Hello, Is there a ADC and DAC of 20 bit operating frequency of 2.4 MHz. If yes which company provide this. If not how can i get 20 bit resolution using lower number of bit say 16 bit.... any suggestion is welcomed waiting for reply, with regards praveen
ADC and DAC of 20 bit with operating frequency of 2.4 MHz???
Started by ●September 29, 2003
Reply by ●September 29, 20032003-09-29
"praveen" <praveenkumar1979@rediffmail.com> wrote in message news:ff8a3afb.0309290611.53312138@posting.google.com...> Hello, > Is there a ADC and DAC of 20 bit operating frequency of 2.4 MHz. If > yes which company provide this. If not how can i get 20 bit resolution > using lower number of bit say 16 bit.... > any suggestion is welcomed > waiting for reply, > with regards > praveenSaying "ADC *and* DAC" suggests a misconception. ADCs are generally slower than DACs for a given word length because of the underlying technical approaches demanded. Some ADCs use DACs as an internal component. So, the ADC in this case could not be as fast as the DAC - as a general comment. Different companies specialize in each I believe. Analog Devices says they have 17 bits up to 1MHz (msps) and 16 bits up to 10Mhz (msps) at: http://www.analog.com/Analog_Root/sitePage/mainSectionContent/0,2132,level4% 253D%25252D1%2526ContentID%253D8777%2526level1%253D117%2526level2%253D118%25 26level3%253D122,00.html Maxim has 16 bits at 1msps etc.... Fred
Reply by ●September 30, 20032003-09-30
praveen wrote:> Hello, > Is there a ADC and DAC of 20 bit operating frequency of 2.4 MHz. > If yes which company provide this. If not how can i get 20 bit > resolution using lower number of bit say 16 bit.... > any suggestion is welcomed > waiting for reply, > with regards > praveenI own an evaluation board of a 14bit ADC which operates up to 55MS/s. It's two years old, thus faster ones should be available yet. If you'd take such an ADC with let's say 48MS/s at 14bit, this would mean that you can oversample your 2,4MHz: Instead of having at least one sample per half period, you'd have 10. If you intended to say 2,4MS/s instead of 2,4MHz, you get 20. Easily spoken this means that you sample the signal more precisely (averaging on the error). This leads to a gain in resolution which can be up to ln(oversampling factor)/ln(2). In the above case of 10 this would be ln(10)/ln(2) = 3.3. You'd end up with a theoretical resolution of 14+3.3 = 17 bit. HTH Bernhard
Reply by ●September 30, 20032003-09-30
Those 14 bitters can now go up to 105 MS/sec (see AD6645) so you'd get some more resolution. ADI also has some multichip modules where they combine adcs to get increased resolution (see AD10678 for 16 bit 80 MS/sec). There's also the AD9260, which they spec as 2.5 MS/sec 16 bit, but it's really sampling at 20 MS/sec with optional decimation (up to 8). I don't think you're going to find a 20 bit ADC that will run at 2.5 MS/sec - I seem to recall any with that resolution will top out around audio rates ... Have a look around at some manufacturers web pages - I just looked at ADI's but there are others :-) and no, we don't exclusively use ADI ADCs or DACs, it just seems to be on the high speed stuff we do ... ----- Ron Huizen BittWare "Bernhard Holzmayer" <holzmayer.bernhard@deadspam.com> wrote in message news:1402364.Gg6ggl9MlZ@holzmayer.ifr.rt...> praveen wrote: > > > Hello, > > Is there a ADC and DAC of 20 bit operating frequency of 2.4 MHz. > > If yes which company provide this. If not how can i get 20 bit > > resolution using lower number of bit say 16 bit.... > > any suggestion is welcomed > > waiting for reply, > > with regards > > praveen > > I own an evaluation board of a 14bit ADC which operates up to > 55MS/s. It's two years old, thus faster ones should be available > yet. > If you'd take such an ADC with let's say 48MS/s at 14bit, this would > mean that you can oversample your 2,4MHz: > Instead of having at least one sample per half period, you'd have > 10. If you intended to say 2,4MS/s instead of 2,4MHz, you get 20. > > Easily spoken this means that you sample the signal more precisely > (averaging on the error). This leads to a gain in resolution which > can be up to ln(oversampling factor)/ln(2). > In the above case of 10 this would be ln(10)/ln(2) = 3.3. > You'd end up with a theoretical resolution of 14+3.3 = 17 bit. > > HTH > Bernhard
Reply by ●September 30, 20032003-09-30
Hello Praveen, Do you really need 20 bits of dynamic range? Or can you use a prescalar? Digital radios are quite content with fewer bits provided they have an AGC before the converter. When I think of 20 bit ADCs, I'm thinking of instrumentation convertors which will be way too slow for you. They operate in the kHz range. Tell us more about your app, and we may come up with a way to do it with obtainable devices. Clay "praveen" <praveenkumar1979@rediffmail.com> wrote in message news:ff8a3afb.0309290611.53312138@posting.google.com...> Hello, > Is there a ADC and DAC of 20 bit operating frequency of 2.4 MHz. If > yes which company provide this. If not how can i get 20 bit resolution > using lower number of bit say 16 bit.... > any suggestion is welcomed > waiting for reply, > with regards > praveen
Reply by ●October 1, 20032003-10-01
Hello Mr.Turner,
We need this ADC to generate a phase shift of 0.1 dgrees in a
200 Khz signal sampled at 2.4 Mhz. In our application the phase shift
is generated using DSP and we need to collect this signal through an ADC
and send it through a DAC.If we do not represent the signal with 20 bits
the 0.1 degree phase shifted signal can not be generated.
Regards
Praveen Kumar
%=====================================xxxxxx=========================
"Clay S. Turner" <physicsNOOOOSPPPPAMMMM@bellsouth.net> wrote in message news:<Z5geb.4973$md5.2909@bignews6.bellsouth.net>...
> Hello Praveen,
> Do you really need 20 bits of dynamic range? Or can you use a prescalar?
> Digital radios are quite content with fewer bits provided they have an AGC
> before the converter. When I think of 20 bit ADCs, I'm thinking of
> instrumentation convertors which will be way too slow for you. They operate
> in the kHz range.
>
> Tell us more about your app, and we may come up with a way to do it with
> obtainable devices.
>
> Clay
>
>
>
>
> "praveen" <praveenkumar1979@rediffmail.com> wrote in message
> news:ff8a3afb.0309290611.53312138@posting.google.com...
> > Hello,
> > Is there a ADC and DAC of 20 bit operating frequency of 2.4 MHz. If
> > yes which company provide this. If not how can i get 20 bit resolution
> > using lower number of bit say 16 bit....
> > any suggestion is welcomed
> > waiting for reply,
> > with regards
> > praveen
Reply by ●October 1, 20032003-10-01
praveen wrote:> Hello Mr.Turner, > We need this ADC to generate a phase shift of 0.1 dgrees in a > 200 Khz signal sampled at 2.4 Mhz. In our application the phase > shift > is generated using DSP and we need to collect this signal through > an ADC and send it through a DAC.If we do not represent the signal > with 20 bits the 0.1 degree phase shifted signal can not be > generated. Regards > Praveen Kumar > ...0.1 degrees: relative to what? If you have a 200kHz sine wave: the longer you sample, the better the precision - with a pure sine signal you have time enough. If you haven't a sine signal, what is 0.1 degrees? Do you mean a constant time delay, then. If so, you wouldn't require a better ADC, but only a fast enough buffer and then (I guess two) DACs which construct the signals from different buffer locations. Or, just another idea: maybe, you can set up an analog delay line which is only adjusted by the DSP. Depends on your app. Bernhard
Reply by ●October 2, 20032003-10-02
Ron Huizen wrote:> Those 14 bitters can now go up to 105 MS/sec (see AD6645) so you'd > get some > more resolution. ADI also has some multichip modules where they > combine adcs to get increased resolution (see AD10678 for 16 bit > 80 MS/sec). > ... > Ron Huizen > BittWare >Thanks, Ron, for this knowledge update. I'm going to need this information, soon. Bernhard
Reply by ●October 2, 20032003-10-02
"praveen" <praveenkumar1979@rediffmail.com> wrote in message news:ff8a3afb.0309302339.50d85e14@posting.google.com...> Hello Mr.Turner, > We need this ADC to generate a phase shift of 0.1 dgrees in a > 200 Khz signal sampled at 2.4 Mhz. In our application the phaseshift> is generated using DSP and we need to collect this signal throughan ADC> and send it through a DAC.If we do not represent the signal with 20bits> the 0.1 degree phase shifted signal can not be generated. > Regards > Praveen Kumar >%=====================================xxxxxx=========================> > "Clay S. Turner" <physicsNOOOOSPPPPAMMMM@bellsouth.net> wrote inmessage news:<Z5geb.4973$md5.2909@bignews6.bellsouth.net>...> > Hello Praveen,<snip>> >> > Tell us more about your app, and we may come up with a way to doit with> > obtainable devices. > > > > Clay > > > > "praveen" <praveenkumar1979@rediffmail.com> wrote in message > > news:ff8a3afb.0309290611.53312138@posting.google.com... > > > Hello, > > > Is there a ADC and DAC of 20 bit operating frequency of 2.4 MHz.If> > > yes which company provide this. If not how can i get 20 bitresolution> > > using lower number of bit say 16 bit.... > > > any suggestion is welcomed > > > waiting for reply, > > > with regards > > > praveen >The bit width of the ADC/DAC has no direct bearing on the resolution of an applied phase shift or time delay. Your 0.1 deg at 200kHz comes out at 1.39nsec (assuming I've not fat-fingered the calculator ;-), you may need to pay lots of attention to group delay and stability of your reconstruction and antialiasing filters. (Omitting assumptions about jitter, reasonable number of bits, averaging of the phase shift etc) Regards Ian
Reply by ●October 2, 20032003-10-02
Bernhard Holzmayer wrote:> Ron Huizen wrote: >=20 >=20 >>Those 14 bitters can now go up to 105 MS/sec (see AD6645) so you'd >>get some >>more resolution. ADI also has some multichip modules where they >>combine adcs to get increased resolution (see AD10678 for 16 bit >>80 MS/sec). >>...=20 >>Ron Huizen >>BittWare >> >=20 > Thanks, Ron, for this knowledge update. > I'm going to need this information, soon. >=20 > BernhardDon't you also need to add in a dither signal for best results? Maybe if the signal is small enough, thermal noise suffices. Your formula, ln(10)/ln(2), works with logs of any base. Since precision goes up as the square root of the degree of oversampling, I think it should be log(n)/2*log(2). Using base 2 makes it particularly convenient, and gives log2(N)/2. (N, of course, is the oversampling ratio.) Stated simply, adding in one-LSB pseudorandom noise oversampling by 256 times, them averaging, resolution is improved by sqrt(256)=3D16: 4 bits. = I once had a piece of AD equipment that did this. It allowed them to use one converter either as 12-bit, high speed, or 16-bit, low speed. Jerry --=20 Engineering is the art of making what you want from things you can get. =AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF= =AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF= =AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF
