Power Amp Analysis and Tuning Help Request

[silverfox]

I've completed, about 3 months ago, a slightly modified power amp section from the Weber 6s100. This design is based on the Sunn Sceptre S-100. The only significant changes are the use of 1/2 of a 12AX7 instead of the parallel 12AU7 driver tube, the output transformer (6.6K) and output tubes from KT88 to 6L6GC as well as running at a lower overall voltage. Part of the lower voltage level is due to the line voltage here being 110, through no fault of the power company. The amp is currently being driven by either an amp emulator straight into the power amp or the emulator into a mixer board with the mixer EQ and Boost giving the best overall sound, perhaps due to causing the EF806 to distort. In either case, the amp can get very loud and sounds good so there appears to be no substantial defects. My future plans are to add an Effects Loop or just another tube to get the input signal level up to where it should be if that can't be obtained by changing the existing circuit. I may also add Boss GE-7 Equalizer circuit and that puts out -15DB of signal gain, whatever that means. I need to make sure this is running at the best specs first however before adding anything else.

What I would like to know is

1. How to figure out what voltage level needs to be present at the PA input to drive the amp to full power, (maybe adding another tube with a level control);

2. How to determine the various drive signal levels within the circuit from the EF806 and the 12AX7 driver tube, (from what I understand that can't be directly measured at the grids of the power tubes);

3. Should I change any of the resistors to maintain the proper voltage relationships on the tubes in the phase inverter;

4. Have I missed something that needs changing.

Thanks for any assistance with this project.

Regards,

silverfox.

[shooter]

I'll take a stab at #1 n 2;
Your bias voltage gives you a good start at where your drive should be.  The tube data sheet gives you "peak AF drive G1", the data sheet I looked at shows bias at -18, grid drive at 18v AC Peak, or 36Vac p-p.  I typically want about 25% more than that, maybe 50% depending how *bent* I want my PA section.  I like cleanish pre, so I try and get a cleanish sine wave to the PA grid(s) at say 40-50Vacp-p for this example

[HotBluePlates]

I'll take a stab at #1 n 2;
Your bias voltage gives you a good start at where your drive should be.  The tube data sheet gives you "peak AF drive G1", the data sheet I looked at shows bias at -18 ...

Correct approach Shooter! However, remember to use Silverfox's actual bias in his amp; he may not be doing exactly what the data sheet says, so best to conform to the real amp.

1. How to figure out what voltage level needs to be present at the PA input to drive the amp to full power, (maybe adding another tube with a level control);

Assume the output stage of the circuit you copied is correctly-designed; if true, the output tubes will deliver their maximum clean output power when the control grid is driven by a signal whose positive peak brings the grid to/near 0v. So that's a peak voltage which equals the bias voltage (average of -28v bias in your diagram), because 28v peak plus -28v bias is 0v on the grid.

Most commonly used tubes have grid current which commences when the grids are slightly-negative of 0v, and typical guitar amps don't use driver circuit capable of sourcing significant grid current into the output tubes. The net result is a peak signal which brings the output tube grid to 0v.

Your question said, "How to figure out what voltage level needs to be present at the PA input to drive the amp to full power ..." Normally, I'd consider the "power amp input" to be the phase inverter, unless the inverter is a split-load design (your V3A is a split-load inverter). Because this type of inverter provides no gain, I usually consider the gain stage before it the power amp input; this conforms to where you've chosen to place your input jack and volume control (so good call!). A second factor deciding "power amp input" is the point where any speaker/OT negative feedback loop is returned; this comes back to your 1st stage cathode, so again on-target.

With that out of the way, let's consider how to figure "signal voltage for full output power" throughout your amp. We know how to arrive at that figure for your output tubes (noted above). For each prior gain stage, you take the signal level needed by the output tubes and divide by the gain of that stage to find the signal needed at that gain stage's grid for full output power at the speaker.

Phase Inverter:
Output Signal Required: 28v peak (to drive the output tubes)
Stage Gain: 1 (or less, 0.95)
Input Required: Output / Input = 28v peak / 0.95 = ~29.5v peak

The split-load and cathode follower are special cases where there will be unity gain or loss through the stage, giving a required input signal bigger than the output signal.

For the EF86, we need to estimate the gain of the stage. We'd take the 29.5v peak we just found, divide by the EF86 gain and know how much signal needs to be at the Master Volume wiper for full output. This is complicated by the unbypassed cathode resistor (and overall negative feedback, as we'll see later). The easiest way to know a pentode's actual gain is to apply a known small signal voltage (as from a sine wave generator), measure the output voltage and divide output/input to find gain.

Analyzing the pentode from a schematic is a bit harder. There's multiple factors which impact a pentode's gain, and even after we calculate it we'd have to consider how gain reduction from the negative feedback changes our result. But there is an easier way if we know the output section and feedback were properly designed... We just assume the whole output stage is a perfect opamp and figure what input voltage the feedback must have to deliver the full output power at the speaker.

[HotBluePlates]

Signal for Full-Output, with Feedback
Let's assume your amp makes 50w output when fully-driven. Your negative feedback is taken from the 16Ω tap, with a series feedback resistor of 4.7kΩ (R4) and a shunt feedback resistor of 47Ω (R15). The closed-loop gain of an ideal opamp with negative feedback is 1 + Rseries/Rshunt. We'll pretend that your amp is an ideal opamp over most of its frequency range.

Speaker Volts (RMS) = √(Power * Impedance) = √(50w * 16Ω) = ~28.3v RMS for full output power.
Gain due to Feedback = 1 + Rseries/Rshunt = 1 + 4.7kΩ/47Ω = 101
Input to EF86 grid for Full Output Power = 28.3v RMS / 101 = ~280mV RMS or ~396mV Peak

Usually, we would apply more than this to the Master Volume control, to allow use of that volume to adjust for louder and softer signals. This control is an audio taper pot, so half-volume is also a loss of 1/10th (10% taper), so if you wanted full power output at half-volume to allow the ability to turn up for weaker inputs, you'd need 280mV / 0.1 = 2.8v RMS at your input jack.

Said a different way, you'll need an RMS input for your power amp between 280mV to 2.8v RMS for the power amp to seem like it has reasonable sensitivity for the Master Volume position.

[DummyLoad]

about 20Vrms to the grids of the output stage. 


--pete

[jjasilli]

+1 to prior posts.But, as Hotblue has mentioned in a companion thread, you're not giving us full info.  I.e.:  We know the B+ is down at the power tube plates, but is it also down at the earlier tubes?  If so, that can be corrected with smaller dropping R's in the B+ supply.  Without full voltage @ Weber's specs, the earlier tubes may be under-performing.  Presumably your power amp starts at the EF86 tube.  If so, you should have enough potential for drive, especially because the EF86 comes after signal loss from the tonestack in the full Weber design.  Hotblue give pointers re biasing at the various tubes.You can test empirically:*  If you plug an el guitar directly into your amp, does the amp fully perform or is it anemic?  *  If you have a signal generator, insert signal at some guitar frequencies into your amp.  Turn up the signal generator until it begins to cause the amp to overdrive or distort.  (1st make sure your wife isn't home.)  Then measure the voltage output form your signal generator and you have you answer.  Caveat:  this does not tell us at what stage(s) in the amp the distortion is occurring.  For that you need a listening amp, a/k/a signal tracer, or a 'scope.  If you don't have these things, which may require a learning curve, then the process of design is hampered.

[HotBluePlates]

... The only significant changes are the use of 1/2 of a 12AX7 instead of the parallel 12AU7 driver tube ...

If you changed the 12AU7 to 12AX7 for "more gain" hopefully you've seen that it's not worth the bother in this circuit. The split-load topology always has a gain of 1 (or actually a slight loss) regardless of the tube used.

3. Should I change any of the resistors to maintain the proper voltage relationships on the tubes in the phase inverter;

No, it's fine as-is. The inverter just has plate and cathode loads, and is biased by direct-coupling from the EF86 stage's plate. The inverter settles in on whatever current it requires to operate. Looking at your schematic, the voltage across the loads and the tube itself are plenty to allow a 20-30v RMS output signal to each output.

... My future plans are to add an Effects Loop ...

There is no sensible place to add an effects loop within this circuit. Typical guitar amps insert such a loop between the preamp and the output section (which you can take to mean "input to the phase inverter"). In your amp, this loop insert point is your input jack before the Master Volume control.

... The amp is currently being driven by either an amp emulator straight into the power amp or the emulator into a mixer board with the mixer EQ and Boost giving the best overall sound ... My future plans are to add ... another tube to get the input signal level up to where it should be ... I may also add Boss GE-7 Equalizer circuit and that puts out -15DB of signal gain, whatever that means. ...

Now that the input sensitivity of your power amp has been established, the best course would be to understand your complete signal chain from guitar to power amp. This probably includes knowing what type of pickups you use (single coil/humbucker, etc). That would answer the question "do I have the right levels throughout?" which is what I was getting at with the Opamp Thread.

[DummyLoad]

there is no way Vg2 of the EF86 has 277V with a 1.5M Rg2 connected: likely that is about 30V.


--pete

[HotBluePlates]

there is no way Vg2 of the EF86 has 277V with a 1.5M Rg2 connected: likely that is about 30V.

I estimate it at 80v, but I assumed these were measured voltages.

I figure screen current for the EF86 should be ~1/6th of cathode current. Using plate current as 5/6th of cathode current makes the math work out well for the marked plate and cathode voltages.

If there is a measured 277v on the screen, and using screen current is 1/6th cathode current, I'd guess there is a 500kΩ screen resistor instead of 1.5MΩ.

[silverfox]

Thanks for all the information and any to come. I've read it and gained some new insights and I'll be re-reading it several more times and then submit the amp to a signal generator and scope, check the 1.5 M resistor value on the Pentode and go from there.

One thing that is different in the schematic, but I don't think makes any difference at this point, for the voltage measurements: The 1uf EF86 screen capacitor in the schematic is in reality a .1 due to the fact I didn't have a 600 volt cap on hand at the time. I've got some now and will be changing that but like I said I don't think it is making much difference if any only slight to the actual tone.

Greatly appreciated,

silverfox.

[DummyLoad]

my estimates are looking at the schematic's telemetry - Ia is 1V/870 for ~1.15mA.  Ig2 is about 1/5 of 1a or ~ 230uA so then 230uA * 380V = Rg2 or roughly 1.65M: we're close, actual part is 1.5M, so drop across Rg2 is 230uA * 1.5M or 345V. 345V vRg2 minus 380V supply, so then Vg2 is about 35V.

approximate gain of pentode is  S * Ra: looking at the datasheets S is about 1.4mA/V with Vg1 = 1V and Vg2 at about 30V, so .0014 * 270000 ~ 380 we need about 22.2Vrms to push through the loss of the cathodyne and we have a gain of 380, 22.2/380 or about 58mVrms to drive to full power without NFB.

--pete

[HotBluePlates]

my estimates are looking at the schematic's telemetry - Ia is 1V/870 for ~1.15mA.  Ig2 is about 1/5 of 1a or ~ 230uA ...

I'm with you on 1.15mA, but this is Ik instead of Ia. Hence the 80v vs 35v guesses.

... approximate gain of pentode is  S * Ra ...

Yes, for fully bypassed screen & cathode. But the cathode here isn't bypassed, and the formula to predict gain below that for the bypassed condition is more than I wanted to deal with, since we have overall negative feedback. So that's why I just defaulted to assuming input sensitivity based on the feedback loop values (which holds as long as gain reduction of feedback is less than the total gain of EF86, output tubes and attenuation due to OT).

[silverfox]

If it helps any I liked the raw sound of the FB disabled so I consistently run the amp that way. I'll check that screen value now since it seems to me it should have burnt up the tube if it was that high; long ago.

I rechecked the voltages and somehow last night I got a bad reading or misread the meter. The voltage on pin 1 of the EF86 is 73.9 V. Everything else is the same. I will update the schematic.

silverfox.

[DummyLoad]

my estimates are looking at the schematic's telemetry - Ia is 1V/870 for ~1.15mA.  Ig2 is about 1/5 of 1a or ~ 230uA ...

I'm with you on 1.15mA, but this is Ik instead of Ia. Hence the 80v vs 35v guesses.

... approximate gain of pentode is  S * Ra ...

Yes, for fully bypassed screen & cathode. But the cathode here isn't bypassed, and the formula to predict gain below that for the bypassed condition is more than I wanted to deal with, since we have overall negative feedback. So that's why I just defaulted to assuming input sensitivity based on the feedback loop values (which holds as long as gain reduction of feedback is less than the total gain of EF86, output tubes and attenuation due to OT).

i meant Ik not Ia, sorry. 

g2 is bypassed: C11 connects to cathode. the presence network has more effect on AC bias than g2.

EDIT: the 10uF/820/5K pot affects cathode bypass more so than screen bypass since the screen bypass is connected to the cathode. in this case, the network is varying the effect of the Ck: in this case, it is local NFB and does not affect Ik. with the pot at or near zero ohms to ground, our analysis should be close.

--pete

[HotBluePlates]

g2 is bypassed: C11 connects to cathode. the presence network has more effect on Ik bias than g2.

I'm tracking with you on that. But the 820Ω cathode resistor is not bypassed, and will cause reduced gain due to local negative feedback (even before the overall feedback). That's the "unbypassed" part I was talking about.

[PRR]

> 22.2Vrms ... 22.2/380 or about 58mVrms

Having a hard time swallowing gain of 380 in an audio pentode. Not absurd but unlikely.

Philips EF86 data:
http://www.mif.pg.gda.pl/homepages/frank/sheets/010/e/EF86.pdf

Top of page 3. 400V, 220K load. Not quite the same values, but close.

Voltage gain is 210.

I would expect not-bypassing to give half of that. 100.

So a quarter-volt before NFB.

The NFB also says quarter volt.

So it has "insufficient NFB" to obey the simplifications. Actual input required will be higher, around 0.4V or half-Volt.

Which maybe makes the "Master Volume" useless here. Not sure what comes before, but it is likely to be able to hit MUCH higher signal without strain. If a plain tube stage, 50V is possible, you have to turn down Master by 100:1 so  the stage before the pot will overload before the power stage.

Anybody have a plan for a *pentode* gain power stage with useful Master Volume right before it?

It may be possible if there's a hi-loss tone-stack with the master volume.

[HotBluePlates]

... So it has "insufficient NFB" to obey the simplifications. Actual input required will be higher, around 0.4V or half-Volt. ... Anybody have a plan for a *pentode* gain power stage with useful Master Volume right before it?

We've been getting a piece-of-a-piece of the story. 

The schematic piece Silverfox posted is just the output section of WeberVST's 6S100 amp kit. And that kit is basically the Sunn 100-S minus reverb & tremolo, while replacing the 6AN8 with an EF86 & paralleled 12AU7, a modified feedback circuit and pentode rather than ultralinear output.

This is a big part of why I kept saying, "assume the power amp was designed properly." I sure didn't want to work out the details from scratch... 

[silverfox]

"I've completed, about 3 months ago, a slightly modified power amp section from the Weber 6s100."

I hope I haven't confused anyone here. The schematic and gut shot above represents the entire amp. There is no preamp section yet. I'll be able to get some scope shots up but not until likely later in the week. I will say the Presence control has little effect on the tone.

What sort of information would help resolve any questions? See also full schematic which I didn't want to confuse the issue with since I've made some changes to the original design.

silverfox.

[jazbo8]

Back-tracking a bit, it appears that the 6L6's are running way too hot, at least on paper, instead of 70% of Pda_max, it's about 118%. Also the 6.6k load is a bit high, so you are not going to get the maximum output from the 6L6's. I would recommed that you sort them out before working on the rest of the circuit...

[silverfox]

I just checked my handy little bias setting chart for 6L6 running at 390 and the bias for that value is listed as 30- for Class AB. Let me know if I've misinterpreted something. Wait, I'll see what the Internet says further on the subject: Bing.com bias setting 6l6 390 VDC "enter"' working.... What I found is, to get the tubes operating at about 70%, the bias should be -44 so if that is correct, you're right. Up to this point I've been using a chart, however you've just gotten me to do the math as follows- 25watts x 70% divided by plate voltage = bias value. At -29 those tubes should be running really hot? (fortunately I've still got the cheap Chinese tubes in that I've used, until I felt the amp was set up correctly.)

I guess the chart I'm using is bad information and or I don't understand how to use it. See chart below.

What gives?

silverfox.

[jazbo8]

Not sure where that chart comes from, but it's not so helpful... the most important parameter Eg2 is missing altogether. Anyway, since you have the 1R current sense resistors installed, you can dial in the bias precisely.

[Willabe]

Not sure where that chart comes from, .......

Isn't that from Dan Torres 'Inside Tube Amps' book?

[DummyLoad]

plate dissipation of 5881 (that chart) is 23 watts. chart also states that it's for 6L6 which has a 19 watt rating. 400V seems to be the limit for class A operation.

V*I=p/pmax= %*100 = % of pmax

the class A for 390V is about 88% of 19W.
the class AB high for 390V is about 75% of 19W.
the class AB (more B) for 390V is about 63% of 19W.

chart seems to be plotted for 19W dissipation?

--pete

[silverfox]

Well fortunately the tubes that I used came out of a Peavey Road Master 150 watt amp. They are 6L6 "GC". I've built this amp design before and didn't like the wiring job or the bias circuit so ripped it apart and rebuilt it. Before I did I seem to recall it Red Plating and I guess that's why. The other tubes were most likely not the GC series and the plate voltage was even higher. Also I mentioned above the line voltage here is at 110 but in the past I've tested the amp builds with a variac and ran the voltage up to 120 which had a significantly higher output voltage on the power supply- 425 as I recall on the B+.

I've learned quite a bit over the past few years but obviously have a long ways to go. Thanks for all your patience in assisting me and to HBP: I guess this is another piece, the bias and tube types that is to say, I don't think I mentioned the "GC" designation previously. I was under the impression a 6L6 tube was a 25 watt tube and that the "GC" rating was perhaps related to resistance to a flash over at higher voltages.

If the bias is supposed to be set at around -44 that might explain why it is so loud at half volume. I should get to work on the amp Thursday and get more details.

Regards and thanks,

silverfox.

[jazbo8]

Not quite -44V, more like -35V if all the B+ voltage remains about the same. You can dial it in by using the formula that you and DL posted earlier. For class AB operation, anywhere from 60-70% Pda_max (which is 30W for the "GC") will be fine.

[PRR]

> bad information and or I don't understand

You may be mixing-up Current and Voltage.

Easier to do here because they happen to look like similar numbers.

I would expect about 45 VOLTS grid bias in a 400 Volt amplifier.

For a 6L6-size tube, this may be around 44 milli-AMPS. Or it may not.

You generally want to get the CURRENT in a good zone, let the bias Voltage fall where it may.

> impression a 6L6 tube was a 25 watt

Original 1937 6L6 was rated at 21 Watts then quickly re-rated 19 Watts. Unless you buy olde junque off eBay you will never see one of these.

For historical marketing reasons, 6L6G and other types through the 1950s are rated 19W-21W but may stand a little more. These are all rare now.

KT66 was specifically aimed to cover 6L6 uses in Europe but has always claimed 25W. This may be conservative.

6L6GC is in fact an All-New tube. The H-sweep in a TV set does hard work. 807/6L6 was tried, but could be better. There was MONEY in TV production so massive development work was done improving the 6L6 guts. Several variants for specific designs. The 6BG6 retains 6L6-like gain but much more rugged. The top-cap needed in TV H-sweep work is omitted in the 6L6GC for audio/servo work. The H-sweep variant has a 20W Pdiss because when a H-sweep goes bad it goes VERY bad, so they need a lot of safety-margin. Audio amps don't tend to go totally wild so easily (unless bias fails or is over-adjusted without watching the current); 6L6GC in audio work is rated 35W Pdiss.

And then tube production collapsed.

And then it came back.

AFAIK, "all" present "6L6 types" are based on 6L6GC or KT66 internal dimensions and detailing, as modified by unhappy customers (Fender, GT, et al; not dweebs like you and me). They should take 35W for 30 days straight. The ones I have abused showed a hint of gas, I would not like to idle for years over 30W; but maybe I was unlucky, and 30W is still a big tube.

[silverfox]

PRR- "You may be mixing-up Current and Voltage." I use the 1 ohm resistor measurement method but confused the issue by stating the parameter in volts rather than mv or ma.

Testing Phase.

I did some work on the amp today, primarily getting the last bugs out. C11, the screen bypass capacitor was changed to the value in the schematic, i.e. 1uf, (previous value was .1uf). Not sure if I'm best suited to determining if  a change from .1uf to 1uf resulted in a significant change in base response. Would that be significant enough to hear or is it very subjective on my part? Sounds better to me and the presence control seems more responsive.

I've got my scope and signal generator set up but before I conduct the test this is my general plan. Let me know if I should change it.

Set bias to -43 to -44 mv using the 1R cathode method.

Do a listening test to determine what amp sounds like; (my thoughts are this is going to significantly change both the output power level relationship to the volume control position, and also to the tone of the amp?)

With the volume control turned all the way up, inject 1 Khz sine wave and as the signal level is increased watch for distortion at the output of the EF86(?), then record the voltage level of the signal generator and the output of the PI tubes? Measure the output voltage at the OT tap in use, for the purpose of determining output power just below the distortion level? Not sure if I should calculate where the output volts should be to achieve 35 watts and then watch for that to appear on the output tap. The reason for that question is, the iron for this build came from a 35 watt amp. I suspect it's good to 40 watts but don't want to burn it up.

This test should be performed with both FB engaged and with FB disabled?

Don't know if there is anything else.

Regards,

silverfox.

[jazbo8]

Not sure if I should calculate where the output volts should be to achieve 35 watts and then watch for that to appear on the output tap. The reason for that question is, the iron for this build came from a 35 watt amp. I suspect it's good to 40 watts but don't want to burn it up.

This test should be performed with both FB engaged and with FB disabled?

Vo = sqrt (PxR), since you have a scope, you can observe the output waveform by slowly turning up the volume until it shows slight rounding/clipping at the peaks and bottoms. A 40W dummy load is going to get very hot quickly at full power, so it should be cooled down with good ventilation and/or mounted on a large heatsink.

You should do the tests without FB first, assuming everything looks ok, then apply NFB and take the measurements again. You can check the AC balance of the PI, by comparing two outputs' amplitude - they should be roughly the same with and without FB.

[shooter]

for the purpose of determining output power just below the distortion level?

I like doing this either into a speaker dummyload(so I don't have listen) or the speaker, the voltage is way lower than tube plates, which *might* exceed your scope/meter rating.  Another reason, the OT might cause some distortion.

[silverfox]

  Originally, I must have set up the bias using the output from the pots rather than measuring the voltage at the 1R cathode resistor as the bias circuit output was -40 volts on both pots but the actual cathode current was 23 ma and 35 ma on the output tubes. It is currently reading 41.5 ma using the cathode resistor method but that is due to the B+ voltage being down at the moment. I had set it to 44 ma.

 

 If you have a signal generator, insert signal at some guitar frequencies into your amp.  Turn up the signal generator until it begins to cause the amp to overdrive or distort.  (1st make sure your wife isn't home.)

So I've got my daughter operating the camera while I observe the wave form and monitor the voltages, inject a 2 V 1khz sine wave into the  input, and crank the volume pot until the phase inverter starts to distort. The others in the house, in a separate room start complaining about a loud obnoxious noise- Is that you making that noise they inquire at the door? It was brief... No family members were injured during the testing of this amp. And of course it was so loud and obnoxious we were wearing hearing protection.
 

I'll take a stab at #1 n 2;
Your bias voltage gives you a good start at where your drive should be.  The tube data sheet gives you "peak AF drive G1", the data sheet I looked at shows bias at -18, grid drive at 18v AC Peak, or 36Vac p-p.  I typically want about 25% more than that, maybe 50% depending how *bent* I want my PA section.  I like cleanish pre, so I try and get a cleanish sine wave to the PA grid(s) at say 40-50Vacp-p for this example

Measurements taken with BK Digital Multimeter, (rms readings)

B+ 385 VDC; Bias setting at: 41.5 ma. The amp sounds much improved. I depend on my wifes FB since the only thing she knows is a difference in tone and is not influenced by theory or adjustments.



FB engaged at point 'G' the signal starts to flatten at 35 volts and point 'H' , 21 volts


No-FB engaged, at point 'G' the signal starts to flatten at 37 volts and point 'H' ,  27 volts



The amp sounds good and the numbers look good. I wasn't able to turn the volume pot all the way up and didn't get the input voltage at the grid of the EF86, so I'm not sure how valid the data is without that. As I recall it may have been a little over half way at the distortion point.


I didn't get an output reading although that would be easy enough now I suppose as all I'd have to do would be to crank it up to the distortion point and measure the input voltage as well as the speaker voltage on the secondary of the OT.

Does anyone have any comments or do I have to conduct the test differently?


Thanks for all the assistance, I've learned a great deal from this exercise,



silverfox.




.

[jazbo8]

I didn't get an output reading although that would be easy enough now I suppose as all I'd have to do would be to crank it up to the distortion point and measure the input voltage as well as the speaker voltage on the secondary of the OT.

Yes, you should measure the output voltage at the secondary of the OT, across the dummy load - the output could be very distorted. There seems to be very large imbalance between the two phases from your G and H readings. They really should be within a few volts from each other. The figures that you show on the schematic, are also a bit confusing, if both tubes are biased at 41.5mA, then the bias voltages should be different for both tubes (assuming they are mis-matched), but you show -41.5V for both. Could you please clarify the actual bias voltage for each of the tubes, as well as measure the output voltage across the dummy load?

[silverfox]

I have separate bias pots for each tube so the setting is not dependant on one pot. The actual bias voltages on the tube may very well be different. I used the 1R cathode resistor measurement method. I'll get a chance to take a look at things again today. I figured the mis match on the PI was wrong but it sounds okay. Perhaps the lower resistor needs changed out. I'll also use the variac so the line voltage is consistent.

Thanks for the fb.

silverfox.