USB DAC input filter

Status
Not open for further replies.
Jump to Latest
#23
Thanks for that, very useful. It proves the ferrite will do everything the Jitterbug does - but at the fraction of the cost. The ferrite will also allow experimentation with number of USB data cable loops.

Also, the ferrite does not introduce an additional, undesirable contact resistance point like the Jitterbug does.

I'd forgotten to say that the USB cable shown in post #14 photo (with 4 loops), carries data signal only. The power comes from a dedicated low noise power supply. So, the ferrite does not affect the power supply in any way.

Nick
 
#25
Well, the ferrite should affect the spikes coming from the source of USB data (computer / laptop), not the rise / fall time of an USB data stream. Further to this, the ferrites have certain, maximised attenuation for a specific range of frequencies. Try to put the oscilloscope between data + and data - with a differential probe, and then between data + and shield, and data - and shield with an ordinary probe... and see what needs to be done... the reading you posted in #22 is in fact very useful.
Throwing a ferrite, blindly, and expecting it to work -> is a gamble, unfortunately.

Regards,
Nick
 
#26
Angelm said:
Unfortunately, feritte does not do same job. I did try feritte, and don't like the sound in my system. Same in my friends. For me Jitterbug is cheapest way to improve sound from computer. I will try Würth Filter soon.
Click to expand...

If on the USB lines I would be very curious as to how this could be? having seen ferrites used a wide variety of equipment with zero change to the data passing through.....
 
#27
#28
I found the attached schematic in ti.com's forum.
I think I will start experimenting with this.
choosing values is a bit trick for me tho.
Also an idea that popped in my head, what about isolating grounds with a 0 ohm resistor
instead of a bead?
 

Attachments

  • 6355.USB.PNG
    6355.USB.PNG
    68.4 KB · Views: 198
#29
A zero Ohm resistor is a link it does not isolate the ground. The whole point of using the bead is to split physically the grounds, the high frequency noise is then attenuated by the bead... Have a look at the design I posted on the other thread, I believe I did a detailed description of the reasons why things were done as they were.... The zero R above is to allow the USB shield to be separated from the USB 0V or for a cap to be used in place of the resistor to provide a low impedance connection at high frequencies whilst isolating the screen from USB ground at lower frequencies.
 
#30
marce said:
Interesting and very similar to the design I posted last August🙂
It has a clear break between input and output, which means no capacitive coupling of the noise from the upstream device.
Click to expand...

Yes I noticed that. Would you say ground isolation this way is the main way for improving the USB interface?

Anyway, I've ordered one as it's so cheap. If it makes no difference I'll put it in a matchbox, paint a logo on it and sell it on ebay for £40.
 
#33
Spartacus said:
Yes I noticed that. Would you say ground isolation this way is the main way for improving the USB interface?

Anyway, I've ordered one as it's so cheap. If it makes no difference I'll put it in a matchbox, paint a logo on it and sell it on ebay for £40.
Click to expand...

We are talking EMC engineering and removing coupling mechanisms for the noise, if you have one contiguous ground then any noise will couple capacitivly to any routes and power lines that are over the ground plane. The scheme I showed was from some work I did where we had to have immunity up to 18GHz, not from a commercial design, where I worked with an rf engineer for two weeks on all aspects of noise reduction... and learned an awful lot regarding coupling mechanisms especially for high frequency noise, which is prevalent when you have digital switching such as in a PC, look up simultaneous switching noise.
 
#39
A ferrite on the USB data line would do more harm than good, same goes for other digital high speed connections like HDMI and LAN.

You know how USB is such an iffy connection that always has spoiled connectors/cables and have to fallback to lower speeds because the cable degraded till it cannot support high speed?

USB 2.0 480Mbps has a clock of 240MHz and bandwidth of 1.2GHz. This is higher than lots of airborne signals (e.g. on-air TV) and higher than most SMPS switching frequency. The ferrite will block out more signal than noise.

In addition, high speed connections are impedance controlled in order to hit those speeds, so adding a ferrite choke (which is a high impedance in series) will cause signal reflections.
 
#40
wwenze, what about the choke i posted from coilcraft?
That one supposedly is meant for USB 3.0 communication.
Also from the comments section:
"
bogatin
There is nothing fundamental linking the clock frequency and bandwidth. As we saw above, with three different rise times, you could have a BW that is 1x the clock frequency to 10x the clock freq or higher, depending on the shape of the rising and falling edges.

This is why the term bandwidth is so ambiguous.

It is not surprising that if you ask 5 people for the bandwidth of a signal based on the clock frequency, you get 6 different answers.

It's really all about the assumption of the rise time in the signal, and then, assuming the signal has a gaussian edge.

If it is important to know 3.6 or 4.5x the bit rate, then don't use the figure of merit of bandwidth, use the whole spectrum."
 
Last edited:
Status
Not open for further replies.
Top Bottom