Thank you for your assessment! What errors, in particular, do you notice? What in your opinion could be improved here?
For some reason they use a 470p cap in parallel with the 1M resistor for the high frequency correction. Replacing that cap with a 220p cap (as in the U87i) corrected the problem. The polystyrene looking cap in the middle of the PCB has nothing to do with the audio circuit, it's a 470p PSU bypassing cap. Best to replace that with COG ceramic.
I went pretty deep into it a couple years ago until I had the response very close to my Andreas Grosser modded U87ai:
1. Replace the tiny trafo with a UTM0587.
2. The fake TLM103 has a headbasket with U87 dimensions. The capsule needs to be raised higher and ideally a dome or some foam (like Basotect) added below to deflect/dampen reflections from the metal bottom.
3.1. Remove the two 2.2k resistors, replace them with through hole components of the same value mounted upright and summed together. From there take a 56k resistor (I used 47k, which is fine, you need to isolate the stage otherwise there can be oscillation on the outputs) to cathode of a 24V zener, which should be in parallel with the cap filtering the drain voltage. I added an additional filtering stage with a 10k resistor + ca 5uf cap, this is all per the v16 U87 schematic (you can find it online). From there it goes through a 100k resistor, which also needs to be taken out and replaced with 47k (I used through-hole again). Now you have a clean voltage feeding the drain side of the FET.
3.2. Remove the source resistor and replace it with a 10k variable resistor (trimpot). Set the bias per the instructions (availible online) via the Messeingang (6.8k and 560 ohm resistor were flipped on all of my mics, so exchange them).
I've attached a photo of the fully modded PCB with the parts named as in the old Neumann schematic also attached.
3.3. A simple pad (as in the U87) can be achieved by adding a cap in parallel with the capsule. Neumann used 560pf.
Much of the secret lies in being very diligent with the fet bias setting. I used a 256B transistor, like Andreas Grosser.
Ultimately this mic sounds very close to the real thing, but lacking some of the detail sonic sophistication. This would be down to the capsule (which is much better on that mic already than the average cheap chinese mic) and maybe the mesh structure, which is a little different on the Neumann mic.