XFMR output impedance

[Samuel Groner]

Hi

Working on some sims for a mic pre with a JT-13K7-A; I noticed that with the 2SK389 there is considerable variation in high freq. response with different source impedances. I wonder how the output impedance of a mic input transformer looks like at freqs above a few hundred kHz? I assume pretty low due to stray capacity? At least the usual equivalent circuits do say so...

Thanks!
Samuel

 

[Samuel Groner]

Transformer experts, where are your statements? OK, one :guinness: for ya..!

Samuel

 

[CJ]

I just tear em up, I don't design the darn things.

 

[PRR]

Capacitance is proportional to size, and typically a few hundred pFd for a mike transformer.

Leakage inductance is proportional to winding inductance and thus bass impedance and nominal working impedance (times how much the windings are interleaved).

The two produce a resonance that drops in frequency as secondary nominal impedance rises. A HiFi "50K" winding may be ringy at 20KHz. Almost any few-hundred ohm winding wound directly on the primary will ring above 50KHz.

Above resonance, yes it is capactive. But the real question is how much does the triode or FET's grid/gate capacitance add to transformer stray capacitance, how much does it shift the resonance, and where does that leave you? A 1K winding tolerates some more capacitance without any harm in the audio band. A 50K winding may be marginal to begin with, and suck badly with a fat FET gate hung on it.

Since FETs have zero noise current, noise figure improves with secondary impedance. But the frequency response will suffer. With small clean FETs, 5K or so may be a good compromise, low NF and bandwidth past 20KHz. With the fat 389, you may want a lower impedance to preserve frequency response, but the low voltage noise means you might still get good NF. (Note that over some range of quality, the small FET at hi-Z gives the same result as a fat FET with lower-Z, and maybe a penny cheaper. That's why you never see massively parallel grids, but rather one grid working at low current lightly loading a hi-Z winding.)

 

[Samuel Groner]

Thanks for the answer, PRR.

So you think I might better add a cascode to my input transistor if I want to use the given transformer?

My first Q was more related to stability of the first active stage; response above 1 MHz varies considerably with source impedance. Beeing aware that SPICE does not tell the hole truth, I still think it is a good idea to get as close as possible. If I understud your post correctly, a low (say 1k) source impedance should be well suited to model a worst-case response (stability gets worse with higher impedances).

I guess the cascode would help stability as well?

Samuel

 

[bcarso]

Cascoded or even bootstrapped drains will help a lot, if as well the gate-source C is being bootstrapped with enough negative feedback. You will get at some point a negative input C, but that may be swamped out by the transformer's capacitance. Beware of parasitics that will allow the devices to scream at a few hundred MHz.

 

[PRR]

> add a cascode to my input transistor if I want to use the given transformer?

No idea. What is this given transformer? What are its parameters? Where do you want to go today: 15KHz or 150KHz?

 

[Samuel Groner]

I mentioned the JT-13K7-A and that's the one I want to use. I would be happy if I had not to deal with custom zobel networks, so the input C should be low enough to allow performance given in the datasheet with the standard loading.

I have small drain resistors already. Can I use bipolar stuff for the cascode? Probably a silly question, but I did not study cascoding that well yet...

Samuel

 

[bcarso]

Use PNP common-base stages and you should be a happy camper imo---if you are convinced that you really need that low an input C to begin with. I don't know what the circuit looks like right now (unless it is the same as the 389 input op amp you were developing earlier).

 

[Samuel Groner]

unless it is the same as the 389 input op amp you were developing earlier

It's basically the same topology, with a few add-ons. I'm not happy with it yet and I want to build the prototype of the original version first. For sure I'll post the design sometime.

Overall project is a mic pre with the mentioned Jensen input, then the opamp I'm talking about now as a first stage with 6 to 42 dB gain, and then my first opamp in inverting configuration as output driver (into JT-11-BMCF) with 2 to 10 dB gain.

Samuel

 

[Samuel Groner]

PNP common-base

Wait, what's that? Can you tell my where to connect what? Drain to collector? And what with the base? Some bias?

Samuel

 

[PRR]

I am so sorry I missed the mention of JT-13K7-A, and sorry again that I do not have all the Jensen part numbers and data memorized, and sorry that I would rather chew data than run around the Internet finding info that you probably have at hand.

Call Jensen. They charge enough for those little lumps, they should be able and happy to tell you how much capacitance is "too much", based on what HF response you consider sufficient(?).