@John & Scott: remember the discussion about mike preamps? Last week I brought it up in a talk with Joerg Wuttke, chief designer at Schoeps. He looked at me with pity: "the limiting factor in mike noise nowadays is the thermal bombardment of random air molecules on the diaphragm. Preamp or buffer noise is less than that and no longer something we worry about"
So, do we worry about the wrong thing?
Jan Didden
So, do we worry about the wrong thing?
Jan Didden
/OT
Last week I was spending my holidays on Crete. Driving a car over there is really fun, totally anarchistic. Cretans ignore every traffic rule. So you must be very alert and don't be amazed when you see a bicyclist, jogger or even someone in an electric wheelchair on the shoulder of a highway. The positive side is that Cretans, at least in traffic, carefully keep an eye on each other. So if one wants to overtake, no matter if it's prohibited, in a curb or whatsoever, they get out of the way by driving on the shoulder and let you pass. Nice and gentle people.
Last week I was spending my holidays on Crete. Driving a car over there is really fun, totally anarchistic. Cretans ignore every traffic rule. So you must be very alert and don't be amazed when you see a bicyclist, jogger or even someone in an electric wheelchair on the shoulder of a highway. The positive side is that Cretans, at least in traffic, carefully keep an eye on each other. So if one wants to overtake, no matter if it's prohibited, in a curb or whatsoever, they get out of the way by driving on the shoulder and let you pass. Nice and gentle people.
AKG C214 Equivalent noise level 13 dB-A to IEC 60268-4 (A-weighted)
Signal/noise ratio (A-weighted) 81 dB
AKG C451B Equivalent noise level 18 dB-A Signal/noise ratio (A-weighted) 76 dB
AKG C414 Equivalent noise level 6 dB-A (0 dB preattenuation) (IEC 60268-4) Signal/noise ratio 88 dB
AKG C4000 Equivalent noise level 8 dB-A Signal/noise ratio (A-weighted) 86 dB
john curl said:That's the problem with models!Still, they can be useful, but I don't count on them, since I have solderless breadboards that can tell me about as much, with about the same amount of effort.
With regards to the comment about the Blowtorch. Most here will never hear a Blowtorch, nor should it be necessary for them to do so. Only 40 or so were ever made, and are spread throughout the world, usually owned by very wealthy people, who can also afford the same quality added components necessary for useful listening quality.
I have never driven a Porsche 959, but I can well imagine what it might drive like. I don't ever have to see one to understand its conception and performance.
That statement proves absolutely nothing about the product. Apart from your own opinion do you actually have any tangible evidence about the products' worthiness ??
Any pictures of it or the guts of it etc ?? Performance measurements etc ?? Any testimonies etc ?? Anything ??
janneman said:
This is true, you can direct people to the literature on this all you want and they keep thinking that there is some DIY JFET mod that will make a .60$ capsule as quiet as a 1" B&K instrument mic.
Many of my associates have experimented with fet inputs for condenser microphones.
First, the self noise of the capsule is a limiting factor, especially in the midrange. Directional mikes have more trouble with this than omni-mikes.
We have found that some early devices, with their low input capacitance, actually work better than some newer devices, with slightly lower input noise.
The second major noise contributor is the size of the gate bias resistor. This should be 1 Gig Ohm or more. This is important in order to short out the noise of the gate bias resistor, itself, and shows up as low frequency noise. This usually limits DC voltage across the input fet, due to leakage considerations.
There are some interesting devices that we have experimented with, such as the Sony 2SK152 that was designed for this sort of service. It is pretty darn good, but it does have some 1/F noise that could preclude it from use as a normal input device for typical amps with lower drive impedance.
First, the self noise of the capsule is a limiting factor, especially in the midrange. Directional mikes have more trouble with this than omni-mikes.
We have found that some early devices, with their low input capacitance, actually work better than some newer devices, with slightly lower input noise.
The second major noise contributor is the size of the gate bias resistor. This should be 1 Gig Ohm or more. This is important in order to short out the noise of the gate bias resistor, itself, and shows up as low frequency noise. This usually limits DC voltage across the input fet, due to leakage considerations.
There are some interesting devices that we have experimented with, such as the Sony 2SK152 that was designed for this sort of service. It is pretty darn good, but it does have some 1/F noise that could preclude it from use as a normal input device for typical amps with lower drive impedance.
john curl said:
I don't get it. if the preamp noise is already lower than the diaphragm noise, what is the purpose of your experimenting?
Jan Didden
janneman said:
Jan - a free download, "B&K Technical Review 1972-3" Yes that means 1972 (37years old) microphone noise derived from first principles, recommended R is 10G ohm BTW.
Jan, you have been blindsided. I worked with B&K, decades ago on this problem. Schoeps is a compromiser, when it comes to SOTA. I know this from real contact, and schematics.
Scott, I am the guy who convinced B&K in 1973 to change in input resistor of the 1/2" and 1" preamp at the time to 1gig or more. They may have known about it in the lab, but not on the production line. 10 gig is possible, but rarely practical. Even 100 gig is available.
For proof of this, I have in my possession actual measurements made by B&K dated 3/21/1974 that show the difference between an ordinary 2619 mike preamp, and a modified one to my specs, and even better measurements up to 100 gig ohm.
Scott, I am the guy who convinced B&K in 1973 to change in input resistor of the 1/2" and 1" preamp at the time to 1gig or more. They may have known about it in the lab, but not on the production line. 10 gig is possible, but rarely practical. Even 100 gig is available.
For proof of this, I have in my possession actual measurements made by B&K dated 3/21/1974 that show the difference between an ordinary 2619 mike preamp, and a modified one to my specs, and even better measurements up to 100 gig ohm.
john curl said:
Shoeps cut costs of transformers introducing nasty distortions in turn. I believe Scott's FET output version sound better; lodaded on well balanced input it may give much less distortions due to quadrature characteristics that cancel each other (theoretically).
john curl said:
What confuses, John? If Shoeps compromised an output stage that does not automatically mean that they are wrong about dominant sources of noises.
Joerg Wuttke is a highly respected member of the technical community it does not aid your cause to say these things.
Wavebourn, do the numbers, first you will find that low noise high capacitance JFETs' like 2SK170 buy you nothing on noise and second you will find that the seconds due to capacitance modulation are swamped by the distortion of the typical source follower/phase splitter. The distortion (I posted actual measurements a few weeks ago) can be trimmed to a -90dB null by simple bias adjustment on the JFET and this trim absorbs all the sources of seconds. Mr. Wuttke has confirmed that this is done on every production mic. The null is so sharp that I remain a little skeptical on how well a -90dB null will "age" but some improvement must be maintained.
B&K uses a different approach stressing ultra low (or negative) input C, there are good reasons for this and it makes good reading.
Being "analytical" instruments B&K mics are not liked by some recording engineers. I really have no comment on this or on the large number of engineers who won't use anything but Neuman mics with VF-14 valves.
Wavebourn, do the numbers, first you will find that low noise high capacitance JFETs' like 2SK170 buy you nothing on noise and second you will find that the seconds due to capacitance modulation are swamped by the distortion of the typical source follower/phase splitter. The distortion (I posted actual measurements a few weeks ago) can be trimmed to a -90dB null by simple bias adjustment on the JFET and this trim absorbs all the sources of seconds. Mr. Wuttke has confirmed that this is done on every production mic. The null is so sharp that I remain a little skeptical on how well a -90dB null will "age" but some improvement must be maintained.
B&K uses a different approach stressing ultra low (or negative) input C, there are good reasons for this and it makes good reading.
Being "analytical" instruments B&K mics are not liked by some recording engineers. I really have no comment on this or on the large number of engineers who won't use anything but Neuman mics with VF-14 valves.
Wavebourn, they already use 1 gig input resistance, I am pretty sure, but many do not. They could do better, but it would be difficult. Why don't YOU check out the noise of their capsules yourself?
I skinned the cat by another way, John.
The exercise was, to take original devices, add a transformer, and arrange such a way distortions will be introduced less (not compensated, because compensating the 2'nd order of a complex function you gain nothing sonically, especially when it is modulated by real signal's envelope).
Variant 1: original. Varactor between D and G, large swing on D in respect to G. Varactor between G and S is effectively bootstrapped: the same swing, but opposite, on S as on D.
Variant 2: Cascode on top. Varactor loaded on a naked diode, C of top transistor loaded on b-e junction of a PNP transistor. Feedback by voltage bootstraps Cgd, but Cge is loaded on a naked eb junction.
Variant 3: No cascode. Сds varactor loaded on be junction of PNP transistor. Higher voltage -> lower capacitance. Diode shunted by a resistor. Feedback by voltage bootstraps Cgs, and Cgd is loaded on a eb junction with resistor in parallel.
Tradeoff: higher Igsx, but it is shunted by the capsule.
Result: lovely sound, compared to harsh original.
Your preferences may be different, however. My cat, I did skin it my way.
The exercise was, to take original devices, add a transformer, and arrange such a way distortions will be introduced less (not compensated, because compensating the 2'nd order of a complex function you gain nothing sonically, especially when it is modulated by real signal's envelope).
Variant 1: original. Varactor between D and G, large swing on D in respect to G. Varactor between G and S is effectively bootstrapped: the same swing, but opposite, on S as on D.
Variant 2: Cascode on top. Varactor loaded on a naked diode, C of top transistor loaded on b-e junction of a PNP transistor. Feedback by voltage bootstraps Cgd, but Cge is loaded on a naked eb junction.
Variant 3: No cascode. Сds varactor loaded on be junction of PNP transistor. Higher voltage -> lower capacitance. Diode shunted by a resistor. Feedback by voltage bootstraps Cgs, and Cgd is loaded on a eb junction with resistor in parallel.
Tradeoff: higher Igsx, but it is shunted by the capsule.
Result: lovely sound, compared to harsh original.
Your preferences may be different, however. My cat, I did skin it my way.
john curl said:
I hear what I hear. Should I measure something I don't hear, if I don't write an article to impress readers, instead of impressing listeners? 🙂
One capsule is very noisy when it is wet. Dirty one... I have 2 more in my barn, I'll give them a try.
- Status
- Not open for further replies.