Continue to Site

Welcome to EDAboard.com

Welcome to our site! EDAboard.com is an international Electronics Discussion Forum focused on EDA software, circuits, schematics, books, theory, papers, asic, pld, 8051, DSP, Network, RF, Analog Design, PCB, Service Manuals... and a whole lot more! To participate you need to register. Registration is free. Click here to register now.

instrument troubleshooting

Status
Not open for further replies.

aredhel_vlsi

Member level 4
Joined
Aug 21, 2009
Messages
72
Helped
1
Reputation
2
Reaction score
1
Trophy points
1,288
Activity points
2,013
Hello people,

I am back to ask a question of general interest. It is tricky to answer but I would like to ask for your experience because it s hard for me to detect the cause, maybe you've come across it. I have an instrument (with a robotic arm) that is connected to my pc with usb cable. The usb data bus is also inside the arm, which might influence the connection. It seems that sometimes, (maybe due to the fact that the arm moves) the camera connection is lost. I noticed that the Data+ channel drops from 5V to 0V when I slightly move the arm. I was wondering how I can troubleshoot and overcome this difficulty. I know it could be may things. I examined seriously two scenarios:

1. A not robust usb cable (it seems that the response of the signal with the oscilloscope differentiates from one end to the other of the cable, ie. double signal frequency at the other end). The cable is comprised of four parts with connections. So the metallic connections might affect the voltage of the signal transferred? Possibly but decreasing the number of connection cables to two didn't help much in the signal response.

2. A not robust pcb that connects to the camera cable. Of course redesigning the pcb is an easy answer, but before putting effort and resources to that I would like to ask how you would approach it. It seems difficult to check it because the pcb has no hardware description, and there not many output pins on it and it's hard to measure every single pin that would be related.

On top of that, by connecting from one computer to another sometimes it fixes but doesn't guarantee it will work. Could be the computer, could be static electricity. I know it's a bit abrupt, but based on the above information what would you advise in the first place? Also by changing the power supply from 240V to 110V also affects the performance. Thus I'm inclined to believe that it's a pcb issue.

Have you encountered a similar case? Thank you in advance.
 

Since the problem is observed 'sometimes' and that too when there is a movement ,
changing the suspected (usb or other) cable and observation will be the first step.
 

Hello, thank you for your reply. Of course I have changed cables, usb cables, pcbs, instruments but it always happens. My question is how to approach more specifically and what kind of phenomena you think might occur here.
 

Guys, I measured the current consumption at the USB3 port, it's 149 mA, which is a lot below of the max consumption of a USB3 port. When I touch the pins with the Amper-meter, the camera connection is lost, it resets. Is that normal?
 

Based on your observation , it can be due to the USB port power capability problem or some dry soldering or connection in that port.

check the usb port power related circuit.
 

Here's my update, please comment on the following. There is a voltage regulator in the schematics: https://www.ti.com/lit/ds/symlink/lm1086.pdf

It seems that in normal operation, Vin= 5V, Vout =3.3V and Vadj =2 V. R3 = 120 Ohms and R4= 200 Ohms. However:

If I try to plug the power supply of the laptop, the Vout shows an overshoot of 20V peak-to-peak instantly. Do you find it normal? Could it mean that the protection of the circuit doesn't work properly? Maybe that's the reason some pcbs are burnt. If I adjust the Resistors values would it improve?
 

Hello, I tried to simulate the behavior I described above but I can't. LT-Spice shows that the voltage of the output never exceeds 3.3 V. I selected DC sweep. schem.PNGcurrent.PNG
However the real response of the voltage is this: Figure_3_Power_Supply_Connection_3.jpg Am I doing anything wrong?
thank you !
 

Guys, I measured the current consumption at the USB3 port, it's 149 mA, which is a lot below of the max consumption of a USB3 port.

That is the max available, until your device asks for more via a protocol known as enumeration.

https://en.wikipedia.org/wiki/USB

When I touch the pins with the Amper-meter, the camera connection is lost, it resets. Is that normal?

The Ammeter range contains a low resistance shunt internally. It's almost a short. If I did this sort of testing I'd be afraid of losing my usb port permanently.

- - - Updated - - -

the Vout shows an overshoot of 20V peak-to-peak instantly.

Does this in any way send 20V into your usb port? Need for caution because it's another way to ruin a usb port. To be safe I think your equipment needs its own power supply.
 

Thank you for your reply. To clear things up: The pcb is powered with 12 V. There is also the usb port that provides 5V.

Yes, I measured 20V on the input (vcc pin of the usb port on the pcb at the input). At the attachment vcc is the yellow channel. I also measure the spike on the Vout of the voltage regulator (blue channel). I found that by adding diodes and Rload I could improve that according to this link: https://www.digikey.ch/en/articles/techzone/2012/apr/protecting-inputs-in-digital-electronics

I simulated the Voltage regulator only in Lt-spice. It seems it works with a diode at the output, as in input of the circuit as well. However, it doesn't work in reality, unless I have some bad connection/soldering ( I have spike again). I will try again.

yes I agree with you it's dangerous. I also use a laptop and I noticed a leakage. Sometimes I measure 3.5 V at the USB port, even if there is no usb cable connected to the laptop. Could it be that this leakage can cause damage to the circuit? I am just reserved to blame that for my faulty pcbs. I don't believe that all of them are damaged by that.

IMG_2162.JPG
 

You can make a simple experiment and connect the oscilloscope probe tip to ground instead of power supply output, with same ground clip connection, you'll seeing probably the same "spike" that isn't actually existent in the circuit.

It's a common mode interference, coupled through oscilloscope earth connection, causing a respective voltage drop along the probe cable and ground clip.

There are a number of options to make a cleaner measurement in presence of arcing contacts and similar interference sources, if you really need it. E.g. differential probes, short ground clips, toroid cores as common mode choke around the probe cable, coax cables soldered directly to the circuit, battery powered or otherwise isolated oscilloscope supply.

But I guess you don't actually need it, just should know that measurements will be disturbed under circumstances.
 

Thank you for your reply. I see, well I hoped it was related, because other than that, which I hoped it was, I can't identify any other cause. I have two pcbs that are identical ! How is it possible that they behave differently? the issue is always on one of the cameras that uses 1.5 m long cable.

What else can I do to identify the problem? My last resort is to desolder each cap and resistor and see if anything is burnt, but this is time-consuming and I am not sure if I will still find the answer there.
Here is the pcb schematics: power.PNGUSBhub.PNG
 

This sounds very much like a problem of bad USB wiring, either PCB design not suited for high speed USB or wrong cables or connectors.
 

Thank you for your reply. When you say bad USB wiring what do you mean? I don't see any issue in that on my pcb. And if it's the case, how do I prove it?

Update: Its impossible to desolder the components, they're so well soldered and so close to each other that I am afraid I will damage them and the board.

What I would say regarding your comment, is that for sure I have very dense routing on the pcb. For example I compared with an older pcb version that I didn't have issues with the cameras. Some routes are on the bottom part and in general it looks more neat, the components are closer to each other.

Maybe if I try to redesign it? In that case I will need guidance for the physical layouts, and I also need to add more pins to be able to monitor what's happening in many nodes. Would you advise any book/guide for that case?

- - - Updated - - -

I found this very interesting article regarding the design of the pcb. There are specific rules for the design of the USB pcb:

**broken link removed**


I also found that the characteristic impedance for usb 2.0 high speed channels, thus Data+ and Data -, should be 90 Ohms. How will I measure this?

If there is some error in the design of the pcb, how can you design your pcb to use impedance controlled tracks ?
 

PCB transmission line impedance will be primarily calculated based on geometry (differential micro strip, differential coplanar strips with or without ground plane). If you can show the details of the USB-Hub to connector wiring (layer stack-up, trace width and separation), we can check it. Best you show a layout screenshot with stack-up information or post gerber files.

I notice that the connectors to cameras are apparently not standard USB connectors. I wonder how you connected the USB cables and if this connection keeps the 90 ohms differential impedance.
 
You made a very good point, I was also doubtful regarding the connectors. We have only four channels, Vdd, GND, Data+ , Data-. There is no shielding. It is true that USB standard cable has two more shields, but I think that for the 2.0 that we use we don't need it. I found that these additional shields (purple-orange, blue-yellow)are for USB 3.0. Not sure why this was chosen. There is no shielding on the pcb. The connector of the camera has a 5th channel but we dont use it. Do you think this might be the cause?

I don't have physical layout available right now, but I could show some part in a photo. I hope they provide it to me soon so that I calculate this impedance. I measured the ohmic resistance, would that help? The problem is on the long cable for the one of the cameras. The following values are for a good working pcb. Normally I loose the streaming on the camera with the long cable.


For the long cable, the ohmic resistance is:
when pcb off: Data + channel: R =0.2 Ohms, Vdd channel: R = 0.2 Ohms
when pcb on and not streaming camera: Data + channel: R =0.3 Ohms, Vdd channel: R = 3 Ohms
when pcb on and streaming camera: Data + channel: R = 0.8 Ohms, Vdd channel: R = 87.6 Ohms

For the short cable: the ohmic resistance is :
when pcb off: Data + channel: R =0.0 Ohms, Vdd: R = 0.1 Ohms
when pcb on and not streaming camera: Data + channel : R = 0.0 Ohms, Vdd: R = 1.2 Ohms
when pcb on and streaming camera: Data + channel:R = 0.1 Ohms, Vdd: R = 36.8 Ohms
 

It's true that USB high speed uses a twisted pair for data without individual shield. But the wire and insulation dimensions are specific to achieve 90 ohm differential impedance. The common foil and copper braid shields around the four wires, designated inner and outer shield in the picture are also required. Review USB 2.0 spec paragraph 6 in case of doubt.

usb cable.jpg
 
Thank you very much for your reply. I know that there is a braid as you very well mentioned, but the truth is I wouldn't imagine it can cause such an issue. I will take this into consideration. I will post layout of the pcb when I have it available.
 

Status
Not open for further replies.

Part and Inventory Search

Welcome to EDABoard.com

Sponsor

Back
Top