Winny_Puuh
Junior Member level 1
Hi guys,
I'm trying to design a neural recording amplifier with some crazy specs (in my opinion at least and unfortunately I'm pretty unexperienced). The task is to amplify amplitudes as low as 1 µV (up to some mV) while having a bandwidth of 50 kHz. Furthermore the bandwidth, gain and noise should be tunable depending on the application. Oh and the dynamic range, whatever that may be, should be laaaarge (probably has sth to do with the range of my input signals from 1 µV to mV).. I'm using a 0.18 mm technology and here we go.
At the moment I'm totally, really totally at a loss on how to solve this problem. Even after reading papers I'm still a beginner in CMOS design and I could need help from someone who has a clue.
Well, here are my thoughts... for the worst case (1µV amplitude) I think I need like a gain of 120 dB to get 1V at the output. Having a bandwidth of 50 kHz makes the specification for the GBW like 50 MHz. That wont work for a single stage... and still there is nothing tunable about it. So I thought maybe I should use 2 or 3 stages. (Is 2 actually better than 3 for some reason?) First stage has like a fixed gain of 60 dB, the next stage is tunable from 0dB to 60 dB (is that actually possible? This is quite a large range in my opinion). I think I need for both (if we say 2 stages) a GBW of 5 MHz, right? To get my 50 kHz bandwidth.
Then I do not know which topologies or feedback to choose... first stage should have low noise, high gain, second stage large output swing? Then I dont know what feedback actually does to my opamps and whether I could just use a parallel combination of RC in the feedback path to set my cutoff frequencies somehow because - oh yeah - tunable bandwidth is a requirement... as you can see, I'm in a total mess right now. I hope someone out there can help...
I'm trying to design a neural recording amplifier with some crazy specs (in my opinion at least and unfortunately I'm pretty unexperienced). The task is to amplify amplitudes as low as 1 µV (up to some mV) while having a bandwidth of 50 kHz. Furthermore the bandwidth, gain and noise should be tunable depending on the application. Oh and the dynamic range, whatever that may be, should be laaaarge (probably has sth to do with the range of my input signals from 1 µV to mV).. I'm using a 0.18 mm technology and here we go.
At the moment I'm totally, really totally at a loss on how to solve this problem. Even after reading papers I'm still a beginner in CMOS design and I could need help from someone who has a clue.
Well, here are my thoughts... for the worst case (1µV amplitude) I think I need like a gain of 120 dB to get 1V at the output. Having a bandwidth of 50 kHz makes the specification for the GBW like 50 MHz. That wont work for a single stage... and still there is nothing tunable about it. So I thought maybe I should use 2 or 3 stages. (Is 2 actually better than 3 for some reason?) First stage has like a fixed gain of 60 dB, the next stage is tunable from 0dB to 60 dB (is that actually possible? This is quite a large range in my opinion). I think I need for both (if we say 2 stages) a GBW of 5 MHz, right? To get my 50 kHz bandwidth.
Then I do not know which topologies or feedback to choose... first stage should have low noise, high gain, second stage large output swing? Then I dont know what feedback actually does to my opamps and whether I could just use a parallel combination of RC in the feedback path to set my cutoff frequencies somehow because - oh yeah - tunable bandwidth is a requirement... as you can see, I'm in a total mess right now. I hope someone out there can help...
