USB Low Speed Hardware Specifications

[alberich]

Hello (from France, so I want to apologize for the errors in my text, but I hope somebody will understand my question).

I am developping (as a hobbyist) an interface board using the IOWarrior24 manufactured by CodeMercs, Germany (which has a USB Low Speed interface). It is in fact a Cypress microcontroler with the appropriate software.

In the specs (by CodeMercs) I read that the USB Low Speed norm didn't allow detachable cables at peripherals and so the cable had to be "captive".

I know that this point is not important for a personal development which will not ne sold, but if this is true it would mean that no USB 1 peripherals had the "B-type" plug and that there were no detachable USB cables before USB 2 ! So, where do the old, non shielded USB 1 cables come from ?

This seems to be VERY strange. Or am I misunderstanding something ? Perhaps this specification applies only to HID-Class peripherals, like the IOWarrior ? Or perhaps this spec didn't exist before USB 2 appeared?

Thanks for any answer ...

[Jan Axelson]

USB 1.0 defined both low and full speeds.

Low speed requires a "captive" cable. As defined by the USB spec:

Assemblies are considered captive if they are provided with a vendor-specific connect means (hardwired or
custom detachable) to the peripheral.

So a captive cable can be removable, but it can't use a standard USB connector on the device. This requirement is to prevent users from attaching low-speed cables to higher-speed devices.

[alberich]

Thanks for Your answer ...
But I think the specification is (from a technical point of view) silly : the problem could be solved if all low-speed devices with a detachable cable used the B-Plug and were delivered with a high-speed cable ! So no user could attach a low-speed cable to a high-speed device, because no low-speed cables exist (yes, I know, more expensive ... But a proprietary non-standard connector is not cheap, too ! )  
Alberich

[Barry Twycross]

I'm not sure what you're getting at.

I'm also not sure you're understanding the situation. USB 1.x defined Low (1.5Mb/s) and Full speed (12Mb/s). USB 2 further defined high speed (480Mb/s). I'm not sure you're using "low speed" correctly, you seem to be meaning all USB 1 speeds.

Full speed devices could and can use detachable cables. There is no such thing as a high speed cable, the full speed cable specified by the 1.x spec is basically the same as the high speed cable specified by the 2.0 spec. It was found that if a cable complied with the 1.1 spec, it would also pass high speed traffic and complied with the 2.0 spec. The cable spec didn't change it was just further specified (which may be basically table 7-8).

Cables for low speed devices (1.5Mb/s ones, not 12Mb/s ones) are much cheaper and are usually soldered to the device, which is cheaper than providing a connector of some sort.

Cables for full speed devices (12Mb/s) can be detachable and quite often are.

[Jan Axelson]

Yes, Barry is correct that low-speed cables are cheaper, and cost is a factor for inexpensive devices like mice. Low-speed cables can also be more flexible (important for mice) though USB 2.0 added a requirement for shielding.

[alberich]

Thanks for Your answers again.
I did APPROXIMATELY  understand the problem ; but the "gag" is that even PROFESSIONAL developments (by CodeMercs themselves) use a B-Plug, for example on their evaluation kits (but they write that this is not in conformity with the norm ! ). Evaluation modules with FTDI chips are using the same plug. This is ... for cost reduction, because they are simply delivered WITHOUT any USB cable ! I saw some prototypes in electronics magazines with the same "problem".
So I think this norm is not, as I wrote, a problem in a hobbyist environment, it becomes important for a device to be sold, in order to put the USB-conformity logo (which includes many other specs) on it.
Alberich

[Guido Koerber]

Hi Alberich,
I am the one who wrote the documentation for the IO-Warrior24 (and did the chip).

As Barry already pointed out there is a low speed cable spec for a simpler cable that works only with low speed signaling.

The idea behind the USB specs definition that low speed devices need a captive cable is to make sure there are no USB cables that meet only the low speed specs but have standard connectors on both ends. Low speed cables don't need to have shield which allows to make them much cheaper and lighter but it also means they will not work with full speed or worse high speed.

So this definition has been made to prevent problems with non-low speed devices.

In case of the IO-Warrior24 starter kit we decided to bend this spec by following it in the intention and not in the words. We do not provide a low speed cable but use a standard B plug which allows you to use any off the shelf cables that are available in all lengths (even beyond the maximum 5m...). Of course it also eliminates a point form our BOM and reduces logistics.

[Jan Axelson]

In my experience, it's not unusual for development boards and tools to use non-standard connectors (or in this case, using a standard connector in a non-standard way) to save cost and maybe be more convenient for users. I see no harm in it.