I have outputs from 4 A/D's. Each output is 8 bits ( // output ) and will likely be sampled at 40MHz. I need to manipulate that data in some kind of device. This includes level shifting, multiplying the data against each other and possibly applying some FIR filtering to the data. The output will be three 8 bit data streams running at the 40 MHz rate. Can anyone suggest a part to consider using for an application such as this? Thanks. Brent
Suggest a part to Use
Started by ●January 17, 2009
Reply by ●January 18, 20092009-01-18
On Sat, 17 Jan 2009 15:28:49 -0800, bulegoge wrote:> I have outputs from 4 A/D's. Each output is 8 bits ( // output ) and > will likely be sampled at 40MHz. > > I need to manipulate that data in some kind of device. This includes > level shifting, multiplying the data against each other and possibly > applying some FIR filtering to the data. The output will be three 8 bit > data streams running at the 40 MHz rate. Can anyone suggest a part to > consider using for an application such as this? > > Thanks. > > BrentThe 40MHz suggests an FPGA, or perhaps a large CPLD if they're small FIR filters. -- Tim Wescott Control systems and communications consulting http://www.wescottdesign.com Need to learn how to apply control theory in your embedded system? "Applied Control Theory for Embedded Systems" by Tim Wescott Elsevier/Newnes, http://www.wescottdesign.com/actfes/actfes.html
Reply by ●January 18, 20092009-01-18
bulegoge@columbus.rr.com wrote:> I have outputs from 4 A/D's. Each output is 8 bits ( // output ) and > will likely be sampled at 40MHz. > > I need to manipulate that data in some kind of device. This includes > level shifting,What do you mean by level shifting bits?> multiplying the data against each other and possibly > applying some FIR filtering to the data. The output will be three 8 > bit data streams running at the 40 MHz rate. Can anyone suggest a part > to consider using for an application such as this? > > Thanks. > > Brent-- Engineering is the art of making what you want from things you can get. �����������������������������������������������������������������������
Reply by ●January 18, 20092009-01-18
bulegoge@columbus.rr.com wrote:> I have outputs from 4 A/D's. Each output is 8 bits ( // output ) and > will likely be sampled at 40MHz. > > I need to manipulate that data in some kind of device. This includes > level shifting, multiplying the data against each other and possibly > applying some FIR filtering to the data. The output will be three 8 > bit data streams running at the 40 MHz rate. Can anyone suggest a part > to consider using for an application such as this?The 40MHz sample rate is out of league of programmable DSPs. This is a domain of FPGAs; which to chose depends on how much processing do you need. BTW, it may be the easiest way to do all of the processing in analog, and sample the data at the very end. Vladimir Vassilevsky DSP and Mixed Signal Design Consultant http://www.abvolt.com
Reply by ●January 18, 20092009-01-18
Tim Wescott <tim@seemywebsite.com> wrote:>On Sat, 17 Jan 2009 15:28:49 -0800, bulegoge wrote:>> I have outputs from 4 A/D's. Each output is 8 bits ( // output ) and >> will likely be sampled at 40MHz. >> >> I need to manipulate that data in some kind of device. This includes >> level shifting, multiplying the data against each other and possibly >> applying some FIR filtering to the data. The output will be three 8 bit >> data streams running at the 40 MHz rate. Can anyone suggest a part to >> consider using for an application such as this?>The 40MHz suggests an FPGA, or perhaps a large CPLD if they're small FIR >filters.I tend to agree. The data rate requirement significantly beyond that of, say, a GPIO-to-USB converter. Although if you could perform the I/O, a fast general-purpose computer might be able to perform the required DSP. Are there any USB 3.0 / GPIO converters yet? Any host adapters? Steve
Reply by ●January 18, 20092009-01-18
On Jan 17, 11:30�pm, Jerry Avins <j...@ieee.org> wrote:> buleg...@columbus.rr.com wrote: > > I have outputs from 4 A/D's. �Each output is 8 bits ( // output ) and > > will likely be sampled at 40MHz. > > > I need to manipulate that data in some kind of device. �This includes > > level shifting, > > What do you mean by level shifting bits? > > > � � � � � multiplying the data against each other and possibly > > applying some FIR filtering to the data. The output will be three 8 > > bit data streams running at the 40 MHz rate. Can anyone suggest a part > > to consider using for an application such as this? > > > Thanks. > > > Brent > > -- > Engineering is the art of making what you want from things you can get. > �����������������������������������������������������������������������I think I should have said scaling operations.
Reply by ●January 18, 20092009-01-18
On Jan 17, 6:28�pm, buleg...@columbus.rr.com wrote:> I have outputs from 4 A/D's. �Each output is 8 bits ( // output ) and > will likely be sampled at 40MHz. > > I need to manipulate that data in some kind of device. �This includes > level shifting, �multiplying the data against each other and possibly > applying some FIR filtering to the data. The output will be three 8 > bit data streams running at the 40 MHz rate. Can anyone suggest a part > to consider using for an application such as this? > > Thanks. > > BrentThanks everyone for your replies! Here is some more info. If anyone would like to add thoughts I would appreciate (If you don't want to that is ok too :-) It has been proposed to use a 12 bit A/D convertor to process signals at IF at 30 MHz. Our system needs to process pulse signals (dynamically) from -10 dBm to -78 dBm. However, the signals at -78 dBm typically have S/N raio > 15 dB. In other words, our composite noise power is about -95 dBm (approx 10 MHz BW + sys NF). Will a 12 bit A/D work in this application, when the A/D is likely not to really see all the way down to the noise level? We have historically used a log amp. Brent
Reply by ●January 18, 20092009-01-18
<bulegoge@columbus.rr.com> wrote:>Here is some more info. If anyone would like to add thoughts I would >appreciate (If you don't want to that is ok too :-)>It has been proposed to use a 12 bit A/D convertor to process signals >at IF at 30 MHz.>Our system needs to process pulse signals (dynamically) from -10 dBm >to -78 dBm. However, the signals at -78 dBm typically have S/N raio > >15 dB. In other words, our composite noise power is about -95 dBm >(approx 10 MHz BW + sys NF).>Will a 12 bit A/D work in this application, when the A/D is likely not >to really see all the way down to the noise level?It might not. It is on the edge, and whether it will work (that is to say, whether that A/D's quantizing noise will not start to dominate) will depend upon the peak/average ratio of the signal, and how ideal of a 12-bit converter it is. Is the strongest signal -10 dBm average, with higher peaks, or is -10 dBm the highest peak? An ideal 12-bit converter which can handle a -10 dBm (RMS) sinusoid signal without clipping will have an equivlanet noise of -96 dBm, and a real 12-bit converter will have at least 2 dB more than that.> We have historically used a log amp.That's not a bad approach if it preserves the desired character of the signal. Or use some form or AGC, if feasible. Or just use a 14- or 16-bit converter. I would say you want a so-called ENOB (effective number of bits) of about 13. Steve
Reply by ●January 19, 20092009-01-19
On Sun, 18 Jan 2009 14:58:31 -0800, bulegoge wrote:> On Jan 17, 6:28 pm, buleg...@columbus.rr.com wrote: >> I have outputs from 4 A/D's. Each output is 8 bits ( // output ) and >> will likely be sampled at 40MHz. >> >> I need to manipulate that data in some kind of device. This includes >> level shifting, multiplying the data against each other and possibly >> applying some FIR filtering to the data. The output will be three 8 bit >> data streams running at the 40 MHz rate. Can anyone suggest a part to >> consider using for an application such as this? >> >> Thanks. >> >> Brent > > Thanks everyone for your replies! > > Here is some more info. If anyone would like to add thoughts I would > appreciate (If you don't want to that is ok too :-) > > It has been proposed to use a 12 bit A/D convertor to process signals at > IF at 30 MHz. > > Our system needs to process pulse signals (dynamically) from -10 dBm to > -78 dBm. However, the signals at -78 dBm typically have S/N raio > 15 > dB. In other words, our composite noise power is about -95 dBm (approx > 10 MHz BW + sys NF). > > Will a 12 bit A/D work in this application, when the A/D is likely not > to really see all the way down to the noise level? We have historically > used a log amp. > > BrentUnless you can oversample and average, your ADC needs enough resolution to see the smallest signal and enough range to not get overwhelmed by the largest. As suggested elsewhere, why not use a 14-bit or 16-bit ADC? They're out there, at higher data rates than 30MHz. Or, if you're bound and determined to use a less than stellar ADC, why not your log amp followed by the ADC? -- Tim Wescott Wescott Design Services http://www.wescottdesign.com
Reply by ●January 19, 20092009-01-19
On Jan 19, 3:33�am, Tim Wescott <t...@seemywebsite.com> wrote:> On Sun, 18 Jan 2009 14:58:31 -0800, bulegoge wrote: > > On Jan 17, 6:28�pm, buleg...@columbus.rr.com wrote: > >> I have outputs from 4 A/D's. �Each output is 8 bits ( // output ) and > >> will likely be sampled at 40MHz. > > >> I need to manipulate that data in some kind of device. �This includes > >> level shifting, �multiplying the data against each other and possibly > >> applying some FIR filtering to the data. The output will be three 8 bit > >> data streams running at the 40 MHz rate. Can anyone suggest a part to > >> consider using for an application such as this? > > >> Thanks. > > >> Brent > > > Thanks everyone for your replies! > > > Here is some more info. �If anyone would like to add thoughts I would > > appreciate (If you don't want to that is ok too :-) > > > It has been proposed to use a 12 bit A/D convertor to process signals at > > IF at 30 MHz. > > > Our system needs to process pulse signals (dynamically) from -10 dBm to > > -78 dBm. �However, the signals at -78 dBm typically have S/N raio > 15 > > dB. �In other words, our composite noise power is about -95 dBm (approx > > 10 MHz BW + sys NF). > > > Will a 12 bit A/D work in this application, when the A/D is likely not > > to really see all the way down to the noise level? We have historically > > used a log amp. > > > Brent > > Unless you can oversample and average, your ADC needs enough resolution > to see the smallest signal and enough range to not get overwhelmed by the > largest. > > As suggested elsewhere, why not use a 14-bit or 16-bit ADC? �They're out > there, at higher data rates than 30MHz. �Or, if you're bound and > determined to use a less than stellar ADC, why not your log amp followed > by the ADC? > > -- > Tim Wescott > Wescott Design Serviceshttp://www.wescottdesign.com- Hide quoted text - > > - Show quoted text -Thanks (both Steve and Tim) I am trying to make the design tradeoff. The log amp is an amazing device (AD8309). We are also processing the compressed IF outputs to make a phase comparison on two channels. I am starting to think I would be better off keeping the log amp and using 8 bit A/D's, rather than trying to bypass the logamps and use 12 or 14 bit A/D's. We currently do the phase processing in analog and then digitze all the signals after they have been brought to video via analog means.
