BMK2 used in cathode-biased amp

[K O'Connor]

Hi Champ81

Yes, delete it.

Regarding post-42:

If you are going to refer to any book, please give a page number or figure number.

Note that a superscript "2" following a quantity means that quantity is squared. A root-sign followed by a 2 is the square-root of 2 = 1.414.

It seems you are looking at the Dissipation Method for setting idle current (TUT2 pg.3-25). The equation quoted is to specifically set the idle heat to 50% of the tube rating. Frankly, it is of more practical use to simply calculate the maximum idle current and know that every value lower is going to extend tube life.

Pa / Va = Ik
25W / 442Va = 56ma

At higher Va, the maximum idle current would be lower, and at lower Va the current can be higher. This just follows Ohm's Law and the Power Equation, therefore:
THERE IS NO SPECIFIC IDLE CURRENT FOR EACH TUBE TYPE

There is no advantage to the sound or to the tube or anything else in the chassis in running the tube flat out. In fact it is mostly detriments. There is no magic idle percentage, 50%, 70%, etc, that all tubes work best at.

When testing an amp (fixed-biased) you will get a clean sine wave with 2mA or so of idle current. The tone may not be as warm as you want, but it does not take a lot to warm it up and beyond that point there is no benefit.

[Champ81]

It has been deleted.

Yes. I was referring to P3-25 Dissipation Method.  P3-27 is where the formala is.

I am reading the next part regarding screen resistors.  TUT 2 Figure 3-23 on Page 3-29 there are three methods to wire in the recommended 1K screen grid.  Currently I have 470R which is low as stated in TUT2.  The enhanced method is the one I am going for in Figure 3-23.

I have attached a layout for el84 tubes.  Is this correct?

[K O'Connor]

Hi Champ81

Screen resistors in guitar amps should be wired as in TUT2 Fig.3-22 - which is the way they are wired in pretty much every guitar amp and every schematic in the TUT-series. There should be individual 1k per tube. A few amps use a shared resistor, which is fine for hifi but not for MI, which TUT also shows.

Fog.3-23 is labelled "Triode-connected tetrodes and pentodes" and does not apply here. The "enhanced" figure applies ONLY to the next section, which uses the screen as a control grid with signal applied to it.

Rewire the screen resistors as stock but use the safer value.

[Champ81]

I am posting another layout regarding the adjustable cathode bias circuit variation with no MV.

If there are any errors in this layout please let me know.

[Champ81]

I am using 50Kohm linear pots and for 10X the value I'd need 500K resistors to go across wiper to ground. I guess they don't have alot of selection for 1W 500K resistors. I only see one on mouser. On digikey it's 1/2W max. I was hoping to go higher in power rating to 1W.


According again to ohms law
P=V^2/R
= 44.7VX44.7V/R
= 0.039962W

A 1/8W would work even with a double safety factor.

Would you suggest going higher?

The next option would be 470k resistors I have at 1W which is close to 500K

[K O'Connor]

Hi Champ81

In a previous post I demonstrated that none of the safety resistors or range resistors need to be high-wattage, and in fact can be 1/4W or less.

Usually, if you are building many things and/or servicing amps, you have an inventory of the most common values required, which for resistors will typically be half-watt ratings. This covers use in many circuit positions, although your experience may be that you use certain values only as, say cathode resistors in gain stages, or as plate resistors. The Rk's could be lower-wattage while Ra-s might be 1/2W or higher. This will also depend on the supply voltages you use or encounter.

You will see published projects where a person uses all-1W resistors, for example, purely as an aesthetic, or to ensure longer leads for hand wiring. This is usually of no electrical benefit as most power resistors have worse thermal noise and drift than smaller ones do. of course, in the parts selection process, you can filter for whatever performance you deem acceptable.

[Champ81]

Thanks Kevin. That's good to know. I didn't know that larger components can be noiser.

[K O'Connor]

Hi Champ81

"larger components can be noisier" is not what I said.

Materials used in the most common forms of power resistors lead to them having high thermal drift and other noise. It is not the fact they are 1W versus 1/4W; rather, the priorities are different, so the construction is different.

You can get metal-film 1W resistors and have uniformly low-noise/drift with the <1W metal-films used elsewhere. Similarly, you can get low-watt resistors with just as much drift/noise as metal-oxide or other types of power resistors. Most people simply are unaware of all the parameters there are for most components they buy - and it does not help that most vendors do not make this information known.

Note that not all of the parameters are relevant to every application.

For example, a resistor has a resistance value. Maybe you are aware it has a tolerance, the variation of resistance from nominal. Maybe you are aware it has a power rating. Maybe you know there are different materials used for the resistive element. It also has a voltage rating, and a temperature stability rating, and possibly a mean-time-between-failures rating, and whether it is inductive or not. If you are buying resistors at a discount store, or even at premium prices in blister-packs, very little of this information will be listed.

This is why I prefer to use full-line electronic suppliers like Mouser and Digikey. They stock a wide variety of brands and types of each component, usually with links to manufacturer's data sheets. Some of those data sheet links are not very good - usually the fault of the manufacturer - and some of the searches can be dead-ends, but they try to stock components that have a high level of proof of their integrity, including the specifics about the part and about the supplier and the supply chain.

You can take a chance on most parts for a guitar amp and there are unlikely to be catastrophe.

[Champ81]

Sorry. I did read that wrong. In general you are saying that power resistors are where more thermal drift occurs rather than in another section of the circuit that uses less power and hence less thermal drift and hence smaller sized resistors?

I use digikey and mouser for almost every component. And I trust that most of the products have the integrity and tracking of where they are from. I agree.

[K O'Connor]

Hi Champ81

That was not quite what I said either, except that it is true that thermal drift of the resistor value will be higher with higher dissipated power and/or ambient heat.

One of the parameter columns is 'temperature coefficient' with values that look like 25ppm/C, as an example. Most inexpensive power resistors will be 100ppm/C or higher, wire-wounds being mostly 300ppm/C. The small resistors used in most parts of the circuit are usually selected (by me or you) to be 50ppm/C, but also range up into the 100s of ppm/C.

For example, if you have a 100k resistor, 1ppm of the value is only 0.1 ohm, or 100mR. 50ppm is then 5R and 100ppm is 10R. If this is a resistor with 5% tolerance, its actual value can be off by 5k, and even as a 1% the value can be off by 1k. So, you can see that the temperature coefficient of the value stability represents a very small variation in practical numbers, but it still reflects something about the integrity of manufacture.

For a power resistor, maybe you have a 1k 5% wire-wound. Its tempco might be 300ppm/C, which is 300mR. If the resistor heats up to 5W its temperature might rise by 170C, which could change the value by 51R, or 5% of the nominal resistor value. We have not even counted the external heat sources.

Meanwhile a 1k resistor used as a cathode bias resistor may only dissipate 4mW and heat is minuscule, so the value drift will be also negligible even with the same high tempco as above. It is hard to find a metal film resistor with that high a tempco in the lower power ratings.

I choose components for reliability and circuits for tone. There are limits as to how far you have to go with these selections.

[Champ81]

Wouldn't this mean that a resistor with lower temperature coefficient is better in terms of resistance stability?

[Champ81]

I came up with a schematic for 4 X EL84 tubes starting with the first schematic by Kevin.
Basically each power tube has its own grid leak and coupling cap pair.  For individual adjustment.  The other way I thought was one bias pot per pair but it would be much better if it works to have a dedicated bias pot for each tube.

[Champ81]

I just got a digital oscilloscope and was using it to trace signal. With multimeter hooked into the amp to measure the bias current and with signal applied into the amp through a dummy load.. I noticed the bias currents go up significantly when the signal is applied and volume increased. This is one of the amps that have the adjustable cathode bias.

I've learned that for cathode biased amps a signal increase should increase voltage across the cathode bias resistor hence decreasing the bias current into the tubes. But in my case its doing the opposite. This would mean if I were to play at louder volumes I'd have to bias the tubes alot colder.

Is it normal for the amp to do this in my current setup?

[makinrose]

I just got a digital oscilloscope and was using it to trace signal.  With multimeter hooked into the amp to measure the bias current and with signal applied into the amp through a dummy load.. I noticed the bias currents go up significantly when the signal is applied and volume increased. This is one of the amps that have the adjustable cathode bias.

I've learned that for cathode biased amps a signal increase should increase voltage across the cathode bias resistor hence decreasing the bias current into the tubes. But in my case its doing the opposite.  This would mean if I were to play at louder volumes I'd have to bias the tubes alot colder.

Is it normal for the amp to do this in my current setup?

If I'm reading your post correctly it's completely normal.  When you set the bias be fixed bias or cathode bias you are setting the idle condition.  This is why when you set the bias you turn the volume down so you know you are setting the amp without any signal applied to the power tubes.  When you apply signal, you will see an increase in dissipation.

[K O'Connor]

Hi Guys

Bias is set without any signal present.

We stated this several places on the forum and in the TUT-series. No signal needed or wanted, so turn the MV or Volume controls to zero.

Yes, there are methods of bias setting for lowest THD where a signal is used, but initial setting is always done without a signal present.

In a cathode-biased amp, the idle current is ideally the same as the peak current ever drawn from the supply at maximum audio output. However, almost universally we see an increase in power draw at full output because the output stage is under-sized for the PSU and OT combination, requiring a class transition from class-A to class-B. The AC-30, for example, is only 22W as class-A, but produces 30W before clipping.

In a pure cathode-biased amp - one without a bypass cap - the voltage across Rk will vary with the signal. As mentioned previously, adding a high-value Ck will reduce or eliminate this voltage variation and was the first attempt to "fix" or stablise the idle condition. 22uF does not cut it as far as stabilisation goes and you see schematics where the voltage is given for idle and full output, as say, 10V and 13V respectively.

Champ81: Note that the 450V rating you placed on Ck is completely wasteful, as I commented previously. Maybe it is what you have in stock? Definitely will do what you expect, which is as much as it does in an AC. A 100V cap is physically a lot smaller and can be a much higher value if you want good results.

Personally, I do not like cathode-bias for output stages except for tone AND set for far less than maximum power in this bias mode. For the complexity of the bias adjustment circuit, it could be a lot easier and more stable using true fixed-bias with a proper negative bias supply. In general, there will be less interaction between the individual adjustments using negative grid voltage than with positive cathode voltage.

In my previous post where I mentioned the dissipation of a 1k cathode-bias resistor being about 4mW, that is for a preamp stage. A 1k used for EL-84s et al would dissipate watts and need to be a power resistor.

[Champ81]

Hi everyone.
The idle current I set it to was around 27mV per tube with the MV at zero. I'm still new to using the oscilloscope and realized I set the signal amplitude to about 2V. I think that was too much signal being fed into the amp and hence the significant increase in dissapation on the tubes. I set it to about 150mV I guess in range of a typical guitar signal?
Only issue I have with the digital oscilloscope is I have some trouble triggering the signal at that low of a signal.

From there if i remember correctly with increasing volume and constant signal applied i saw the current go from 27mV to as high as 78mV for the EL84s which i think is above the max operating range. I would say in reality the signal wouldn't be constantly and consistently fed in like that when playing. The current draw would jump around. But would it ok for the tubes to pass their typical tube rating when high amplitude signal is breifly applied? Maybe this is normal and not unsafe for the tubes?

Apart from all that I see what you are saying Kevin. It would be more efficient to reserve that high voltage rating and use a smaller voltage rating with a higher capacitance? And hence the higher the capacitance the better the stability in the output section if I read that correctly?

I currently finished a liverpool trainwreck build and increased the 50R cathode bias bypass capacitor at the power section from 22uf to 220uf. Also increased the main power filter cap from 47uf to 80uf. From my ears at least it tightened up the bass response and I liked what it did. Would it be safe to say that using high capacitance for those parts of the circuit be overall beneficial in that at least it improves bias stability in a cathode bias amp? As well as tighten the bass response a bit?(if that is what is wanted).

I agree that the 450V is way too high of a rating as the voltage drop is only about 45 times less than that in that portion of the circuit and wild be better to use a higher capacitance and lower voltage rating.

[K O'Connor]

Hi Champ81

You say you have 27mA per tube. What is the plate voltage in this amp at that current? With EL-84 rated at 12W Va should be 444V or less. In post-42 or so, you referred to EL-34, which is fine at 12W since it is rated at 25W max.

Please STOP using signal to set idle.

It is only of benefit for learning and satisfying curiosity to monitor current with a signal present. Obviously, you can become easily confused by the numbers you see. You have to remember that "heat" is what watts means, and this is a quantity that is averaged over time. With an audio signal, and even with a steady sine wave, the instantaneous power / heat can be way higher than the idle heat or even the maximum rated heat, but there are also points in the signal cycle where the heat is far lower or even zero.

Tube ratings are "average". This means that the 25W plate rating for the tube is satisfied if, for example, the voltage across the tube is 500V and the current through it is 500mA, and also if you pulse the current to 100mA at 500V half the time, and the current is zero half the time.

The current rating of 100mA you referred to is likely for a specific application; it is certainly not the cathode current rating for the tube as listed by the manufacturer. An EL-34 cathode can emit substantially higher current than that. You should refer to TUT6 about this.

Back to the cathode biasing... The 50R resistors in your circuit seem a bit low in value (no doubt they are perfect as 50R resistors go ). The implication would be that a single tube could be biased with 200R, which is not the case with EL-34. As stated previously, the single-tube average Rk value is 470-500R, call it 500R to be safe. Two tubes sharing Rk can therefore use 250R and 4-tubes 125R. The total Rk value with the adjustment scheme MUST be higher than what would be typical without the adjustment, otherwise there is no room for adjustment other than hotter than stock - the opposite of what we want.

About Ck: I forget that some readers do not see the subtleties of what I write, especially when it is being read on a glowing screen. So, to be clear:

Ck only has to withstand the voltage that is across it. Typically this might be Vs/10. In the case of the adjustable circuit, we need a bit more voltage than that, so maybe rate the cap for around Vs/5.

The cap voltage rating and its value have nothing to do with each other.

I simply referred to the fact that going from a 450V cap to a 100V cap should reduce its size AND that if this frees up space it is an opportunity to increase the value of Ck.

Higher-value Ck provides better stability of the idle point BUT you have to go significantly higher than the values you see in any guitar amp. Look at hifi examples, where 1mF is common, although even that is too low for a quad of tubes.

Personally, I never use Ck in places I have self-biased output tubes. Including Ck diminishes the sonic difference between fixed-bias and cathode bias.

[Champ81]

I completely understand what you said about the voltage rating of the cap. I meant to add that it's about saving the space. So a lower voltage means you can go higher in capacitance since size will free up the space. Higher voltage ratings in this situation would be a waste of real estate.

Everytime I bias I do set the idle bias at idle with the MV at zero. It was only that I injected signal and monitored after setting the idle at zero master volume I saw the current fluctuations and was a bit freaked out to see some high readings. I'm I'm rest assured now.

For the amp in question has 2XEL84's. That is the one I set to about 27mV of idle current at MV zero. Va (Measured at pin 7 of power tubes) is 347.8V. At first power filter cap is 355V. I was always not sure what is Va. Sorry for the dumb question. I always saw it as V (anode) so the anode (plate) of the power tube.

[K O'Connor]

Hi Guys

The limitations of the forum text formatting does not allow accurate depiction of the generic quantity names with respect to the use of subscripts. In general, when the subscript is an upper-case letter, the quantity is external to the tube or device; a lower-case subscript represents an internal quantity or parameter.

In ASCII (web text) an asterisk is often used to denote that the next number/letter is to be a superscript. For example, c*2 would represent the quantity 'c' raised to the second power (squared). I'm sure there is a corresponding standard for subscripts but I have not seen it. Below, I am going to use an apostrophe, as V'A. It might be more usual to use a hyphen, but then V-A looks like an equation and the reader has to understand by context that it is not.

In any case, specifically with voltage designations in tube circuits, TUT3 lists the preferred supply node designations. Other tube parameters are listed throughout the TUTs as they are needed. So, for an output stage, there is always a plate (anode) supply node called V'A. If the tube is a tetrode or pentode, there is usually a supply node for the screen, that we call V'S. The actual voltages at the tube terminals would be V'a and V's, respectively.

Often for convenience the internal / external aspect of the parameter may be set aside if there is an understood context for the presentation and the quickest form of the designation helps move things along.

When there is a voltage that uses a reference other than ground, the subscript will have two letters where the second designates the reference point, as in V'gk. This is the voltage between the grid (g) and the cathode (k), where the cathode is the reference.

In old texts, a "battery" supply designation is used, as would have been the case for simple experimenting, or for prior to the widespread availability of an AC mains distribution system:

A heater supply battery
B anode supply battery
C grid supply battery

D and beyond can be used randomly as required. Most other circuit voltages would be derived from B or C. A polarity designation is often added to B and C as B+ and C- to indicate how the battery is connected.

In an output stage where a choke or transformer load is used, the voltage at the plate will be more or less the same value as at the related filter cap. The screen voltage will be similar regardless of the screen resistor value.

In a preamp stage with a resistive plate load, there will be a significant voltage difference between the filter cap voltage and the tube anode voltage due to the drop across R'A. Preamp tubes have a deceptively low plate voltage rating as an inductive load is assumed, as was the case with very early circuits and applications. What percentage of circuits a tube is used in that have inductive loads is not known, but the rating is antiquated. Careful reading of the complete data sheet often indicates the true limit for the tube.

For example, when you look at 12AX7 ratings, you might see 300V listed for maximum plate voltage. Should that be what the filter cap voltage is, or should it be the voltage at the tube? Most people would assume the former just to be safe. The fact is that a 12AX7 has an arc voltage of 540V, which allows a supply voltage of 300V with an inductive load that can produce close to 540-300=240V peak. With a resistive load there are no flyback voltages, so B+ in this case can be close to 540V. This is a good thing since a cold start-up can potentially charge every supply filter cap to the highest unloaded voltage of the rectifier output.