Difference in sound using different carts when digitizing vinyl record?

With 44kHz sampling I doubt sound quality will be good enough to appreciate the capability of a true hifi cartridge. Your best return on investment will be from using a good, clean stylus on any reasonably capable cartridge (AT95E will do quite well for this pursuit), and clean records. 

If you took your digitizing platform up to DSD, the 192kHz sampling rate would be good enough to faithfully reproduce the sound quality of a true hifi cartridge. However, the investment in DSD is very steep, and the choices of equipment for consumer applications is limited. 

@wcfeil ?

@wcfeil, the definition of “average” is subjective. Would you consider an average cartridge a $60 AT95E or a $260 Grado Gold 2 or $750 Ortofon 2M Black, or a $1200 Sumiko Blackbird?

Similarly for turntables and arms, what would you offer as “average”?

My belief is that the law of diminishing returns applies here like in most other areas of hifi. While exceptional results could be obtained for a total outlay of $10K, the sweet spot is probably around $1.5~$2K. 

@chakster I agree with you. The biggest limiting factor is the cartridge and stylus profile. My experience with going from a very good low end MM cartridge with bonded elliptical stylus (AT95E) to a mid-range MM cartridge with nude mount line contact stylus Audio-Technica (VM540ML) on the SL-1700mk2 (same motor/platter/tonearm, but vastly superior chassis) is exactly as you stated.

@orpheus10 the notion that a twice processed signal has in your words “absolutely no difference” after being sampled, digitized, stored and the converted back to analog makes no sense at all. How do you account for the staircase effect?

@inna nope, stylus is the only thing that reads the groove. That has infinitely more impact on sound quality than anything else that is not in the chain of custody of the signal. 

@orpheus10 the staircase effect is the result of trying to approximate the analog signal from the turntable into a fixed scale of values that is used when digitizing. That approximation, or quantization, while good enough for some, is still only an approximation that can never faithfully reproduce the infinite number of values that only analog can convey.

If I could use an analogy, consider two ways of drawing a half circle with the flat side facing down.  The first is with a compas, which is pretty much foolproof at making a nice, smooth, linear curve. But what if we had to represent that circle in terms of taking measurements vertically (quantize) in the Y axis, as we move along the X axis? Now, what if we had to map (sample) that curve only in finite increments along the X (time) axis? That sampling process gets only one measure for each whole number value along the X axis, (1,2,3,4,5...). Take those points and plot them as a bar chart. You might say that half circle all of a sudden looks pretty poor. You might say that half circle resembles a staircase as you proceed from 9 o’clock to 12 o’clock and then continue from 12 o’clock to 3 o’clock. That’s what happens when an analog signal is digitized. You can double the rate of sampling to reduce the incremental step sizes, but It still is an approximation that comes with harmonic artifacts as unwelcome passengers.

The choice is yours: accept the obviously noticeable inherent flaws of the quantization process at low sample rates of 44kHz or even 96kHz or make the jump to high resolution digital (DSD) at 192kHz or stay in the analog domain.