Guiding is always a critical subject for obtaining great astronomical images. To improve the star shapes and punctiformity, a high quality camera had to be used. This test will examine two of the best options on the market (in spring 2011), both manufactured by UK based Starlight Xpress.
By the way, I'd like to remark a false myth along amateurs: "for guiding also a bad camera, even color, will suffice". This is false, for many reasons: the main one is that for high accuracy guiding the SNR of the guide star is fundamental; moreover a more accurate camera can be used with up to 5 times lower image scale, allowing to get a far more easy search of the guide star.

Lodestar
Lodestar guider is my actual guiding camera from late 2010. I'm very happy with it: this is a big step forward respect to my previous guiders: the old SBIG ST4 and a Philips Vesta Pro modified with a BW chip. The image quality is outstanding and I've successfully used not only for guiding. Maybe this is not a perfect camera (see some remarks here), but it comes very near. One of the first noted characteristics is simplicity: simple to mount (only 70 g weight, 31.8 mm external diameter), simple to connect to PC and power (single USB2 cable), simple to connect to the telescope (single ST4-compatible cable), simple to connect to software (all the most used software packages are supported).

SXV-H5
This camera is not strictly speaking a pure guider, as - when it was designed - I suppose it was mainly intended to substitute the successful HX-516, a great  imaging camera that in the late '90s produced great deep-sky and planetary works. Respect to Lodestar, the dimensions, mass and cabling are quite larger, making the usage less simple. Moreover, being cooled, the power requirement is more heavy.

Comparison
Characteristics	Lodestar	SXV-H5
Chip	Sony ICX429ALL interline	Sony ICX424AL  progressive
# pix X	752	660
# pix Y	580	494
Total pixels	436160	326040
Relative total pixels	134%	100%
Dim pix X [um]	8.6	7.4
Dim pix Y [um]	8.3	7.4
Relative average dim	114%	100%
Sensor dim X [mm]	6.47	4.88
Sensor dim Y [mm]	4.81	3.66
Area [mm^2]	31.1	17.9
Relative area	174%	100%
Cooling	no	yes
ADC bit	16	16
Download time [s]	<0.2	<0.2
Readout	two independent readouts for even and odd lines	single reading for all lines
Quantum efficiency	See figure below
Weight of head [g]	70	311
Weight of power supply [g]	0	671
Needed cables	1 USB 1 ST4	1 USB 1 ST4 2 power
Current [I]	0.5	0.7
Voltage [V]	5	12
Power [W]	2.5	8.4
Relative power	30%	100%
Price (new 2011) [€]	510	-
Price (used 2011) [€]	?	500
Links to manufacturer	Lodestar	SXV-H5

Test set-up
A simple test setup was used, composed of a typical guiding refractor, mounted on a bigger imaging refractor. Both were mounted on a heavy duty equatorial mount. Here are the instruments characteristics.

• Guiding scope (with the guiding cameras at direct focus): Vixen 80ED, apochromatic refractor, 80 mm diameter, f/8, 640 mm focal length.
• Main scope (not used): Takahashi FS152, apochromatic refractor, 152 mm diameter, f/8, 1216 mm focal length.
• Mount: 10Micron GM2000 QCI, German equatorial mount, 45 kg max load
• Location: Saronno (VA), a very light polluted city in the Po Valley, Italy.
• Temperature: about +20 °C

Test protocol
A series of images of the M82 galaxy were made with both cameras, in a total time of about 30 minutes, when the object was very high in the clear sky, about 70°. The sky conditions were stable and the whole sky was clear. Here are the registered images:

Exposure	Lodestar	SXV-H5
1 s	5 exp	5 exp
10 s	5 exp	5 exp
30 s	5 exp	5 exp
dark 1 s	5 exp	5 exp
dark 10 s	5 exp	5 exp
dark 30 s	5 exp	5 exp

The dark images were median combined, while the light images were average combined.
Image analysis
The analysis of the recorded images was conducted with respect to a few aspects.

• SNR of the guide star: the better the SNR (signal to noise ratio), the better the guiding accuracy. This is very frequently a highly underestimated factor in guiding, but this is the key to a good subpixel accuracy. Cameras with high sensitivity and quality, usually connected to a higher number of bits, produce better results. Both cameras have 16 bits and produced very similar SNR, so that there is virtually no difference in the two, regarding guiding accuracy due to subpixel centroid measurement. See "Reference star SNR" in the table below.
  
• Background noise: to measure this quantity, the standard deviation of the background must be compared relative to the sensitivity, measured by means of the flux intensity of a reference star. The result is similar in the two, being only a little better in Lodestar. Oddly the relative std of Lodestar improves with longer exposures. See "Std relative to flux" in the table below.
  
• FWHM of the guide star: the pixel dimension is only a little different, being the average dimension of the rectangular pixels of Lodestar only 14% larger than the square pixels of SXV-H5. Thus the latter is able, in theory, to get smaller stars (measured in arcseconds). The measurements showed no sign of this but maybe a better focusing of both can enhance the difference. Or, more likely, the seeing was limiting the measurements, to about 3.5", a value quite usual from the observing location. See "Fwhm ["]" in the table below.
  
• Hot pixels: the approximate number of hot pixel was measured in the 1 s dark frame. This is the most significant  exposure, as a guider image frequently is used without dark calibration, and so a hot pixel can produce a spurious signal that can ruin the guiding. Both cameras were found to produce very few hot pixels (Lodestar ~30, SXV-H5 ~2), and the SXV-H5 is nearly perfect, being cooled. The brightness of the hottest pixel was found in the Lodestar to be 273 times the standard deviation of the background and 26 times in the SXV-H5. See "Number of hot pix (approx)" in the dark data table below.
  

The above analysis was derived from the measurements and computations shown in the below tables.

M81 average	5 x 1 s		5 x 10 s		5 x 30 s
Camera	Lodestar	SXV-H5	Lodestar	SXV-H5	Lodestar	SXV-H5
Reference star flux	5020	10224	100255	103371	258849	274615
Flux relative to SXV-H5	49%	100%	97%	100%	94%	100%
Reference star SNR	70.6	64.35	562	499	677	762
Std of background	6.48	14.52	12.6	19.4	20.6	29.8
Std relative to flux	91%	100%	67%	100%	73%	100%
Fwhm [pix]	1.208	1.289	1.199	1.319	1.724	1.63
Fwhm ["]	3.29	3.51	3.26	3.59	4.69	4.43

And from these dark data.

Dark median 5 x 1 s	Lodestar	SXV-H5
Std of background	8.7	13.9
Ave	1894	1478
Max	4266	1846
Max-Ave	2372	368
Max-Ave normalized	273	26
Ave+20std	2068.0	1756.0
Ave+10std	1981.0	1617.0
Number of hot pix (approx)	30	2

Conclusions
Both cameras was found to have similar (great) sensitivities and characteristics. The Lodestar has many advantages such as small size, small weight, small power consumption, few cables, larger sensor area. On the other hand SXV-H5 has fewer hot pixels thanks to cooling and a little better resolution, yielding to a theoretical little better guiding accuracy, when the same focal length is used. From an image quality point of view, the cameras finished in a draw: the SNR of the stars, the most important quantity for subpixel measurement, was found to be nearly the same. The test results can be summarized in the table below.

Test results
Image quality	tie
Sensitivity	tie
SNR of guide star	tie
Number of pixels	Little better Lodestar
Hot pixels	Little better SXV-H5
Pixel dimension (smaller is better)	Little better SXV-H5
Dimensions	Much lighter Lodestar
Cable complexity	Much better Lodestar
Power consumption	Much better Lodestar
Sensor area	Much better Lodestar

Acknowledgments: I'd like to thank my friend Giosuè for testing together; Mauro and Giancarlo for added tips.