TUSB7320: High speed disconnect detection threshold

Hi Ryan

Thank you for your reply.

I understand that this is a USB2.0 specification.

Rather than the USB2.0 standard, could you please tell me about the voltage specification value of the TUSB7320 at which it detects a disconnect?

In actual testing, when measured in the SOF section, it exceeded 525mV, but a disconnect was not detected (see waveform)).

I would like to know the actual voltage value at which the TUSB7320 detects a disconnect.

Regards

Yoda

Hi Yoda-san,

The actual disconnect threshold can vary between devices based on the device process. The USB2.0 spec typically defines that within the range of 525-625mV, the signal can disconnect. So while 536mV may not disconnect, other devices may disconnect at that voltage. I can't say for sure at what voltage exactly this device begins to disconnect, but it is guaranteed to disconnect at 635mV and above, at the very least.

Thanks,

Ryan

Hi Ryan

Thank you for your reply.

I have four questions.

- You answered that a disconnect occurs at 635mV or above, but isn't the compensation 625mV or above?

- Is the threshold range of 525mV to 625mV due to component variation?

    Please let us know if there are any other factors.

- Measurements were made with a differential probe close to pins B12 and A13 of the TUSB7320.

    Please let us know if there are any discrepancies in the measurement method.

- Please tell me the mechanism of how TUSB7320 detects the Disconnect threshold.

Regards

Yoda

Hi Yoda:

HS disconnection must be indicated when the amplitude of the differential signal at the downstream facing driver’s connector ≥ 625 mV, and it must not be indicated when the signal amplitude is ≤ 525 mV.

  for TI test setup: The voltage on DP is swept while DM is held at 0V then the voltage on DM is swept while DP is constant at 0V. Typical V-V plots are shown  below:

Best

Brian

Hi Brian:

Thank you for your reply.

We tried increasing the amplitude value to check the disconnect detection, but the disconnect detection did not occur even when it exceeded 625mV.

The waveform below is an example of the disconnect detection not being detected even when it exceeded 680mV.

The USB2.0 standard states that disconnect detection occurs when it exceeds 625mV. (Red frame below)

We would like to hear your opinions on the measurement method, location, timing, etc. carried out here.

We are investigating the cause of the difference between the standard and the actual measurement.

Measurement location: Measured near the B12 (USB_DP_DN2) and A13 (USB_DM_DN2) terminals of the TUSB7320 with a differential probe

Measurement timing: Sampling between the Start Of Frame and the End Of Packet

Regards

Yoda

Hi Yoda-san,

This is odd, it should be disconnecting. The measuring points seem good to me. The main things I would like to see, if possible, are eye diagrams of the USB2 signaling and if disconnects can be seen when testing with devices, such as USB2 flash drives if possible. That way, we can see if any other devices disconnect or not.

I will look in the meantime internally to see if we have any tracked data on a disconnect threshold for this device.

Thanks,

Ryan

Hi Ryan

Thank you for your reply.

These are the conditions for obtaining a waveform where the disconnect detection exceeds 625mV.

- Connect to a USB memory (with built-in flash ROM).

- Change the resistance between R1EXT (pin A24) and RIEXTRTN (pin B23) of the TUSB7320 to increase the voltage amplitude.

- Waveform at the time of disconnection (cannot be recognized as a packet)

- USB2.0 eye diagram

You previously answered my question about the disconnect detection threshold.

The TUSB7320 should be complaint to the USB2.0 spec, meaning that the disconnect detection threshold should be indicated until at least 525mV, and that at 625mV, it must be detected:

The USB ENGINEERING CHANGE NOTICE for USB2.0 mentioned a change to the disconnect detection threshold. Has there been a change in thinking?

usb.org/.../usb_20_20240927.zip

Regards

Yoda

Hi Yoda-san,

Is the disconnection being done by you? Or is the signal being disconnected via the disconnect threshold? When you say it could not be recognized as a packet, I'm guessing you mean that the signal had to high of a voltage to be recognized by the flash drive.

hisashi yoda said:

Change the resistance between R1EXT (pin A24) and RIEXTRTN (pin B23) of the TUSB7320 to increase the voltage amplitude.

This resistor should be only 9.09 KOhm typically for the sake of calibration. It may have some effect on this, but I will reach out internally to confirm whether that's the case or not.

hisashi yoda said:

The USB ENGINEERING CHANGE NOTICE for USB2.0 mentioned a change to the disconnect detection threshold. Has there been a change in thinking?

That's a good point, there is potential that this device was created with this ECN in mind for creating the disconnect threshold. I can reach out internally to confirm this one as well.

Thanks,

Ryan

Hi Ryan

Thank you for your reply.

Ryan Kitto said:

Is the disconnection being done by you? Or is the signal being disconnected via the disconnect threshold?

->The disconnection is intentionally created.

The standard specifies that the threshold for disconnect detection is 525mV to 625mV,

so we are trying to see at what threshold the disconnect detection occurs.

In terms of design, a 9.09kΩ resistor is used between R1EXT (pin A24) and RIEXTRTN (pin B23) of the TUSB7320.

In that case, it indicates 530mV to 540mV and no disconnect detection occurs.

Please confirm the following two points within your company.

-Is the reason why disconnect detection does not occur even when the threshold exceeds 625mV related to the calibration resistance value?

-What is your company's opinion on the change in the disconnect detection threshold due to the USB ENGINEERING CHANGE NOTICE for USB2.0?

We would like to know the results of the confirmation as soon as possible, so please reply by 10/21.

What I want to know most is the voltage value that TUSB7320 detects as Disconnect (about the specifications of TUSB7320,

not the standard value of USB2.0) If we understand this, our doubts can be resolved.

Regards

Yoda

Hi Yoda-san,

I spoke to some members on our team for more information, here's what I found related to your questions:

hisashi yoda said:

-Is the reason why disconnect detection does not occur even when the threshold exceeds 625mV related to the calibration resistance value?

When we talked regarding the disconnect threshold/detection, we found that there is a setting for a disconnect threshold for each port, with the setting for the disconnect threshold being set to the max setting. While I can't say what the range of the disconnect is exactly, it is likely a good bit greater than what you tested. The default at the lowest setting puts the threshold for disconnect between 550mV and 610mV, so a high setting is likely past that, potentially explaining why you didnt see a disconnect at 683mV. At the normal recommended calibration of 9.09KOhms, I believe you will not pass this disconnect threshold. I don't believe the calibration resistance value has any effect on this.

hisashi yoda said:

-What is your company's opinion on the change in the disconnect detection threshold due to the USB ENGINEERING CHANGE NOTICE for USB2.0?

This device was released back around 2011. The ECN shown was drafted in 2018 it looks like, so this device is not compliant to that ECN, as no changes have been made to adhere to it. As such, I would stick to that original description for this in the USB2.0 spec, and not the ECN. Any devices we have made after this ECN was added should reflect this ECN.

Unfortunately at this moment I can't say exactly what the disconnect threshold is, as this device is a good bit older, and I'm not sure if there are any members on our team who have access to the related in depth files currently.

Hopefully the answers I gave today provide some clarification on the questions you've asked. Overall, I would say that at the intended calibration resistor setting, you should not need to worry about disconnects unless there is an extreme external factor.

Please let me know if you have any other questions.

Thanks,

Ryan

Hi Ryan

Thank you for your reply.

Ryan Kitto said

When we talked regarding the disconnect threshold/detection,

we found that there is a setting for a disconnect threshold for each port, with the setting for the disconnect threshold being set to the max setting.

While I can't say what the range of the disconnect is exactly, it is likely a good bit greater than what you tested.

The default at the lowest setting puts the threshold for disconnect between 550mV and 610mV,

so a high setting is likely past that, potentially explaining why you didnt see a disconnect at 683mV.

I have three questions.

1.In the previous specifications, the maximum threshold for disconnect detection is 625mV.

Are there any TUSB7320 parts for which disconnect detection does not occur even if the threshold exceeds 625mV?

2.The threshold for disconnect detection is set to the maximum.

The default minimum value for the threshold is between 550mV and 610mV,

so am I correct in understanding that the threshold for disconnect detection for the TUSB7320 is 610mV or higher?

3.Could you please tell me the default value for the maximum setting for the disconnect detection threshold?

I understand that the TUSB7320 is not ECN compliant.

Regards

Yoda

Hi Yoda-san,

hisashi yoda said:

1.In the previous specifications, the maximum threshold for disconnect detection is 625mV.

Are there any TUSB7320 parts for which disconnect detection does not occur even if the threshold exceeds 625mV?

As far as I am aware, there are not TUSB7320 parts where the disconnect threshold should exceed 625mV.

hisashi yoda said:

2.The threshold for disconnect detection is set to the maximum.

The default minimum value for the threshold is between 550mV and 610mV,

so am I correct in understanding that the threshold for disconnect detection for the TUSB7320 is 610mV or higher?

hisashi yoda said:

3.Could you please tell me the default value for the maximum setting for the disconnect detection threshold?

This is not information we have easily on hand, as the engineers that originally designed this device have since left the company or the databases with this information have been archived. I would say that the minimum point at which a disconnect can be detected would be starting at 550mV, which should help narrow down the range a bit. I am not sure if the calibration resistor would affect this or not.

Please let me know if you have any other questions.

Thanks,

Ryan

Hi Ryan

Thank you for your reply.

Ryan Kitto said

As far as I am aware, there are not TUSB7320 parts where the disconnect threshold should exceed 625mV.

In response to this answer, we have adjusted the resistance between R1EXT (pin A24) and RIEXTRTN (pin B23) of the TUSB7320,

but there are some TUSB7320s for which disconnect detection does not occur even when the voltage exceeds 625mV.

Could you please comment on this result?

Regards

Yoda

Hi Yoda-san,

Is this with the 9.09 KOhm resistor on R1EXT? It looks like you have it set at 4.7Kohm per this diagram, which is outside of what we recommend.

If you have the resistor at 9.09 KOhm, the device should perform as expected.

Thanks,

Ryan

Hi Ryan

Thank you for your reply.

I have summarized the answers I have received so far in a diagram.

Which threshold range is guess 1 or guess 2?

Ryan Kitto said

-When we talked regarding the disconnect threshold/detection,

we found that there is a setting for a disconnect threshold for each port, with the setting for the disconnect threshold being set to the max setting.

-As far as I am aware, there are not TUSB7320 parts where the disconnect threshold should exceed 625mV.

From these two sentences, does the maximum setting for the disconnect detection threshold be 625mV?

If it is different, please comment.

Regard

Yoda

Hi Yoda-san,

hisashi yoda said:

Which threshold range is guess 1 or guess 2?

hisashi yoda said:

 

From these two sentences, does the maximum setting for the disconnect detection threshold be 625mV?

I would say guess 2 is likely closer to the actual disconnect threshold, however, I would not say the minimum disconnect is at 610mV. It's likely less than that, within that 550-610mV range. As for the maximum, yes, that is likely 625mV.

Please let me know if you have any other questions.

Thanks,

Ryan

Hi Ryan

Thank you for your reply.

 Ryan Kitto said

I would say guess 2 is likely closer to the actual disconnect threshold, however, I would not say the minimum disconnect is at 610mV.

It's likely less than that, within that 550-610mV range. As for the maximum, yes, that is likely 625mV.

The maximum value for disconnect detection is said to be 625mV, but when I changed the connected device,

the threshold exceeded 625mV, but disconnect detection did not occur.

In this measurement, the resistance value between R1EXT (pin A24) and RIEXTRTN (pin B23) of TUSB7320 is 9.09kΩ.

Could you please comment on this measurement result?

I am wondering why it is not detected based on the voltage threshold.

Are there any conditions other than the voltage value for how the TUSB7320 detects disconnection?

For example, is there a regulation for the sampling time of the voltage value?

Regard

Yoda

The time to detect a disconnect event is a minimum of 40 bits per USB spec.  The signal amplitude over the threshold itself may not trigger a disconnect.

Hi Undrea

Thank you for your reply.

Undrea Fields said

     The signal amplitude over the threshold itself may not trigger a disconnect.

Under what conditions will a disconnect detection be triggered?

Will it not be detected if the time exceeds the regulation for an instant at the [ns] level, as shown in the red circle below?

Regard

Yoda

Hi Undrea

I checked it further.

Could you comment on the results?

Undrea Fields said

The time to detect a disconnect event is a minimum of 40 bits per USB spec.

Measured at 40 bits of EOP (End of packet) in the USB specifications.

Is the threshold for disconnect detection the peak of that section (40 bits) (red frame), or the gentle part (blue frame)?

waveform A(Disconnect not detected)

waveform B(Disconnect detected)

Comparing the potential of the gentle part (blue frame) (green frame), waveform A does not detect disconnect at 533mV,

but waveform B detects disconnect at 560mV.

*A cursor is positioned at 40 bits (83.3[ns], which is 40 times 480MHz).

Regard

Yoda

I believe the voltage level is sampled at that specific time (After the ~40 bits), not the peak.

Hi Undrea

Thank you for your reply.

Undrea Fields said

I believe the voltage level is sampled at that specific time (After the ~40 bits), not the peak.

I have two questions.

1.

Could you please let me know if there is any information about the above judgment in the USB specifications?

2.

Please tell me how 40 bits are judged in the TUSB7320 specifications.

This is an example from another company, but there are also cases where the judgment is based on specific bits.

https://ww1.microchip.com/downloads/aemDocuments/documents/OTH/ProductDocuments/DataSheets/00002105A.pdf

The following passage is found on page 15.

"In Host mode the USB3450 will sample the disconnect comparator at the 32nd bit of the 40 bit long EOP during SOF packets."

Is there similar information on the TUSB7320?

Regard

Yoda

I'm not clear about what you are asking for #1.  The threshold comparator is sampled to determine if the disconnect voltage has been reached or not.

#2, we would need to locate the design spec for this device(This will take quite a bit of time).  As it is so old the design locations have been moved.  The device complies to the spec, so the detect occurs after 40 bits.

The disconnect can now been seen by the customer, is there another issue related to this?

Hi Undrea

Thank you for your reply.

Undrea Fields said

#2, we would need to locate the design spec for this device(This will take quite a bit of time).

-> How long is it likely to take to find the design specifications?

I need a reply quickly. I would prefer it by November 5th.

I will withdraw question #1.

Regard

Yoda

The exact voltage at which the disconnect occurs will vary from device to device with process variations.  There is a disconnect test in the test program so devices are tested.  

As observed, the disconnect does occur within the spec limits at the given 40 bit time sampling point.  

Please help to close this question at the customer, there does not appear to be any issue with this functionally(No disconnect issues to my knowledge have been reported in the 15+ year life of the device), do you agree?   

Hi Undrea

Thank you for your reply.

I disagree.

I understand that this part has a history of success and that testing has been performed.

Someone previously commented that it takes time to find the design specs for a device, so I asked how long it takes.

Are there no design specifications for the device?

I found the following in the USB2.0 specification.

usb.org/.../usb_20_20240927.zip

Based on this information, can we say that disconnect detection is determined after the 32nd bit?

Specification name: Universal Serial Bus Specification Revision 2.0

Section: 7.1.7.3 Connect and Disconnect Signaling

When a downstream facing port is transmitting in high-speed mode and detects that it has sent 32 bits without a transition,

the disconnection envelope detector’s output must be sampled once during transmission of the next 8 bits at the transceiver output.

I need a reply quickly. I would prefer it by November 7th.

Regard

Yoda

Hi Undrea

A customer asked us how to detect Disconnect for our design using TUSB7320.

Therefore, as a designer, I am inquiring about the specifications of TUSB7320.

Undrea Fields said

  You can reply that the device will sample for disconnect by the 40th bit.

Does it detect at the 40th bit, or somewhere within the 40 bits?

If it is the latter, I would like to know at what bit number it is being detected.

Regard

Yoda

Hi Undrea

A customer asked us how to detect Disconnect for our design using TUSB7320.

Therefore, as a designer, I am inquiring about the specifications of TUSB7320.

Undrea Fields said

  You can reply that the device will sample for disconnect by the 40th bit.

Does it detect at the 40th bit, or somewhere within the 40 bits?

If it is the latter, I would like to know at what bit number it is being detected.

Regard

Yoda

I need a reply quickly. I would prefer it by November 12th.

Hi Yoda,

The detect happens after 40 bits. Any point past the 40th bit, if there is a disconnect, it should be detected and there should be a disconnect. We are unable to test on our side for the exact bit which prompts the disconnect.

Thanks,

Ryan

Hi Ryan

Thank you for your reply.

I will change the question to one that was not an answer about the detection method.

This is the answer I received two months ago. (Blue text)

Ryan Kitto said

The TUSB7320 should be complaint to the USB2.0 spec,

[New question]

Does the TUSB7320 also comply with the following sections of the USB2.0 standard? (Red text)

Section: 7.1.7.3 Connect and Disconnect Signaling

When a downstream facing port is transmitting in high-speed mode and detects that it has sent 32 bits without a transition,

the disconnection envelope detector’s output must be sampled once during transmission of the next 8 bits at the transceiver output.

usb.org/.../usb_20_20240927.zip

 usb_20.pdf

Regard

Yoda

Hi Yoda,

Yes, the TUSB7320 should be compliant to this as well.

Thanks,

Ryan