From: Joerg H. <in...@jh...> - 2006-06-26 01:42:32
|
Hi, so...@pk..., 25.06.2006 23:51 : > I have breadboarded an EEG system similar to ModularEEG, but with a > switched cap filter replacing the op-amp "Besselworth". Since my filter > has a gain of 1, if I don't want to add another stage, I need to put the > extra gain in the inst-amp and op-amp gain stage. Using round numbers, if > I want an adjustable gain of 1000-20000, I could use: > > instrumentation amp 10, op-amp 100-2000 This one should be preferred. > > or > > instrumentation amp 100, op-amp 10-200 > > Is there a suggestion of one distribution over the other (or somewhere in > between)? From the INA114 datasheet, I don't see a spec problem going with > a gain of 100. The CMRR is better. The slew rate is longer, but not a > problem at EEG frequencies. The biggest problem with using a higher gain > at the inst-amp that I can figure is that it would take a smaller dc > offset on the input to make the ouput go to the rails. Yes. This could make the input stage quite useless. The brainmaster BM2E has a tricky method to have high gain in the input stage and block DC before it can saturate the amplifier: Instead of a simple gain setting resistor a R-C series combination is used that lets the instrumentation amplifier input side act as a highpass. >>From the INA114 datasheet, I see an alternate way to ac couple the output: > drive the reference input with an integrator fed by the output. Has anyone > experience with configuring an instrumentation amp like this? Yes. See MonolithEEG http://openeeg.sourceforge.net/doc/hw/#othersites and my analysis of reference integrator performance http://www.jhansmann.de/eeg/simulations/simulations.html INA118 with active highpass (Reference Integrator) In short: The reference integrator performance is often over estimated, because it has only effect on the output amplifier. It can not avoid saturation of the input stages. > Would it be > helpfull in reducing problems with dc offsets on the inputs? No. > Right now I > have the op-amp stage ac coupled as with ModularEEG (1uF,1M) so that all > dc problems are pretty much confined to the inst-amp stage. Yes. Regards, Joerg |