Erm, I wouldn't wish to discourage you, but you might find that running ten ADCs in parallel comes with a few issues.
Firstly, unless they are exceptionally well matched, they will all give different readings for the same input voltage - although with care, you might be able to calibrate that out.
Secondly, you might find that distributing the same signal to ten ADCs is more complex than it sounds, particularly if it's got some high frequency components. Aim to drive them from a good, high slew-rate op amp with a low impedance output. Even so, each input is likely to have some non-trivial input capacitance (all of which will add in parallel), which will interact with the output impedance of your op amp to give you phase shifts - and these can start to bite at only about one tenth of the 3dB point (1/ 2.PI.r.c Hz).
Remember of course, that you need to put a decent anti-aliasing filter upstream of your ADCs to prevent higher frequency components appearing as unwanted artefacts in your digitised baseband (check out Nyquist's sampling theory). You want to filter out components that are higher frequency than half the sampling rate - but bear in mind that the sharper the filter you use, the sooner phase distortion will creep in).
Remember that even with several converters running in parallel (each with, I presume, its own phase of the sampling clock), you won't actually get a faster sample rate on each converter. What you'll get is interleaved samples, so each ADC will need its own independant anti-aliasing filter... each of which will, I'm afraid, need to be super well matched, otherwise you will trouble re-constituting the signal.
Sorry if it sounds like I'm throwing cold water on your idea, and by all means give it a try, but sampling fast signals is no easy job I'm afraid (speaking as an electronic engineering graduate).
You might find that getting one of your faster converters working is a far easier approach...
oscilloscope/bus sniffer
-
mjcross
- Member++
- Posts: 16
- Joined: Wed Jun 29, 2011 5:39 pm
-
mjcross
- Member++
- Posts: 16
- Joined: Wed Jun 29, 2011 5:39 pm
Erm, I wouldn't wish to discourage you, but you might find that running ten ADCs in parallel comes with a few issues.
Firstly, unless they are exceptionally well matched, they will all give different readings for the same input voltage - although with care, you might be able to calibrate that out.
Secondly, you might find that distributing the same signal to ten ADCs is more complex than it sounds, particularly if it's got some high frequency components. Aim to drive them from a good, high slew-rate op amp with a low impedance output. Even so, each input is likely to have some non-trivial input capacitance (all of which will add in parallel), which will interact with the output impedance of your op amp to give you phase shifts - and these can start to bite at only about one tenth of the 3dB point (1/ 2.PI.r.c Hz).
Remember of course, that you need to put a decent anti-aliasing filter upstream of your ADCs to prevent higher frequency components appearing as unwanted artefacts in your digitised baseband (check out Nyquist's sampling theory). You want to filter out components that are higher frequency than half the sampling rate - but bear in mind that the sharper the filter you use, the sooner phase distortion will creep in).
Remember that even with several converters running in parallel (each with, I presume, its own phase of the sampling clock), you won't actually get a faster sample rate on each converter. What you'll get is interleaved samples, so each ADC will need its own independant anti-aliasing filter... each of which will, I'm afraid, need to be super well matched, otherwise you will trouble re-constituting the signal.
Sorry if it sounds like I'm throwing cold water on your idea, and by all means give it a try, but sampling fast signals is no easy job I'm afraid (speaking as an electronic engineering graduate).
You might find that getting one of your faster converters working is a far easier approach...
Firstly, unless they are exceptionally well matched, they will all give different readings for the same input voltage - although with care, you might be able to calibrate that out.
Secondly, you might find that distributing the same signal to ten ADCs is more complex than it sounds, particularly if it's got some high frequency components. Aim to drive them from a good, high slew-rate op amp with a low impedance output. Even so, each input is likely to have some non-trivial input capacitance (all of which will add in parallel), which will interact with the output impedance of your op amp to give you phase shifts - and these can start to bite at only about one tenth of the 3dB point (1/ 2.PI.r.c Hz).
Remember of course, that you need to put a decent anti-aliasing filter upstream of your ADCs to prevent higher frequency components appearing as unwanted artefacts in your digitised baseband (check out Nyquist's sampling theory). You want to filter out components that are higher frequency than half the sampling rate - but bear in mind that the sharper the filter you use, the sooner phase distortion will creep in).
Remember that even with several converters running in parallel (each with, I presume, its own phase of the sampling clock), you won't actually get a faster sample rate on each converter. What you'll get is interleaved samples, so each ADC will need its own independant anti-aliasing filter... each of which will, I'm afraid, need to be super well matched, otherwise you will trouble re-constituting the signal.
Sorry if it sounds like I'm throwing cold water on your idea, and by all means give it a try, but sampling fast signals is no easy job I'm afraid (speaking as an electronic engineering graduate).
You might find that getting one of your faster converters working is a far easier approach...
-
mjcross
- Member++
- Posts: 16
- Joined: Wed Jun 29, 2011 5:39 pm
(duplicate comment deleted)
-
aortiz557
- Member++
- Posts: 27
- Joined: Fri Jan 06, 2012 3:55 pm
i do not take any offense, this kind of criticism gives me new search terms to look up. i've accumulated a box of sample adc's over the past few months, i do suppose i should just be less lazy and use that instead. i don't think i will exceed a bandwidth of 20mhz with this project, since that is the limit of my analog scope and without a better scope i can't really verify any accuracy. i shall do some more research and see if i cant get those other ADCs working.
-
mjcross
- Member++
- Posts: 16
- Joined: Wed Jun 29, 2011 5:39 pm
No criticism intended :-) I think that what you're doing - trying stuff in a spirit of adventure and learning - is a great approach. Stick with it and you will get there...
If you want to get your head around analogue to digital sampling, the following tutorial might be helpful
If you want to get your head around analogue to digital sampling, the following tutorial might be helpful
-
aortiz557
- Member++
- Posts: 27
- Joined: Fri Jan 06, 2012 3:55 pm
its been too long guys, i've fried my xc-1a board a long time ago, but now i have a new job (mechanic) and i shall save up some money to get a new board sometime soon. before putting it away i finished up some pcbs for a tlc5510ains, a pga117, and some dacs connected to some op-amps, alltogether this makes an adc, level shifters, and scaling all combined. expect progress within the month