SEX change operation

[audioengr]

  Just for fun.

What I'm after here is some modding advice.  I have completed the SEX amp (with lots of mods going into it) and I'm simply blown away by the SQ.  It is approaching my reference ribbon system, which has never been challenged by headphones before.  I am using HE-400 with it.

I only have limited tube design experience.  What I'm interested in doing is improving the bass response by beefing-up the decoupling caps on the plates of the final stages.  They currently have 33uFd 450V caps for filtering and power decoupling.  Is there any downside to making these say 100uFd 450V like the buffer stages?

Also, I'm curious why this filter/decoupling is not right at the plates, but behind the chokes...

Thanks,
Steve N.

 

[Doc B]

Have you installed the C4S yet? That will improve the bass.

 

[audioengr]

Have you installed the C4S yet? That will improve the bass.

Not yet.  This is the best next step?

Why is the cap not on the plate, but behind the choke?

Steve N.

 

[Maxwell_E]

I'd go with the C4S next. For only 45 bucks it's a steal at twice the price. I feel it did more for my sound than the ClarityCap ESA's I put in the parafeed and coupling positions and it was cheaper to boot.

 

[audioengr]

Okay, C45 it is.  I actually used the same signal coupling caps that I use in my Overdrive DAC that I manucture, V-Cap CUTF and oil caps in parallel.  Sounds really sweet.

Steve N.

 

[jrebman]

Steve,

The cap is between the plate and the output transformer, not the choke.  This is a parafeed topology and the opts are ungapped and small in size as they carry not DC bias -- the plate choke abosorbs that and the parafeed caps (supplied with 1.5 uF) block the DC from  the OPT.  Much higher inductance on the  OPT this way (ungapped).  The formula for determining parafeed caps size is 2 * plate choke inductance in henries / primary impedance of opt ^2 (and in this case the impedance in the s.e.x. kit is 8k.  You can go half this or twiice this but beyond 2x you gain no benefit, but lower than target values tend to tighten up the bass but rolloff higher.  So, 100 uF wouldn't buy you anything that about 3.3 uF would.  In theory .

HTH,

Jim

 

[jrebman]

Steve,

I just re-read your message and I'm a bit confused.  I think I missed the boat on what you were asking about in my last post, and I don't have the  benefit of having a schematic, and I've only built the 2.0 version before (though I got started on my 2.1 version today.

Anyway, my explanation above relates to the plate choke/OPT circuit, not the ps decoupling or cathode bypass caps.

-- Jim

 

[Paul Joppa]

The 450v power supply cap is at what I think of as the end of the power supply. The plate loading choke is part of the parallel-feed output circuit - again, this is the way I think of it. It's all interconnected after all!

The other way to see it is to think of the power supply as a current source, not a voltage source. In that case, having the high-impedance choke in series at the end of the power supply is perfectly natural.  :^)

 

[audioengr]

The 450v power supply cap is at what I think of as the end of the power supply. The plate loading choke is part of the parallel-feed output circuit - again, this is the way I think of it. It's all interconnected after all!

The other way to see it is to think of the power supply as a current source, not a voltage source. In that case, having the high-impedance choke in series at the end of the power supply is perfectly natural.  :^)

I understand why the cap is located there, but I want to understand the function of the cap.  If its for decoupling, then making the value larger to stabilize the voltage for high di/dt makes sense.  If the cap is for a filter comprising the choke and cap as a resonant circuit, then the value might be critical.

 

[Paul Joppa]

Ah! OK, the 100uF/450v capacitor is a power supply filter. One of the virtues of parallel feed is that the plate choke presents a high impedance load to the power supply, reducing the slew rate in the power supply.

The 22uF/450v cap is another stage of filtering for the driver. In stock form, the relatively large plate resistor (150K) serves some of the function of a plate choke; the C4S increases that impedance greatly.

Nevertheless, bigger and/or better caps do make a difference in both places, just not as large a one as normal.

 

[Paul Birkeland]

If you put a 33uF capacitor from the plate of the output tube to ground, you will short all your signal to ground, thus getting no output.

When you say you're interested in improving the bass response, you might want to clarify what you're looking for.  The SEX amp is designed to provide flat frequency response, so if you want to boost the bass, you'd want to do some equalizing before the input of the SEX amp.

Increasing the 22uF capacitors before the driver supply plate load resistor will not have much if any influence on the output section of the amplifier, though it should be somewhat audible on the driver stage without the C4S modification.

Also, just some advice, doing repairs at Bottlehead periodically involves removing "improvement modifications" that lead to operational problems.

 

[audioengr]

If you put a 33uF capacitor from the plate of the output tube to ground, you will short all your signal to ground, thus getting no output.

I get this after examining the circuit again.

When you say you're interested in improving the bass response, you might want to clarify what you're looking for.  The SEX amp is designed to provide flat frequency response, so if you want to boost the bass, you'd want to do some equalizing before the input of the SEX amp.

Its not steady-state FR that I want to affect.  Its the dynamic current, di/dt at lower frequencies that I want to improve.

Also, just some advice, doing repairs at Bottlehead periodically involves removing "improvement modifications" that lead to operational problems.

Im not too worried about that.  I have over 30 years design experience and a EE degree.  I also modded other companies products for over 10 years for customers worldwide.  No operational problems yet.  Everything measures perfectly and looks good on my scope and other audio analyzers.

 

[Doc B]

The short answer is yes, you can put a larger value cap at the final node of the power supply. How much larger it will need to be before you hear a difference (if you hear a difference) is something you will have to work out - forgive me - empirically. The old rule of thumb was around 5X, John Camille used to propose more like a 10X increase. The only potential issue I see is that at some point you may stress the fuse with the startup current inrush.

Material may have an impact as well, that is, film rather than electrolytic. Obviously available real estate for a beer can sized film cap becomes an issue.

I suspect that the C4S may give you a lot of what you are looking for sonically.

 

[audioengr]

The short answer is yes, you can put a larger value cap at the final node of the power supply. How much larger it will need to be before you hear a difference (if you hear a difference) is something you will have to work out - forgive me - empirically. The old rule of thumb was around 5X, John Camille used to propose more like a 10X increase. The only potential issue I see is that at some point you may stress the fuse with the startup current inrush.

Material may have an impact as well, that is, film rather than electrolytic. Obviously available real estate for a beer can sized film cap becomes an issue.

I suspect that the C4S may give you a lot of what you are looking for sonically.

Great.  I plan to give the C4S a try, but Ill probably change these caps too.  I have some low ESR Nichicons that are rated for 105 degrees to try.  I would normally use Panasonics, but they are not available in high voltages.

 

[Paul Joppa]

Shunt-regulating the high voltage (essentially an active capacitor) provides audible improvement even when the power supply is followed by a current-source plate load. That makes hardly any sense, but the real world does what it does anyhow.  :^)

But in parallel feed, the impedance of the plate load has a big impact on the subjective bass, similar to what larger power supply caps do for series feed - in this sense the two circuits are sort of duals of each other.

This is a big part of why the C4S on the driver seems to improve the bass; its impedance is around 5Meg. No way to say whether it's the improved PSRR, the greater load line linearity, or what - but the effect is audible.

I have never tried it, but Magnequest makes the "Dowdy" choke, 125 henries at 40mA. That's the biggest I know of short of replacing the power stage load with an active current source.

 

[audioengr]

Okay, I have the C4S board and I'm ready to assemble and install it.  I have a question:

Is it critical to leave the board in one piece and connect the ground wire to the chassis lug 23L?

With my space constraints, I would rather split the board into two identical pieces and ground the boards at lug 22 and lug 13 respectively rather than both at lug 23.  It appears that using lug 23 causes this ground current to mostly run through the chassis rather than the wires.

 

[Grainger49]

I can't think that there would be a difference electrically.  As long as the boards are well grounded you should be fine.

 

[audioengr]

I think it is much more than just a ground.  There is current running from the current sources I believe.  If this is the case, then it should be better to ground it to the ground return wires for each tube circuit rather than the other chassis lug which is only connected to the other ground return wires through the chassis.  Currents through the chassis should be avoided IME.

 

[Paul Birkeland]

There is a tiny bit of biasing current through the C4S, but this is not signal current, and it is not in the signal current loop.

If you split the board into two pieces, you will have to mount two pieces in two places.  I would be much, much more concerned about a floating board grounding out the high voltage supply over anything else mentioned in this thread.

 

[audioengr]

There is a tiny bit of biasing current through the C4S, but this is not signal current, and it is not in the signal current loop.

If you split the board into two pieces, you will have to mount two pieces in two places.  I would be much, much more concerned about a floating board grounding out the high voltage supply over anything else mentioned in this thread.

I understand your concern.  I modded over 40 different manufacturers components from 2000-2009 and most of these components are still in use world-wide.  They don't short-out.  No worries.

For instance, one thing that I do is insulate all of those bare metal wires from the resistors and caps with Teflon tubing of different sizes.

Steve N.