We use cookies to provide our visitors with an optimal site experience. View our privacy notice and cookie notice to learn more about how we use cookies and how to manage your settings. By proceeding on our website you consent to the use of cookies.
I’m looking for some help in narrowing down an ADC search. The spec’s I’m looking for:
100MSps minimum. I’ve found that 105 and 125MSps are the commonly available values, but have some wiggle room on this.
14bits minimum. 16 bit is also an option, but depends on details and availability.
Parallel CMOS output. I might have some wiggle room to implement LVDS, but many of the ADCs I’ve found swap CMOS for LVDS odd/even bit pattern, which doubles the bitrate, and moves it out of reasonable for the FPGA I’m using.
3.3V operation. I might be able to get lower voltage operation working, but the signal coming in will most likely be centered around 1.65V.
Single channel, Diff input. 2-channel is fine, as long as it’s not a major cost-add.
Size, power consumption, and simultaneous sampling aren’t a constraint in this.
Using this, I’ve narrowed the selection down to ~10 part families. However, I’m trying to filter further, but I’m not sure what spec’s are really important vs not. (listed in Alphabetical order)
ADS5500
ADS6145/ADS6144
ADS62P45/ADS62P44
LTC2207/LTC2207-14
LTC2208/LTC2208-14
LTC2217
LTC2255/LTC2254
LTC2285/LTC2284
Also, I found that many of these have an effective number of bits <12, so would it just make sense to get a 12bit ADC and save some money?
What should my next level of part selection filtering be?
As far as further filtering for what you already have A parameter that can narrow further is the operating temp range.
The below link shows what I had found for the items you had referenced in your post.
As for items that are 12bits to start with the below link shows what I had found so far.
And some of these look to have similar pricing to the items you were looking at.
inputs filter(s): Do you mean what filters are going to be used as the last stage before the ADC? The re-centering caps act as a high-pass filter, but I haven’t dug into the filter portion yet.
signal amplitude: There is a transimpedance amp outputting 240mV max, but centered around 3.175V (no idea why that chip chose that value). I’m re-centering it and amplifying it 10x. I have control over both the re-centering position (3.3V/2, 1.5V, etc.) and the amplifier gain.
precision voltage reference: Would be very helpful to have internal, but can work around otherwise.
socket vs PCB: PCB mount, strong preference for SMD.
matched length PCB traces: I’m designing the board, so I will be matching the lengths.
synchronization of multiple channels: Single channel only.
future expansion: Not needed.
lifecycle of the ADC: I’m avoiding anything other than “active” status.
Assuming you want to know the signal’s absolute amplitude, recommend a close examination of the reference in terms of quality and voltage. For example, if you select the “perfect” 1.024 VDC internal / external voltage reference, the input is bound between by 0.512 +/- 0.4 volts (allowing for a bit of headroom to prevent clipping).
Recommend an analog low-pass filter that stays well within the Nyquist sampling criteria to prevent aliasing. Once digitized, we are stuck as there is no meaningful way to separate signal from the alias.
Watch out for added noise in the level-shifting circuitry.
Be sure to visit the individual MFG pages to obtain the latest life cycle status.
It may be useful to cycle this conversation around a few more time. Others may also join in.
As the person picking the part, you get to/have to decide what’s important for your purposes and what isn’t.
Unfortunately, the further distinguishing criteria of your present slate of candidates aren’t surfaced in the DK parametrics, meaning that it’s time to get out the spreadsheet and miner’s lamp to do some datasheet spelunking.
Beyond basic structural considerations, folks doing megahertzy things are often interested in AC performance described in the frequency domain, and looking to specs such as SFDR, INL and DNL as measures of merit.
Again, the burden of choice is yours.
ENoBs characterize a device’s noise and distortion performance relative to an ideal device, all of which are on backorder owing to difficulty sourcing the unobtanium oxides necessary for their production and failure of government-sponsored programs to develop a costslotsium-based substitute.
