Pentodes as pentodes

[Dr. Toobz]

I'm curious to hear how pentodes really sound, as I've only built things with triodes. Plus, I'm interested in seeing if I can squeeze more gain from the Quickie. Looking at the schematic, the Quickie ties grid 2 of the 3S4 pentodes to the anode to "make" a triode. I was thinking of putting a triode/pentode switch in the circuit and on one side of the switch, taking grid 2 to the B+ with a resistor. As for the value, I've heard that the same value as the load resistor (in this case, 4k) is a decent starting point. But what is the proper way to determine this?

The extra gain might be useful on some of my quietest LP's going through the C4S'ed Seduction. I know that the output impedance will be a lot higher, but the interconnect going to the S.E.X. amp (which has a 100k input impedance) is only a foot long, so I wouldn't expect too many problems. I may end up not liking the sound, but there's only one way to find out! Plus, I just can't resist messing around with this circuit, as it's so simple and provides a great springboard for learning about how tubes work. I've already played around with the cathode bias and plate loads...

[Paul Joppa]

Just take the screen to the high voltage. The screen supply should be stable, so if you use a resistor you'll want to add a capacitor from screen to ground as well. But the better way is  to experiment with a tap on the batteries, give the screen 18v or 27v or 36v.

Don't do this with a current source plate load - you need the plate resistor to keep the source impedance low, since a pentode's effective plate resistance is very high.

[Dr. Toobz]

I like the idea of just going to the HT supply. I have a bunch of pentode schematics laying around, so maybe I should try to figure out if they use a common figure (or range of figures) in terms of the difference between the B+ and screen voltages (which I can figure out by calculating the voltage drop across the resistor if I know the value of r and the HT voltage).

For plate loads, I can already change between resistors and anode chokes with just the flip of a DPDP switch. The resistors are 4k; chokes are 3.7k. I've added three switches to the Quickie now and might do a fourth. (Monitor/Tape, Plate Load, Bias Setting, and maybe Pentode/Triode). What a fun kit to play around with!

[Dr. Toobz]

Ok, so I did this! Some interesting observations here, though things didn't work out 100% correctly (as I'll explain):

1) I "permanently" put back the chokes as the plate loads (150H, 3700 ohm DCR) rather than having them on a switch. Then, I wired a DPDT switch for both tubes so I could switch from triode to pentode. One side simply completes a circuit between the plate (pin 2) and screen (pin 4) for triode mode, as in the stock Quickie. The other side of the switch takes "high" voltage to the screens instead. I played around with different voltages, given that the terminal strip for the 9V cells has several options available (9V, 18V, 27V) in addition to the full 36VDC. Interestingly, the full B+ sounds best - the lower voltages start to distort earlier.

2) In pentode mode, gain is huge. While the pre-amp is at unity in triode mode at about half open, it only takes a quarter of a turn of the pot to do the same thing with the pentodes. So, it has at least twice the gain.

3) Here's the problem: one channel is a few dB louder than another in pentode mode, but yet when I throw the switch, is perfectly balanced in triode mode! What's strange is that each screen gets the same voltage, which I verified with my meter. What's not the same are the plate chokes. The quieter side shows a larger voltage drop across the choke than the other, even though they're the same type of chokes (Hammond 156C). One side is about .3V down from the other. In triode mode, this seems to make no difference, but is noticeable in pentode mode, in that the image shifts towards the louder channel. Why is this? Could one of my joints be cold, or do chokes really have that much variance between them? Should I go back to 4k resistors instead?

4) The 3S4 tubes actually sound nice as pentodes. Things are very "clear" - the triode mode sounds a lot darker in comparison, though I happen to like that most of the time. But, the pentode mode has a much higher output impedance (I think the Rp for these tubes as pentodes is .1M). I'd assume this means that the stock coupling caps (2uF?) end up being too small to avoid a rolloff in the bass frequencies. Correct? There's a lot less bass with the pentodes - I'd estimate they are starting to roll off around 100Hz. What size caps can I use to keep a low -3dB point with both modes? I remember somebody posting about how to figure this out one time.....

As pentodes, these little suckers put out enough juice to bring my Heresy III's (99dB/1W/m) up to listenable levels (note that I said "listenable," not "loud") at full power through some Specos. Pretty cool! In fact, the spec sheet on these suggests up to 270mW output power, as pentodes, if the B+ and screen voltages are increased to the maximum rating. I don't intend to do that, of course - this was more an exercise in learning about tubes vs. trying to jury rig an amp, since I already have the S.E.X. amp as my main rig.

EDIT: Thought I'd mention that the channel imbalance in pentode mode does not follow the tubes - it's always in the channel where I measured the B+ to be slightly lower (27.6VDC vs. 27.9 for the other one). Swapping cables also doesn't make a difference - same imbalance, same place.

[Paul Joppa]

Because the tub's impedance in pentode mode is very high, the gain is controlled by the (lower) plate load impedance. In this case, that is the inductive reactance of the choke, below the choke's resonance - probably a few kHz. The gain increases with frequency because the choke's reactance increases with frequency.

Above the choke's self-resonance, the impedance of the choke is capacitive and the gain will fall as frequency increases.

For this kind of choke, the inductance depends on the air gap built into the choke. That will be set by a thin material, usually paper, and about 2 thousandths of an inch thick - or it may even be simply "butt stacked". The laminations will not be identical nor will they be perfectly stacked, so the residual effective gap due to these imperfections will be quite erratic, especially if that is the only source of a gap. You can test this theory by swapping chokes between channels; you'll probably see the gain follow the choke.

By putting a resistor across the choke you can extend the frequency response and reduce the gain, since the high choke impedance in the midband will be swamped by the lower value resistor.

[Dr. Toobz]

I've noticed that most schematics using 3S4 or 1T4 tubes as pentodes usually has them loaded with 1M resistors, and I only have a plate choke with an effective resistance of 3.7k and AC impedance of (probably) 25k at certain frequencies. So I'm assuming that it would be better to load the pentodes with a high-value resistor? If I were to use 1M resistors as plate loads when in triode mode, wouldn't that cut down the gain too much? Maybe a should wire the switch to load the triode-tied arrangement with the chokes and pentode mode with resistors....

[Paul Joppa]

I've noticed that most schematics using 3S4 or 1T4 tubes as pentodes usually has them loaded with 1M resistors, and I only have a plate choke with an effective resistance of 3.7k and AC impedance of (probably) 25k at certain frequencies. So I'm assuming that it would be better to load the pentodes with a high-value resistor? If I were to use 1M resistors as plate loads when in triode mode, wouldn't that cut down the gain too much? Maybe a should wire the switch to load the triode-tied arrangement with the chokes and pentode mode with resistors....

Sorry to say, there are too many questions, each of which needs too long an answer, to respond well. So I'll just give you a short and crappy response!

When we speak of the tube's "load" we are referring to the impedance seen by the plate. The 1 Meg resistors you see are almost always going to be from the control grid to ground.

Pentodes act (at the plate) like a current source, with the current controlled by the grid voltage.  So their AC output voltage increases as the load increases, for the same input (grid) voltage - i.e. gain is increased. But if you use a large resistor as plate load, the current must be small or the supply voltage high. For example, a 3S4 running 2mA (as it does in the Quickie) would need 2000 volts of battery if it had a 1 meg resistor in series with the plate. And, since the plate voltage is limited to 90 volts, the peak output current into that 1 meg load would be less than 0.000090 amps. Neither is practical ...  :^)

Triodes on the other hand are closer to voltage-controlled voltage sources, and the load impedance (in the normal range of values) has little effect on the output voltage, or on the gain.

[grufti]

I wouldn't consider that a crappy answer. That's about as concise as anyone could ever respond, and in most cases that is extremely hard to do.


Blaise Pascal, Lettres Provinciales XVI: "Je n'ai fait celle-ci plus longue que parce que je n'ai pas eu le loisir de la faire plus courte." / "I would not have made this so long except that I did not have the leisure to make it shorter."

[JC]

I had just read this, and while Paul's answer was "short", it was by no means "crappy." 

More thought-provoking than anything, I would have to say.

[xcortes]

Angelo,

When you go to pentode mode, do you get more noise? I've been experimenting with amplifying a cassette head signal with tubes and need a super silent high gain stage to go between the head and the tape preamp.

Thanks

[Paul Joppa]

Pentodes are noisier in theory than triodes - I believe the difference is usually 3 to 5dB, equivalent input noise. For triodes, the equivalent noise source is an imaginary resistor, approximately equal to the inverse of the transconductance and at the cathode temperature.

But this is all simple "shot" noise. (Pentodes have both screen and plate shot noise sources, plus something called "partition noise".) There is also 1/f noise, a.k.a. flicker noise - a low-frequency pink noise that can dominate, especially when phono or tape equalization is applied. This has numerous sources, not all that well understood. There are many external sources of noise, such as power-supply noises, electromagnetic interference, microphonics, external circuit resistors, etc. So it's not always true that a high transconductance tube is quieter.

[braubeat]

If you don't mind going to a little trouble and taking the chokes apart (I am assuming you have those inexpensive hammond units). Then you can get some fine sandpaper and smooth out the but ends of the choke where the air gap is. If you put the sandpaper on a very flat surface and carefully sand down the core but ends until they are smooth then put it back together (you probably don't even need any paper in there for just 2 ma.) then the performance of the chokes should be better matched. I have done this before with good results. Also put some rubber bands or tape around the core to make sure the gap is tight.
I would be interested to see if it works better.

Michael