[Slightly different #2] Oktava MC/MK-012 circuit elegance ?

[clintrubber]

[quote author="Rossi"][quote author="PRR"]

It is a JFET source follower driving a PNP emitter follower. [/quote]

I'm relieved, cause that's what I always thought it was until this thread came up. :grin: [/quote]
Hey, that's not a valid statement anymore now - you had your chance to react but didn't until now :wink:

Kidding aside, will further read the recent messages, promises to be interesting info.

 

[PRR]

> Are you sure?

I was, until I looked again.

I could whine and moan about output pins being hidden in the top-left corner; but I was being stupid.

Sorry. It will happen again.

 

[clintrubber]

[quote author="PRR"]> The 1 uF cap makes it a push-pull.

No. It's not push-pull[/quote]

I'd say that if we consider this one here a push-pull then the Oktava is also one: the 1uF cap arranges a more or less identical drive for the PNP.

Connected its top-plate to JFET-gate i.s.o. to JFET-source as it does now is of course problematic w.r.t. adequate drive for the PNP-base and its connected resistors, but apart from that the connecting-point is not essential for the pp-label to be correct or not.

But there's indeed the 7k5, I guess the presence of this one made you say it isn't a pp.

The Gefell is not push-pull either. Triple Darlington,

Sorry, don't understand the triple.


BTW, note that I'm not after an argument about correct naming or something like that. I do enjoy the discussions though that are caused by (mis-)naming & such and that's where the real valuable info is (circuit analysis: what's going on in the less obvious circuits).

And we have that going here, so thread-intention realized I'd say, let's have many more like that :thumb:

Regards,

Peter

 

[Rossi]

I don't know, I'm still with PRR's assessment. In the push-pull example you just posted, the signal reaches both bases at the same time. In the Oktava, the signal goes to the FET, configured as a source follower, and from there to the PNP emitter follower output stage. Simple yet elegant zero gain circuit, if you ask me.

BTW. I did a sim with the 2N3819 and two different 2SK170 models (the ones from Linear systems, I can't seem to find an original Toshiba model). Bias points and (simulated) performance are almost identical. So one cool thing about this circuit is that it seems to be able to accept a great variety of components, which is probably a good thing for any Russian made mic.

One other thing I like about the circuit is that it doesn't require a zener. Probably a good thing for a small size mic. I recently found out that zeners can radiate quite some noise.

 

[clintrubber]

[quote author="Rossi"]I don't know, I'm still with PRR's assessment. In the push-pull example you just posted, the signal reaches both bases at the same time. In the Oktava, the signal goes to the FET, configured as a source follower, and from there to the PNP emitter follower output stage.[/quote]
I'd say the signal (as fed thru the JFET and cap) reaches the PNP-base still fairly instantaneous for the frequencies at hand.
But which doesn't mean that I state that the Oktava is a push-pull as we know it :wink:
Let's forget about the labels, the discussion about the stuff actually happening in the circuits far more interesting.

Simple yet elegant zero gain circuit, if you ask me.

Sure, like it :thumb:

BTW. I did a sim with the 2N3819 and two different 2SK170 models (the ones from Linear systems, I can't seem to find an original Toshiba model). Bias points and (simulated) performance are almost identical. So one cool thing about this circuit is that it seems to be able to accept a great variety of components, which is probably a good thing for any Russian made mic.

I don't know how the model-parameters are set for those two devices. I saw I_DSS-ranges are about identical and the '170 has a much smaller VT-range. Perhaps the typ values are put close and setting them to their extremes might reveal larger differences. But maybe the two different SK170-parameter sets you threw at the simulator already were doing exactly that ?

One other thing I like about the circuit is that it doesn't require a zener. Probably a good thing for a small size mic. I recently found out that zeners can radiate quite some noise.

Yes, the dislike for Zeners because of that is universal.
Yet I'm wondering how noisy it still can be after filtering: when a zener is used you see it usually in combination with lot's of uF & some nFs in // to filter. Applications allowing it could also use a cap-multiplier.

The numerous mic-mod stories on the web 'and I got rid of the zener and I heard a better stereo separation in the upper bass-frequencies' are often not very well founded I'd say though, people associate zeners with noisy (they sure can be) and replace and unsurprisingly it's an improvement, you know how tests can be (note I'm NOT hinting that your experiments are among these).
So I expect that only a part of these mods would be a real improvement because of the zener-swap/mod when further scrutinized.

BTW, according to Motchenbacher the noise-mechanism for Zeners above ~5V is different and 'worse' than for those below (<: Zener effect, >: avalanche breakdown & Zener effect, but avalanche dominant).
So use multiple stacked lower-voltage ones one would conclude from that. It'll increase BOM-cost, but replacing a ~12V zener with four ~3V ones is very feasible, at least for DIY.

Yet I'm wondering if it still all matters when adequate filtering is applied.

Regards,

Peter

 

[Rossi]

[quote author="clintrubber"]
I'd say the signal (as fed thru the JFET and cap) reaches the PNP-base still fairly instantaneous for the frequencies at hand.
But which doesn't mean that I state that the Oktava is a push-pull as we know it :wink: [/quote]

I think the point is that the two devices in your example are driven by the same signal at the same impedance. In the Oktava circuit the signal runs through two consecutive impedance converters.

I don't know how the model-parameters are set for those two devices. I saw I_DSS-ranges are about identical and the '170 has a much smaller VT-range. Perhaps the typ values are put close and setting them to their extremes might reveal larger differences. But maybe the two different SK170-parameter sets you threw at the simulator already were doing exactly that ?

I'm not thinking about the two 170 models so much as about the difference between the 2n3819 and the 2SK170(s). In a lot of circuit it makes a difference and the bias points change considerably. In the Oktava cirucits everything stays pretty much the same.

Yet I'm wondering if it still all matters when adequate filtering is applied.

I wasn't actually talking about that. I think you can pretty much filter out the electrical zener noise, but I found that a zener also radiates noise through air. So physical distance to the FET and the high impedance area is required. Preferably more than an inch or so. At least that's what I found.

 

[clintrubber]

[quote author="Rossi"][quote author="clintrubber"]
I'd say the signal (as fed thru the JFET and cap) reaches the PNP-base still fairly instantaneous for the frequencies at hand.
But which doesn't mean that I state that the Oktava is a push-pull as we know it :wink: [/quote]

I think the point is that the two devices in your example are driven by the same signal at the same impedance. In the Oktava cirucuit the signal runs through two consecutive impedance converters.[/quote]
'Unequal paths' wouldn't be a criterium imho.

Im not thinking about the two 170 models so much as about the difference between the 2n3819 and the 2SK170(s). In a lot of circuit it makes a difference and the bias points change considerably. In the Oktava cirucit everything stays pretty much the same.

But that could be because the used model-parameters for the 2N3819 & 2SK170 are close; their ranges overlap for various parameters.
So to be able to conclude on the amount of sturdyness of the Oktava circuit to handle component-spread, we need to know how much apart the used parameters were actually set for those simulations.

but I found that a zener also radiates noise through air. So physical distance to the FET and the high impedance area is required. Preferably more than an inch or so. At least that's what I found.

Cool, wasn't aware of that path. Would be interesting to compare a stack of more 'low-voltage' zeners to a single higher ones here, maybe the non-avalanching zeners spit less (see above). Anyhow, shielding zeners would be going to far and that distance is impractical. If possible, zeners out ! :wink:

Regards,

Peter

 

[PRR]

> But there's indeed the 7k5, I guess the presence of this one made you say it isn't a pp.

And 7K5 is very large. In most cases, larger than the load.

Simulate it. If the load current is 1mA peak, the lower device is supplying 0.9+mA of that. Probably 0.98mA, with the upper device supplying 0.02mA. So at best it is push-PULL, so very asymmetrical that we get no "push-pull advantage".