Q: JFET swap in microphone
Martin Norberg
I want to put a 2SK170 instead of a 2N3819 in a condenser microphone.
The original circuit has 8,2 k between source and ground, 1G between
the gate and ground, and 10k from drain to supply voltage. I measure
1,4v at the source, gate at 0v and 15v at the drain. Supply voltage
comes from +48v phantom power through a voltage divider with 57k/47k
and also a 1uF decoupling cap.
Will the operating conditions work with a 2SK170? I have the GR
version on hand. I could modify the circuit to get the best
performance out of the new FET.
I'm not very good with solid state, have built a lot of stuff with
tubes though, so I know the basics. But I'd sure be glad if someone
who knows this stuff could give me a hint...
Martin
David Curtis
Could you possibly sketch out the schematic and email it to me? I'll do
a Spice simulation with both of the jfets and see what happens. I took
a look at the specs, and noticed a few things: The Vgs (off) for the
2SK170 is typically -.5, max -1.5, whereas that of the 2N3819 is
typically -3, max -8. You might need to adjust (decrease) the
gate-source bias voltage. Otherwise, the Idss looks about the same, the
gate current looks close, the 2SK170 has higher breakdown voltage; you
ought to be able to make it work, but you'll probably have to tweak a
little. You might try this: measure the quiescient drain current with
the 2N3819 in place, then put the 2SK170 in and decrease the source
resistance until you get the same quiescent drain current. That might
work. It's hard to say how much you'll need, even with a simulation to
guide you, since there is so much variation among individual devices.
Cheers,
Da5id
Martin Norberg
schematics at this adress:
http://web.telia.com/~u31427269/extra/New-2.jpg
It is a bit unclear, but the important parts are visible. The
schematics is common for two models, it is the DC-20 version I am
working on. Also, the 2*100M resistors are now 1G instead.
It seems that the current is very low, is this normal for microphone
use?
Martin
(switched to Agent newsreader for posting now)
On 23 Jan 2005 01:21:59 -0800, "David Curtis"
jwill...@audioupgrades.com
difference is amazing as the 170 is pinched off in the transients. It's
like after your ears pop on a long flight. Just adjust the source
resistor. A trim pot can make the selection quicker as all these fets
vary a bit in optimum bias levels.
Jim Williams
Audio Upgrades
Mark
Why do you want to change them...
Adjust the source resistor for the desired operating current and you
should be good to go, but what do you expect will be different?
Mark
David Curtis
than the 2N3819.
Phil Allison
"David Curtis"
>
>I don't know what Martin's reasons were, but the 2sk170 has less noise
> than the 2N3819.
>
** Err - that would only be apparent with low impedance sources - which a
condenser mic clearly is not.
............. Phil
jwill...@audioupgrades.com
3819 types. Self bias designs like AKG451 inputs use a voltage divider
off the + rail to set bias and this is a prefered method for these fast
305 fets. The real reason to swap to the 305 is you will hear more
detailed top end performance. These are very low noise as the noise
spec is optimized for very high input impedances and the best part is
the fets have extremly low leakage or stray capacitance when used at a
very high input impedance. (I use them with 10 gig ohm input
impedances). 170's stray off top end details due to die leakage. Try it
and see, don't believe me!
Jim Williams
Audio Upgrades
David Curtis
Can you help me understand why? Is it because the device's
input-referred voltage noise is eclipsed by the Johnson noise of
resistors at the input?
I've recently been working with noise models for condensor mic input
stages, and making some measurements. Models suggest that Johnson noise
washed out the voltage noise of the input device at low frequencies,
but as frequency increases, the device voltage noise predominates. The
frequency at which this crossover happens depends (of course) on
several parameters. So far, my measurements (though crude) seem to
confirm this. I may be missing something...
[This isn't intended as an argumentative challenge, merely a request to
dispell my ignorance]
Thanks,
Da5id
Pooh Bear
Mark wrote:
2SK170 is quieter ( and could work with adjustment of source and drain Rs )
but a quick glance at the data sheet suggests a DC operating point of Vgs
= -0.25V is about the best you'll get. Cutout is around 0.5V typical.
As Jim Williams says, this'll restict the dynamic range it can handle
compared to the 2N3819. 0.5V pk-pk at the gate compared to 1.4V pk-pk.
Approx 10dB degradation in level handling.
Graham
Martin Norberg
puts out in terms of voltage? Could I measure that with my scope (Tek
310A) and the right probe, or is it too sensitive? I also have an AC
millivolt meter, HP 400D, maybe that has a high enough impdance input
as it is tubed.
Martin
Phil Allison
"David Curtis"
** Reversing the top posting !!!
>>>I don't know what Martin's reasons were, but the 2sk170 has less noise
>>> than the 2N3819.
>>
>> ** Err - that would only be apparent with low impedance sources - which
>> a
>>condenser mic clearly is not.
>
>
> Can you help me understand why? Is it because the device's
> input-referred voltage noise is eclipsed by the Johnson noise of
> resistors at the input?
>
> I've recently been working with noise models for condensor mic input
> stages, and making some measurements. Models suggest that Johnson noise
> washed out the voltage noise of the input device at low frequencies,
> but as frequency increases, the device voltage noise predominates. The
> frequency at which this crossover happens depends (of course) on
> several parameters. So far, my measurements (though crude) seem to
> confirm this. I may be missing something...
>
** A 1 Gohm resistor certainly does generate a lot of *white* noise -
around 500 uV in a 20 kHz bandwidth.The mic capsule's capacitance will
attenuate this noise by creating a 6 dB / octave LP filter starting at some
sub sonic frequency - the more the capsule capacitance the greater the
attenuation at any given frequency.
As you said, at some frequency the attenuation of the resistor noise is
sufficient to bring it below the inherent FET voltage noise level. This
will typically happen at a frequency around 3 to 5 kHz and above that
frequency FET (white) noise will still remain. However, below the
"crossover" frequency you are left with a lot of white noise filtered by 6
dB /octave or "red noise" - ie much deeper sounding than pink noise.
In most cases, the total *audible* noise energy in the band below this
"crossover" frequency exceeds the energy in the band above. The higher the
"crossover" frequency the more so this will be the case.
Condenser mics typically have a softer sounding hiss, due to all the
additional low frequency energy, compared to dynamics with their simple
*white* noise characteristic.
............ Phil
Pooh Bear
> Thanks for your great ideas. Hmm, I wonder what the capsule actally
> puts out in terms of voltage? Could I measure that with my scope (Tek
> 310A) and the right probe, or is it too sensitive? I also have an AC
> millivolt meter, HP 400D, maybe that has a high enough impdance input
> as it is tubed.
> Martin
Not enough input impedance either way.
You can easily measure the voltage on the fet source though. You'll need to
remove the C bypassing the source to gnd R and use the stage like a source
follower to see the volts on the gate.
Oh, I meant cutoff - not cutout in my previous post btw !
Having to put my answer here shows how much more useful it is to 'bottom post'
than top-post too for your consideration too !
Graham
Martin Norberg
<rabbitsfriend...@hotmail.com> wrote:
>Not enough input impedance either way.
>
>You can easily measure the voltage on the fet source though. You'll need to
>remove the C bypassing the source to gnd R and use the stage like a source
>follower to see the volts on the gate.
I think I'm with you. Is it like with tube cathode followers, that you
get a gain of a little less than 1 in this configuration?
>
>Having to put my answer here shows how much more useful it is to 'bottom post'
>than top-post too for your consideration too !
>
>
Lost you there. Maybe it depends on the newsreader? In agent 1.9 it is
displayed as a tree, and I just post my follow-ups to the last
speaker. Are you saying that is not how you do it anyore?
Martin
Martin Norberg
On Wed, 26 Jan 2005 12:01:57 +1100, "Phil Allison"
<phila...@tpg.com.au> wrote:
>** A 1 Gohm resistor certainly does generate a lot of *white* noise -
>around 500 uV in a 20 kHz bandwidth.The mic capsule's capacitance will
>attenuate this noise by creating a 6 dB / octave LP filter starting at some
>sub sonic frequency - the more the capsule capacitance the greater the
>attenuation at any given frequency. <a lot of good information snipped>
>
So my small diaphragm capsules might not have enough capacitance to
attenuate the noise? What role does the DC blocking C in series with
the capsule play in this - setting the LF rolloff, I guess, but does
it affect noise?
Another thought, would varying the polarizing voltage have a chance of
improving things? I have an adjustable DC source in the preamp.
I'm beginning to see that improving old mikes isn't just swapping a
FET or two. :)
Martin
Phil Allison
"Martin Norberg"
> Now this was interesting.
"Phil Allison"
>>** A 1 Gohm resistor certainly does generate a lot of *white* noise -
>>around 500 uV in a 20 kHz bandwidth.The mic capsule's capacitance will
>>attenuate this noise by creating a 6 dB / octave LP filter starting at
>>some
>>sub sonic frequency - the more the capsule capacitance the greater the
>>attenuation at any given frequency. <a lot of good information snipped>
>>
>
> So my small diaphragm capsules might not have enough capacitance to
> attenuate the noise? What role does the DC blocking C in series with
> the capsule play in this - setting the LF rolloff, I guess, but does
> it affect noise?
** Go figure what two Cs in series does.
> Another thought, would varying the polarizing voltage have a chance of
> improving things?
** Yep - the more V the better.
But watch out for weird noises.
>
> I'm beginning to see that improving old mikes isn't just swapping a
> FET or two. :)
** Indeed.
.............. Phil
Pooh Bear
Martin Norberg wrote:
> On Wed, 26 Jan 2005 04:26:06 +0000, Pooh Bear
> <rabbitsfriend...@hotmail.com> wrote:
>
> >Not enough input impedance either way.
> >
> >You can easily measure the voltage on the fet source though. You'll need to
> >remove the C bypassing the source to gnd R and use the stage like a source
> >follower to see the volts on the gate.
>
> I think I'm with you. Is it like with tube cathode followers, that you
> get a gain of a little less than 1 in this configuration?
Exactly so.
> >Having to put my answer here shows how much more useful it is to 'bottom post'
> >than top-post too for your consideration too !
> >
> >
> Lost you there. Maybe it depends on the newsreader? In agent 1.9 it is
> displayed as a tree, and I just post my follow-ups to the last
> speaker. Are you saying that is not how you do it anyore?
It's a question of where you type the text. I've rarely used Agent but I'm sure
it's not significantly different to any other newsreader in this respect.
In short - you should place your 'follow-up' *below* ( or interspersed with ) the
original text so as to create a 'flow of thought'.
Graham
Martin Norberg
electronics suppliers in my town. Are there any online sources thet
would be willing to sell me just a few? I'm not involved in the
industry, so I don't think the manufacturer would send me any samples.
Martin
bruce seifried
Martin Norberg <svar...@hotmail.com> wrote:
Try Mouser Electronics. I just placed an order for some; they have them
in stock. Made by Vishay.
bruce seifried
eclair engineering
Martin Norberg
a General Radio bridge, the capsule taken out of circuit) and found it
to be around 150 pf. Is that normal for condenser mikes?
Martin
Martin Norberg
shielded connecting leads, which was as high as 125 pf. So measured
capsule capacitance would be 25 pf. So, my question again: is *this*
normal for a small condenser capsule?
Martin
Phil Allison
"Martin Norberg"
> Correction, sorry. I forgot to subtract the capacitance of the
> shielded connecting leads, which was as high as 125 pf. So measured
> capsule capacitance would be 25 pf. So, my question again: is *this*
> normal for a small condenser capsule?
>
** Yep - 20 to 30 pF sounds right, large a capsules run to 50 to 80 pF.
............... Phil
Martin Norberg
condenser microphones. Sorry to see that nobody seems to have hands-on
experience with fixing noisy vintage mikes, but it's no big deal. I'll
just experiment with FETs R's and C's and see where it leads. Perhaps
a pencil tube instead of the FET if I get really crazy :)
Martin
Richard Crowley
Did you see Scott Dorsey's article on upgrading the Oktava 012
microphone (in Recording magazine)? There are also many
schematics for tube/valve, FET, and RF microphones out there
on the internet.
Martin Norberg
Now I have, very interesting.
I have also tried 60v as phantom power (taking care not to exceed the
ratings of the caps in the mic) with great result: Much better s/n,
headroom, everything...
But looking at some recent dicussions here I realize that this might
not be the best place to talk about signal to noise... Silly, I just
got here...
*aiming killfilter*
Martin
JP Gerard
they just might but I wouldn't trust them to...
Also, make sure you don't polarise your FET with more than about 30VDC
between terminals, or you'll eventually cook the thing.
I haven't found a FET that stands more than 30VDC but they could exist of
course...
My main worry would still be the capsule... I don't have the schemo handy,
is there a Zener reg. in there for the capsule or semiconductors supply
rails?
JP
"Martin Norberg" <svar...@hotmail.com> a écrit dans le message de
news:bhv6019oufn5qfd3i...@4ax.com...
Scott Dorsey
>Errr... I don't think the Oktava MK012 capsule will handle 60VDC very well -
>they just might but I wouldn't trust them to...
>
>Also, make sure you don't polarise your FET with more than about 30VDC
>between terminals, or you'll eventually cook the thing.
>I haven't found a FET that stands more than 30VDC but they could exist of
>course...
Not if you want low noise and low input capacitance anyway. Although getting
ANY kind of high voltage jfets is a real problem these days. There are some
switching parts that go up to around 40V but I can't find anything that will
go much higher.
>My main worry would still be the capsule... I don't have the schemo handy,
>is there a Zener reg. in there for the capsule or semiconductors supply
>rails?
Nope. Just resistors. So as the phantom voltage changes, the transistor and
FET biases change and so does the polarization.
--scott
--
"C'est un Nagra. C'est suisse, et tres, tres precis."
JP Gerard
and
> FET biases change and so does the polarization.
> --scott
I'm looking at the 012 schemo now... I'd forgotten how basic this thing is -
clever though!
No regulation whatsoever, you're right... ugh...
JP
Scott Dorsey
>> Nope. Just resistors. So as the phantom voltage changes, the transistor
>and
>> FET biases change and so does the polarization.
>
>clever though!
>No regulation whatsoever, you're right... ugh...
It's kind of clever, but it really doesn't have much output drive, and of
course the output is actually unbalanced. But hell, for two transistors and
no iron it's pretty impressive.
Martin Norberg
wrote:
>Errr... I don't think the Oktava MK012 capsule will handle 60VDC very well -
>they just might but I wouldn't trust them to...
Hehe, I guess my original post got lost in the noise... I'm not
working on an Oktava, but an old (almost classic) PML DC-20 - Small
diaphragm, 2N3819 FET and transformer output. 1969 vintage or so, a
bit like a KM84 I guess. Was terribly noisy, so I wanted to try
another FET. I've ended up changing most carbon resistors and tantalum
caps for modern types, but I still get the best gain and 'sound' with
the original 2N3819. The noise floor is a lot lower now though, after
the old caps are gone.
The FET is fed from a voltage divider, so in reality (with a 60v
supply) it has 20v on the drain. 60 is just 12 volts over 48,
shouldn't be enough to fry the capsule? Or...?
Schematic of the PML (Pearl) is here:
http://web.telia.com/~u31427269/extra/New-2.jpg
Scott, great article! How about cramming a tube in the first stage?
Does it have to have really high gain or can a triode connected
pentode (like an EF86) work there? I was thinking of getting one of
the Russian or Chinese mics just to try...
Martin
Scott Dorsey
>
>Scott, great article! How about cramming a tube in the first stage?
>Does it have to have really high gain or can a triode connected
>pentode (like an EF86) work there? I was thinking of getting one of
>the Russian or Chinese mics just to try...
EF86 is way too big to fit into the case of a pencil mike. The triode
connected pentode thing will work, though, with a submini like a 5899
or a nuvistor, although the pentodes will tend to be more microphonic
than a triode with the same gain.
Martin Norberg
>EF86 is way too big to fit into the case of a pencil mike.
Yeah, I was thinking of getting something like a large Oktava MK319 or
a cheap ADK or something for that experiment.
On the other hand, I am staring at a cople of 5718 pencil triodes that
I found in some junked lab machine. They're really small! Input
capacitance only 2,5 pf and a gain of 27... But they were probably not
made for audio use, so who knows if it would work. Anyway, converting
a solid state mic to tube would involve building a special power
supply and sorting out a lot of things. But it would sure be
interesting to try :)
Martin
JP Gerard
Even if the FET will most likely be fine (there's a potential divider
feeding the drain so even a 60VDC at the transfo center tap would be OK),
the capsule could - possibly? - see its diaphragm collapse with 60VDC
applied.
You could either change the potential divider to run the FET at a higher V
and keep the capsule at 48VDC from a +48 Phantom line or lower your 60VDC
supply to 54 VDC, which is within tolerance, without changing anything
inside the mics.
These look like good mics, don't kill the capsule.
JP
"Martin Norberg" <svar...@hotmail.com> a écrit dans le message de
news:l5v701ti8qto12bm8...@4ax.com...
Scott Dorsey
>
>On the other hand, I am staring at a cople of 5718 pencil triodes that
>I found in some junked lab machine. They're really small! Input
>capacitance only 2,5 pf and a gain of 27... But they were probably not
>made for audio use, so who knows if it would work. Anyway, converting
>a solid state mic to tube would involve building a special power
>supply and sorting out a lot of things. But it would sure be
>interesting to try :)
You want more gain than 27, I think. Look at the 5899, which is about
the same size. It's still available NOS for reasonable prices, and it
worked well enough in the B&K mikes. It _is_ very microphonic, though.
Len Moskowitz
Martin Norberg <svar...@hotmail.com> wrote:
>I have also tried 60v as phantom power (taking care not to exceed the
>ratings of the caps in the mic) with great result: Much better s/n,
>headroom, everything...
DPA uses 130 Volts in their top-of-the-line mics. And Bruel & Kjaer
(DPA's precursor company) uses 200 Volts for their measurement mics.
--
Len Moskowitz PDAudio, Binaural Mics, Cables, DPA, M-Audio
Core Sound http://www.stealthmicrophones.com
Teaneck, New Jersey USA http://www.core-sound.com
mosk...@core-sound.com Tel: 201-801-0812, FAX: 201-801-0912
Kurt Albershardt
> Martin Norberg <svar...@hotmail.com> wrote:
>
>
>>I have also tried 60v as phantom power (taking care not to exceed the
>>ratings of the caps in the mic) with great result: Much better s/n,
>>headroom, everything...
>
>
> DPA uses 130 Volts in their top-of-the-line mics. And Bruel & Kjaer
> (DPA's precursor company) uses 200 Volts for their measurement mics.
...Which can be nicely powered from P48 by the Josephson C617 amplifier body.