I will mesure the capsule capacitance and calculate so :
capsule-capa*(5,5-1) = C in pF
For -15dB > x0,18 > 1/0,18 = 5,5
t. bone SC1100 aka Feilo Z3300 modding, anyone?
- Thread starter theaterguy
- Start date
Help Support GroupDIY Audio Forum:
capsule-capa*(5,5-1) = C in pF
For -15dB > x0,18 > 1/0,18 = 5,5
It's not how it works.
Your calculation would be correct for a capacitor across the capsule, but not for a charge amp.
The input circuit goes from a non NFB common source stage to a voltage-to-current NFB type.
It may be possible to calculate - with a precision of ca. -50/+100% though - but it's typically the kind of things you adjust by trial-and-error.
Anyway, I'm not sure the value of 15dB is optimum.
Let your ears (and soldering iron) do the work.
jeremynothman
Well-known member
Thanks guys for your brains here!
I think the 15dB was essentially given the Max SPL at 125dB. I'd imagine that 20dB would be a good amount of pad to add, to bring it in the range of a 47Fet or similar.
Which value would you begin with for a significient attenuation ?
I've in stock : 27 / 47 / 68 / 150 / 220 pF all polystyrène
Try with 68, but it's a shot in the dark. It would probably result in too much attenuation, but it's a starting point.
thor.zmt
Well-known member
Guys,
OMG, I am having a serious Skywalker moment.
EVERYTHING IN THIS SCHEMATIC IS WRONG! Well done Scheena! (sorry for my Ap Lai Chau accent).
Why not change for something with a sensible turns ratio?
I did NOT believe it was even REMOTELY possible to disimprove the Chinese "Design" (I use this term in the LOOSES POSSIBLE sense).
But you guys actually managed.
By removing the 3k resistor you just took a VERY HIGH GAIN circuit and you made it EVEN HIGHER GAIN!
The original circuit uses the 3K resistor you guys removed to "degenerate" the J-Fet transconductance.
Yes, it adds noise. This is the result of using an totally inappropriate J-Fet for a microphone.
But it limits gain.
With 1.8k and a real 2SK170BL we can expect ~0.3mA current in the FET. The transconductance will be around 8mA/V and noise will be 1.3nV|/Hz.
So without any resistance in the source of the J-Fet the source impedance is 125 Ohm and the transconductance 8mA/V.
Once we use a 20k drain resistor the gain will be Transconductance times resistance, or 43dB!
By inserting the 3k Resistor we add 1uV (7nV|/Hz) noise, but we reduce transconductance to 0.32mA/V, so the gain is reduced to 6.4 and then the output transformer is 1.75:1 so gain is reduced to 3.65 or 11dB, which is still WAY HIGH, but probably almost ok.
So you guys boosted the mic internal gain by 32dB! OMG!
Seriously, this is NO WAY to modify circuits.
Now the solution would be to change the circuit into a charge amplifier.
To do that do this:
The 22pF capacitor will introduce a feedback loop (hence from after the output coupling capacitor and before the input cap directly to the capsule) and with 66pF microphone capsule results in 9.5dB Gain with is reduced to 4.5dB by the output Transformer. Of course, the LF Cut no longer works. It is now inside the feedback loop.
Now gain will be very sensible (~4.5dB), a pad is probably not needed (but switching in an extra 68pF into circuit results in a 10dB Pad) and our feedback loop will improve overall objective performance.
All with one correctly placed small value capacitor.
If 3dB gain is low, a 12pF capacitor gives around 10dB Gain (including Transformer).
Thor
OMG, I am having a serious Skywalker moment.
EVERYTHING IN THIS SCHEMATIC IS WRONG! Well done Scheena! (sorry for my Ap Lai Chau accent).
Why not change for something with a sensible turns ratio?
I did NOT believe it was even REMOTELY possible to disimprove the Chinese "Design" (I use this term in the LOOSES POSSIBLE sense).
But you guys actually managed.
By removing the 3k resistor you just took a VERY HIGH GAIN circuit and you made it EVEN HIGHER GAIN!
The original circuit uses the 3K resistor you guys removed to "degenerate" the J-Fet transconductance.
Yes, it adds noise. This is the result of using an totally inappropriate J-Fet for a microphone.
But it limits gain.
With 1.8k and a real 2SK170BL we can expect ~0.3mA current in the FET. The transconductance will be around 8mA/V and noise will be 1.3nV|/Hz.
So without any resistance in the source of the J-Fet the source impedance is 125 Ohm and the transconductance 8mA/V.
Once we use a 20k drain resistor the gain will be Transconductance times resistance, or 43dB!
By inserting the 3k Resistor we add 1uV (7nV|/Hz) noise, but we reduce transconductance to 0.32mA/V, so the gain is reduced to 6.4 and then the output transformer is 1.75:1 so gain is reduced to 3.65 or 11dB, which is still WAY HIGH, but probably almost ok.
So you guys boosted the mic internal gain by 32dB! OMG!
Seriously, this is NO WAY to modify circuits.
Now the solution would be to change the circuit into a charge amplifier.
To do that do this:
The 22pF capacitor will introduce a feedback loop (hence from after the output coupling capacitor and before the input cap directly to the capsule) and with 66pF microphone capsule results in 9.5dB Gain with is reduced to 4.5dB by the output Transformer. Of course, the LF Cut no longer works. It is now inside the feedback loop.
Now gain will be very sensible (~4.5dB), a pad is probably not needed (but switching in an extra 68pF into circuit results in a 10dB Pad) and our feedback loop will improve overall objective performance.
All with one correctly placed small value capacitor.
If 3dB gain is low, a 12pF capacitor gives around 10dB Gain (including Transformer).
Thor
Last edited:
Voyager10
Well-known member
The actual gain will be limited by the drain-gate capacitance (i.e. Miller effect) but this is very nonlinear (depends a lot on drain-gate voltage), so expect lots of distortion.
Adding a 22pF cap as Thor suggests will swamp this capacitance and remove a lot of the distortion.
If the SC1100 uses a single 180K resistor (R7) to bias T1 this will limit the maximum negative output swing, because T1's emitter is only 1-2V above ground. You can improve this by increasing R7 to 330K and adding a 470K from T1 base to the +ve supply.
Adding a 22pF cap as Thor suggests will swamp this capacitance and remove a lot of the distortion.
If the SC1100 uses a single 180K resistor (R7) to bias T1 this will limit the maximum negative output swing, because T1's emitter is only 1-2V above ground. You can improve this by increasing R7 to 330K and adding a 470K from T1 base to the +ve supply.
thor.zmt
Well-known member
Last edited:
thor.zmt
Well-known member
Much more simple Indeed !
To where we are now, why not change K170 for a more appropriate Jfet ?
Your last schem. keeps the 1,8KΩ source R (no 3KΩ back here) : is it intentional ?
I believe that following your schem. no more need for a pad = benefit and less components
BR
Last edited:
thor.zmt
Well-known member
The the charge amp no longer works, not enough open loop gain.
Here the point is to make minimal and mostly subtractive changes and get a better result. Quick fix, minimal effort.
If we want a to design a competent solid state microphone, that is a whole other ballgame.
Thor
Ok will try your schem. as it is drawn...
thor.zmt
Well-known member
once it all works ok (and not before) and for laughs and giggles, connect C3 to the Emitter of 2SA1015, not GND. This makes the "open loop" gain very high.
It is not really usable without the charge amp.
Another trick, cascode the 2SK170 (google it).
Thor
So no more pad necessary, neither low-cut according to your schem (less is better)
R source stays at 1,8KΩ
I must ordre E-152 and 22pF mica silver and "cut" a new perf-board
will mod only one of the 2 mics to make some tests...
Another task (plus 6AK5 / Jfets pre / HT PSU in both pre / and... some more I guess)
R source stays at 1,8KΩ
I must ordre E-152 and 22pF mica silver and "cut" a new perf-board
will mod only one of the 2 mics to make some tests...
Another task (plus 6AK5 / Jfets pre / HT PSU in both pre / and... some more I guess)
thor.zmt
Well-known member
Almost.
Gate to lower FET source.
And use a higher pinchoff voltage FET for upper. But we are exiting "reuse existing PCB and stuff" and redesign.
If we do that, why go polish on a solid piece of excrement (polish a turd) or Tuna Fish, if we can do mo betta?
Thor
rock soderstrom
Tour de France
independent of the existing PCB and in a new thread.like a J304 which has a VGS(off) from -2v to -6v ? or (again) a J113 ?
a new pcb will be created from a perf-board (6cmx4cm seems a good option for mic bodies and it's a standard size)
Attachments
Last edited:
thor.zmt
Well-known member
thor.zmt
Well-known member
Folks,
After a chat with @Emmathom let's do a cascode charge amp:
Note this is a simulation, so it shows components not part of the microphone and signal probes. The transformer is based on the Monacor LTR-110 which is 2:1 600R : 150R.
The box in front simulates the microphone, 60V Bias. At the back there is a simulated Preamp with Phantom power and a signal probe.
The "BIAS" labelled current sink simulates 2mA current draw for the bias board. If anyone knows actual current draw, let me know, so I can update.
I assume a "bogey" capsule with -40dB(V)/PA sensitivity, or to normalise to something sensible 10mV @ 94dB. Gain is 10dB (so -30dB/PA sensitivity for the microphone). This is still quite high gain.
Adjusting C4 and C5 can reduce gain. With C4 = 33pF and C5 = 100pF gains is 0dB and microphone system sensitivity is same as the Capsule, or down 12dB.
The simulated noise levels are at 26dB unweighted.
Maximum output level before obvious clipping (THD > 1%) is around +7dBu, but the transformer is likely the limiting factor. So overload should not occur until ~126dB SPL in high gain (10dB) with a "bogey" capsule. With the 12dB pad max SPL goes up equally.
Distortion at nominal 0dB (1PA) in high gain mode looks < 0.03%, H2 dominates.
So this will be a relatively speaking "very clean" mic with little saturation or colour when given a loud signal. In "low gain + Pad" mode it should b ok in kick drum or other high SPL application.
The circuit is entirely derivative and generic. Nothing to see really.
As the circuit involves loop feedback to the gain, gain is maximised by "bootstrapping" the drain load resistor from the follower BJT Emitter.
To minimise wasted current while keeping things simple, a 1.5mA "current regulator diode" is the emitter follower "tail" load. The electronics in total consume around 1.75mA Nothing really to see here.
All signal capacitors are inside the feedback loop, so their sonic signature is reduced by the action of the looped feedback. In this day and age I think SMT C0G capacitors for C4/5 are the best choice.
Styroflex are microphonic, as are most other options. Silver Mica are the alternative to C0G..
Thor
After a chat with @Emmathom let's do a cascode charge amp:
Note this is a simulation, so it shows components not part of the microphone and signal probes. The transformer is based on the Monacor LTR-110 which is 2:1 600R : 150R.
The box in front simulates the microphone, 60V Bias. At the back there is a simulated Preamp with Phantom power and a signal probe.
The "BIAS" labelled current sink simulates 2mA current draw for the bias board. If anyone knows actual current draw, let me know, so I can update.
I assume a "bogey" capsule with -40dB(V)/PA sensitivity, or to normalise to something sensible 10mV @ 94dB. Gain is 10dB (so -30dB/PA sensitivity for the microphone). This is still quite high gain.
Adjusting C4 and C5 can reduce gain. With C4 = 33pF and C5 = 100pF gains is 0dB and microphone system sensitivity is same as the Capsule, or down 12dB.
The simulated noise levels are at 26dB unweighted.
Maximum output level before obvious clipping (THD > 1%) is around +7dBu, but the transformer is likely the limiting factor. So overload should not occur until ~126dB SPL in high gain (10dB) with a "bogey" capsule. With the 12dB pad max SPL goes up equally.
Distortion at nominal 0dB (1PA) in high gain mode looks < 0.03%, H2 dominates.
So this will be a relatively speaking "very clean" mic with little saturation or colour when given a loud signal. In "low gain + Pad" mode it should b ok in kick drum or other high SPL application.
The circuit is entirely derivative and generic. Nothing to see really.
As the circuit involves loop feedback to the gain, gain is maximised by "bootstrapping" the drain load resistor from the follower BJT Emitter.
To minimise wasted current while keeping things simple, a 1.5mA "current regulator diode" is the emitter follower "tail" load. The electronics in total consume around 1.75mA Nothing really to see here.
All signal capacitors are inside the feedback loop, so their sonic signature is reduced by the action of the looped feedback. In this day and age I think SMT C0G capacitors for C4/5 are the best choice.
Styroflex are microphonic, as are most other options. Silver Mica are the alternative to C0G..
Thor
