High speed ADC with ethernet output

Hi all

I want to develop a system for my next project. I have to use a high speed ADC with ethernet output so that I can transmit the digital data through wifi. The ADC speed should be above 5 Msps and I think it will be transmitted at around 1-2 Mbps through Wifi. Is there any module which can perform this task? If there is any solution that can transmit data faster than wifi, I will change that as well.

Do I need to go for FPGA boards or a microcontroller development board will do fine? If you have done similar projects, plz help me choose the best hardware for my requirement.

You've got a problem. You can't sample at 5Msps and then send your data at 2 Mbps. Assume, for example, that your ADC is an 8-bit part (you didn't specify), that means you will need a 40Mbps interface.

barry said:

You've got a problem. You can't sample at 5Msps and then send your data at 2 Mbps. Assume, for example, that your ADC is an 8-bit part (you didn't specify), that means you will need a 40Mbps interface.

Click to expand...

Is there any wireless interface that can support 40 Mbps? Please suggest me if you know. I was considering that if we do ADC conversion in 5 Msps, then some time will be lost during some other operations and then I can send data at around 2 Mbps through wifi. Do u think it is possible? If not, please suggest me some ideas. It would be a great help.

If you use a 5-mega-sample-per-second ADC you will get, TA-DA!! 5 million samples per second. If you want to throw away some of those samples (decimation) that's fine. But you need some specifications. What is your REQUIRED output sample rate? What is your sample width?

I'm not an expert, but I know that certain versions of Wifi can support 40 Mbs. There are probably other wireless protocols that might work.

barry said:

If you use a 5-mega-sample-per-second ADC you will get, TA-DA!! 5 million samples per second. If you want to throw away some of those samples (decimation) that's fine. But you need some specifications. What is your REQUIRED output sample rate? What is your sample width?

I'm not an expert, but I know that certain versions of Wifi can support 40 Mbs. There are probably other wireless protocols that might work.

Click to expand...

Actually I need to get fastest data transfer rate that I can get. So I need to design ADC accordingly. Till now I found some wifi modules like Xbee Wifi which can transfer up to 2 Mbps, so I took it as a baseline and wanted to design ADC circuit accordingly. So I was curious about what kind of ADC will be good for this requirement. The 12 bit ADC will be fine for my use. Can u plz suggest some ADC for this requirement?

So if your maximum data rate is 2 Mega-bits/sec, your maximum a-d sample rate would be 2,000,000/12=166K-samps/sec. There are many, MANY, ADCs that can meet this requirement, look a LTC1860, for example. Look at Analog Devices, Linear technology, Texas Instruments.

Hello!

So if your maximum data rate is 2 Mega-bits/sec, your maximum a-d sample rate would be 2,000,000/12=166K-samps/sec.

Click to expand...

Even this figure is optimistic.

1. If XBee announces 2 Mbps, there is some overhead. I don't know what payload you can have, but usually it's
quite far from the max theoritical rate.

2. Using radio devices, you need to add some security for sample validation. For example a CRC. You might also
want to have markers (e.g. begin of packet),etc... So even if you have a 2 Mbps payload, you will get less with all
this overhead.

3. You have to decide what to do when you receive an invalid frame. Discard? Resend? If you discard, then you
will get less data. If you want to resend, then you have to send a message to your device, saying please resend
packet xxxx.

4. By the way, you may also need a packet ID, some kind of timestamp, etc...

The final throughput will really depend on what you want to get.

Dora.

doraemon said:

Hello!



Even this figure is optimistic.

1. If XBee announces 2 Mbps, there is some overhead. I don't know what payload you can have, but usually it's
quite far from the max theoritical rate.

2. Using radio devices, you need to add some security for sample validation. For example a CRC. You might also
want to have markers (e.g. begin of packet),etc... So even if you have a 2 Mbps payload, you will get less with all
this overhead.

3. You have to decide what to do when you receive an invalid frame. Discard? Resend? If you discard, then you
will get less data. If you want to resend, then you have to send a message to your device, saying please resend
packet xxxx.

4. By the way, you may also need a packet ID, some kind of timestamp, etc...

The final throughput will really depend on what you want to get.

Dora.

Click to expand...

Thank u for this valuable suggestion. So if i set my ADC to sample at 1 Msps, what will be the best way to transfer the ADC output through the wireless medium. Please help me on this. After looking at your comment, I think Xbee wifi will not be enough to transfer the data if the ADC sampling rate is 1 Msps. Please kindly suggest me some ways to do it.

I think you are getting confused between sample rate and data rate. You have to multiply your sample rate by the number of bits-per-sample to get your data rate.

Hello

Thank u for this valuable suggestion. So if i set my ADC to sample at 1 Msps, what will be the best way to transfer the ADC output through the wireless medium. Please help me on this. After looking at your comment, I think Xbee wifi will not be enough to transfer the data if the ADC sampling rate is 1 Msps. Please kindly suggest me some ways to do it.

Click to expand...

Barry is right. 1 Msps, if you set your ADC to 8 bits, is 8 Mbps. You cannot make it.
Again, it really depends on what you want to do, but starting from a bit rate of 2Mbps, if you can send 100k samples
of real (and valid) data, then it's a fairly good design.
But before deciding the physical medium to transfer your data, you should specify clearly
what your specs are. Do you really need 5Msps? Do you need this sample rate continuously?
For instance, there are cases where you indeed need a high sample rate, but bursty.

Dora.

Just to repeat: You need some kind of specification. You've got numbers flying all over the place here without any clear definition of what you really want to do. Do you need to sample continuously? How many bits? How fast is the signal you're looking at (that can help determine the required sample rate)? How fast must you transfer the data, i.e, how many samples per second do you NEED at the receiver end?

Before you even THINK about design, you need to establish some requirements.

Thank u for all your valuable suggestions. I am new to this thing so I am really confused. My objective is to transmit 1 MHz transducer output in a real time using wireless medium. How can I make it possible? I think for 1 MHz transducer output, the ADC sampling rate must be higher than that. How can I choose the best ADC and wireless medium according to my requirement? Please help...

If you are sampling a 1MHz signal (and that's assuming a 1MHz sinusoid), your minimum sample rate is 2MHz. You still have not defined what your sample size is , so we still can't tell you what your minimum data rate is.

I might be a little harsh here, but it sounds like you've got yourself into a project with a lot of elements of which you know very little: sampling, wireless, ethernet, etc.. While I admire your ambition, it takes people a lot of time, energy and experience to get to the point where they can design a system such as you are describing. Is this a project for school? Can you, perhaps, pick something less challenging? If this IS a student project, can't you get some guidance from your professor?

Hi
I know I m just a novice but still I need to do this to complete my project. Please help me on this or show me the path. I've calculated my minimum sampling rate to be 1 MHz and the sampling size is 12 bits. So does that mean I need 12 Msps ADC? and how to get this to transfer through wireless medium.

I really want to learn this things and be able to do myself. I just need you guys to show me path so that I can choose which devices to use. Please help me on this.

Hello!

We were asking for SPECS and all we know is the sampling rate and data precision.
No, sampling at 1MHz means 1 Msps (mega sample per second).
Now as you have 12 bit data, you need a payload of 12 Mbps (mega bit per second).
This will be extremely difficult with a 2 Mbps connexion. But before thinking further:
- Do you need to ensure that all your data is valid, or can you afford to loose data?
- If you can afford to loose data, how much can you afford to loose?
- If you cannot afford to loose data, what will be your error recovery policy?
- What do you do if you loose data (which WILL be the case with a radio connexion)?
- Can you estimate the bit error rate of your connexion?
-> Depending on the bit error rate, you will have to resend more often...
-> The size of your buffers will depend on the bit error rate (if you have many
errors, then you have to resend often. If the buffer is too large, you will have
no valid data at all...

Hints: if you are stuck to a 2 Mbps connection, why not trying a lower bit rate,
and if it works, upgrade?
If you are not stuck to a 2 Mbps connection, why not choosing a better one?

Dora.

To expand on what doraemon said, can you send bursts of data? In other words, do you need to send 1 Msps continuously, or can you just capture blocks such that you can have an AVERAGE data rate of 2 Mbps? For example, you can capture 10 samples(at 1MHz) every 100 u-sec(100 cycles), which would give an average DATA rate of 1.2Mbps.

What is it you are actually sampling?

Thank you for the reply..Actually I can not afford to loose data...for error recovery, can't we use some type of channel coding...although I am not sure that is the right way...Can't we use ADC and then save the output data in buffer till it gets filled and then transmit it by wireless. And it is not necessary that I should use 2 Mbps connection. I am looking for other options of wireless as well which is faster. Please help me...