DIY FET mic with very low output.

[Gus]

http://www.groupdiy.com/index.php?topic=217

5.6K should be 56K

 

[saxtim]

I tried using a software oscilliscope I downloaded, but didn't have much luck getting it working - will try more later. I did however do the following which may be just as useful, though I have no idea never having used oscilliscope to debug things (even if I did have one I wouldn't have the skills to figure out whats going wrong):

I fed a 1000khz sine wave into the FET amp with capsule not attached.

I measured with AC voltmeter to confirm I had 30millivolts going in.

I then recorded the signal into cubase at various points in the circuit using a simple audio probe.

I've only recently just got my recording interface, a fireface 800 so I'm a bit new to the whole thing so I'll comment on how I recorded the signal which should hopefully help the more knowledgable figure out was is going on.

I sent the audio probe into a line input (+4dbu). I noticed that when I recorded the signal with the software mixer set to 0db I didn't get -30db (30mV) signal at the FET input. (which was what I was feeding in, and measured with a AC voltmeter) I can't remember what exactly dB i got with the software mixer set to 0 - but when I measured other parts of the circuit it clipped, so I set the software mixer to -10.10dB which allowed me to recorded the other parts of the circuit without clipping.

At the fet input I recorded this:

http://www.geocities.com/mejafu/test1.jpg

(copy and paste link in to new window to open)

peak of -20.67dB

at cap 110, output of Fet preamp (before the transformer) I recorded this:

http://www.geocities.com/mejafu/test2.jpg

peak of -10.10dB No longer looks too sine wavish though - don't know if that's a bad thing.

Finally, I recorded the output of the mic (through transformer). I set the mic preamp to the bottom setting (-10dB) and the software mixer to 0 and I got this:

http://www.geocities.com/mejafu/test3.jpg

peak of -11.97 dB

Bo Hansen commented:

Apply a sine tone 1 kHz, 30 mV (-30 dB) from the oscilator and then measure with a AC mV meter what you have after the C110 capasitor to ground, you shall have approx. 1,2 to 1,3 volts (+4dB)
This is approx. 35 dB gain in the FET amp.

I'm definately not getting 35dB gain (at least from my calculations - 20.67 to -10.10 is around 10dB).

Does this help anyone to provide more suggestions where the problem might be? Are there any other tests I could do with the same setup that will help, or any voltages (other than the ones posted below?)

Thanks!

tim

 

[saxtim]

I should comment that the scale on the test pictures - it's not a dB or voltage scale. Its marked as percentage.

tim

 

[tk@halmi]

The signal out of the FET amp looks really bad. Remove the transformer from its output and see if you still get that sort of mangled sine wave.

 

[saxtim]

Hope everyone is having a good festive season.

I've just come back to this after a few days off and I have a little progress to report.

The signal out of the FET amp looks really bad. Remove the transformer from its output and see if you still get that sort of mangled sine wave.

Ok - I disconnected the transformer and repeated the process I described a couple of post earlier, with the one exception that I set the software mixer to 0 this time, not -10db.

Anyway, this is what I got:

at the input to the Fet amp:

http://www.geocities.com/mejafu/test1a.jpg

(copy and paste in new window to open)

I got a peak of -39.97 dB

at cap 110, output of Fet preamp I recorded this:
http://www.geocities.com/mejafu/test2a.jpg:

Peak of -18.93dB

To summarise, since disconnecting the transformer:

-I have a clean looking sine waving coming out of the Fet preamp
-around twice the gain (approx 21dB compared to 10dB with the transformer - still short of 35dB suggested by Bo Hansen though)

Looking back at the figures from when I tested with the transformer attached - I see I had the software mixer set to -10.10dB (everything else was the same), however the dB readings I got in circuit (-20.67 at input, -10.10 at ouput) are louder even though I was recording the signal with less volume - that must also mean something?

So - any suggestions to what this all means? Where I should start looking next?

thanks
tim

 

[Bo Hansen]

Saxim/Tim,

If you use a computer soundcard as a "oscilloscope" you have a input that not have enough high resistance to use in testpoints as the T1 FET transistors Drain (the C110 connection) the level will be attenuated because the soundcard input will load down the point you measure.

If you whant to use the soundcard input you must build a impedance converter to connect before the soundcard mike or line input to get a hi-Z input like a oscilloscope or a AC millivolt meter input.

This is very simple, it is only a FET transistor connected as a source follower with 3 resistors, 2 capasitors, 9 volts battery and a small metall box with RCA connectors in and out.

--Bo

 

[saxtim]

Bo,

thanks - I'll give this a go.

This is very simple, it is only a FET transistor connected as a source follower with 3 resistors, 2 capasitors, 9 volts battery and a small metall box with RCA connectors in and out.

Can you point me in the direction of a schematic with parts/values?

In terms of the soundcard loading the signal down - this may make sense in explaining the following - I tried measuring C110 with my AC range on my DMM again today. When I first did this some days ago I got no reading, mostly likely due to the distorted sine wave when the transformer was connected. Since I disconnected the transformer and now I'm getting a sine wave at the output, I tried measuring again with the DMM.

I got 1.5V at C110 - about what is expected (if not a little higher). This was with my meter on 20V AC range. However, when I switched my meter to 2V AC range, I consitently got readings of .7V. I don't know what to make of this, other than to think my meter (which is a cheaper model) maybe is not up to the task. User error could be a factor, however I got the 1.6V reading and the .7V readings (with the different ranges) consistently throughout the circuit, on more than one occasion, so I don't think it was a fluke? Is there an explaination for this?

Anyway, if someone could point me in the direction of a schematic for the impedance converter, I'll build one of those and see if I can get some more readings from my soundcard.

Thanks again,

tim

 

[rafafredd]

make sure you use a 60Hz to 400Hz wave when using a DMM to measure AC. at 1kHz, most models won´t work...

 

[saxtim]

make sure you use a 60Hz to 400Hz wave when using a DMM to measure AC. at 1kHz, most models won´t work...

Thanks - that makes sense now I think about it. I'll try and remeasure my voltages with a sine wave in that range and see if I get more stable voltages.

thanks

tim

 

[saxtim]

Ok, even with the sine wave at 120Hz I still get the same erratic readings when using different ranges if my DMM. I'm still getting around . 7V when I set it to 2 volts AC and around 1.5V when I set it to 20V AC (both readings taken at the same spot, output of the FET amp).

Guess I'll have to pursue the impendance converter and try again with the sound card. Anyone got a schematic and parts values for what Bo mentioned above?

thanks

tim

 

[Bo Hansen]

Tim,

Here is a simple FET impedance converter that you easy can put together.
It have only five components, two phono connectors and a battery, so you can build it around a solder tag strip and mount in a small metal box.
You can use a 2N3819 or any type of N-channel JFET transistor,
but it is very important that you adjust the 10 K preset resistor to get +5 volt on the FET transistor source.

The converter have a 1 Mohm Hi-Z input and a output that can drive a 50 Kohm load with +10 dB/2,5 volts output swing. (it can drive lower loads down to 10 Kohm but with lower output swing)

It will run from a normal 9 volts (PP3/6LR61) battery, and draw only 0,5 mA.

--Bo

 

[saxtim]

Bo,

Thanks for the schematic. I built it, however using it I'm still getting more or less the same readings when I record signals from in circuit as before.

I retried measuring with my voltmeter again though, I still get .6-.7V at C110 with the 2V range, however with the 20V range I get 1.50V and with the 200V range I get 1.3-1.4V, so maybe I can discard the 2V range reading as the other two readings are close to each other, and perhaps confirm the voltage being around 1.3-1.5V. In this case it looks like the FET amp is working ok and has plenty of gain.

I'm inclined to think this is the case not just because of the voltage measurements, but also because when I had the transfromer attached at the output and I recorded the sine test signal directly from the mic's output, the levels were fine.

Earlier in the tread I removed the transformer from the circuit as it seemed to be causing trouble. I've still got it removed at the moment, which allowed me to take the readings above. When the output transformer was wired up I got this:

http://www.geocities.com/mejafu/test3.jpg

This was recorded with a 1000Hz sine wave going in, and recorded directly from the mics output, through the transformer, with my mic preamp set to -6db (lowest setting). Plenty of level.

The signal before the transformer was also distorted when I recorded it with a probe.

Removing the transformer gave me a clean sine wave, and that's what I've been testing the last week or so, trying to ensure that the gain of the FET amp is correct.

So I guess the questions are:

1. Why does adding the transformer to the circuit distort the signal?
2. Does the answer to question one explain why I get no signal at all with the mic capsule connected?

The data sheet for the transformer is posted earlier in the thread, and I'm now sure that when I wired it up as red/brown as primary it was wired correctly. (This was how it was wired when I got the distorted signal above).

thanks

tim

 

[saxtim]

just bumping to see if anyone had any thoughts

thanks

tim

 

[Bo Hansen]

Tim,

The primary is red and brown, (red connected to C110 and brown to ground)

The CineMag CM-2480 seem to be ok to feed from a single FET transistor drain output.

If you disconnect the transformers primary red and brown and instead for the FET amp connect a oscillator with 0 dB/1 kHz sine tone, and if everything is ok, you shall have aprox. -20dB on the secondary side.
If not, you have a short between the secondary and the XLR-connector or some wrong inside the transformer.

-Bo

 

[saxtim]

Latest update:

I removed teh 2n3819 and placed a 2sk170 in its place and reattached the transformer.

I know get a clean sine wave with transformer attached through to the output of the mic (from the transformer). Plenty of level is available at output with a 30mV input to the FET amp.

Gus has told me that I need to get a FET to bias between 2.7-4.7k for the circuit. To bias the 2sk170 I only needed 1K though.

With capsule attached I now get some clean signal from the mic (compared to none at all) but output is still extremely low. I basically have to yell into the mic with my preamp on full to get a signal.

I know the capsule works as I had it in another circuit.

It seems that the FET is the problem. I'm not sure what caused me to get the distorted sine wave with the transformer attached before, maybe something to do with the source R being so high (10k).

Still - what I don't understand is that both instances, with the 2n3819 and the 2sk170, testing the head amp alone I get plenty of signal (even if the 2n3819 was distorted). Obviously adding the capsule changes things again?

I think the next step might be to find a different FET to bias around the right source R. I don't really want to buy 1000 2n3819 like Gus did - is there another FET that's more tightly spec'ed that will work well in this circuit? Preferably something I can get from Farnell.

Thanks

tim

 

[Bo Hansen]

Tim,

Have you test so the transformer is ok, as I explain for you in my last post ???

I have not ask you before, but I suppose and hope that you have check so all dc supply voltage is ok inside the microphone, the +33 volt on the D101 zener diode and that the voltage step up switch circuit for the polarization voltage is ok.
Because in otherwise we can not discuss FET transistor gain or bias point or the microphone output level before you are absolute sure that all dc voltage supply rails is ok.

You can easy find acceptable bias point for the 2N3819 FET, (or 2SK170) as following adjustment:
(but the best way is to measure the high level distortion on the output)

1. Instead for the source resistor, insert a preset resistor on 10 Kohm.

2. Connect a oscillator between C103 and ground without the capsule and polarization voltage feed, apply a 1 kHz sine tone and connect a oscilloscope to the microphone output between pin 2 and 3 on the XLR connector. (of course you have also apply +48 volts phantompower via two 6,8 kohm resistors to pin 2 and 3 and -/ground 48 volt to pin 1)

3. Turn up the oscillator level so the output signal start to clipping on the upper or lover sine end, adjust the source preset so you get the sine to clipping symetrical in both end, adjust the oscillator level and repeat the preset adjust so ju get highest output swing just before it start to clipping symetrical.
(the U87 clipping have not sharp edge, it is more like round tube clipping)

4. Disconnect the preset resistor and do a resistans measurment on it, take a normal resistor on same value and put it back as a selected source resistor.

The symetrical clipping is dependent of the output transformer secondary load that it will sense from different microphone preamps, so when you test the clipping behaviour without a microphone preamp you only have the two internal 2,2 kohm R115 and R116 phantom receive resistors and the two external 6,8 kohm phantom feed resistors as a transformer secondary load, (approx. 3,3 kohm total load) so you can do a compromise and connect a extra 2,2 kohm resistor between the XLR pin 2 and 3 to simulate different microphone preamps.

--Bo