All cheap USB sound card adaptor or cards have microphone input enough to drive a dynamic microphone. I wonder what is the preamp chip used inside and the gain. Any idea?
Mind giving some examples for what kind of devices you are thinking of?
For basic 1-chip affairs (often CMedia or Realtek) with consumer-level 3.5 mm I/O, they generally have selectable gain preamp stages built-in (10, 20, 30 dB being typical). Input noise density is on the order of 20 nV/√(Hz), maybe 10 if you're lucky. Adequate for electret mics but not particularly great (rather noisy) for dynamics.
There are some inexpensive USB audio interfaces with noise performance that's actually adequate for dynamic mics, notably the M-Audio M-Track Solo/Duo or Behringer UMC22. AFAIK they are using an NJM2122 opamp (much like the Focusrite Scarletts and other low-midrange audio interfaces), a quirky little part designed for mic inputs in portable (DAT) recorders once upon a time. Cost is being kept down by using a single-chip PCM2902 USB codec and running everything off of +5V only, plus a plastic enclosure in case of the M-Audios (which seem to be a bit sensitive to EMI). Add a Behringer XM8500, and you've got a "cheap and cheerful" little podcasting setup.
BTW, you can
drive a speaker or headphone but not a microphone. A microphone, being a signal source, is driving the microphone input. And as outlined above, a dynamic mic is passive and does not require a bias voltage or phantom power supply to work. It'll generate signal on just about any input, the question is just what kind of signal to noise ratio you're getting and how much the frequency response is changed by an input impedance outside a normally-expected range.
Also how do you test the maximum gain in -dB?
You mean system gain,
@Julian Krause style (-dBV/dBFS or -dBu/dBFS)?
You need a lowish-impedance audio output, a software signal generator (the one in REW will do) and a multimeter suitable for measuring low-level AC voltages on the order of 2 Vrms (prefereably True-RMS and suitable to at least 1 kHz, but you can make do with a more basic device if you stick to 50-60 Hz or so).
The idea is to first set the signal generator to a sine output at -3 dBFS or thereabouts and turn up (sound device) output volume, so the signal amplitude can easily be measured by the multimeter. Do so and convert the result to dBV. Let's say we are getting 0.8 Vrms or -2 dBV.
Then the signal generator would be set up for much lower output, say -60 dBFS - this needs to be low enough to be well within the acceptable range for the input at maximum gain. Sound device output volume setting would be left as before.
Then you'd be connecting the output to the input, look at the spectrum in order to make sure you're not blatantly clipping (input level should be well below 0 dBFS), and take note of digital input level on the recording side. Let's say it is -12 dBFS.
Approximate system gain can then be calculated as follows:
Gs = -2 dBV - (-3 dBFS) + (-60 dBFS) - (-12 dBFS) = -47 dBV/dBFS
I also heard some explaining USB sound card doesn't have any preamp but just ADC. If so, how come the dynamic microphone can still work in such pure ADC USB sound card?
Again, please give an example for such a device. It would certainly not work (well at all) with a Behringer UCA202 or a Soundblaster X-Fi HD, or a trusty M-Audio Audiophile 2496 or ESI Juli@. I suspect the device in question may have been a 1-chip affair that is, in fact, not just a fixed level ADC but does contain a preamp stage. You could probably pull it off decently enough with a Cosmos ADC or RME ADI-2 Pro.