>Marty Bigos wrote- > >>I think Howard hit upon the major issues here, but I will take a stab >>at obfuscation. >> >>In general THERE IS NOTHING WRONG WITH MEASURING PEAK HEIGHTS in an >>analog system. There are limitations, and the primary one is dynamic >>range. To measure the peak height one needs a way to detect the peak. >>In analog circuitry, these are known as "peak detect circuits", "peak >>sense and hold" circuits, etc. They do not have a four decade dynamic >>range. So for linear measurements, e.g. DNA, or ratios that vary by >>less than 20-fold or so, these circuits are fine. For >>immunofluoresence, the dynamic range is too great, so range >>compression is needed. This is where the log amps come in. The >>outputs of the log amps can then be measured by peak detection >>circuits, yielding a four decade range at a lower resolution than the >>linear scaling. > >Well, not quite. There is something wrong with measuring peak >heights in an analog system when the beam height is close to or >smaller than the cell diameter, because, under those conditions, the >peak height doesn't give you the answer you're looking for, i.e., >how much fluorescent stuff is there in or on the cell, because the >peak height isn't proportional to the area. You are correct. I was assuming that the optical layout resulted in a beam height larger than the measured object. > > >>In analog circuitry, area measurements suffer similar design >>limitations as peak detection circuits. So if your measurements are >>scalable to a linear domain, analog area measurements of linearly >>amplified signals will work fine. This is very useful, as pointed >>out, for doublet discrimination. If the dynamic range is too great, >>there is no range compression fix; as Howard pointed out the area of >>the log is not meaningful. > >And, as I noted, one can get the log of the area by low pass >filtering the preamp signal, in effect integrating it, and making >the peak height proportional to the area or integral of the original >signal, before putting the signal through the log amp and detecting >the peak of the log signal. > >>I also want to point out that many preamplification systems used in >>flow electronics are bandwith limited, which, in essence, means they >>are doing some integration as well. In general, this has mostly >>helped the accuracy of the measurements. > >It decreases the accuracy of peak measurements, but makes them >closer to area measurements - two wrongs sort of make a right, in >this instance. Almost all amplification systems are bandwith limited, so in that sense are low pass filters. The question is, what is a reasonable limitation for flow electronics. That will depend on the maximum jet velocity and the size of the spot on the jet. For current flow systems, these parameters result in a transit time of an object varying from 1 to 10 usec through the interrogation spot. Thus I believe (and the electrical engineers can correct me here) that a bandwith higher than a few mHz will just allow the statistical photoelectron signals to be seen and won't add (and probably subtract) from overall measurement accuracy. In this sense the preamp gives one a moving average output from the higher frequency input. I am not sure if I would consider that design choice a "wrong". Marty > > >>i hope this doesn't confuse the issues more. > >Me too. > >-Howard
This archive was generated by hypermail 2b29 : Sun Jan 05 2003 - 19:01:08 EST