Battery PS Heaven

[Len]

Attached is a sketch of my last mod of the Quickie low pass filter, using the Sowter 9063's as output trannies. They are spec'd at 24 dBu, with good low end response. I made sure not to give away any of the Quickie design in the sketch.

Bass volume is now just a tad shy, but bass quality seems much better, quicker, deeper and just more musical (I check this by turning of the Excite main amps and leaving the Paraglow bass amps on).

Two questions:
1. I'm totally in the dark whether it is a bad thing to load the transformer secondary with a value lower than its own impedance (but higher than its DCR).

2. Also, I am wondering if there is a way to increase Quickie gain, such as maybe a different plate choke or a C4S at higher voltage. I am redoing my phono pre PS and have access to clean HT. If not, can I just increase the input resistor to the Excite to pad it down a little?

Here are my calcs, though I'm sure there are some mistakes in it:

Stock Quickie output impedance 2k and gain 2.25 ( http://db.audioasylum.com/cgi/m.mpl?forum=bottlehead&n=138602).

With Hammond, it should now be 4.5 and output impedance should be 4K as per
(http://db.audioasylum.com/cgi/m.mpl?forum=bottlehead&n=138791)

Now it should be at 4K output impedance at 4.5 times gain.

Putting the 110 ohm DCR in parallel with the 500 ohm loading resistor gives 90.2 ohms (about 82%) , and the ratio of 500/600 ohms also gives 83%. So either way, reflected back to the primary is 82% of  of the secondary's impedance, so the 10K should now be seen as 8200 ohms, about double the Quickie's 4K output resistance for max power. Is that right? It's real muddled in my head.

Gain is reduced by 4, then multiplied by .82 from the loading resistor.

So a 1V signal would become 4.5V, then 4.5/4 = 1.125V, times .82 = .92V, so I need a little more gain.

 

[Paul Joppa]

OK, this will take a while to answer usefully.

First question, can you load a transformer with  lower-than-specified impedance? Transformers don't have impedances of their own, but they do have impedances for which they are designed. Bottom line, you can run them at different impedances as long as it's not too much different. A factor of two difference is often too far to go. A factor of 5/6 which I think you are asking about is not.

Now the hard part - looking at the gain. The 3S4 appears as a voltage source of about 4.4 times the input voltage (that's the mu of the tube triode wired, according to my calculations). It has a series resistance of about 4K ohms, the plate resistance of the triode.

I had to look up the Sowter to find enough details. It is normally a 600:10KCT step-up transformer; you are using it in reverse, so the meaning of "primary" and "secondary" are swapped, making it tricky top read the specs! The voltage ratio is specified, which I assume is the turns ratio, and the total (primary and secondary) DC resistances are referred to the 600 ohm winding. So we'll take the source resistance of 4K and scale it by the turns ratio squared, or 16, to get 250 ohms. The winding resistances add up to 110 ohms, so the total series resistance is 360 ohms.

Your load resistance is 500 ohms, so the gain of 4.4 is reduced by 500 / (500+360). Then you must multiply this by the turns ratio, 1/4. So the net gain is 0.64.

I ran some numbers to see if a different transformer ratio would do any better; the best I found was a gain of 0.7 with a transformer ratio of 2.55 (3250:500 ohms). The difference is insignificant, only 0.75dB.

I don't see any way to increase the gain with this crossover. Changing the input loading resistor of this or the eXcite won't affect the gain, and neither will changing the power supply voltage. I recommend that you decrease the gain of the fullrange amp instead, if you can. There is a discussion of reducing amplifier gain in my white paper on signals and noise.

If that is not possible, I think the simplest approach would be to get a step-up transformer to place after the lowpass filter. Less easy to do, but a simpler circuit, would be to replace the crossover with a higher-impedance version so that the transformer ratio can be reduced.

I can't find any specifications for other tubes in triode mode, but the 3Q4 has the same pinout and slightly higher transconductance. It may or may not have a bit more gain, and may or may not work in this circuit.

 

[Len]

There is a discussion of reducing amplifier gain in my white paper on signals and noise.

Less easy to do, but a simpler circuit, would be to replace the crossover with a higher-impedance version so that the transformer ratio can be reduced.

Here's the formula for normal parafeed cap size:

C = 2*L / (R*R)

C is the capacitance in farads, L is the plate choke inductance in henries, and R is the load impedance seen by the plate (i.e. transformer primary impedance) in ohms. If you use kOhms the C is in microfarads.

This value gives the best bass extension while keeping the impedance seen by the tube's plate relatively constant and resistive to the lowest possible frequency. The value is not especially critical, anything from half to twice this value will work pretty well. For speakers with especially deep bass there may be some interaction with the speaker impedance fluctuations, and experimenting within this range of values can sometimes prove fruitful.

Paul,

Thanks for the very complete answer and for the research you put into my crazy project!

My computer was down over the weekend. I had thought I was at 92% gain, so I put 470K series in front of the 279K shunt (grid loading resistor) at the input to the Excite. This weekend I'll go for 1 Meg instead, so that I'll  have some room for adjustment using the pot.

I ran the numbers for the same filter, now at 80 Hz, at 4K input and output impedance. Chokes are 12.7H and 8H! I should be able to get away with such a high output impedance going into the Paraglow. I'll look for the chokes.

One last question, if I may. What is the impedance of the Sowter's 10K winding at this point? I ask because of the parafeed cap formula above. If it's 8K, then I would need about a 5uF cap (2*150/(8000*8000). In this case, I want to optimize for bass. If I lose top end, so much the better.

Thanks again.

EDIT: I just realized that the Hammonds' 150 Henries is at 8ma. At 2 ma it should be higher, though figuring that number is way beyond me.

 

[Paul Joppa]

Not to worry!

The Hammond's inductance won't increase more than 20% and probably less, at zero current - inductance is dominated by the air gap.

And you are correct, the Sowter is 10K:600, or 8.3K:500, so 8K is a pretty good estimate when it's loaded with 500 ohms.

Good luck on the large-choke front. Marchand might have some, though those inductances are pretty high for signal-level chokes.

 

[Len]

Latest update on the low pass filter:

I changed the parafeed caps to solen 6.5uF's that I had in the parts box, then found my audio alchemy CD player and put on Dark Side of the Moon.

Bass goes way down there, is not boomy, and is pretty quick. I'm happy. I'll hold off on further padding to the full range, and start working on changing its driver to a 6J5GT instead.

Thank you PJ for all the help!

 

[Len]

I changed the drivers in my main amp to 6J5GT's and left the 470K/279K pad in. Now I get full bass at about halfway on the volume control of the Quickie and everything sounds great.

I may change the chokes for a different LP frequency one day, but I think I'll play with putting caps before the Excite this winter instead. We'll see if it messes with the Lowthers.