Photomicrography and macro photography are both popular subjects in the InterWeb,
including among better writers than I.  Consequently, content here will focus (heh) on specific experiences,
rather than more general theory and technique,
which for macro photography see e.g. Allan Walls.
While not aspiring to create aesthetically pleasing images,
facilitating capture of clear and useful phono stylus image wear images seems achievable.
Ray demonstrated that Nikon's long working distance CF objectives are well-suited to this,
while I already owned American Optical (AO) microscopes lacking long working distance objectives, oh well.
Generalities
Microscopes traditionally shared some mechanical conventions:
RMS (Royal Microscope Society) Objective threads
23.2 o.d. eyepiece (ocular) tubes
160mm "tube length"
Consequently, optically-matched objectives and oculars could be swapped among frames.
Practically, after frames evolved from straight tubes to separate heads with prisms,
additional lenses might be introduced, specifically to allow for e.g. vertical illuminators
and/or polarizing filters to be inserted into optical paths.
These optics are typically optimized for some manufacturer's objectives optical formula.
Most modern microscopes use other than RMS threads; here is a list. Here is a list of threads used in astronomy.
Attaching microscope objectives directly to cameras is considered macro photography,
rather than photomicroscopy, and will mostly be addressed elsewhere.
However, attaching a camera body to a microscope body
with nothing but air between objective and sensor
is generally considered photomicroscopy.
This works with so-called finite objectives.
Infinity objectives want a so-called tube lens to make rays finite,
either for oculars or directly focusing on a sensor,
in which case that tube lens may be a telescopic camera lens focused at infinity.
Since some infinity microscopes depend on their tube lens
to correct objective aberrations,
direct projection from microsope's tube lens
to camera's sensor may yield better images.
AO pioneered infinity microscope production in the U.S.;
read here for more history
Practically, for optical magnifications much more than 1x,
technique and patience exceeding mine are wanted for hand-held macrophotography.
For inanimate as well as opaque subjects, hand-holding has few advantages.
Most conventional (diascopic biological) microscopes can be used with external illumination
for some opaque microscopy, but their objectives are optimized for viewing thru glass slide covers,
which becomes increasingly important for higher magnification objectives, e.g. more than 10-20x.
Live View, focus bracketing, Abbe
Using a digital camera with zoomed Live View eases focusing.
Beyond convenience, best viewfinder focus with DSLR mirror down may differ from best focus on the sensor.
Silent shutter (Canon's mode 2) and remote shutter release minimize vibration during exposures.
Replacing halogen illumination with blue LEDs should improve image resolution, with Raleigh's criterion having Abbe diffraction limit depend on illumination wavelength and Numerical Aperture (N.A.).
Practically, blue LEDs not only have shorter wavelengths than do red and green
but excluding those longer wavelengths also usefully reduces chromatic artifacts from older and simpler lenses.
However, blue sensors are relatively sparse in digital cameras;  high resolution is required to compensate.
Canon DSLRs with Live View and silent shutter (e.g. 50D) can be had used for less than $100.
Even better, free Magic Lantern firmware adds
focus stacking for many such cameras.
Eyetube adapters typically have T2 or M42 threads,
requiring a camera-specific adapter.
A projection ocular and extension tubes should cost less than $100;
monocular microscope heads typically cost much less than trinoculars,
while having less light loss than binocular heads,
and used eye tube camera adapter clamps go for around $15.
Obtaining a more or less complete used microscope
with epi illumination and bright/darkfield objectives can be tricky.
Starting from scratch, consider
Nikon 210mm B/D objectives in an Optiphot...
Initial stylus image capture employs a 10x epi microscope objective with darkfield illumination and Olympus NFK 3.3x projection ocular into a Canon EOS sensor.
To hold the stylus at ~45 degrees, the cartridge was mounted in a headshell
with no finger lift and clamped at the bayonet connector using a clothespin:
Although it facilitates sorting spacing between camera sensor and objectives, a bellows is not necessary;
extension tubes suffice, with more length yielding greater magnification.
The resulting image has decent resolution, but poor contrast.
This microscope has short working distance objectives;
the 20x objective would not clear this stylus' plastic grip at this angle.
90 degree shots will clear but require image stacking;
whether objective darkfield illumination will then
highlight stylus wear spots remains to be seen.
To investigate low contrast, Ray Parkhurst suggested
removing the camera and looking down into the tube.
Internal surfaces were not reflecting appreciable light,
but a small bright spot and dim blob are seen in the optic at the bottom of the eye tube.
Lacking ground glass, wax paper was laid over the camera adapter:
The dim blob is revealed as the projected stylus.
The bright spot appears less focused than when viewed directly and,
depending on viewing angle,
more or less distinct from the projected stylus image.
Removing the microscope head and looking down into the arm
reveals a bright spot centered in an epi illuminator lens
in a port off to one side
as well as a less bright spot on the lower lip of that port.
That very bright area below the epi lens is the brightfield mirror.
An A.O. #3002 image erector spaces the head further from the epi illuminator
and slightly improved contrast, despite adding magnifying optics in the path:
Infinity objectives can be used with a camera lens focused @ infinity,
in this case, a Reichert EPIStar 10x
with Canon EF-S 55-250mm STM zoomed to 250mm.
To prevent this zoom from collapsing to 55mm,
the camera body needs support.
Fortunately, focus is internal.
Some observations:
EPIStar darkfield illumination wants alignment
Kenko 1.4x TELEPLUS HD DGX reportedly works with this lens,
but Kenko TELEPLUS PRO 300does not.
About 50% more focal length is wanted to eliminate vignette:
monochromatic illumination or fully corrected objective is wanted.  
1:1 crop of focus near center
CF PL2.5X suspended
with Novoflex bellows
and Canon 6D over tube
to AO trinocular head: 
Microscopes better resolve using blue illumination, since shorter wavelength
and less opportunity for chromatic aberrations.
A 40mm RGB LED halo ring
fits around microscope objectives and can be supported by a 20mm i.d. o-ring.
That o-ring below LED halo blocks some illumination...
Bayer-filtered camera sensors have relatively sparse blue pixels,
aliasing is liable for images with less than 4x oversampling.
Dark wedge along the center of the above blue M55E
stylus image is partly surface not in focus,
but also stylus angle not 45 degrees (twisted cantilever).  
Glued a curled strip of business card
as a half-cylinder to the halo backside...
...an o-ring secures LEDs to the objective:
Rotating the headshell in a clothespin brings
more of the stylus upper surface nearly in focus,
and using green light instead of blue employs more camera pixels.
Contrast could be improved by putting something flat black between the cantilever and its light plastic grip.
AKA longitudinal chromatic aberration correction
Microscope fine focus is indexed in microns, 200 per revolution.
Images of 0.01mm test slide (with cover slip) using MEIJI S.PLAN M 20X
with Series 10 tube lens and Olympus PE 2.5X; no matching corrections.
red 144
green 153
white 148
blue 153
Red focus is out about 9 microns from blue and green..
With an AO head, AO objectives are substantially corrected, better matching a Nikon CF PL relay lens.
Headless, an NFK relay lens approximates AO tube lens corrections.
Finding the stylus with a microscope objective is easier using conventional oculars
than with camera Live View and relay lens.
Since the head has an infinity tube lens, this requires infinity objectives.
Since Series 10 AO Spencers often include a 10X objective, how well does one work on a stylus?
Because they are smaller in diameter than MEIJI and Leitz objectives
for which cardstock half-cylinders were epoxied to LED halos, a shim was cut from clear vinyl tubing.
This particular objective appears to have about 4mm working distance.
With halo illumination, a stylus tip is better illuminated with cartridge vertical centerline
rotated more nearly 60 than 45 degrees from horizontal, reducing surface in focus.
With NFK 3.3X relay lens and about 200mm projection distance,
full field of view with APS-C sensor is usefully in focus;
both of these images (downsampled, but not cropped) were captured after central focusing (zoomed Live View):
 
After visually confirming that minimal detail is lost by downsampling original (5472x3648) images by 3x,
then, given optics are NOT matched for color correction,
compared whether blue (left) or green (right) illumination yields more detail:
 
Approximate IrfanView color corrections settings for recovering dynamic range after grayscale conversion
and before 3x downsampling: brightness ~25, contrast ~99, gamma ~1.65
A set of infinity objectives acquired years ago
also included a LEITZ WETZLAR 160/- EF 10/0.25
with 6.8 mm working distance (Laborlux K-D brochure page 14)
A recently acquired RMS to T2 adapter enables macrophotography experiments.
Having been attached using hot glue,
the RGB LED ring light mounting sleeve failed during an extended session;
illumination temporarily reverted to white LED gooseneck while epoxy cures.
Practically, focusing a 10X objective by bellows is harder than by headless microscope.
 
 
Increased working distance
impacts numerical aperture,
but trials with bellows did not indicate any drastic effects,
with results compromised by difficulty focusing.
Eliminating all optics but objective should increase contrast,
but illumination differences and reduction in magnification confound comparison.
Here is a 1:1 crop of that stylus tip from an APS-C Canon sensor:
 
With camera body attached to a microscope, zoomed live view eases focusing,
but using objectives with bellows mostly demonstrated
limitations of conventional photography focus rails.
Resolution in this setup may also be constrained by sensor pixel density..
50.5mm dovetail to M42 adapter arrived from
RAF Camera.
This secures camera body and extender tubes to A.O. arm without head (and tube lens).
Finite objectives can then be used in A.O. arm with much better positioning and focus control than by bellows.
 
10X cannot fill an APS-C sensor with stylus:
This image was easier than using bellows,
except positioning LED for similar illumination:
Leitz color calibration on Series 10 with 7D
white
blue
green
red
74
65-8
72-5
67-70
Field of view: full frame 20X and 3.3X relay vs APS-C 10X
These are both 480*320 downsamples of full sensor images in the same microscope arm
with similar ring illumination:
Zooming the Leitz image
(right-click, "Open image in new tab", then [Ctrl]&[+] keys)
reveals that only about the center third is nearly focused;
this objective wants to be used with a 3.3X relay lens..
.. or perhaps 2x telextender.
Repurposing available bellows, EF and T2 extension tubes with adapters
yields nearly 140mm parfocal distance for Leitz objective to NFK 3.3X relay lens
and 150mm projection distance to APS-C sensor.
NFK 3.3X is perched at the T2-EF adapter:
This stack is decidedly less than rigid
and really needs electronic shutter and remote release.
Field of view is usefully improved: (downsampled but uncropped)
blue 109
"white" 107
red 97
green 101
NFK 3.3X corrections designed for Olympus objectives cause
chromatic focus differences from Leitz 10X alone.
M55E stylus is better illuminated by a 40mm RGB LED halo on Leitz objective.
Additional 3.3X magnification allows use of blue LED for better optical resolution
followed by about 4x downsampling after converting blue to grayscale and adjusting dynamic range:
The stylus in the above image was rotated less than 45 degrees from horizontal.
This image has the stylus rotated somewhat more than 45 degrees,
illuminated by white (R+G+B)LED halo, no contrast adjustment:
Leitz objective has working distance and nose profile for 45 degree illumination.
Calibration shot with Leitz EF and 3.3X photo relay to APS-C sensor:
Much of the camera's sensor is wasted on out-of-focus objective.
Reducing blue images from 5472x3648 to 800x533
imperceptibly impacts detail:
...while green images noticeably lose detail below 1280x853:
relatively sparse blue pixels in Canon sensor's Bayer matrix
evidently matter more than shorter light wavelengths at this lower magnification.
Introducing a (white) "helper" LED:
Rotating M55E stylus from 45 to 60 degrees improves illumination
...because of twisted stylus:
Pickering D400 stylus shows more typical 45 degree illumination: