I'm seeing everything from 3000-8000 as the appropriate Primary Impedance on the OPT for a single ended 6v6 amp. Seems that higher plate voltage calls for higher impedance? Many say "5000." The OPT I bought a while ago for a "Champ" has 7000 on the primary (4 on the secondary.) I use a 125D Hammond for a homebrew SE 6v6 since I added a switch for 4 or 8 ohm speakers. Problem is, I use connections resulting in about 8500 on the primary. This because one data sheet I saw said 8100. Today I tried the cab's 4 ohm on the current "8" setting which means 4100 was on the primary instead of 8800. It was a little brighter which is good for this amp. Not much though. So what's the REAL impedance I should use on the primary with about 300V on the plate and 280 on the screen in class A SE? My choices are 2800, 3400, 4100, 5800, 6400, 8800.

They are all right, depending what you want out of the amp... Highest power? Highest 2nd order harmonics? Best compromise? BK

Ideally I'd like: -max power -more treble -to keep the nice breakup -low risk of tranny damage -long tube life I know that some of these are contradictory. Can you tell me which direction one goes for each result? And what the value was in the Tweed Champ? Thanx!

Try using the 4100-Ohm tap with a 350-Ohm cathode resistor. Here are some load lines I drew up on top of the RCA data sheet. Let me know if you need me to explain how I came up with these numbers. If your tube happens to match these curves, you'll be idling around 12W or 85% of 14Wmax. I wouldn't run the amp hotter than this for tube life. You should get close to max power with good harmonics at this point. After I reviewed my post, you might actually cross the 'Max Dissapation' threshold at certain points of the signal swing with a 4100 load. You can either use the 5800 ohm tap (which will change the tone, harmonics, etc.) or use a 500-Ohm cathode resistor. Raising the cathode resistor effectively shifts all your load lines down. BK

Another bit to be mindful of is that those curves are drawn for Vs=250V. At Vs=280V the knee of the Vg1=0V curve is closer to 50v and 115mA.

Exactly. Designing for max power doesn't necessarily mean best tone. It's also just a starting point. I also don't subscribe to: 'Fender used this value so it must be right'... Generally, 2nd order harmonics dominate lower load resistances and higher order harmonics dominate higher loads (for SE amps). I'd also experiment with the cathode bypass cap (see if it sounds better by removing it). BK

Just for fun... If you follow the methods suggested in the RCA Receiving Tube Manual (RC-30) and simplify to assume a pure resistive load (i.e. straight line instead of an ellipse), then you end up with load lines more like this: I've used a SPICE simulation to reflect Vs=280, so the results are subject to deficiencies in the 6V6 model, but still interesting. The red dots represent max dissipation of 14W, so by RCA's method you'd go with the 3400 ohm primary and a bias point around -17V. The astute reader will also note that in RCA's listing for a 6V6 in Class A operation they suggest, under different conditions, plate loads of 5K, 5.5K, and 8.5K. The 8.5K value seems odd when compared to the curve above -- until you realize that they've used Vp=315V and held Vs to 225V.

OK, I understand the lines at various grid voltages and the red dots representing 14W. I understand that each of the 3 straight lines represent one of 3 load resistance choices. But I don't understand what those 3 lines actually represent. What relationship is represented by moving along one of those 3 lines?

I think OTM's response is more along the lines of "what does it mean", but in the context of the meaning of life I can explain how the graph works, but I can also see the next question around the corner -- "Out of the graph, which one's the right answer" and for that one, the answer is, basically, "all of 'em or none of 'em depending on your particular design goals". Pretty sure that this is what OTM is getting at. Anyhow... What the line is supposed to represent is the change in plate voltage and plate current that results from a change in control grid voltage when driven into a pure resistive load for a given screen voltage. So using the 3K4 line if we start with a quiescent point of -17.5VDC on the control grid, we would expect the tube to idle at approximately 38mA with a plate voltage of 300VDC. Raise the grid voltage to -15VDC and we'd expect to see plate current go up to around 48mA and plate voltage to drop to 270VDC or so. There are many aspects of the circuit operation that are not modelled here. These include the fact that we are driving into an inductive load in real life (which turns the straight line into an ellipse) as well as a variety of corrective factors that help predict where plate voltage will actually end up for a given B+ (here we've just assumed that Ep=B+ and ignored the corrections. We also haven't modelled any of the side effects of changes in current draw on voltage due to "sag" in the power supply). The thing I'd keep in mind is that you can spend a LOT of time trying to perfect your model's accuracy, but it will always be wrong (variations from tube to tube and other influences). How I tend to use them is to get a feel for how a proposed design will behave (am I exceeding ratings? will a particular load produce wildly asymmetrical output? Am I diving into the non-linear portion of the tube's characteristic? What range of voltages will my fixed bias supply need to provide? What size cathode resistor should I start with?) and then discover the actual operation through prototyping.

Thanx Todd for the explanation of the straight lines. I’m generally into graphs and tubes and all but never made the jump to fully understanding how some of the lines on these graphs represent their behavior. I also understand that tube amps (especially for guitar) often do not follow the “rules.” My Harmony runs the el84’s “too hot” and my SFDR runs the plate voltage “too high” under the rules. Another stupid question. Since I’m using cathode bias, do the absolute values of the Vg figures correspond to the cathode voltage? The Vg value is when compared to the cathode yes? And what about the “load resistance” vs “primary impedance?” I am changing the primary impedance (AC) when I change the secondary connections to the speaker since the OPT merely “reflects” that speaker impedance back to the primary. Changing this connection on the secondary has no affect on the DC (bias) current since I am not changing the DC resistance in the primary. Are all these 3-8k figures really AC impedance? Am I opening Pandora’s box? Thanx for the education guys and please pardon my ignorance.

Yes. Vg is not an absolute value, but to be interpreted as the difference between cathode and control grid. Since you're cathode biased Vp (plate voltage) is affected as well, but now we're starting to get into gory details. What the graph tells you is that for a bias point (or more accurately quiescent point) with Vg = -17.5, we'd expect approximately 38mA of plate current. Going back to the original curves posted by BK-Amps, it looks like we can expect screen current to be approximately half of plate current, so the cathode current (Ik) is the sum of plate (Ip) and screen (Is) currents: Is = 0.5 * Ip Ik = Ip + Is Ik = 1.5 * Ip Ik = 57mA Using Ohm's law: I = E/R 0.057A = 17.5V/R R = 17.5V/0.057A R = 307 ohms So if our various assumptions are correct, then a good first value for the cathode resistor required to develop 17.5V at the cathode with a 3K4 plate load and 300V at the plate is 300 ohms. Further, power is resistance times amperage squared. Safety/reliability tells us to double the rating of the resistor relative to the expected power dissipation it will see, so we get: (2 * P) = 2 * R * (I^2) (2 * P ) = 2 * 300 * (.057 ^ 2) (2 * P ) = 1.94W So you'd want at least a 2 Watt part. Now the evil part: Experience combined with some of the evil details that we've elected to ignore for simplicity's sake suggest that the actual screen current won't be that high. Less current requires a larger resistor to develop the same voltage. Less current is also a safe way to test. A decade box loaded with power resistors is always a sweet way to test, but I'm betting you don't have one of those. So, for this set of parameters, I would have a collection of cathode resistors ready to try. Standard values in that range are going to be 470, 430, 390, 330, etc. Start with something larger than you expect and tweak downwards until you hit the bias point you want. Not familiar with that particular transformer, but it's supposed to be set up so that if you use the 4K1 primary connections and the 16 ohm secondary connections and then actually connect it to a 16 ohm speaker then you should be able to treat the primary as having 4K1 impedance for the purposes of your design. Same goes for other consistant combinations. You can calculate the results of connecting things mismatched from what's labeled because, as you've correctly stated, the transformer is just an impedance matching device in this application. Pandora's box is opened when you start looking at the actual instantaneous impedance of the system in live playing conditions. The speaker's impedance is dynamic and depends on frequency, power applied, etc. The transformer has all sorts of parasitic attributes that make it frequency and power dependent too. Not to mention that the plate resistance of the tube changes as voltage swings Again, the trick is to keep in mind "right tool for the job". Interpreting characteristic curves is a great design tool, but will never tell you with any detailed accurately how the amp will end up behaving. It's a starting point. Same goes, by the way, for SPICE simulations. Great for finding ugly mistakes and testing for odd bits in the overall frequency response (and really handy for testing oscillation... take your SPICE amp model and apply a 100kHz signal and see if it oscillates ), but in the end just a tool. Everything proves out when you actually solder the stinkin' thing together and start tweaking.