Author Topic: Capsule amplifier in electret microphone (Read 11867 times)

Roger Gustavsson · « **on:** July 07, 2007, 03:48:38 PM »

I am new to this forum. Just found out that there are quite a few threads on the Nakamichi microphones (=Teac, Tascam, Altec, all manufactored by Primo). I would like to modify my own set of CM-300. The output level is too low in my set up. The output transformer reduces the output from 8 to 1.6 mV. Please, look at the schematics below:

A is the original circuitry of the CM-300.

B is a revised circuitry with a raised supply voltage as suggested by Stephen Sank.

C would be the front end of B. Without transformer not the very best for driving long cables. What difference will there be between C and D?

What about E, can it be used? Sure, also problematic with long cables. F is taken from the homepage of Scott Helmke. It uses a P48 phantom power supply. As I would like to run my microphones on internal batteries maybe a P12 would be suitable? Is my schematic G correct? The components are sometimes added, for what reasons? What about the values of the drain and source resistors to suit the 2SK118-O with an Idss=1mA?

Roger Gustavsson

SparkE! · « **Reply #1 on:** July 07, 2007, 11:16:18 PM »

Schematic E looks OK to me.

Schematic F is exactly what Scott Helmke uses and other DIY folks have used that schematic with good results with the Transsound TSB-165 mic elements.

Schematic G would need some sort of path for current to flow from the collectors of the PNP transistors. In Scott's schematic, the current would flow through the zener diode. The way you have it connected, it would put something close to 48 V across your JFET and it probably wouldn't like that too much.

Before building anything, you might want to try modeling the circuit with Spice.

Roger Gustavsson · « **Reply #2 on:** July 08, 2007, 04:11:28 AM »

So E would also work but with limited capability in driving long cables?

About the schematic G. There will not be 48V anywhere, the phantom supply voltage is just 12V in scheamtic G, that is why I deleted the 12V zener. I think I forgot a resistor, like the 330R in Scott's schematic. I am only a layman in electronics, so all help is welcome!

I have no wish at all to run on outboard phantom power supply. Therefore, a lower supply voltage would make things easier, 9-12V. I also would like to use the 2SK118 JFET that is already there and get rid of the tiny output transformer.

Roger Gustavsson

Roger Gustavsson · « **Reply #3 on:** July 08, 2007, 01:41:19 PM »

Well, Team Nakamichi seems to concentrate more on how to modify the microphones to "actives" and/or how to run them on phantom supply. The "actives", actually not any more active than the originals, are not suited to drive long cables. I think it is the possibility of "stealth" recording that attracts many. I am not doing such recordings so it is of less importance to me. Phantom supply is not very diffcult at all (I have a schematic from Stephen Sank) but it is a no help if your equipment does not have it. All you need are some piece of wire, two resistors and a zener diod.

I am usually running two Nakamichi CM-300 (ORTF or A/B) into the Nakmichi MX-100 (on batteries) and then to a Digigram VX Pocket V2 soundcard in my laptop. The Digigram soundcard does have a micpreamp but it is too noisy at higher gains with the CM-300 while recording acoustic music. The original CM-300 needs a gain of 500 in my setup. The MX-100 will do that but I would like to do it without the help of the MX-100.

I would certainly prefer to shorten the route for the weak signals from the microphones to the soundcard. Skiping the output transformer will give about 14 dB more output and a gain of 100 would do the trick. For better "transport", and some amplification, of the signal a Chargeamp à la Scott Wurcer can be used just near to the capsule and it will run on an internal battery (there are various alternatives to the original battery). It means soldering surface mount devices, not very easy for an amateur like me but I have the components... The Helmke way seems to be a bit easier with its "normal" components. Less gain but probably okay for my soundcard. The P93 by Rod Elliot can also be used. The gain is raised in my schematic.

Roger Gustavsson · « **Reply #4 on:** July 09, 2007, 08:07:14 AM »

I have made some changes to my schematics. Would like to ask for some comments!

SparkE! · « **Reply #5 on:** July 09, 2007, 10:15:58 AM »

You still have no DC path from the collectors of your PNP transistors to ground in Schematic G. That one will not work. Schematic E will still work and it would be the easiest to implement.

Roger Gustavsson · « **Reply #6 on:** July 10, 2007, 01:10:55 AM »

Quote from: SparkE! on July 09, 2007, 10:15:58 AM

You still have no DC path from the collectors of your PNP transistors to ground in Schematic G. That one will not work. Schematic E will still work and it would be the easiest to implement.

Here we go again with a new schemtic.... Are there any components that can be taken out?

Roger Gustavsson · « **Reply #7 on:** July 10, 2007, 01:20:32 AM »

Quote from: O-Canis on July 09, 2007, 01:49:46 PM

So, with all this hi-tech digestion of drawings with little squiggly lines on 'em ... is there a chance that someone holding a pair of Leggedy moded Naks could find that there is a way to get them to really "put out"?

I was rather... no, was really disappointed when I ran them the first time at moderately high SPL (read, louder than my stereo) and that they really didn't come up until I was at 9.5 (to 10) on my UA5 gain. further control between 9.5 and 10 is nearly impossible to manage. Here's a coupla' Ts for the entertaining dialog... carry on.

I am not familiar with Leggedy's modification. Is it just adding a phantom supply? A phantom supply just replaces the internal battery. The output of the CM-300 is low. Removing the balancing output transformer will raise the output by five times (14 dB) but driving long cables will be diffycult.

Roger Gustavsson · « **Reply #8 on:** July 10, 2007, 12:29:06 PM »

Then you have about 5 times (14 dB) more level than I have! The Edirol UA-5 seems to have a maximum gain of 50 dB (317 times). Sure you will have to crank up that level control. Depending on what you are recording, somewhere between 20-40 dB of gain is probably needed. Are you having problems with the background noise level at high gain settings? That is what I have with Nak300 into my Digigram soundcard. Running the Nak300 with MX-100 is without problem.

Roger

SparkE! · « **Reply #9 on:** July 10, 2007, 12:33:55 PM »

Quote from: Roger Gustavsson on July 10, 2007, 01:10:55 AM

Quote from: SparkE! on July 09, 2007, 10:15:58 AM
You still have no DC path from the collectors of your PNP transistors to ground in Schematic G. That one will not work. Schematic E will still work and it would be the easiest to implement.

Here we go again with a new schemtic.... Are there any components that can be taken out?

I'd start by getting rid of the 12 V zener and replacing it with a wire. Then I'd move the wire that goes to the 330 ohm resistor over to connect directly to the 12 V battery. Then I'd add two resistors, each 10k or so, between the base and emitter of each of the PNP transistors. (You had these added in a previous version of the schematic.)

Roger Gustavsson · « **Reply #10 on:** July 11, 2007, 02:05:26 AM »

Thank you, SparkE!

Here is a new schematic. Is it correct? Are the components in red needed?

SparkE! · « **Reply #11 on:** July 11, 2007, 09:49:47 AM »

Quote from: Roger Gustavsson on July 11, 2007, 02:05:26 AM

Thank you, SparkE!

Here is a new schematic. Is it correct? Are the components in red needed?

Yes, the components in red will reduce the signal-induced power supply ripple in the second amplifier stage.

Now you don't have any DC path to the emitters in your second stage, though. Do you have any circuit design experience?

Roger Gustavsson · « **Reply #12 on:** July 11, 2007, 10:19:55 AM »

I am a complete layman in electronics.... I want to run the F circuitry on batteries, 9-12V. That would be equal to P12 phantom power? Should the two 6k8 resistors be changed to 680R? What about that 12V zener? I do not know if anything else has to be changed to suit the 2SK118-O.

SparkE! · « **Reply #13 on:** July 11, 2007, 12:23:21 PM »

How about this?

Roger Gustavsson · « **Reply #14 on:** July 11, 2007, 02:21:06 PM »

Great!

Why R4/R5 = 10k? With C4/C5 lower cutoff frequency = 34 Hz. Should I increase the resistors or the capacitors for a lower cutoff frequency?

SparkE! · « **Reply #15 on:** July 11, 2007, 06:26:09 PM »

Quote from: Roger Gustavsson on July 11, 2007, 02:21:06 PM

Great!

Why R4/R5 = 10k? With C4/C5 lower cutoff frequency = 34 Hz. Should I increase the resistors or the capacitors for a lower cutoff frequency?

I reduced R4 and R5 to set the DC bias point of the second amplifier stage. If you want a lower cutoff frequency, increase the size of the coupling caps, C3 and C4. Also please note that C5 and V2 are the signal equivalent of the electret mic element. I put them in to be able to run the SPICE simulation.

You still will not have a lot of signal level out of this because there's not a lot of gain. The main things you are getting with this topology is a lower source impedance and balanced output. You may need to add a third amplifier stage if you need more signal level. I guess you'll find that out when you hook it up though...

Roger Gustavsson · « **Reply #16 on:** July 15, 2007, 01:27:59 PM »

Quote from: SparkE! on July 11, 2007, 06:26:09 PM

You still will not have a lot of signal level out of this because there's not a lot of gain. The main things you are getting with this topology is a lower source impedance and balanced output. You may need to add a third amplifier stage if you need more signal level. I guess you'll find that out when you hook it up though...

Another DIY:er made an simulation of the Helmke circuitry. I think the gain was about 1.5 times. That will correspond to 12 mV/94 dB. A lot more than the original 1.6 mV. The noisy floor of my soundcard, would probably not be a problem then. Your circuitry have perhaps simular gain?

Roger

SparkE! · « **Reply #17 on:** July 15, 2007, 07:37:31 PM »

Quote from: Roger Gustavsson on July 15, 2007, 01:27:59 PM

Quote from: SparkE! on July 11, 2007, 06:26:09 PM

You still will not have a lot of signal level out of this because there's not a lot of gain. The main things you are getting with this topology is a lower source impedance and balanced output. You may need to add a third amplifier stage if you need more signal level. I guess you'll find that out when you hook it up though...

Another DIY:er made an simulation of the Helmke circuitry. I think the gain was about 1.5 times. That will correspond to 12 mV/94 dB. A lot more than the original 1.6 mV. The noisy floor of my soundcard, would probably not be a problem then. Your circuitry have perhaps simular gain?

Roger

The gain of the circuit I simulated was about .7 and I think that's the gain of the Helmke circuit too.

Roger Gustavsson · « **Reply #18 on:** July 16, 2007, 04:36:00 AM »

Thank you, SparkE!

So there is no gain. Your circuitry even reduces the output somewhat? But there are two out of phase signals, would it not mean that these should be added to each other? The output impedance will at least be low?

SparkE! · « **Reply #19 on:** July 16, 2007, 09:31:38 AM »

Quote from: Roger Gustavsson on July 16, 2007, 04:36:00 AM

Thank you, SparkE!

So there is no gain. Your circuitry even reduces the output somewhat? But there are two out of phase signals, would it not mean that these should be added to each other? The output impedance will at least be low?

I'm including the gain associated with adding the two outputs. The main advantages you get from this type of topology are the differential output, low distortion and low output impedance. So, you can drive longer cables without much degradation, except due to resistive losses in the cable.

Roger Gustavsson · « **Reply #20 on:** July 16, 2007, 10:30:43 AM »

But there will still be 10-11 dB more level from this circuitry than the original Nakamichi with its output transformer. Not a bad thing!

I have been told elsewhere that the schematic E in the beginning of this thread, should have drain and source resistors in the range 2-3 kOhm for a 9V supply. Would give an output impedance of 590-650 Ohm. Could still drive up to 2 nF of cable capacitance? Can you say if the output level will change from the schematic C or D?

SparkE! · « **Reply #21 on:** July 16, 2007, 11:27:47 AM »

No, I think that if you use 2k to 3k, it will give each output an impedance of 2k to 3k in schematic E.

Schematics C and D will give a lower output impedance at the expense of higher harmonic distortion.

Roger Gustavsson · « **Reply #22 on:** July 16, 2007, 11:58:48 AM »

So the simulation below with a different JFET is all wrong? Load is 10 nF and 10 kOhm.

SparkE! · « **Reply #23 on:** July 16, 2007, 12:21:09 PM »

Well, according to the upper cuttoff frequency shown in your simulation, that appears to have an output impedance of about 122 ohms per side, so I guess I had that wrong... I was thinking that the output impedance would be dominated by the drain and source resistors.

Roger Gustavsson · « **Reply #24 on:** July 16, 2007, 12:36:26 PM »

The schematic C will give this: 1/gm // R source

(1/1.2ms) // 1200 = 492 Ohm

I have no idea of the phase splitted output, like in E but with source and drain resistors = 2 kOhm. Maybe the different legs will be (1/1.2ms) // 2000 = 588 Ohm, using the same formula?

SparkE! · « **Reply #25 on:** July 16, 2007, 01:14:37 PM »

When I load the output of the circuit that I simulated (NakMod.jpg) with a 10 nF cap, it implies an output impedance of about 13 ohms per side. So I guess that the emitter follower amps further reduce the output impedance.

Also, I changed my JFET model to the JFET you are using and the gain went up to 1.4, so that's about a 3 dB gain. The harmonic distortion is about 0.5% at 1 kHz.

Roger Gustavsson · « **Reply #26 on:** July 16, 2007, 01:21:42 PM »

Thank you for helping me out!

0.5% distorsion at what level? Are we talking input or output level? My recording situation seldom sees anything above 100 dB. Sure your circuitry should lower the output impedance compared to the nonbuffered one.

SparkE! · « **Reply #27 on:** July 16, 2007, 01:32:47 PM »

I was using 1.6 mV at the input since that is what you had mentioned previously. I don't know what signal level to use.

Roger Gustavsson · « **Reply #28 on:** July 16, 2007, 02:42:30 PM »

0.5% is high compared to the modified P93 http://taperssection.com/index.php?action=dlattach;topic=87397.0;attach=67399. It seems to have 0.024% at 8 mV input and 208 mV output. It takes up more space than your circuitry.

SparkE! · « **Reply #29 on:** July 16, 2007, 06:48:29 PM »

I'm not sure if I'm buying that claim of .024% distortion. Is that what you're getting from a simulation? What version of SPICE are you using?

Roger Gustavsson · « **Reply #30 on:** July 17, 2007, 04:27:37 AM »

I was not doing the simulation myself. It was made by a friend from another forum. He is not using SPICE as far as I know.

SparkE! · « **Reply #31 on:** July 18, 2007, 11:30:49 AM »

The problem with the high distortion turned out to be a problem between my ears. I needed to reduce the time step size in the simulation. Once I did that, I found that the distortion is negligible. When I do the FFT, an artifact of the turn-on transient is what sets the noise floor of the plot and all of the harmonics are below that. From what I can tell, the distortion has to be below 0.05%, but I don't yet know how far below. Whatever the case, it's clear that the microphone distortion will be dominated by the response of the mic element itself and not by the amplifier.

SparkE! · « **Reply #32 on:** July 18, 2007, 12:48:25 PM »

OK, I finally got it figured out. I needed to use both a smaller time increment and a longer sample. The simulation took about 10 minutes to run and the distortion appears to be on the order of .007% which is clearly negligible compared to the distortion that will be introduced by nonlinearities in the mechanical deflection of the mic diaphragm itself. Unintentional coupling within the layout will have a larger effect than that!

I knew that as a differential amp, it had to have a lower distortion than a single ended amplifier, but it surprises me at how low the distortion turns out to be.

Roger Gustavsson · « **Reply #33 on:** July 18, 2007, 02:49:58 PM »

You are still talking about your "NakMod.jpg"? What are you using for Q1 and Q2?

Nick's Picks · « **Reply #34 on:** July 18, 2007, 03:01:53 PM »

man...
I love people who know more than I do.

+T guys...

SparkE! · « **Reply #35 on:** July 18, 2007, 03:29:24 PM »

Quote from: Roger Gustavsson on July 18, 2007, 02:49:58 PM

You are still talking about your "NakMod.jpg"? What are you using for Q1 and Q2?

Yup, that's what I'm talking about. Q1 and Q2 are modeled right now as 2N3906 transistors. Nothing special required since they are just emitter followers. You could probably choose something quieter.

Roger Gustavsson · « **Reply #36 on:** August 08, 2007, 04:38:06 AM »

I have found a simular circuit. Please, see below. I have adapted it for my Nakamichi capsules. Will it work? Are the 10 kOhm resistors necessary? What happens if I lower the supply voltage to 7.2 or 9V?

SparkE! · « **Reply #37 on:** August 08, 2007, 09:11:21 AM »

It'll work just fine. You need to just build some circuits and try them. I keep expecting you to come back to the list with some actual information and you just keep coming back with more questions.

The 10k resistors could probably be omitted, but they don't hurt anything. They may actually lower the harmonic distortion a little, but it would not be a dramatic decrease.

Roger Gustavsson · « **Reply #38 on:** August 09, 2007, 08:55:19 AM »

Yes, I will build this circuitry... The JFET and the 1 GOhm resistor will sit very close to the capsule. The rest fill fit on a tiny board, just 25 x 12.5 mm. Like my attached drawing. I take your word that it will work, as I cannot confirm it with meassurements. It is also difficult to find the JFET and the highvoltage resistors but I have them now.

Roger Gustavsson · « **Reply #39 on:** August 30, 2007, 03:55:45 PM »

I just realized that it will be a lot easier to stay with a 2-pole connection between the microphone capsule and the main body, just like in the original Nakamichi CM-300. That will not let me use the Hicks' circuitry. Went back to the P93 designed by Rod Elliot. Are there any way I can use it with a 2-wire/2-pole connection? Maybe like the Kudelski (Nagra) supply? Not with +62 V but +9 or so...