A few years ago I made a lengthy post (page 22 of this thread) looking at how the frequency response of the wet path (i.e. BBD onwards) in the EM3207 compared to an original EM using an SAD1024.
I found that the frequency response in the "wet path" of an original EM drops off more strongly than in the EM3207.
This can explained by the presence of the transistor buffer Q1 at the BBD output in the EM3207, and the lack of such a buffer in the original EM.
I believe Thomeeque put in Q1 because the EM3207 has (47k || 680nF) on its BBD output rather than (10k || 3nF) as in the original EM. The buffering effect of Q1 causes more highs to feed into the path that follows (i.e. C8 onwards).
That is not what happens in the original EM. The highs in the original EM wet path get shelved off (i.e. lowered with increasing audio frequency).
I suggested some mods for the EM3207 to make the circuits a closer match (e.g. by adding extra components before/after Q1) but I have now found what I think is a far simpler and better way of doing this.
I asked myself why Thomeeque used the values he did on the output of the BBD in the first place, instead of using 10k like the original EM ? There is nothing in the MN3207 datasheet saying that a resistor of about 47k is necessary. The example circuit in the datasheet shows (100k || 100k) on the BBD outputs, so maybe 47k was chosen based purely on that circuit?
Anyway, I got out my old EM3207 build to try something and guess what?
You can change the parts on the BBD output to match the original EM and drop the Q1 buffer completely!!!
- Change (R12, R13, R14, C7) from (4k7, 4k7, 47k, 680nF) to (1k, 1k, 10k, 3n3).
- Remove Q1 and R15 completely.
- Connect the top of C7 directly into C8
The maximum BBD gain of the EM3207 is then slightly reduced by doing this (about 1dB), and this is actually a good thing if you want a closer match to the SAD1024.
Note that maximum gain of the MN3207 is slightly over 2dB (according to datasheet and my measurements on the pre-modded EM3207), while SAD1024 with 10k BBD output resistor has gain of about 1dB.
So to summarise. If you make the changes mentioned here, you'll have a closer match to the SAD1024 based circuit, and lower noise than the original EM3207 too (since the shelving effect in the wet path once Q1 is removed reduces hiss).