At 11/27/2009 15:27, you wrote:
> > Great! Too bad the manufacturer didn't do this; > > would've saved you & others the trouble of having > > to characterize a brand new component. > >I actually ended up verifying what another person on >a similar path did before me. > > > I didn't miss that at all. You seem to be missing > > my point that the leading, small resonators in front > > of the GLB will either degrade the NF far more than > > a 1/4 wave coaxial resonator, or offer far less > > out-of-band rejection. > >We were talking about two different directions. Yes to >all the above if the topic is pre-amplifiers and external >cavities. > >A GLB Pre-Selector Pre-Amplifier at a high RF Site >parked bare naked between a duplexer and a receiver can >out perform some bare preamplifiers. Of course - no argument there. But as you mention above, I'm not making that comparison. > If you have the >luxury of the extra typical High Q band-pass cavities >then the NF will be better... but again the 3rd order >performance might not be and what happens after the >active device plays a much more important part of the >realized high signal level performance. ...if you're considering overall system performance, which is largely a function of the RX you're using. If you're using one of the "bulletproof" older commercial RXs, the preamp's dynamic range becomes much more important that the RX. > > Only if the GaAsFET preamp is maldesigned. Some > > will break into oscillation at different source/load > > complex impedances. > >The key is the manufactures data sheet for the device >and the completed pre-amplifiers real world measured >3rd order performance. This reply appears to be in the context of one designing their own preamp. I was thinking in terms of commercially available preamps. > > I don't worry about my RX's dynamic range - I use GEs :) > > But if you're not, more pass cavities, or even your GLB > > preselector, after the preamp are an easy fix, since > > loss is less of an issue there. > >Receiver/antenna system design and construction can be >just this side of voodoo magic. I think this is where/why we seem to be missing each other's points. To me it's not magic at all: it's simple, straightforward engineering. If you have a ballpark antenna noise temperature, RX noise temperature (deduced from 12 dB SINAD), knowledge of nearby (in freq. & location) potential "problem" radiators, you can calculate the necessary hardware (gain, required noise figure & filtering) to get the most noise-free signal detected. I suppose the more unknowns you have in the above equation, the more it becomes "voodoo magic". > > Well, again we're talking (OK, typing) but simply > > not communicating. P1dB & 3rd order intercept are > > closely related, as they are both measures of a > > widget's dynamic range/linearity, & I use the terms > > somewhat interchangeably for the purposes of this > > discussion. > >Closely related but not necessarily the same. Where >they are different in high level operation can be >and sometimes is a big factor in the system performance. Irrelevant; we're really getting off track on this one. Neither P1dB nor 3rd order intercept of an amplifier is affected by filtering placed after it. > > Again, the ONLY filtering that will improve an > > amplifier's resistance to IMD is filtering on its > > input, not its output. The tuned stages that are > > after the GaAsFET in the GLB serve only to protect > > the following device (RX or another preamp) from > > overload by out-of-band signals. > >Nope, the trailing stages do contribute to the >GLB IMD Performance. The out of band issue is an >additional side benefit. Incorrect. If a large signal is presented to the active device, how do tuned stages on the OUTPUT of the active device protect it from overload? They can't. > > In the case of a good commercial RX like our GE > > Mastr IIs & Motorola Micors, this is almost > > always unnecessary, since they already have integral > > high Q (& lossy!) helical resonators. > >The key issue is the band-width of the above radio >front end circuits, which are fairly wide in the >real world. The measured 3 dB bandwidth of a G.E. MVP front-end helical resonator assembly is 1.8 MHz. I don't consider that "fairly wide". > > FWIW the BF981 is a dual-gate MOSFET, not a GaAsFET. > >Yep, but the GLB trailing tuned circuits would also >improve the realized GaAs Fet device equipped 3rd >order performance... as would properly set cavities >between a pre-amplifiers output and the receivers >input. No, they don't. They only improve the 3rd order performance of whatever is downstream. >I suspect the point being missed here... is when a >preamplifier is generating spectral buckshot in a >very toxic condition... the filtering after >the active device improves/reduces what the receiver >sees/has to deal with. If the preamp is generating "spectral buckshot", you're already screwed. The typical symptom here is on-channel crap being generated by the preamp, which of course cannot be filtered out. > Mucho better to have high >Q cavities if you can but the GLB trailing tuned >circuits are better than the nadda of a bare preamp >dumping a wider mix of poop direct into a receiver >front end. No, the correct solution is to not allow the nonlinearity in the first place. Letting the preamp generate "poop" & then filtering the off-channel garbage you've already generated in that preamp is a poor solution; the in-band garbage generated in the preamp goes right on through. >No one wants an overloaded preamplifier condition >but there are interesting ways to manage and improve >the situation. The GLB preamplifier design is one >such layout that receives little credit for doing >the mentioned. Well, IMO it gets all the credit it's due. No published NF + no published P1dB or 3rd order intercept = suspect product. Let's move on - I think this thread has run its course. Bob NO6B
