ADP3250 Input Bandwidth

[cagri]

I am trying to understand the Analog bandwidth: 55+ MHz @ 3 dB  of the ADP3250.

I want to capture a 50 MHz frequency and I am wondering if that is possible with ADP3250.

What does the Analog bandwidth: 55+ MHz @ 3 dB mean?

 

[zygot]

Someone from Digilent needs to explain how they arrived at that specification to answer to your question but there's an additional consideration. It depends on what you want to measure with a digitized signal captured by Fs = 2X the signal of interest. -3 dB represents 1/2 the energy in a signal and is a standard representation of analog bandwidth, especially as it relates to ADC. Often meeting the minimal Shannon requirement for sampling is OK, but not for every purpose.

[JColvin]

Hi @cagri,

My understanding is that the 55 MHz bandwidth specification should be at the -3 dB, at least as per the specifications document: https://digilent.com/reference/test-and-measurement/analog-discovery-pro-3x50/specifications. Listing it at (+)3 dB does not make sense to me, so I'll make sure to request to get that store page detail updated. The plus portion of the 55+ MHz analog bandwidth (which is achievable with appropriate probes, as noted here: https://digilent.com/reference/test-and-measurement/analog-discovery-pro-3x50/specifications#refnotes:1:note1) is to indicate that this is a conservative value being quoted since actual analog bandwidth will depend on tolerances, the selected gain range, and other details, so a 'minimum' bandwidth value was provided with the potential of users seeing better results.

Zygot is correct of course that you will sometimes want to have more than 2x the bandwidth to properly capture the signal of interest or that you would rather be interested in the 0 dB bandwidth instead of the -3 dB value where the signal is attenuated by ~30%. I do not know if the "cutoff point" for 0 dB has been specifically characterized for the ADP3250/ADP3450 though.

Thanks,
JColvin

[cagri]

thank you both for the response!

On the other hand, the sampling rate can be set to 125 MS/s maximum as well as 0.5 GS/s oversampling, what is the downside with these two options?

Do you know if there Is a setting option where I can set the input range ( for instance 2 Vpp) so that I can use the full 14-bit resolution for a 2 Vpp input signal?

Which ADC IC is the Analog Discovery Pro using? 

 

[zygot]

Perhaps you should ask how Digilent achieves and defines 0.5 GS/s oversampling; and even more importantly, how they measure and validate this, and what effect it has on ENOB and noise. This is concept that has been around for low cost scopes and digital analyzers and has been much abused. Few companies using it actually provide enough information to allow prospective users the opportunity to evaluate the claims.

[attila]

Hi @cagri

See the application Help. 200mV/div = 2Vpk2pk

[zygot]

When you advertise a specification the words that you use are important in order for a prospective customer to understand what it is that they are buying.

Historically, companies presenting products that do funky things to increase Fs specifications would use terms like 'equivalent time sampling' to indicate that something different is happening with their product in a particular mode. It would be entirely reasonable for a customer, or potential customer, to assume that 0.5 GS/s oversampling refers to an actual Fs at the sampler. It certainly wouldn't occur to the not-so-sophisticated customer to even ask what that means. If the specification referred to 0.5 GS/s Equivalent Time Sampling, then at least this might prompt some questions. When something is sold as an instrument it's entirely reasonable for customers to draw inferences to specifications for high priced instruments of a similar function. In the context of these instruments 'oversampling' has no real meaning but is likely to be misinterpreted.

As it turns out, even 'equivalent time sampling' isn't very useful as there are a lot of ways to do that, not all of them particularly useful. So, more information needs to be provided.

Let's say that a company offers two oscilloscope products. They both use the same sampler, but different speed grades, and the same sampler analog front end. I contend that the performance of the product using 1 sampler at Fs=500 MHz is not equivalent to a different product using 4 samplers at Fs=125 MHZ. Moreover, how the 4 sampler product implements it's clock management matters. If control IC ( let's says that it's an FPGA ) outputs 4 separate sample clocks to each of the samplers, that is quite different than if it outputs 1 sample clock to an external programmable clock generator device with phase control over individual clock outputs, each going to separate samplers. This scenario is part of the analysis, but still indicates that more information is needed in order to avoid misleading advertising claims.

The top commercial instrument vendors provide detailed performance specifications and in general have good application notes for customers wanting to interpret those specs. Their performance is usually generally commensurate with the cost. ( at least there was a time when this was true ). Part of the high cost is not only performance verification but meeting a lot of EMC and other requirements. Seasoned engineers generally have some level of expertise to make a reasonable purchase choice given the proper amount of information. Other people aren't as fortunate. The low end instrument market is a black/grey area where guaranteed specs are traded for a lower cost. This isn't so obvious to everyone. If you are going to market inexpensive products to an unsophisticated public then perhaps you can dispense with detailed guaranteed performance specifications, but then, I believe, that you need to be more careful about causing confusion, intentional or otherwise. Edited June 9, 2022 by zygot

[JColvin]

In the interest of clarity, the page that Attila linked has a bullet point that indicates the oversampling can combine multiple repetitive captures and that the captures are shifted in four phases. This is explained a bit further in the oversampling guide available on the ADP3450/ADP3250 Resource Center (https://digilent.com/reference/test-and-measurement/analog-discovery-pro-3x50/oversampling).

So yes, the 500 MS/s rate available from the oversampling on a repetitive signal from multiple inputs on the same signal is not the same as a flat 500 MS/s input that would be able more accurately capture non-periodic signals.

[zygot]

15 hours ago, JColvin said:

In the interest of clarity

I've read the linked content. In the interest of clarity and and honest advertising Digilent needs to provide more information. It already does this with it's Zmod boards.

The HW that Digilent uses in it's (relatively) high cost instrument product are not doing anything novel or patent worthy so why not just explain how you actually implement what you coyly, and misleadingly, refer to as an oversampling Fs. And fix up the wording on any advertised specifications so that any potential user can understand what they are buying. Considering the cost of these instruments relative to the Zmods ( and It's reasonable to assume that they share the same circuitry ) they would seem to deserve the same, or at least similar, level of documentation  That's all I'm suggesting. 

BTW in the linked documentation the 25 MHz clock graphically rendered example purportedly demonstrating the difference between 125 MHz and "oversampled" only serves to raise questions instead of serving as an explanation.

There's certainly a place for this class of instrument considering the substantial added-value engineering that comes with them. It's that added-value that makes them worth buying... but only if the customer gets what they are expecting after reading the advertisement and available documentation. I can't see the down side to being completely up-front about the details that are need in order to evaluate a potential purchase.