ADC Converter Ic with high bit rate

[dineshbabumm]

Can anyone please suggest a good ADC convertor IC with high bit rate.. It can be used to build an oscilloscope.. We are thinking about it.. So,please suggest one good ic.. Will be nice if it got high bit rate..

 

[hjames]

There are a large number of 40-80Msample/sec ADCs that are used for digitizing video signals. The current set of high end ADC's runs in the Gsamples/sec range and are used for radio and specialized sensor applications. Look for parts from Analog Devices, Maxim, TI, National, (and probably any other company which makes analog parts). Finding a fast ADC isn't difficult - dealing with the 40-1000MB/sec data output is.

Fer example, the Analog Devices, AD9200 is a 10 bit 20MS/s ADC. There looks like there are higher speed variants - which go up to 100MS/s.

So here's a question back: what are you connecting the ADC to?

 

[dknguyen]

Dealing with the output usually requires an FPGA- expensive, hard to get, and extremely user unfriendly which is the main thing that seems to have driven people on this forum away from trying to build their own oscilloscope (they cost $2000+ for a reason).

 

[Papabravo]

What you have to connect the fast ADC to is a boatload of very fast memory whose care and feeding is by no means an easy task. You'll probably learn more and have more fun with an analog scope. I started one in 1974 and I achieved sufficient financial resources to buy one, a Tektronics 2264, before I finished it.

 

[dineshbabumm]

There are a large number of 40-80Msample/sec ADCs that are used for digitizing video signals. The current set of high end ADC's runs in the Gsamples/sec range and are used for radio and specialized sensor applications. Look for parts from Analog Devices, Maxim, TI, National, (and probably any other company which makes analog parts). Finding a fast ADC isn't difficult - dealing with the 40-1000MB/sec data output is.

Fer example, the Analog Devices, AD9200 is a 10 bit 20MS/s ADC. There looks like there are higher speed variants - which go up to 100MS/s.

So here's a question back: what are you connecting the ADC to?

Thanks for your reply.. We are planning to connect the ADC to computer through ports and use it like an oscilloscope using the softwares avilable.. So,we can have our own oscilloscope.. So to make it work at atleast 100 MHz,we need ADC ICs.. It is the first time we are going to use ADCs.. So no idea about which IC to go about..

 

[dknguyen]

Is your computer port even able to handle at 100MHz?

 

[dineshbabumm]

Is your computer port even able to handle at 100MHz?

I dont know much about computer ports.. My friend knows about this well.. So i don't know about this.. I think it must be able to handle high frequency.. Correct me if i am wrong..

 

[Papabravo]

I believe that your little friend is misinformed. 100 MHz is an A/D conversion every 10 nanoseconds. The access time of your memory needs to be less than that and I don't think you're going to make it. The bandwidth of the signals you can observe will top out at maybe 10-15 MHz. Not a very capable scope.

Oh, 1 msec worth of data will fill 100K words or 200K bytes, but hey memory is cheap right? especially the kind with sub 10 nsec write times.

According to the Analog Devices Website you can get an AD9461, at 130 Ms/sec with CMOS outputs for $65.00 ea in 1000 quantity. The LVDS version is $200.00 in 1000 quantity. Maybe if 10% of the people on the forum wanted one we could place an order of 1000 pieces and get the quantity price. I shudder to think what the price will be if you want a handful.

 

[Oznog]

Is your computer port even able to handle at 100MHz?

No, but you don't need to. Typically a scope gets triggered and then takes like 256-1024 samples. Then it displays it or in this case sends this page of data through the serial port. It won't make another page until the transfer is complete, and the computer probably has no use for a continuous stream of 1000MS/sec anyways.

But the computer might not display as smoothly as you'd like at serial port speeds... you might be able to speed that up by using USB.

 

[JimB]

I shudder to think what the price will be if you want a handful.

If he sends them a nice e-mail, maybe they will send him a couple of freebies!

JimB

 

[Nigel Goodwin]

Thanks for your reply.. We are planning to connect the ADC to computer through ports and use it like an oscilloscope using the softwares avilable.. So,we can have our own oscilloscope.. So to make it work at atleast 100 MHz,we need ADC ICs.. It is the first time we are going to use ADCs.. So no idea about which IC to go about..

Your idea is flawed, because you can't simply transfer such large amounts of data at such speeds.

The technique for a high speed computer scope is to have all the hardware in the scope part - memory and everything, so the entire operation is internal to the scope hardware. All the computer does is produce a display based on the small amount of slow data sent it from the hardware.

This, unfortunately, makes the hardware complicated and expensive, there are numerous projects about it on the net, it's a common final project at uni.

 

[hjames]

"If I had enough time..."

Here's the components I'd choose:

Xilinx Spartan 3E FPGA with > 100 IO/s. sadly this would probably need to be a BGA package

Cypress PLL clock generator

stick of DDR memory (64 bit bus, would be used underclocked)

Cypress USB interface chip (bog standard, and easy to use)

ADC converter (any of them would work)

Analog section - termination, buffering, and high-speed comparators for triggering

So the general idea is that the FPGA reads in data from the ADC and shovels it to memory. A quick FPGA implementation should be able to run the memory at 100MHz => 100MHz * 8bytes (64 bit bus) = 800MBytes/sec - overhead. The PLL chip will generate the clocks, and the USB will simply dump the memory to the computer.

Cost estimate:

Parts - ~$300 (USD)
(I'd guess > $150 is wrapped up in the USB + FPGA + RAM + ADC)

Board - ~$400 (USD)
4-6 layer boards aren't cheap

Design time -
2 weeks of picking components
1 week of FPGA implementation
2 weeks of debugging the mess
1 week of assembling the monster

This would be ~1 month of full-time effort, plus I'd budget at least $1500, for the inevitable board screwup, and other unforseen expenses.

One thing to note is that if the target is in the 20-30MHz range, things do get *much* cheaper and easier too...

 

[oandac]

There are some ready made oscilloscopes. You can find similar devices in state.
https://www.picotech.com/oscilloscope.html
You can also control it with your own software!

You can use PCI or PCIE DAQ card which has onboard clock, memory and ADC. NI, advantech. IOtech and many other vendors produces nice and smootly working cards for PC.

As far as I know you can not access to USB port not less than 1 msec intervals. Windows can not handle it ! less than 10 msec is not accurate on windows. Keep in mind that, if you want to get data from memory you used via USB, acquisition rate and the amount of memory should fit or you will have a buffer over flow problem.
Read more on PCI and USB https://forums.ni.com/ni/board/message?board.id=70&message.id=4700