this has been tried and has failed? Maybe placing a Thunderbolt 3 device (with two Thunderbolt 3 ports) somewhere in the chain could help.

These do work:
Studio Max -> one of the 3 OWC TB ports TB hub <- (master port OWC TB hub) <- M1 mac mini
So the OWC TB4 hub belongs to the M1 Mac mini?

Studio Max-> (master port) Caldigit elements hub -> one of the 3 caldigit elements thunderbolt ports <- M1 Mac mini
The CalDigit TB4 hub belongs to the Studio Max?

Studio Max-> <- M1 Mac mini
No hubs works as expected since this is the simplest connection.

So daisy chaining this way:
Studio Max-> (master port) Caldigit elements hub -> one of the 3 caldigit elements thunderbolt ports <-> - (master port OWC TB hub) works. On the system report, both hubs are reported.
Logical. Chaining two TB4 hubs should work. Did you try reversing the order to make sure the OWC can be upstream of the CalDigit?

Tried this next:
Studio Max-> (master port) Caldigit elements hub -> one of the 3 caldigit elements thunderbolt ports <-> (master port OWC TB hub) <- one of the 3 OWC TB ports TB hub <- M1 mac min
This works. But the OWC ports act as a hub to the studio Max, not to the M1 mac mini which defeats the purpose of the hub
This seems expected. Since the OWC was first connected to the Studio via the OWC's upstream TB4 port, it should belong to the Studio Max.

This doesnt work:
Studio Max-> (master port) Caldigit elements hub -> one of the 3 caldigit elements thunderbolt ports <-> one of the 3 OWC TB ports TB hub <- (master port OWC TB hub) <-nothing connected

System report on Studio Max doesnt show the OWC hub. So it isnt negotiating between the 2 hubs/
This seems expected. A TB4 hub requires its upstream port to be connected but here only the OWC's downstream port was connected.
 

whodiini

macrumors regular
Original poster
Aug 16, 2021
157
63
You want one TB4 hub to belong to one Mac and another TB4 hub to belong to the second Mac. Then you want to be able to connect the TB4 hubs together so that each Mac can TB IP the other.

So, while each Mac is on, connect a TB4 hub to each. They should appear to be working with 3 downstream ports with nothing connected to them.

Then you should be able to connect a cable between the TB4 hubs. If things work as expected, then each Mac should be able to see each other. But you haven't been able to get this part to work?

I don't think it should matter what length of devices (chain depth) is connected to each Mac. At least, this has never been a problem with Thunderbolt 1/2/3 devices. You should be able to connect a chain of 5 devices to one Mac and 5 to the other, and still be able to connect the two chains. Or maybe the total length needs to be less than 6. I've never tried connecting two chains that long.


Correct. One port is expected to be upstream (in the direction of the host). The other 3 ports are expected to be downstream (in the direction of peripherals).


The order of connection matters. With power on, connect each hub to each host. Only if they are confirmed to be working connect downstream port of one hub to the other hub. I suppose this has been tried and has failed? Maybe placing a Thunderbolt 3 device (with two Thunderbolt 3 ports) somewhere in the chain could help.
YES! Good idea.

I had a no-name (Rocketek TB3 dock) which has one TB3 host connector and one TB3 peripherial connector. No other TB ports. So this worked instead of the OWC but since it doesnt have more TB ports, it didnt do what I wanted. BUT, i tried to put it in the middle of the chain to negotiate between the OWC and Caldigit:

Studio Max-> (master port) Caldigit elements hub -> one of the 3 caldigit elements thunderbolt ports <-> master port of the Rocketek <-> peripheral TB port of the Rocketek <-one of the 3 OWC TB ports TB hub <- (master port OWC TB hub) <- M1 Mac mini

This worked! I have no idea why inserting yet another TB hub made it work. I also tried to reverse the connections on the Rocketek and it also worked:

Studio Max-> (master port) Caldigit elements hub -> one of the 3 caldigit elements thunderbolt ports <-> peripheral TB port of the Rocketek <-> master port of the Rocketek <-one of the 3 OWC TB ports TB hub <- (master port OWC TB hub) <- M1 Mac mini

It seems that there is a miscommunication of the IP over TB protocol between the Caldigit and OWC and inserting yet another hub between the 2 makes it work. This is quite a kludge so I need to call up Caldigit AND OWC and see if they can do this same test but using 2 of the same type of hubs. The Caldigit is a better hub but is 2x more expensive than the OWC hub. If getting 2 of the same type of hub works, then I will have to go that route.

So the OWC TB4 hub belongs to the M1 Mac mini?
Yes

The CalDigit TB4 hub belongs to the Studio Max?
Yes
No hubs works as expected since this is the simplest connection.

Logical. Chaining two TB4 hubs should work. Did you try reversing the order to make sure the OWC can be upstream of the CalDigit?
no
This seems expected. Since the OWC was first connected to the Studio via the OWC's upstream TB4 port, it should belong to the Studio Max.

This seems expected. A TB4 hub requires its upstream port to be connected but here only the OWC's downstream port was connected.
 

whodiini

macrumors regular
Original poster
Aug 16, 2021
157
63
It gets weirder and weirder...

I got a plugable TB4 hub. Has the latest intel chipset. No difference swapping it for either the OWC or elememts. I then tried putting the hub in between the 2 hubs. Didnt work. Nothing worked. I replaced the caldigit hub. Still nothing. It seems that IP over thunderbolt doesnt pass thru 2 hubs. I had on hand an acasis thunderbolt enclosure which has 2 thunderbolt ports. I put it in between the 2 caldigit and OWC and it didnt work. So there is something magical about my Rocketek that acts as a translator. I tried between the OWC and plugable and that worked as well. So far the only way IP over thunderbolt works thru 2 hubs is to put the Rocketek in between! I need an expert in thunderbolt to explain what is going on..... This makes no sense.
 

steve123

macrumors 65816
Aug 26, 2007
1,103
672
So there is something magical about my Rocketek that acts as a translator.

I wonder if it is power related. You are connecting two downstream ports together so there is a contention with the power. That is, both hubs are attempting to deliver power to each other. So, one or both hubs may be shutting down their downstream port.

The Rocketek may not connect power from the upstream port to the downstream port similar to the way the Apple Thunderbolt 3 to Thunderbolt adapter works. So, in the case where you insert the Rocketek, the two downstream ports that are connected do not have a contention condition and therefore the ports turn on.
 

whodiini

macrumors regular
Original poster
Aug 16, 2021
157
63
I wonder if it is power related. You are connecting two downstream ports together so there is a contention with the power. That is, both hubs are attempting to deliver power to each other. So, one or both hubs may be shutting down their downstream port.

The Rocketek may not connect power from the upstream port to the downstream port similar to the way the Apple Thunderbolt 3 to Thunderbolt adapter works. So, in the case where you insert the Rocketek, the two downstream ports that are connected do not have a contention condition and therefore the ports turn on.
I tried with an OWC TB4 hub, no change. Put it on a different computer, no change
 

steve123

macrumors 65816
Aug 26, 2007
1,103
672
I tried with an OWC TB4 hub, no change. Put it on a different computer, no change

I am not sure I understand what you mean?


Studio Max-> (upstream port) Caldigit elements hub -> one of the 3 caldigit elements thunderbolt DOWNSTREAM ports <-> one of the 3 OWC TB DOWNSTREAM ports TB hub <- (upstream port OWC TB hub) <- M1 mac mini

What I was attempting to say is that you cannot connect two downstream ports together without causing power contention. In the original configuration, the OWC and the Caldigit downstream ports are connected together. You cannot do that because the power conflicts with each other so the hubs power down their port.


Studio Max-> (upstream port) Caldigit elements hub -> one of the 3 caldigit elements thunderbolt DOWNSTREAM ports <-> upstream port of the Rocketek <-> DOWNSTREAM TB port of the Rocketek <-one of the 3 OWC TB DOWNSTREAM ports TB hub <- (upstream port OWC TB hub) <- M1 Mac mini

When you inserted the Rocketek adapter, it likely does not pass power from upstream to downstream so it "isolates" the power between the two downstream ports of the hubs and so the hubs do not power them off. The downstream port of the Rocketek is connected to the downstream port of the hub but because the Rocketek is not powering the port, there is no contention. This is why this configuration worked.

So, inserting another hub between the two hubs does not help the situation.

Intel advises that you can configure the "charging mode" of a thunderbolt port in the PC BIOS (Portable Device Charging Mode). But it is not clear if you can configure a hub in such a way as to disable the downstream power. Moreover, I do not know if Apple has incorporated this capability somewhere into macOS. So, the only way you can accomplish what you want to do is to use the Rocketek between the hubs or some other thunderbolt hub/adapter that does not power the downstream port.
 
Last edited:

joevt

Contributor
Jun 21, 2012
6,843
4,136
I wonder if it is power related. You are connecting two downstream ports together so there is a contention with the power. That is, both hubs are attempting to deliver power to each other. So, one or both hubs may be shutting down their downstream port.
No, that's not how USB-C power delivery works. Voltage and current are negotiated. High current (for charging) is not taken until they agree on the max current.
 

steve123

macrumors 65816
Aug 26, 2007
1,103
672
No, that's not how USB-C power delivery works. Voltage and current are negotiated. High current (for charging) is not taken until they agree on the max current.
This negotiation generally happens when a device connects to a port. The device (power sink) and port (power source) negotiate the power requirements. When you have two downstream facing ports connected, there are two power sources which have to negotiate with each other and decide which one will supply the power. This is only possible if at least one of those ports is a dual role power (DRP) port which can take the power role as either a sink or a source.

When you connect a host port from one Mac to another Mac, this generally works because the host ports on a Mac are Dual Role Power ports. So they negotiate between themselves and determine which port will be the source and which one becomes the sink.

In the case of the hub, unless the downstream ports on the hub are DRP ports, it is not possible to connect them both together and have them properly negotiate PD. I do not think the downstream ports on either the OWC or Caldigit hubs are DRP port. I am not certain are not DRP but I could not find anything in the specifications of either hub that says they are DRP. Moreover, OWC and Caldigit refer to those ports in their documentation specifically as downstream ports.
 
Last edited:

whodiini

macrumors regular
Original poster
Aug 16, 2021
157
63
This raises an interesting question you raised. Read some stuff online and DRP and DRD ports are described. I wonder, if it has something to do with either DRP or DRD (Dual role data) because 2 downstream ports normally do not talk to each other. It would come down to the chips used and implementation. So I looked at the OWC and Caldigit elements. The owc review: https://www.tomshardware.com/news/o...4-hub-three-extra-thunderbolt-ports-and-usb-a
suggests that the chip is a JHL8440. The Rocketek uses a JHL7440 controller. I assume the caldigit uses a JHL8840 as well. No literature associated with intel controllers mention DRP or DRD. However, there must be a chip in front of the controller to handle PD. I googled and there are posts regarding the pairing of a PD chip with the thunderbolt chip. The PD chip would handle the DRP and possibly the DRD functions, not the thunderbolt chip. I tried unsuccessfully to open up the Rocketek to see what chip it had. Possibly it only works because the Rocketek has a DRD and can take 2 downstream ports and communicate between them?
 

joevt

Contributor
Jun 21, 2012
6,843
4,136
This negotiation generally happens when a device connects to a port. The device (power sink) and port (power source) negotiate the power requirements. When you have two downstream facing ports connected, there are two power sources which have to negotiate with each other and decide which one will supply the power. This is only possible if at least one of those ports is a dual role power (DRP) port which can take the power role as either a sink or a source.

When you connect a host port from one Mac to another Mac, this generally works because the host ports on a Mac are Dual Role Power ports. So they negotiate between themselves and determine which port will be the source and which one becomes the sink.

In the case of the hub, unless the downstream ports on the hub are DRP ports, it is not possible to connect them both together and have them properly negotiate PD. I do not think the downstream ports on either the OWC or Caldigit hubs are DRP port. I am not certain are not DRP but I could not find anything in the specifications of either hub that says they are DRP. Moreover, OWC and Caldigit refer to those ports in their documentation specifically as downstream ports.
You only mention two roles: providing power (source) and receiving power (sink). They must have a third option - no power. If they can't do that, then Intel should take away their Thunderbolt certification.

Consider a Thunderbolt device such as the OWC Mercury Helios 3S. It has its own power supply and can provide 85 watts. Are you saying you can't connect a OWC Mercury Helios 3S to a Thunderbolt 4 dock because the OWC Mercury Helios 3S won't take power? That would be silly.
 

joevt

Contributor
Jun 21, 2012
6,843
4,136
I think DRD is mostly USB 3.x

For Thunderbolt (and USB4), every port should be DRD to support cross domain communication. Look at the USB4 spec.
 

steve123

macrumors 65816
Aug 26, 2007
1,103
672
They must have a third option - no power.
Yes, I agree. The power for one of the ports can be negotiated to turn off. So, for instance, a device (upstream facing) can tell the upstream port (downstream facing) it does not require power and so the DFP turns it off.

For Thunderbolt (and USB4), every port should be DRD
2.1.3 USB4 Power
Power for USB4 operation is established and managed as defined in the USB Type-C Specification and the USB PD Specification.


Note, DRD (Dual Role Data) is not the same as DRP (Dual Role Power).

The Type-C specification does not mandate that all ports are DRP.

there must be a chip in front of the controller to handle PD. I googled and there are posts regarding the pairing of a PD chip with the thunderbolt chip. The PD chip would handle the DRP and possibly the DRD functions, not the thunderbolt chip.
Yes. At the moment, the PD controllers are separate chips because they generally integrate the power switches which require higher voltages (up to 48V for an extended power range port) and 5A of current.
 

steve123

macrumors 65816
Aug 26, 2007
1,103
672
Possibly it only works because the Rocketek has a DRD and can take 2 downstream ports and communicate between them?
Do you mean DRP? Perhaps you can let us know exactly what Rocketek product you are talking about and post the technical specifications for it because I am simply speculating without knowing this information.
 

steve123

macrumors 65816
Aug 26, 2007
1,103
672
Thinking more about this. I believe the limitation is the firmware of the hub(s). I do not think there is a hardware limitation here. Further, I do not think the downstream ports need to be DRP. Rather, this particular use case of DFP connected to DFP needs to implemented. The firmware is not open source. So, you would need to get this use case on the radar of the developer which is Intel since that is who makes the controller. You can point the use case out to OWC and Caldigit but I am not sure how much influence they will have with Intel’s firmware team so it may be a long time, if ever, that they get around to it unless there are a lot of people asking for it.
 

whodiini

macrumors regular
Original poster
Aug 16, 2021
157
63
I contacted Caldigit and OWC. They escalated it up, and never heard from either since...

If it was just intel firmware, then there is no reason why the Rocketek would work as an intermediary since it has intel firmware as well. I also see DRP/DRD used in the same sentence so I suspect that whatever chip implements DRP also does DRD. It may well be that this is not an intel issue but the USB chip that negotiates DRP/DRD. if there are 2 DFP then they may not talk to each other, but if one is a DRP/DRD, then it could negotiate to change itself to a UFP.

So far, this makes the most sense to me of the possibilities. I need to figure out how to open up the Rocketek. I got it for cheap on Amazon, thinking if it did not work I can return it. But after using it a while, it operated like most TB3 hubs and was a lot cheaper. It is a no-name Chinese made hub and specs are hard to come by,.
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.