Open source project: Little oscillator PCBs for LDCs shared on Osh Park

Help Support GroupDIY:

midwayfair

Well-known member
Joined
Apr 7, 2015
Messages
535
Location
Baltimore
Khron said:
PS: RuudNL Yes, the RC time-constant with 1G/100nF would indeed be 100sec, but that's "only" the time the voltage reached 63%; i believe the widely-accepted(?) figure for it reaching 100% is 5 times the RC time-constant, so 500sec  :)

Is this the time constant for the capsule itself to come up to full voltage, or an argument for omitting the C6 and C10 filter caps after the 1G resistors (or using, say, 100M there instead)?

I think one of the K47 capsules I measured from the group buy was 170pF ... :/
 

Khron

Well-known member
Joined
Apr 8, 2010
Messages
2,040
Location
Finland
It's an argument for not using 1G resistors to filter the bias voltages, at all :D

Ie. skipping R3 & C6 in here, altogether (and their correspondents on the negative "branch"):

https://jonpattonmusic.files.wordpress.com/2017/04/oscillator-for-schoeps-circuits1.png

See the "rise time" equations over here: https://en.wikipedia.org/wiki/RC_time_constant#Cutoff_frequency

midwayfair said:
Is this the time constant for the capsule itself to come up to full voltage, or an argument for omitting the C6 and C10 filter caps after the 1G resistors (or using, say, 100M there instead)?

I think one of the K47 capsules I measured from the group buy was 170pF ... :/
 

midwayfair

Well-known member
Joined
Apr 7, 2015
Messages
535
Location
Baltimore
Khron said:
It's an argument for not using 1G resistors to filter the bias voltages, at all :D

Ie. skipping R3 & C6 in here, altogether (and their correspondents on the negative "branch"):

https://jonpattonmusic.files.wordpress.com/2017/04/oscillator-for-schoeps-circuits1.png

See the "rise time" equations over here: https://en.wikipedia.org/wiki/RC_time_constant#Cutoff_frequency

Fair enough. I might go in and jumper the 1G to see if there's a noticeable change in noise.

I should add here that the multiple filters aren't pointless. If there's some ripple in the phantom power, that will appear as a wiggle on the oscillator, which will then put out an inconsistent voltage. Multiple stages can ensure that this is never a problem. I guess a compromise would just be another 1M in place of the 1G or something. The 1G was also a carry-over from the Schoeps circuit itself. Phantom power was probably a lot dirtier in those days.

I don't have a good way to measure noise, but I would expect most of us have clean phantom power.
 

Khron

Well-known member
Joined
Apr 8, 2010
Messages
2,040
Location
Finland
A few corrections:

1) The phantom power is usually supplied through a pair of 6.8k resistors in the preamp. Those, in conjunction with the pair of 1-2.2k "summing" resistors in the mic, and the first filter cap after those, form a preeeeeeeeetty "hefty" low-pass filter already.

Using 4.4k (3.4k from the two 6.8k in parallel, in series with a ballpark 2x 2k in parallel) and 10uF over here [ http://sim.okawa-denshi.jp/en/CRtool.php ], gives a low-pass cut-off frequency (-3dB corner) of 3.6Hz, and going down by 6dB/octave from there on. I'm gonna hazard a guess and say that not a lot of ripple's gonna make it to the input of this oscillator. And that's with VERY conservative values, i think you'll agree. Not that the oscillator would / will care all that much, anyway ;D

And that's even before the other voltage-dropping resistor (in this case, that 10-50k trimmer) plus the 220uF or however large it is, capacitor ;)

<later edit>
Oh, but isn't the input to the oscillator taken from after the zener regulator on the mic pcb, anyway?  ::)
That's some extra ripple-"cleaning" right there...
<end of edit>

2) Not quite buying the "wiggle on the oscillator, which will then put out an inconsistent voltage", partially due to the above-mentioned reasons, and partially due to the filter capacitance on the output, coupled with the tiny tiny current it needs to supply.

3) "Carry-over from the Schoeps circuit"? I'm afraid you might've mis-read it - that's assuming we're talking about the original one. The only 1G resistor i see there, is the one from the JFET gate to the wiper of the bias trimpot. The "lone" series resistor that's on the output of the oscillator is a 1M one.

midwayfair said:
Fair enough. I might go in and jumper the 1G to see if there's a noticeable change in noise.

I should add here that the multiple filters aren't pointless. If there's some ripple in the phantom power, that will appear as a wiggle on the oscillator, which will then put out an inconsistent voltage. Multiple stages can ensure that this is never a problem. I guess a compromise would just be another 1M in place of the 1G or something. The 1G was also a carry-over from the Schoeps circuit itself. Phantom power was probably a lot dirtier in those days.

I don't have a good way to measure noise, but I would expect most of us have clean phantom power.
 

emrr

Well-known member
Joined
Apr 12, 2006
Messages
7,923
Location
NC, USA
midwayfair said:
Yes, if you want to generate a negative voltage rather than a positive one, you would reverse the diodes. The diodes only see AC voltage and the direction each pair faces changes whether they rectify to a positive or negative voltages.
May I ask what situation that would pop up in?

Just saw this, thanks.  Hacking in bias for figure 8 in a vintage 87 clone but with a single backplate capsule.  Several changes to make that happen, of course. 
 

midwayfair

Well-known member
Joined
Apr 7, 2015
Messages
535
Location
Baltimore
emrr said:
Just saw this, thanks.  Hacking in bias for figure 8 in a vintage 87 clone but with a single backplate capsule.  Several changes to make that happen, of course.

I see. So you have a mic that already has a bias voltage generator, and you just need a way to create a SECOND voltage for the figure 8? If you have room, it might make more sense to remove the current voltage generator you have an use the version of this board that makes two voltages for you. They'd be identical voltages so you should get a better pattern.
 

emrr

Well-known member
Joined
Apr 12, 2006
Messages
7,923
Location
NC, USA
midwayfair said:
I see. So you have a mic that already has a bias voltage generator, and you just need a way to create a SECOND voltage for the figure 8? If you have room, it might make more sense to remove the current voltage generator you have an use the version of this board that makes two voltages for you. They'd be identical voltages so you should get a better pattern.

True.  I'm cramming a vintage 87 circuit into scavenged Neumann metalwork for use with a modern Neumann single backplate K67, so space is at a premium.  I may have to breadboard the solution.  It'll end up at least partially like a U87AI in terms of bias approach.  There's a number of other things to change as well. 
 
Top