I am using NANOSTAR BRUKER for few years, and the sealed tube is out of order
now and the intensity dropped significantly even after rigorous alignment of
beam. However, there might have a better beam source using microfocus beam that
deliver better flux. Try to think about this. The variable temp sample stage is
also advantageous and useful for other non-protein samples if needed.

We don't have an in-house SAXS system (yet) but I spent quite some time looking at different commercial systems (Rigaku, Bruker, Anton-Paar, Hecus).
            The two systems I liked the most are from Anton Paar and from Hecus, both form Graz, Austria. They both build a Kratky camera, which has the beam focused on the detector and allows for a much smaller footprint (1-1.5 mtrs sample to detector) compared to a system with a collimated beam (as used at synchrotrons and with the in-house system from Bruker). Another advantage of the focused beam is that it doesn't cut off most of the beam so you get more photons on the sample, giving shorter exposure times.
            I sent a protein sample to both companies and also collected SAXS data on the synchrotron for comparison. The data form the in-house systems is very similar to synchrotron data, albeit somewhat more noisy. Exposure time was very decent: 20-30 minutes for a 80 kDa protein at 10 mg/ml using a micro-source. With a modern rotating anode I expect one could get even shorter exposure times.
            As for detectors, Pilatus or a gas detector gave good data.
CCD camera will give more noisy data and an image plate would take too long to scan in my view.

We've tested other instruments, including the previous system from Rigaku, and none of them came close to synchrotron data. The original Rigaku pinhole system was a pain to align and had a kludgy sample system. The AntonParr system did not give good data at all but we've since heard from an Australian group that the US reps are not strong on biological samples and that may be a fluke. We've not tested the Bruker system. Until the BioSAXS-1000 we'd been very negative about the possibility of a lab based system. However, the Kratky geometry (with the flux advantage it provides) coupled to a microfocus system amazed us.  We're frequent synchrotron users for SAXS but see the ability to characterize samples beforehand as useful. We're doing a lot of work and in this case a significant proportion of our data could be collected at home. What is more useful is the ability to characterize solution or sample conditions for the more difficult targets before spending time at the synchrotron collecting data to prepare for the ideal experiment. I'm guessing this is your interest. We're after a home system for this purpose.

Here’s my two cents on our Home SAXS experience at Penn so far.  Feel free to post in entirety:

            Two years ago our department installed a Rigaku PSAXS small-angle x-ray scattering system installed on a port of an existing (and relatively old) RU-H3R rotating anode generator. Currently our setup includes Osmic mirror optics, a 3-pinhole enclosed pre-flight path, an evacuated sample chamber with cryostated sample holder, adjustable sample-to-detector distances, and gas-filled multi-wire detector. Generally, we leave it in one configuration that provides an accessible scattering angle of 0.006 < Q < ~0.25, sufficient for the characterization of most any macromolecule.  (where Qmin*Dmax should be less than or equal to Pi to perform the inverse Fourier transform)

            Our group does a lot of synchrotron SAXS and neutron scattering, as do several groups in our department.  Compared to data from these sources, the data quality from the home source that can be obtained for a well-behaved and monodisperse sample can be very good, on par with the quality of data obtained with neutron scattering (ie publication quality).  This type of instrument is ideal for binding experiments using I(0) or basic characterization of a sample for structural examination. (I haven’t had great luck yet doing I(0) binding experiments at synchrotrons, but I haven’t given up just yet.)

            We’ve done measurements on a variety of systems ranging in size from 660 kD down to 9 kD – as long as you can find the right concentrations to work at to get good signal-to-noise, and your sample is well-behaved for the time course of the experiment, you’re in good shape (pun unintended).  We generally start out at 3-6mg/mL for things between 45-100kD and take it from there. Generally, much higher sample concentrations are needed for small macromolecules with this type of generator.       

            The sample holder provided by the vendor was a capillary system (~100 uL per sample).  For us, this didn’t work out for a general user scheme and for small sample volumes and wasn’t cost effective, so we ended up custom-making our own design that took advantage of existing technologies.  I don’t know what vendors currently provide with their systems, but this is a buyers point that should be examined very closely, from our experience. Ask a lot of questions. In this regard, the current synchrotron experimental sample setups available are pretty far ahead of the curve.

            Another buyers point to consider carefully is the service contract. It’s not like your annual centrifuge bill. We’ve had a few hiccups with our detector, so the service contract has proven to be worthwhile.  

The software provided for data reduction (SAXSGui) works alright. I’m hoping that this software will continue to improve to mirror (or better yet) use a program like BioXTSAS RAW, where you can go from data reduction immediately into a Guinier fit, typical manipulations of the scattering profile, and the inverse Fourier transform in one program.

            The obvious downside of the home source measurement is exposure time and lack of high Q.  With an older generator like ours, you can be looking at single exposure times of 3-6 hours at a given concentration.  We’re pretty excited to be upgrading to a MicroMax-007 HF microfocus rotating anode and Osmic VariMax mirror optics this month.  From what I understand from a colleague at another institution with a Rigaku system, this is going to have a very big impact on our exposure times and the sample concentrations we can work at for smaller particles.  I thought I heard somewhere that Rigaku was working on a two camera system (?), so I suspect that accessible Q will increase with newer generations of instruments.

            I can definitely get much more done in terms of sample queue during a 24 hour shift at a good Beamline than I could at home over a month, but having an instrument at home does provide a great opportunity to characterize samples immediately as they become available, with good precision. It has become a regular part of our workflow for many projects. Some types of samples are simply not compatible with synchrotron sources because of radiation damage (ie: disulfide rich proteins), and some types of samples don’t have the shelf life to wait for the next trip, so a home source is an ideal alternative and complement.