Kajiu/SII - Optimizing Parafeed cap value for planar phones?

[Deke609]

I'm intrigued by a 2011 post from PJ in a thread that discusses the ability of the S.E.X. (pre- 32 ohm version) to drive the LCD-2 or HE-6. Here's what PJ wrote:

[snip] Does anyone have a link to impedance measurements on these planar phones? I'd expect, if they are low efficiency, they might have very flat impedance, in which case the parafeed capacitor can be optimized for that specific load.

I use my SII's and Kaiju with my LCD-4's which have a flat 200 ohm impedance across all frequencies. Each amp is wired for 16 ohms output impedance. And in case it matters, I'll also mention that I am feeding the amps with a Beepre.

@PJ: are there any tweaks to the value of the parafeed caps that you would recommend trying?

many thanks, Derek

[Raymond P.]

When you said optimize, I assume you want to know how para cap values affect frequency response. I've have attached some plots with different cap values: 1u, 2.2u, 3.3u, 4.7u, 5.6u, 8.2u. Load is a 200 ohm resistor.

As you might have guessed, the values mostly affects the bass region. (The 2nd graph zooms in on this region.) The 1uF value is the blue line (with the large hump.) The 8.2uF value is the dark green. You can probably see the trend with increasing cap values.

These plots are from a LTSpice simulation I had setup when I was planning to convert an SII for 45 tubes.

[Deke609]

Many thanks Raymond!

When you said optimize, I assume you want to know how para cap values affect frequency response.

Actually, I was just parroting PJ's use of the term.    I wasn't sure (and still am not) whether PJ was referring to just frequency response or something else (e.g., power transfer as a function of impedance? - seems doubtful to me, as I think that would mostly be governed by the primary/secondary ratio of the OT, but I don't know enough for this to be more than a guess on my part).

I've have attached some plots with different cap values: 1u, 2.2u, 3.3u, 4.7u, 5.6u, 8.2u. Load is a 200 ohm resistor.

As you might have guessed, the values mostly affects the bass region. (The 2nd graph zooms in on this region.) The 1uF value is the blue line (with the large hump.) The 8.2uF value is the dark green. You can probably see the trend with increasing cap values.

These plots are from a LTSpice simulation I had setup when I was planning to convert an SII for 45 tubes.

These are cool. Did the curves change much going from matched impedances of secondary and speaker to mismatched 16R/200R?

many thanks, Derek

[Raymond P.]

These are cool. Did the curves change much going from matched impedances of secondary and speaker to mismatched 16R/200R?

Cpara = 3.3uF
Blue = 16 ohm load
Red = 200 ohm load
The 16 ohm load has about a 4db lower gain. The lower gain makes sense since 16 ohms presents a lower impedance on the primary side to the parafeed cap. The hump is also reduced. I'm not sure what's creating that hump. It's probable some resonance going on created by the RCL network and larger values of C pushes it down to lower frequencies.

[Deke609]

Many thanks again Raymond!

This brings me back full circle wondering whether the 200 ohm impedance of my cans warrants some parafeed cap value tweaking. Another way of stating the question: is PJ's rule of thumb that the parafeed cap value should lie somewhere between 0.5X and 2X 2L/(R*R) based on an assumption that the OT output impedance will match the load impedance? And if so, what kind of a tweak might be in order?

cheers, Derek

[Paul Joppa]

Sorry for the late post, I thought I had answered but it didn't "take".

My post (from 9 years ago!) was off the cuff, and I had expected something close to speaker impedance, like 30-40 ohms. I don't think it applies to 200 ohms.

I have not yet found a satisfactory analysis for unloaded outputs (and 200 ohms on a 16-ohm tap is essentially unloaded). What you have is more akin to a transformer-output preamp.

I'll suggest that you load the amp with the resistance it's set for - 16 ohms? - in parallel with the 'phones, and see whether you hear any difference. When the theory is lacking, experiment is the place to start.  :^)

I'll probably attack the theory again someday, but it won't be soon.

[Deke609]

Many thanks PJ. I think I tried that with one of my SII's sometime and don't recall hearing a difference. But that was before the Beepre and Kaiju - so maybe I'll hear something (or convince myself that I hear something) with the more resolving BP/K setup. I'll try it this wknd if I have suitable resistors on hand - I assume I want something non-inductive? Or doesn't it matter?

cheers and thanks, derek

[Paul Joppa]

At 16 ohms, it's hard to get much inductance. More important, it should have a good power rating, since it will dissipate 12 times as much power as the headphones ...

[Deke609]

Perfect. Thanks PJ. I'll try my 50W 16R dummy loads that I use for breaking in new caps/tubes. They get slightly warm if I run hours of deep bass through them at high volume, but never hot.

[EricS]

Thanks for making and sharing these graphs, Raymond!  The parafeed output cap graphs don't surprise me, but it's good to see actual data.  I'll be doing some measurements with the parafeed cap in my 300B build soon.  It seems the bigger (and more expensive) caps always work better in places like this 

[Paul Birkeland]

It seems the bigger (and more expensive) caps always work better in places like this 

The graphs do also show what happens when you go too big!