Screen resistors - please weigh in

[Roe]

I experimented with a little with my ac30. I prefer around 330v on the plates and 320v on the screens. I'm using individual 100R screen resistor and a collective 470R resistor (instead of choke - I have a big CLC filter)

[gui_tarzan]

This is a fantastic thread! I am building a PP EL84 amp right now for playing mostly blues and medium rock. I am still learning about how certain components affect tone and never thought about the grid resistors playing a part in this but some of the experiments the guys have done really opened my eyes.

I am still amazed at the genius of the original tube makers and how they were able to come up with this one item that makes such a difference in how we hear music.

[gingertube]

Just read thru' this thread - a couple of comments on points raised above.

Hot switching of screens - a tube with no screen supply is effectively shut off, it will idle at 1 or 2 mA so momentary no screen supply will not cause a problem. I have a PP 6V6 amp (own design) with switchable fixed voltage/Ultralinear connection to the screens - There is a little "thump" when hot switching but no damage.

The common screen resistor in a push pull amp - as one side draws more screen current the other will draw less, this reverses on the other half cycle. The result of this is that the average screen current from a push pull pair is reasonably constant. If you use a common screen resistor then you get a reasonably constant screen voltage, almost as good as a regulated supply. The common screen resistor was a Mullard thing and is typical of HiFi designs which is why it is no surprise to set it turn up in a Hiwatt. I used a common 470R in a cathode biased push pull EL34 "Trainwreck" with a rail of +380V (32 Watts max).

The next thing to remember is that the screen is "just a grid" in that the tube has gain from a signal applied to the grid. A small value individual screen resistor is valuable for its "grid stop" function in helping to supress parasitic oscillation , this is what those 47 Ohm, 100 Ohm range value screen resistors are doing. When you start using much bigger INDIVIDUAL screen resistor then you are introducing negative feedback from the signal voltage developed across the individual screen resistor due to signal dependent screen current.

Use a COMMON screen resistor to limit screen voltage and let that feed smaller INDIVIDUAL screen resistors for parasitic oscillation supression and for signal dependent tonal changes.

Be aware that just getting the screen voltage down to say 5 to 10V less than the anode (at idle) will improve the tone.

I'm not sure about the series diode trick - certainly early Ampeg SVT amps used this arrangement.

Thats my "shootgun at the barn door" stuff. Hope there is something of interest in it.

Cheers,
Ian

[Roe]

Any comments on the old Marshall practice of using individual 470 ohms screen resistors on (a duet of) EL34s in series with a (shared/collective) 110 ohms choke? Will this be safe with lower plate and screen voltages (e.g. below 415v)?

[Paultergeist]

What a great thread.......just want to give a shout-out of "Thanks" to all the contributors......

[Ken Moon]

Here's a very different angle - I've used this PS design an about half a dozen 2x6V6 amps, and it sets the screen at only about 1.5 volts lower than the anodes.

However, it's easy to set the screen voltage to whatever you want by setting up a voltage divider (adding a resistor between R4 and GND)

I've found through playing around that I get the most rockin' sound (more punch, volume, cuts through the mix better), if I keep the PS as it's shown here, and the screen is very close in voltage to the anode:

[img779]http://i514.photobucket.com/albums/t346 ... CLCpsu.jpg[/img]

I haven't measured screen current (I get lazy when things work well the first time), but this circuit doesn't limit it using series resistors either.

[pdf64]

Ken, does

it sets the screen at only about 1.5 volts lower than the anodes

refer to the actual screen and plate voltages, or B+1 and B+2?
As the latter won't include the few volts dropped across the OT primary or screen grid stopper (if fitted).

Could you explain the action / benefits of those transistor pairs?

Thanks - Pete

[Andy Le Blanc]

thats slick... have you found any good equivalents for the bul312fp?

theres a couple very simple tube series regulators too but the voltage drop
across the circuit may be better for condition for hifi.

I like the screen around between - 2 to -5 ish to the plates as well.

I found you have to set up screens at the same time you set bias
to accommodate each tube compliments characteristic's .

its trickery-er with simple resistors to get perfect balance, but imbalance
in the push-pull circuit gives tone color

[martin manning]

Yes, very interesting, Ken. What is the source of that circuit? Your design?

Pete, this is a pair of darlington pair regulators. Lots of stuff on the web about this configuration.

[Ken Moon]

Thanks for the interest. Any high-voltage BJT can be used, the BUL312 is nice because it's all plastic and in this configuration doesn't need any more than a clip-on heat sink, and would probably get by with no heat sink at all. MOSFETs can also be used, like the IRF820, in a slightly altered configuration. You can also use op amps, if you want.

The source of this circuit in a guitar PSU was Mr. Merlin (the Valve Wizard) Blencowe, who discusses cap multipliers in his out-of-print Power Supply book. Many discussions on AX84 about 4 years ago also went into this type of design, but I designed this particular implementation of the concept. Google "capacitance multiplier" for many more sources of info.

The capacitance multiplier is a type of series voltage stabilizer. Its function is to reduce power supply ripple. It generally can be used to replace an LC filter in a guitar amp PSU. In the design I posted, it replaces a CLCLC filter, to feed both the plates and screens of the power tubes. Successive stages can use traditional RC filters. The cap multiplier is not a voltage regulator, just a ripple remover.

The noisy raw voltage is applied to the transistor’s collector, and a smoothed voltage is applied to the base. Since the transistor is arranged as an emitter follower, it will follow the base voltage, and deliver a much cleaner voltage supply to the load.

The base voltage is supplied from an RC filter (R3/C2). R3 also sets the bias of Q1/Q2. This filter can be very small, with very little ripple appearing across C2, because Q1/Q2 buffers C2 from the heavy load current. Because the transistor can supply much more load current than the RC filter alone, while maintaining low output ripple, it is as if an RC filter with an enormous capacitor was being used (value of C2 appears to be multiplied by the hFE (current gain) of the transistor). This is the source of the term cap multiplier.

The use of Q1 and Q2 in a Darlington Pair arrangement increases the composite transistor’s gain (gains of the 2 transistors are multiplied together), yielding better filtering using the BUL312FP’s, which each have an hFE of about 9 (IIRC).

A resistor can be added between the bottom of R3 and ground if desired, and this will set the output at a specific voltage level, using the common voltage divider equation Vout = Vin * R3 / (R3 + added resistor).

More details - The emitter tracks the base voltage, which is the integration (by the RC constant) of the input voltage. Being an integration, it will stay around the mean voltage (you might call it the DC component) of that ripple voltage, enough to kill the peaks of that sawtooth, but not the valleys, because it will stay between those values. The diode D2 ensures that the capacitor takes the "most negative" value of that sawtooth , or the value in the "valley".

It won't stay there forever, but we only need the RC time constant to be over 10X to 20X the ripple waveform period. It's the exact value to kill just the ripple with minimal loss of useful voltage.

For example, if the raw B+ voltage is 430V + 30V ripple above it, the instantaneous voltage will go from 430 to 460V. The capacitor will be charged to 430V; the base will take that value, the emitter will follow about 0.7 volt below; the transistor will absorb any voltage between 460V and 430V plus the couple of volts saturation voltage of the transistor pair, for a final output of 427 to 428 volts of super clean DC. I think the output ripple is < 10mV, but I've packed away my notes so I'm doing this all from my rapidly fading old memory

[Ken Moon]

I found a marked-up schematic of an amp using this circuit, with a 300-0-300 PT and no 47R resistors (those were intended to mimic the sag of a tube recto).

With the amp on and no input signal, the B+ was 432V, B+1 was 430V, and B+2 was 428V, and I did use 470R resistors between B+2 and the screens of each 6V6. Unfortunately, I didn't notate the actual screen voltages at the tubes

[Ken Moon]

Way off topic, Pete, but I hung out with a band on and off in 1973 and 1974 called Redwing, when I lived in Sacramento right after high school.

My younger brother used to babysit for Tim Schmit and his girlfriend, and my wife (then girlfriend) got to know her, and introduced me to the band. I was in a band called Earthrock, but I jammed with the Redwing guys several times after Tim left for Poco, but they were a little too countrified for my taste

[danotron]

I just came across this thread and I am having a problem you guys may help me with.

I was measuring my plate voltage to bias my tubes it measured 426v. I then measured the my screen voltage and it was higher 429v or so. I found that my output txer is dropping a few volts on each side.....measured from centre winding to either lead...blue or brown.

How does this affect my amp? Is this ok? How should I fix it?

to summarise:

- B+1 (output txer centre tap) = 431 volts
- voltage measured on plates (pin 3) = 426 volts
- voltage on screen (pin 4)= 429 volts

[Gaz]

Are there screen grid resistors fitted? If not, add some or just make them larger. Like if they are 470R, just make them 750R or 1K. You will loose a couple watts, and maybe a little sparkle, but you can add that back elsewhere.

[Jerryz1963]

Yes, very interesting, Ken. What is the source of that circuit? Your design?

Pete, this is a pair of darlington pair regulators. Lots of stuff on the web about this configuration.

If you want it even simpler, u can use MOSFETs. Zeners in series can be used to set the reference for the gate. I have found that these transistorized regulators vastly improve the sound quality, especially when applied to the preamp stage, and they also dramatically reduce power supply hum.

I did try it first with bipolars, but had to use two transistors, one as a driver to the actual pass transistor because there was enough base current flowing to cause problems. That is not an issue with MOSFETs, however, as gate current is practically nonexistent.

RE: SCreen resistors. I downloaded and translated the German data sheets so I could work with the Russian GU50. THe Russian data sheets were very minimalistic. They espouse a 5K screen resistor. I honestly didn't notice much difference in sound quality as I varied the screen resistance, increasing it well beyond the 5K because I wanted to run in ultralinear mode. However, the GU50 is a pentode, not a tetrode, so, maybe that has something to do with it.

My experience with screen resistors has been that things seem to "balance" better with them. I built a push-pull-parallel amp a few years ago and it would not have worked without the screen resistors. sounded like garbage without them. WITH the resistors, it was just beautiful.