How do I measure bit error rate? 2

[bpelec]

I need to measure the bit error rate between the input and output of a slip-ring. The channel is rated to work up to 1.5 Ghz, so I don't need to go above this frequency.

Having had a quick look at equipment that can automatically do this, and seeing the £40,000 price tag, I would like to learn more about what the measurement is and whether there are other ways of doing it.

My questions are these:

1) What is the definition of bit error rate?

2) Is it possible to measure bit error rate with a signal generator and a two-channel oscilloscope (I assume not, but I think the answer to this question will help me understand the problem better).

3) Is there a way of measuring bit error rate without buying a £40,000 piece of equipment?

Thank you in advance for your help.

Best regards,

BPELEC.

 

[itsmoked]

Bit error rate usually refers to how many correct bits you get for every badly received bit.

Generally every system has bit errors so the channel must be designed to find them and correct them or conversely find them in packets and request packet re-transmissions.

If the channel is one way then the user of the data must be prepared to reuse old data or discard received data if it is found to be in error. Otherwise the channel error rate is required so a suitable error correction system can guarantee corrected data on the received end.

But I'm rambling...

http://www.highfrequencyelectronics.com/Archives/Jan03/HFE0103_Tutorial.pdf

Lots of trade offs since error correction information now clogs every packet or some real capacity of the channel.

 

[bpelec]

Thank you itsmoked, that was a useful article...

To clarify my situation: All I have is the slip ring, which is a passive tranmission medium, just like a cable.. I have no involvement the way data is sent over the slip-ring. All I need to do is to be able to give a bit error rate reading for the slip-ring.

From reading the article, I have some follow-up questions:

1) It seems that BER depends on the way the data is being sent, ie it depends on the modulation technique. Therefore, when quoting BER for the slip-ring, I will have to quote the type of data used for the test.. This leads me to think that I need some pretty sophisticated equipment to generate specific data modulation methods.. Am I thinking along the right lines here?

2) Can I measure the BER by just measuing the noise on the slip ring channel? If I know the energy of the noise, can I express the ratio of the transmissione energy to the noise enegy, and obtain bit error rate that way?

Thank you for your help.

BPELEC

 

[VEBill]

"...BER depends on the way the data is being sent..."

Exactly right.

 

[bpelec]

Ok.. So is there a way to measure it without an expensive, dedicated piece of kit? Or, failing that, is there a way to get a strong indication of bit error rate with some cheaper equipment?

Thanks again!

BPELEC.

 

[zeitghost]

How about hiring in the expensive kit for a week?

 

[sreid]

Perhaps you could send 100 mbs ether net data between two computers through the spinning slip rings and keep tract of the number of packets and the number of packets in error (CRC resend request). If you also did this without the slip rings, you would have some idea of the added BER from the slip rings.

 

[itsmoked]

bpelec; I think the answers to your questions are:

1)Yes you are thinking the right lines.


2)Yes but it might include a ton of "guess work" and make your result suspect.

Also it sounds like you are selling a generic slip ring product that might be used with several different communications schemes (out of your control). This means that if you figure the BER for a particular scheme your customers will be calling and asking for the BER for some other scheme.. (naturally)

So it might actually make a lot of sense to rent a BER measurement tool for £1k for a week or two that lets you measure the BER umpteen different ways with the push of a button. Then you can specify with confidence several different ways which will delight your customers.

I don't see a way to use a cheaper kit since you would still need to use different comm schemes. But I am not a comm guy so there may be someone who does have a method. A "method" though still leaves you without hard data you can quote as an instrument would.

 

[logbook]

BER (Bit Error Ratio) is the modern expansion by the way.

It is impossible to specify the BER for a passive transmission line. Basically your slip ring system is a noise source. You should quantify it as such.

Suppose you are transmitting baseband digital pulses of signal level 0V for a low and 5V for a high with a threshold comparator set at 2.5V. Provided the noise is less than 2.5V peak you will get no errors at all. If the noise spikes are larger than 2.5V you will get occasional errors. You therefore need to know the peak values of the noise. Typically comms systems assume gaussion noise but you have to characterise the system to see what the noise is actually like.

If you have a BER of 1 in 10E10 with a given transmission scheme and you increase the signal amplitude by say x5 you will essentially get no errors at all! This is why BER is such a useful measure of the effective noise in a comms system.

Notice that a BER tester cannot solve your problem at all. It is really up to you to quote RMS and peak noise and let the customer decide how this affects their comm channel.

 

[itsmoked]

Nice explaination logbook, thanks.

 

[bpelec]

Thank you all for your responses. I think I'm going to initially follow logbook's advice and measure the RMS and peak noise and leave the customer to worry about the BER... If I can't avoid measuring BER, I will rent some equipment as itsmoked suggests.

Best regards,

BPELEC.

 

[MrBananas]

Make sure to note that the noise power is often more important than the noise amplitude in most comm systems.