Hi, I have an astronomical camera with a USB 2 interface. I run it on a windows 10 PC and it has a problem connecting to the client software. If it manages to connect, data transfer does not happen and the camera hangs until it is disconnected (unplug USB cable) and reconnected.

The windows device manager persistently shows ''device descriptor request failed". I know a little about the USB comms protocol and understand this is an initialisation comms failure.

If I unplug all other USB devices from the machine, the camera will work as expected, no problem connecting or downloading images.

The camera manufacturer has been very helpful and has indicated that the problem is probably due to EMI.

My astronomical observatory is in the middle of a farmer's field, in an AONB (no mains electricity and no Electricity pylons). All systems are powered from a 12V lead acid battery.

Although I have some understanding of basic electronics and I program AVR microcontrollers, I don't have sufficient knowledge of USB to fault find this effectively.

I would appreciate any ideas of what I could try to solve the problem.

Please let me know if you need any further information.
thanks,
Paul

 

I'd regard the camera manufacturer's statement with a jaundiced eye! A protocol violation is more likely IMHO. It sounds like Windows has asked for a Device Descriptor and the camera has failed to respond or responded incorrectly. If you had a USB protocol analyzer you might be able to develop some insight. There are Windows applications that will do this if you are interested in taking the dive.

 

Thanks for the reply, and yes I agree about being open minded about this. My systematic approach to fault finding shows that the camera does not fail to connect if it is the only device connected to the computer. I have also tried removing USB plugs one by one, then plugging in the camera after each device removal and trying the connection. It nearly always fails if another USB device is connected.

So if it connects reliably without other devices being present, would that indicate there's no protocol violation?

The other USB devices I have connected are two home grown AVR4809 nicrocontrollers (with 4 x serial interfaces via FTDI)
One arduino mega2560
A camera rotator device which uses a stepper motor (from a well known manufacturer)
A DC motor controller (from a well known manufacturer)
A focus controller (from a well known manufacturer)
Keyboard, Mouse
7 port powered hub (this is used to connect all the above devices except the camera which is connected into a port on the computer chassis). The hub USB cable is a USB 3 type connector plugged into a USB3 socket on the computer chassis, but all other devices in the above list are USB 2

Thanks very much for help with this, I'll explore the USB protocol analyser as suggested.

 

7 port powered hub (this is used to connect all the above devices except the camera which is connected into a port on the computer chassis).

Maybe Windows is being fussy about which ports the devices use. I have a card-reader which works in one of my laptop USB ports but not in another.
Does the camera connect if plugged into one of those 7 hub ports?

 

Since the device operates properly when there is nothing else on the bus, my immediate impression is a voltage sag. This is just surmise, but have you tried progressively adding devices and trying them individually to see the effect?

 

thanks for replies. Yes I have tried plugging in the camera connecting it and taking an image. The download time is about 5 seconds under 'normal' conditions and this is what I get. If I then plug in another USB device, the camera appears to stay connected, but when I try to take an image in the client software (reliable, mature, and widely used software), the download hangs and after 10 minutes I end the task in Windows task manager. I have monitored the supply voltage (fully charged 12V lead acid battery) at 13.5 volts. The computer power supply spec is 12 >= supply <+ 19 V dc. I haven't monitored the USB bus voltage directly, but it's a powered hub (i've used different hubs too, as I have been trying to solve this problem for 9 months or so).

Not sure if I have mentioned that of all the connected devices, it is only the camera which is problematic. All other devices work as expected.

Plugging the camera into the 7 port hub or into a port on the computer chassis makes no difference.

I returned the camera to the manufacturer and it was inspected and tried out over three days with no connection errors. To be honest, if I connect it as the only device, I can replicate that test.

thanks again.

 

thanks for replies. Yes I have tried plugging in the camera connecting it and taking an image. The download time is about 5 seconds under 'normal' conditions and this is what I get. If I then plug in another USB device, the camera appears to stay connected, but when I try to take an image in the client software (reliable, mature, and widely used software), the download hangs and after 10 minutes I end the task in Windows task manager. I have monitored the supply voltage (fully charged 12V lead acid battery) at 13.5 volts. The computer power supply spec is 12 >= supply <+ 19 V dc. I haven't monitored the USB bus voltage directly, but it's a powered hub (i've used different hubs too, as I have been trying to solve this problem for 9 months or so).

Not sure if I have mentioned that of all the connected devices, it is only the camera which is problematic. All other devices work as expected.

Plugging the camera into the 7 port hub or into a port on the computer chassis makes no difference.

I returned the camera to the manufacturer and it was inspected and tried out over three days with no connection errors. To be honest, if I connect it as the only device, I can replicate that test.

thanks again.

OK, then given this additional information, my next suspect is line termination. If there is an impedance mismatch, reflections are going to interfere with the data. I am not clear if the camera works alone on the hub or only alone without the hub (or both).

Do you have to use a particularly long cable on the camera? I am assuming you tried different cables but if they are all long perhaps try a shorter one. Also, as a datapoint, if you plug just a cable into one of the hub’s ports, does it cause the problem?

 

The camera will work if it is connected as the only device and this is true if it is the only device on the powered hub, or if it is the only device plugged to the computer chassis.

I also neglected to say that the camera is a high speed USB 2 device.

Re the cable, yes I have tried a few different cables (all 3m length). The one in use is of a type advised by the camera manufacturer. I am not sure if the manufacturer specifies a maximum length but I use one 3 metres long. If I were to reposition the computer (which is not easy but definitely doable) I could shorten this to two metres and will give that a try and report back in a few days.

I'll also try plugging in just a cable and see what effect that has.

Many thanks for these ideas, much appreciated.

 

The camera will work if it is connected as the only device and this is true if it is the only device on the powered hub, or if it is the only device plugged to the computer chassis.

I also neglected to say that the camera is a high speed USB 2 device.

Re the cable, yes I have tried a few different cables (all 3m length). The one in use is of a type advised by the camera manufacturer. I am not sure if the manufacturer specifies a maximum length but I use one 3 metres long. If I were to reposition the computer (which is not easy but definitely doable) I could shorten this to two metres and will give that a try and report back in a few days.

I'll also try plugging in just a cable and see what effect that has.

Many thanks for these ideas, much appreciated.

One other strategy might be to use a longer cable on the hub and a much shorter cable on the camera.

This is all a matter of eliminating possibilities. The cause is still a mystery but if any of this can elicit a different behavior it could help point to a cause.

 

Further update - today I used a new USB2 shielded cable of length 2m to connect the camera to the USB port on the computer chassis. Connection and data download worked reliably. I tried various tests including disconnecting the camera and reconnecting, computer shutdown and restart, insert USB cable - all worked fine.

I then I thought to do the same with the camera connected to the powered USB hub and that all worked fine too.

So it looks like changing to a shorter cable has solved the problem. It's early days and I'll try it all again tomorrow - timelapse shouldn't make any difference

I'll report back,

thanks for help.

 

It looks like the USB cable length is critical. The camera connects reliably if the USB cable is less than 2m in length. I have tried connecting with one new 2m shielded cable and two different shielded 3m cables. The results of 15 tests are consistent - using 3m cables the camera does not connect reliably, with the 2m one it connects every time.
Most peculiar.
I've read a bit about USB and it seems the cable ends are impedance matched using resistors.

Anyway, it seems the problem is solved, thanks for help.
Paul

 

Further update - today I used a new USB2 shielded cable of length 2m to connect the camera to the USB port on the computer chassis. Connection and data download worked reliably. I tried various tests including disconnecting the camera and reconnecting, computer shutdown and restart, insert USB cable - all worked fine.

I then I thought to do the same with the camera connected to the powered USB hub and that all worked fine too.

So it looks like changing to a shorter cable has solved the problem. It's early days and I'll try it all again tomorrow - timelapse shouldn't make any difference

I'll report back,

thanks for help.

Hi,

I've had that problem too. I also noticed that some cables of the same length work better than others.

The best solution though is to move to a different type of system that does not use USB cameras but better quality security cameras. Better performance, better choice of cameras, better cable length choices (100 feet typical), all around better working.
Another thing that works pretty well is to use a good quality security camera combined with a USB converter that converts the camera to USB then use the computer as if you had a USB camera. You can also buy a security system that has HDMI output and convert that HDMI to USB with a HDMI to USB converter. The price on this stuff came down a lot over recent years you dont have to spend 300 dollars (USD) anymore, easily under 100 for everything.
The security cameras are much better too you can get high power telephoto lenses or very wide angle lenses to cover more ground, and even adjustable telephoto to wide angle. You can even get special motion adjustable cameras that allow you to rotate the camera and tilt also to change the direction of the camera lens (more expensive though). Oh and night vision is almost a given.
It's a whole different world once you leave the USB camera market.

 

Hi,

I've had that problem too. I also noticed that some cables of the same length work better than others.

The best solution though is to move to a different type of system that does not use USB cameras but better quality security cameras. Better performance, better choice of cameras, better cable length choices (100 feet typical), all around better working.

Hi, I have an astronomical camera with a USB 2 interface.

You missed something, @MrAl.

 

You missed something, @MrAl.

Oh geeze thanks for telling me, yeah completely missed that.

In the voice of The Church Lady: "Never mind".

 

Oh geeze thanks for telling me, yeah completely missed that.

In the voice of The Church Lady: "Never mind".

Wouldn’t that be Emily Litella?

 

Wouldn’t that be Emily Litella?

Hi,

The Church Lady i was referring to was from an SNL recurring character played by Dana Carvey starting around 1986.
Her (his) famous last words were "Never Mind" after going through an elaborate dissertation that in the end turned out to be completely moot.
I think there was also a funny dance that went with that character too but dont remember that well now.
Here is a Wikipedia reference:
The Church Lady - Wikipedia

[EDIT: This was actually Gilda Radner, see post #18, thanks]

 

I very much admire the persistence that lead to what is a logical solution. . Cable length affects both signal strength and signal timing and so the fix does make sense.

 

Hi,

The Church Lady i was referring to was from an SNL recurring character played by Dana Carvey starting around 1986.
Her (his) famous last words were "Never Mind" after going through an elaborate dissertation that in the end turned out to be completely moot.
I think there was also a funny dance that went with that character too but dont remember that well now.
Here is a Wikipedia reference:
The Church Lady - Wikipedia

I recognized your reference, but you are confusing The Church Lady’s shtick with Emily Litella’s.

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

A typical example:

What is all this fuss I hear about the Supreme Court decision on a "deaf" penalty? It's terrible! Deaf people have enough problems as it is!

The news anchor would interrupt Litella to point out her error, along the lines, "That's death penalty, Ms. Litella, not deaf ... death." Litella would wrinkle her nose, say something like, "Oh, that's very different," then meekly turn to the camera and say, smiling, "Never mind!"

The Church Lady’s famous taglines were “Well, isn't that special?” and "Could it be ...Satan?". Emily Litella was a very early recurring character. In a way, Church Lady is a inheritor of her legacy being a regular guest on Weekend Update but I think by the time Church Lady appeared, the FCC had dropped the fairness doctrine that was the “reason” Emily Litella was appearing on a newscast.

(EDIT: Church Lady had her own interview program, she wasn’t on the newscast.)

And now we are quite off topic, so I will not be posting more about this.

 

I recognized your reference, but you are confusing The Church Lady’s shtick with Emily Litella’s.

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



The Church Lady’s famous taglines were “Well, isn't that special?” and "Could it be ...Satan?". Emily Litella was a very early recurring character. In a way, Church Lady is a inheritor of her legacy being a regular guest on Weekend Update but I think by the time Church Lady appeared, the FCC had dropped the fairness doctrine that was the “reason” Emily Litella was appearing on a newscast.

(EDIT: Church Lady had her own interview program, she wasn’t on the newscast.)

And now we are quite off topic, so I will not be posting more about this.

Yes you are right, so one last comment.
Yes that was the loveable Glida Radner, sorry about that. She was so great. I miss seeing her.

Back on track...
I have had a lot of problems with USB too numerous to mention. I think the reason is the protocol(s) were made too complicated and not all companies understand it well enough to get it to work right every time. Then we have the upgrades to new standards that keep changing. 1.1, 2.0, 3.0, 3.1, 3.2, 9999.9, etc. It never ends yet we use the same connector.
The benefit is of course scalability, the downside is complications. When will it ever end.