An amateurs attempt at cap upgrades

[TomsRhinoplasty]

So, as I am sure most of you will understand, I have recently slipped into a new binge swing of desire to mod and upgrade my setup.  It started after I recently installed my Amperex 7316 1959 Bugle Boy.  It sounds wonderful.  But, it left me chasing more.

Because of what I have read online about certain sonic qualities associated with different capacitor types (potentially a very misleading practice, I know), I have ordered a number of capacitors that I am going to attempt to wire in parallel for a hybrid approach at the output cap.

On the way to my address I have:

2 Nichicon Elko UPM2E101MHD 100uF 250V Electrolytic caps - BASS
8 Russian military 10uF 300V Paper in Oil caps - for the mids man (Over ordered these, I know.  Only planning to use 2 per channel)
2 Mundorf MCap EVO 3.9uF 450V Aluminum Oil caps - to highlight the highs

I think it will all fit, if I am crafty with it.  My only question is... is this actually worthwhile?  Like, I am excited and looking forward to experiencing the difference, but are these choices wrong?  Am I running a fools errand? 

Electrical circuits aren't the most intuitive to me.  I know that I can do it, but the uncertainty surrounding certain variables in the approach has me guessing if it will just work how I imagine.

From what I understand, the low, mid, and high frequencies should all be channeled automatically (with a parallel wiring) into the cap that best handles them, because of their respective impedances.  Is this true and actually going to work?

I feel like I just need to talk it out.

-preston

[Paul Birkeland]

It might help to quantify the impedances at the various frequencies of interest. 

At 1kHz, the 220uF caps are going to present about 0.7 ohms of reactance, and your 4 10uF/300V caps will be more like 4 ohms.  Sure, some current is going to flow through the 10uF caps, but not very much.

Similarly at 10kHz, the 220uF caps will present 0.07 ohms of reactance, and your 3.9uF cap will be more like 4 ohms.

While there used to be a lot of electrolytics that wouldn't handle high frequencies all that well, the specific caps you chose show an impedance of around 0.09 ohms at 100kHz, so it's going to be a challenge to force the signal current in the output stage to go anywhere else. 

Having said all that, of course try it and see what your ears like! 

[TomsRhinoplasty]

So, if I wanted to "handoff" more of the mids and highs to the PIO and Aluminum Oil caps, then I would need to get electrolytics with higher capacitive reactance?  Is that what you are saying?

And through that the other caps become the lower impedance path in the mids and highs?  Is there a downside to that?

-preston

[Paul Birkeland]

Yes, intentionally buying low performing electrolytic capacitors is in itself a questionable proposition.

You could just get one big film cap per channel.

[TomsRhinoplasty]

Is there a specific film cap that you would recommend?  Finding the "best" (I know that's highly subjective) caps with the right specs proved to be somewhat challenging to me.  Perhaps it had a lot to do with my limited knowledge combined with the enormous amount of options available online.

I also have some Nichicon UPM2E470MHD aluminum electrolytic caps from old stuff I've collected rated at 47uF and 250V.  Those in theory would handoff more of the mid range to the PIO caps right?

I guess all of my reading online amounted to me understanding that each cap type has its strengths - the electrolytic being best/most practical for bass, the PIO being best for obtaining warmer, bloomy open mids, and the film caps being best for those clear and transient upper mids - highs.


-preston

[Paul Birkeland]

You're in the general category, so you'd have to let me know which kit you're putting these in. 

There's no listed high frequency impedance for those alternate caps you listed, so you'd have to measure it to see how they perform. 

[TomsRhinoplasty]

I am putting them in a crack w/ speedball.

I still have some time before I receive the ones I've ordered, but once they come in I will measure and update.


-preston

[Paul Birkeland]

https://www.mouser.com/ProductDetail/Panasonic/EZP-V60107MTC?qs=OlC7AqGiEDkd9vetrBxybA%3D%3D

This is the easy option.  You can use black RTV from the auto parts store to just glue them to the chassis plate.

[TomsRhinoplasty]

Thank you for the recommendation! I am going to install the hybrid system I have coming and see how it sounds.  I was trying to intentionally avoid going straight up film capacitor for the entire frequency band because of the claimed "ultra transient, neutral, clean" sounds.  This amp is for fun and enjoyable listening, and I want to push it a little more into the warm and luscious area without sacrificing too much transience and space.

I honestly hate using language like that because of how imprecise and objective it is, but its all you can do to describe the sonic experience you might be after.  I prefer objective and reproducible/verifiable language when it comes to experimenting with technology and hardware like this, but that's just not the nature of it.

I got the small film caps yesterday.  Excited to get the rest!

You might find this interesting. I gave chatGPT the caps I have to play around with and asked if it could come up with some fully technical and deterministic way to understand what caps will process what frequencies and this is what it gave me (I'm pretty sure it guess and made up the ESR values):

Below is a numerical snapshot of how signal-current splits in the four hybrids you asked about.
The only thing that changes from one column to the next is which capacitors ( C ) you parallel on the same two turret lugs; the headphone load is kept at 300 Ω (HD-6XX).

0. Reference values used in the math
branch   type   C (µF)   ESR (Ω @ 120 Hz*)
A   Nichicon UPM 100 µF / 250 V   100   0.15
D   Nichicon UPM 47 µF / 250 V   47   0.20
C   MBGP-1 PIO 10 µF / 300 V   10   0.35
B   Mundorf EVO Oil 3.9 µF / 450 V   3.9   0.05

* ESR rises only a few % from 120 Hz to 10 kHz for these parts—good enough for a first-order estimate.

1. Four capacitor “stacks”
stack ID   parts per channel   total C (µF)   –3 dB corner into 300 Ω
S-1   1 × A + 1 × B + 2 × C   123.9   4.3 Hz
S-2   1 × A + 1 × B + 3 × C   133.9   3.9 Hz
S-3   1 × D + 1 × B + 2 × C   70.9   7.5 Hz
S-4   1 × D + 1 × B + 3 × C   80.9   6.6 Hz

All four keep bass roll-off well below the 20 Hz threshold for 300 Ω loads.

2. Current-share at three useful frequencies
(Percent of the instantaneous signal-current that flows through each branch)

stack                                           100 Hz               1 kHz                           10 kHz
S-1
(100 µF elec)                                   Elec 82 %           Elec 81 %                   Elec 77 %
                                                        PIO 15 %            PIO 16 %                    PIO 19 %
                                                        Film 3 %               Film 3 %                   Film 4 %


S-2
(+30 µF PIO)                                   Elec 75 %           Elec 75 %                   Elec 73 %
                                                        PIO 22 %            PIO 22 %                    PIO 24 %
                                                        Film 3 %               Film 3 %                     Film 3 %


S-3
(47 µF elec)                                  Elec 66 %           Elec 66 %                   Elec 64 %
                                                       PIO 29 %            PIO 29 %                    PIO 30 %
                                                       Film 5 %              Film 5 %                     Film 6 %

   
S-4
(47 µF elec + 30 µF PIO)          Elec 58 %            Elec 58 %                   Elec 58 %
                                                       PIO 37 %             PIO 37 %                    PIO 36 %
                                                       Film 5 %               Film 5 %                     Film 6 %

   



3. What the numbers say
observation   impact on what you hear
The electrolytic still carries the largest share everywhere—because its reactance is lowest.   It anchors deep-bass and keeps phase shift minimal.
Bumping PIO from 20 → 30 µF raises its share of mid-band current by ~6 %.   Upper-bass & vocals become a bit warmer / “woodier.”
Cutting the elec from 100 → 47 µF (S-3 / S-4) gives the PIO roughly double the current it had in S-1.   Noticeably more PIO tone without audible bass loss (f-3dB ≈ 7 Hz).
The film cap never dominates current numerically; its job is to give a very low ESR path above a few kHz.   It cleans transient edges and adds “air,” even though it only carries 4–6 % of the net current.

4. Is the exercise still worth it?
Yes—because ESR and dielectric absorption matter more than current-fraction alone:

Even a small diversion of mid-band current away from a voltage-dependent electrolytic into a linear PIO reduces low-level grain.

The film cap’s vanishing 0.05 Ω ESR damps the composite ESR at 5–20 kHz, sharpening attacks.

Dropping the electrolytic to 47 µF is the single most effective move: it keeps bass intact (> 50 µF total) while letting the PIO own ~200 Hz–3 kHz.

So, in practice:

If you…   You’ll get…
Stay with S-1 (100 µF + 20 µF + 3.9 µF)   Audible polish but still “stock‐ish.”
Add the 3rd PIO (S-2)   A touch more body in mids.
Swap to 47 µF elec (S-3)   Clear, noticeable bump in warmth & depth without bass loss.
Do both (S-4)   The fullest PIO character you can get before bass starts to soften on < 120 Ω cans.

That should give you a precise picture of how each combination will steer bass, mids and highs—and whether the trade-off in parts count and chassis space is worth it to your ears.

I haven't yet gone through to see whether or not it actually makes sense.  But, it kind of blows my mind that a software tool is available for me to run a quick first pass knowledge draft like this


-preston

[Paul Birkeland]

C   MBGP-1 PIO 10 µF / 300 V   10   0.35

I have to admit that I went a little cross-eyed trying to read all of this.  The reactance of a 10uF cap at 120Hz is 132 ohms, definitely specify "reactance" rather than ESR. 

[Doc B.]

I will be curious to know how many fingers the bass player appears to have if you use chatGPT to design your sound system.

[TomsRhinoplasty]

lolol definitely being used purely as a reference/resourcing tool and not the driving force.  Maybe it will send me into sonically uncharted territories that only the AI had access to before my inquiry. 

Or, maybe chatGPT is a hipster and only listens to old 8-bit video game soundtracks.


-preston