Beyer micpre - circuit analysis & sims

Here's more on the BeyerDynamic circuit from this thread:
http://www.groupdiy.com/index.php?topic=7375

I guess I better move this here to the Drawing Board.


I traced the relevant part of one micpre-channel, it goes like this:

The pdf file contains both the circuit & an AC-sim.

The graph in the pdf shows two nodes: the total output (OUT) and
the interm. signal after the ComplFeedbackPair (CFP_OUT).
Input-signal was 1 mV.
The total of six traces is explained by the sweeping of the C1-value,
that's the cap between TX & Q1.
Note this card seems to be intended for a speach/conference-system,
so the bandwidth was poor (with the orig C1-value of 0.47uF).


#1.
Gain is very high, some 70 & 90 dB for CFP_OUT respect. OUT.
But I didn't see any provision on this card for connection to an external gain-control.
I want to add such a thing, likely along the lines of
http://sound.westhost.com/project13.htm

This will also give a more usable situation at the higher frequencies.


#2.
Realizing the general setup of the Beyer-circuit is a bit
like that ESP-circuit (apart from the actual amp),
the absence of the input-resistor pops up
(there's 1k2 in the ESP-circuit, but 'nothing' in the Beyer).
So if I'm understanding this correctly could it be that
the Beyer-circuit is doing the zero-Z-tranny trick then ?
(more on that here:
http://headfonz.rutgers.edu/zero-Z-tranny.pdf
http://www.groupdiy.com/index.php?topic=4864 )
And then there should be pre-TX stuff outside the card,
probably in the form of a stepped pad, realizing the
sought after gain-control.

Any ideas about all this ?

Thanks / Regards,

Peter

> Gain is very high, some 70 & 90 dB for CFP_OUT respect. OUT.

Huh??? NO! Gain is tranny ratio (3.15) times the ratio 15K/680. Looks like 3*20 or around 60. Whoops: the 47K+10K loads it to more like 15 in the amp, ~50 overall.

Ah-- your SPICE lets you write 1mV input, but in fact for .AC runs it always uses 1V. (Stick a probe on the source to check me.)

So assuming a few more fudges, your plot's "38V" is the gain. Not 1,000-3,000 as you seem to read it.

Where did you get K=0.990? That's a terrible audio transformer; you didn't get that from Beyer. You can assume K=0.999 for any half-good audio iron, and some are much more (like 0.999,9). That's your 2.8KHz roll-off right there: K=0.99 implies 100:1 bandwidth.

Does your IND=6 refer to the primary or the secondary? (My SPICE inputter doesn't take numbers this way, no TURNS.)

> I didn't see any provision on this card for connection to an external gain-control.

At gain of 38-50 and dynamic mikes or speech-only, you don't need a gain control.

> this card seems to be intended for a speach/conference-system, so the bandwidth was poor

"Speech" systems today will usually pass 50-15KHz fine. "Fine" may include a bass-cut for intelligibility, but this is often done in a later stage.

> I want to add such a thing, likely along the lines of westhost project13 ... Realizing the general setup of the Beyer-circuit is a bit like that ESP-circuit (apart from the actual amp), the absence of the input-resistor pops up

This is NOT Westhost's #13. That one has high gain and a low input impedance. This one has gain set by 15K/680+fudge, and a moderately high input impedance.

If you did try Westhost's feedback without an input resistor, you would approach zero-Z input but also a shorted-mike. No harm, but may skew the sound.

Simplify your analysis. Pretend the transformer is perfect: take it out and just remember to multiply gain by 3.16. Your 200Ω dummy-mike transforms to 3.16^2 or 10 times higher: 2K. Watch both sides of that resistor: if voltage drops more than 10%, you are putting a "too heavy load" on the mike (most mikes are rated for loads 10 times higher than nominal impedance).

The impedance at Q1 base is roughly 150K bass-mids (0.47uFd is plenty), roughly 10K at the top of the audio band. I could show my calcs but I got up too early and bed calls me.......

Thanks PRR, your comment much appreciated.

from PRR:

> Gain is very high, some 70 & 90 dB for CFP_OUT respect. OUT.

Huh??? NO! Gain is tranny ratio (3.15) times the ratio 15K/680. Looks like 3*20 or around 60. Whoops: the 47K+10K loads it to more like 15 in
the amp, ~50 overall.

Ah-- your SPICE lets you write 1mV input, but in fact for .AC runs it always uses 1V. (Stick a probe on the source to check me.)

Click to expand...

Thanks, it does indeed that.
While it does make little sense I used anything else than 1V for
the AC-run I didn't expect this 'simulator-initiative'.
FWIW, at work I'm using nothing but 1V for AC (inhouse-sim, non-SPICE),
let's see if that sim has this trick as well...

Where did you get K=0.990? That's a terrible audio transformer; you didn't get that from Beyer. You can assume K=0.999 for any half-good audio iron, and some are much more (like 0.999,9). That's your 2.8KHz roll-off right there: K=0.99 implies 100:1 bandwidth.

Click to expand...

, I wasn't paying attention to the coupling - I hadn't realized this to be a crucual factor.
The TX used in the sim is just a generic TX-part from the program-lib, nothing advanced there.
So I put in the ratio & prim-inductance but ignored the K.

That BW is indeed lame, was initially expecting the caps were causing it (like C7) but they didn't.
I've increased K but it doesn't seem to make a difference in the sim though. Will check this further.
I'll read up on why coupling actually influences this to such a noticable extend.
Well, your suggestion about skipping the whole TX takes care of this for now.

Does your IND=6 refer to the primary or the secondary? (My SPICE inputter doesn't take numbers this way, no TURNS.)

Click to expand...

The Beyer-TX-datasheet says 6H prim, and the SIMetrix-simulator has an identical entry.

> I didn't see any provision on this card for connection to an external gain-control.
At gain of 38-50 and dynamic mikes or speech-only, you don't need a gain control.

Click to expand...

You're right, these are modest gains.

> this card seems to be intended for a speach/conference-system, so the bandwidth was poor
"Speech" systems today will usually pass 50-15KHz fine. "Fine" may include a bass-cut for intelligibility, but this is often done in a later

stage.

Click to expand...

The circuit in its present form seems to have a midband input-impedance of around 9k4.
As a consequence 470n for C1 gives f-3dB of around 35Hz,
without doubt fine for the application. Could make this a bit larger for "music".

This is NOT Westhost's #13. That one has high gain and a low input impedance. This one has gain set by 15K/680+fudge, and a moderately high input impedance.

The impedance at Q1 base is roughly 150K bass-mids (0.47uFd is plenty), roughly 10K at the top of the audio band. I could show my calcs

Click to expand...

More about the 9k4 mentioned above: this lower value would be because of R3 & C3.
Removing them brings the input-imped. close to your expected 150k
(sim says ~125k).

If you did try Westhost's feedback without an input resistor, you would approach zero-Z input but also a shorted-mike. No harm, but may
skew the sound.

Click to expand...

I remember that input-impedance issue is also the reason using
those 1:3.16 TXs (the Beyer card has five of these) aren't optimal for using them with the #13-circuit.
As you commented earlier at the old place, something like 1:1 should be used there, otherwise input-impedance would drop too much.
Or for a 1:3.16 TX, increase the 1k2 to 12k (to keep mic-load constant) and live with the non-optimal noise level, but let's not.
Probably other circuit-tweaks would help then, but OK...

Simplify your analysis. Pretend the transformer is perfect: take it out and just remember to multiply gain by 3.16.

Click to expand...

This results in ~21dB, so with the TX added around 31dB. It now all makes more sense.

but I got up too early and bed calls me.......

Click to expand...

Same here - it's beyond me why stuff like this always takes place
at a time when other people have already several hours of sleep
under their belt... :? But OK...


Regards,

Peter