circuit question

[shooter]

Econ101 says my winter build is gonna be a SE KT88 mono block PA.
I'm looking at drivers, below are 2 circuits, the penciled one I used on a QuadSE EL84 and worked well.
the other I found in my archives.

I'm reading Merlin's cathodyne PI, since the "B" side of the 6SN7 seems to be a 1 legged PI.(If I get rid of the bypass cap in the 300b version)

my question, since it's less than unity, would taping the cathode side "appear" as a better? impedance
load for the KT, or doesn't it matter? 

(pretty sure the driver R n C values will need to be tweaked for the KT88)

[HotBluePlates]

Can't evaluate any driver stage without knowing the output stage's details (especially bias voltage and whether negative feedback will be used around the output section) and the supply voltage that will be feeding the driver.

[shooter]

The crayon numbers I'm shooting for;
Typical self biased SE, plate at 400-450, idle about 90%, 80something mA, OT 5k, 40%UL .  Expecting a bias around 16 to 20. Big cathode bypass cap, input Coupling cap .1-.2, Grid R's unresolved, penciled in at 10k, 270k
PT is good for 150mA
It's meant not to get bent

EDIT:  I think I've found my answer from Merlin's site;
http://www.valvewizard.co.uk/cathodyne.pdf

But i'll need AOLT more time in the bathroom!!!

[PRR]

> SE ... cathodyne PI

SE, cathodyne, and PI do not go in the same sentence (or amplifier).

You only got one power tube. Why build a 2-output driver?

Just build a 12AX7 Champ. It does the job. Even with 12AT7, maybe with 12AU7 with changes. This covers most any small triode you find.

Minor quibble- as HBP says, you NEED to define what the driver has to drive. Champ's 6V6 power bottle can stand a 500K grid resistor. The KT88 types want a <220K when used self-bias. If you (foolishly IMHO) go fix-bias, 100K. The 220K is still in the range Leo targeted with his 12AX7 100K 1.5K operating point.

FWIW, I built a SE 6550 (same as) working 520V-600V to plate and 280V-320V for G2 and driver. Driver cribbed right off AA-Champ. NO problem getting ample drive. (Also I observed to real benefit from the heroic high plate voltage, stay with 350V-400V.

Your 6SN7-300B plan is not a cathodyne. It appears to avoid a grid-cap into the 2nd stage, but instead needs a big cathode cap. Wastes about half of available B+. I do not see the point of it.

[HotBluePlates]

The crayon numbers I'm shooting for;
Typical self biased SE, plate at 400-450, idle about 90%, 80something mA, OT 5k, 40%UL .  Expecting a bias around 16 to 20. ...

Gotta yank out a KT88 data sheet. Last page gives a curve for UL operation with 425v plate-to-cathode.

KT88 is 35w plate dissipation, so 35w/425v = ~82mA.  Call it 80mA idle current.  On the attached curve, I drew a horizontal line in red at 80mA, with a red vertical at 425v (split the difference for your range, and is the Vb for the curves).

You say you'll have a 5kΩ OT, and an 80mA peak * 5kΩ = 400v.  25v plate is obviously too low a plate voltage to sustain the current based on the shape of the 0v gridline, so you won't get quite twice the idle current as a peak output.  This means you'll get a bit less current swing, unless you reduced your load to something more like 4.6kΩ.  The 160mA peak plate current (twice idle) is denoted by the blue horizontal line.

The green diagonal indicates your 5kΩ load.  Between this and the 80mA point at 425v plate, you see your bias will not be ~16-20v, but more like 49v.  Even more if you want to idle less than 100% dissipation.  Your driver will need to deliver a 49v peak output, with plenty of reserve headroom and keeping out of the bent portion of the driver's curves.  That is, if you want to stay "unbent".

At this point, you'd want to reconsider whether the design is worth pursuing.

The data sheet condition for push-pull UL in cathode bias has 425v plate & screen, and indicates the bias is -50v, so my scribble checks out.  Also it shows Plate & screen current as 87mA per tube.  So screen current is likely near 7mA, and again checks out.  This suggests your cathode resistor for a single tube will be something like 50v/0.087A = ~575Ω (if you have a 5% resistor you might need to be on the high side of that tolerance range).  600Ω would definitely be safe.

To land at 425v plate-to-cathode after bias, you'll need 475v.  Does that jive with your proposed PT's raw a.c.?  Again time to validate your proposed components and possible design.

Questions one would ask at this point would be, "will I have enough supply voltage after decoupling to enable to driver to deliver +/-50v peaks cleanly?"  And if you haven't already bought the OT (or PT), you'd ask yourself whether a lower supply voltage would be of benefit.  Lower B+ lets you idle at higher current, which means less bias voltage, and also less drive voltage required to push the KT88 to full output power.  It would also mean a likely lower-impedance OT.

The overall lesson is while cut & paste design will get audio to the speaker, if you have to meet any particular performance target you will need to actually go through a design process to verify your assumptions.

[shooter]

Thanks, I will copy and re-read tomorrow!

It appears to avoid a grid-cap into the 2nd stage, but instead needs a big cathode cap. Wastes about half of available B+. I do not see the point of it.

It looked to me like a DC coupled CF, except the signal is off the Anode.  with both Ra and Rk = I looked at the circuit as, either or. 

Gotta yank out

I attached the data sheet I was "guessing" from.

I'll have to hunt up the link,(Le Gourmet), I did a guitar build using 2 6SN7s n a KT88, my B+ was lower, maybe a-k 365, bias somewhere around 20. That got me ~15W clean and ~23 totally bent.

This build is and audio PA section, 1Vrms in, something close to the 15W clean, then I have a JBL pro series horn that starts about 800hz to 20khz 100w 111 db spl.  The low end will be picked up by a Altec voice of the theater 15".  I have all the $ items on hand,  gonna start on a bread-board platform.

one of my oddities, I would like to give any tube BUT an AX7 some lovin

heroic high plate voltage

I have GREAT respect for HV, I've seen plasma arcs when 140Kvdc fails bad, a 400vdc poke is plenty

[PRR]

> driver will need to deliver a 49v peak output

Up-side of your proposed line swings -49V to -10V. So peak drive is 39V up-side. (Will be more on the down-side, off the chart.)

A happy tube can make peak swing of 20% of B+. So we need 39V/20% or 40V/0.2 or 200V of B+. This should be available with 400+V on the power bottle.

Power output looks like 15W (375V and 80mA peak) which aint bad for ~34W Pdiss.

Plate resistance varies over the cycle, but near idle Rp is near load impedance, Damping Factor before added NFB near 1, which is fine for guitar but low for hi-fi.

If you go toward 300V B+ 130mA 2K load, G1 bias is -35V and now you can swing to the zero line. So the peak G1 drive is not much changed (35V vs 39V). The main B+ has changed more than the G1 drive. We can still swing it though.

The "19W" on the Tung-Sol 6550 and Svetlana KT88 sheets is way exaggerated. IIRC it is at 13% THD (Svetlana omits that) and the tube current jumps-up severely (91ma to 123ma), screwing with the self-bias.

[shooter]

A happy tube can make peak swing of 20% of B+.

These are the pearls of wisdom I won't feed to the possums

That's the pencil work I was doing as I faded last night.  I attached a 6SN7 cheat sheet
the right 2 columns show V.G and Eo.  If gain is Vout/Vin, the chart shows Eo 34, Vg 14, so when I do the algebra thing, I get 2.4v in to yield a gain of 14 and Vo at 34.  Now, is that remotely correct?

The "19W" on the Tung-Sol

I have been using 42W max plate dissipation as my red-line when I use the 88, the audio Watts fall where they may.

[PRR]

>> The "19W"
> I have been using 42W max plate dissipation

Sorry. The Svetlana KT88 sheet claims 19W output; the Tung-Sol 6550 (same guts, taller glass) claimed 20W out. But these are very extreme conditions for a sweet speech/music amplifier.

Yes, 42W is a fine Dissipation for any good 6550.

Another possum tidbit: figure a well-loaded power pentode, SE, can deliver a cleanish Sine output 40% of the idle dissipation. (50% for a perfect devce, 40% for real pentodes, ~~25% for triode.) For 42W plate heat, hope for 17W sine output. The Tung-Sol 6550 condition at 38W Pdiss would be a fine 15W amp, but they used a low load impedance so on test-bench it bias-shifts to more DC power and more test output. And they push to a high THD number. Just boastful. (But H-K 257 will whup it.)

[PRR]

> I do the algebra thing

34/14 is indeed like 2.4.

Flip to the first page and see if Eo is peak or RMS. You will need a clearer idea when you figure if this will slam your power bottle. 34V peak may be not-enuff for the conditions we have been discussing. 34Vrms is 47V peak which is ample. Of course you could also go down to the Ebb=300V data (because you will surely have near 300V of B+ available in the chassis) and see that any interpretation gives ample to flog 6550/KT88 at reasonable G2 voltage.

2.4V at the driver, 0.02V typical input sensitivity, you need gain of over 100 between the input jack and the driver grid.

[shooter]

Flip to the first page

That file didn't ref measurements, I did find another that was indeed rms @ 5%THD

SE, can deliver a cleanish Sine output 40%

That one is forehead stamped!

0.02V typical input sensitivity

This build assumes and input of 1Vrms.  So I'm working out if I need 2 gain stages, or if I can do it with 1 and CF that into the grid.  It's looking like 2 gain stages.

I have 2 PT's available, one is 250, the other has 540 or 680, it also has a 120, 125 primary, and I believe I can mix-match the output taps.
got the breadboard started, copper-clad ground-plane, pwr jack, fuse block,  wiring diodes, caps, n R's, on/off switch.  Once that's in place I'll play around with what-works from the PT.

might take a couple days, Christmas II is Friday, and workin the weekend!

Thank you for all the insight, an the twin possums that eat the barn cats food are toping 20lbs each !!!

[shooter]

I remember in a PT when you step up V you step down I.  What I don't know is what happens to current when you put 120vac on the 220vac tap.

[HotBluePlates]

... What I don't know is what happens to current when you put 120vac on the 220vac tap.

Broadly, VA-capability stays the same so now that there is less voltage on the secondary, more current can be pulled from the secondary (and be supported by the core).

The above is limited by wire thickness.  The wire hasn't gotten bigger, so there is some maximum current at which it will melt (and before that, at which the wire will get hot enough to burn off the insulation).

[shooter]

stays the same so now that there is less voltage on the secondary

Thanks HPB, I was guessing that but wasn't sure.  My design Idle I draw should be about 60-70% of PT rating.  I did a quick Dummyload check and loaded, B+ is 390vdc into 5k R, calc I @ 78mA, close enough to target.

[HotBluePlates]

...  I did a quick Dummyload check and loaded, B+ is 390vdc into 5k R, calc I @ 78mA, close enough to target.

With filter caps? Or just the resistor?

If the latter, you'll see the average voltage output instead of the peak voltage output as you'll get when using a filter cap.

Double-check:
 - 220v primary with 120v applied, 540v and 680v (nominal) secondary taps (@150mA).
 - 680v / 220v * 120v = 371vac  ->  371vac * 1.414 = 525vdc
 - 540v / 220v * 120v = 295vac  ->  295vac * 1.414 = 417vdc

The above rectified voltages should be even higher when drawing less than 150mA from the secondary.

The average voltage output of pulsating d.c. (rectified but no filter cap) is 0.9 * vRMS or 0.637 * vPeak.  If you used the Red secondary wires, my estimate above turns to 371vac * 0.9 = 334vdc, but higher due to less than 150mA of loading.

Just want to make sure you'll get the supply voltage you think you'll be getting...

[shooter]

I've wired my PS like post #11

540v / 220v * 120v = 295vac  ->  295vac * 1.414 = 417vdc

out of the caps - no-load (R) I get 440vdc, (didn't look at AC, yet).  then I loaded it with the 5k and it dropped to 390vdc at the caps

[sluckey]

Are you concerned about the reduced filament voltages from the other secondaries?

[shooter]

reduced filament voltages

there you go, peeing in my cornflakes
Yes, I will be!  Thanks for that catch BEFORE I wired the PT!

So it's out to the barn to dig up another tranny

[PRR]

> what happens to current when you put 120vac on the 220vac tap.

My vote is "hardly changed". Current is limited by heat due to wire size, which is not changed. Lower core loss at the lower voltage means less total heat, so you could try 10% more current. But basically you are wasting half your iron/copper.

[shooter]

I played with what I have, and the closest I get, spot on, AND 6.3 fil, I'm back to 120 tap primary, BUT, I'm using secondary CT + RED, (11 and 8).  This, with the updated PS schematic in post 11 yields 400vdc into a 5k R D.L.

I left it on for just a couple minutes, it's late!   400vdc / 5kohm = .08A
If I run it for 10 - 15 minutes without excessive heat, or smoke, from the PT, Can I move on?
Ya, the DL will be diss 32W!  it's 50W so......, I'll just submerge it in my cornflakes

[PRR]

Transformer that large may take over an hour to reach final temperature.

[shooter]

I'm using the UL tap, is a screen R still prudent, (thinking DC and bad things more than sound)?

[jojokeo]

Excuse me for buttin' in but need to ask a question regarding the 88's lower screen voltage as it's listed in the DS. I remember back when SG built wet & dry rigs one of which was SE kt88. I don't recall him lowering the G2 voltage and think he left it up near the plate voltage? ie hooked up in the normal SE manner as most other amps. Can y'all clear up this screen voltage issue please? Is the lower voltage listed in the DS because of a hi-fi thing(?) and it's perfectly fine to not worry or address it being lowered and/or regulated for a geetar amp?

[PRR]

> listed in the DS.

Which datasheet you lookin at?

The Svetlana does not give the load impedance, a critical detail.

Consider: SE power is limited by Pdiss more than anything else. If you raise the plate voltage, you must reduce the plate current. But raising the screen _raises_ the plate current. Now you have to go way-way down on G1 bias voltage to keep the pig under control. Not a real happy plan.

So HIGH-plate-volt amps often lead to a lower screen voltage. But as screen current is very variable, power provisioning becomes an added problem.

I have gone as far as a dual-300V supply to get 580V plate, 300V screen, 10K load, 60mA, on a 6550 which is same-as KT88 except for height. I proved I wudda been happier with more sane values like 300V-350V supply and 2.5K loading at 125mA. At this point screen only needs to be 250V but 300-330V will be fine.

[shooter]

keep the pig under control

And that's a big goal here for hifi app.  I'm looking for the biggest practical clean swing I can get out.  Once that's working, I will probably experiment with seeing what happens when I drive it ugly.

[jojokeo]

> listed in the DS.

Which datasheet you lookin at?

The Svetlana does not give the load impedance, a critical detail.

Not load impedance - was referring to the low G2/screen voltages that I see in all KT88 datasheets (red arrow). On several schematics I've seen there has been a Voltage Regulator tube used for this (thinking it was an amp schematic called Maxim by Timbo about a year ago?).

Can you show/explain how to create a proper lowered regulated screen voltage the modern way (perhaps using a zener diode or two) in how this should be done if say a higher voltage B+ of ~485 was used?

[shooter]

Franco is asking the same, might find your answer here;

http://el34world.com/Forum/index.php?topic=21359.0

I'm more concerned if the tube starts flaking out my OT pays the price, so I was wondering if a G2 R would help, or just make Leo mad cuz I was adding things I didn't need

[PRR]

Why would you want a high B+ on a SE stage??

[jojokeo]

I'm more concerned if the tube starts flaking out my OT pays the price, so I was wondering if a G2 R would help, or just make Leo mad cuz I was adding things I didn't need

Adding too much R to the screen(s) and you start limiting output power which you don't want as it turns into a current limiter in a similar way as one is used to protect a LED. You want a lower voltage as the datasheets suggest but I've seen it mentioned as using a voltage regulator tube off of the high B+ to accomplish this. This likely was before high wattage zeners became available? Looking for PRR, sluckey, or HBP to weigh in?
 

*this doesn't necessarily have to apply only to SE kt88 use but rather also pp AB1 too. Trying to understand the concept and if it mainly applies to kt88s (& 6550) or not?

[shooter]

When my 2nd Espresso wears off some I should start getting answers!

He's the working schematic, and as yet fully tested bread-board.
hope to have initial numbers by this evening.

[shooter]

probably shouldn't have taken a break
2hours of self imposed troubleshooting later, (no grid ref R on the kt!)

I now have a different PT, so I'm lower V, higher I

P 305vdc  K 26vdc  G2 307vdc( )  I 104mA   Pdis 29W

I'll plug in the 6SN7 and warm up the scope.

[shooter]

   I have a mono-block!  needs a couple tweaks; minor hum, but no fil CT.
initial readings;

7.8VAC rms out of pre,  43Vac rms at 10ohm dummy load - clean. 

[PRR]

> 43Vac rms at 10ohm dummy load - clean.

185 Watts. Truly astonishing performance.

[shooter]

Truly astonishing performance

ok, i'll re-check that,
I bought a canvas yesterday and the whole time I was getting the breadboard up, I was like an 8yr old male that forgot to pee til it was almost to late
 

[shooter]

 
How's 3.8Vac rms sound?  I had my scope reference to system ground, but the speaker jack isn't!
so now I have to take my 30watt heater and try and get more than 1.5w sound.......after I paint

[HotBluePlates]

Excuse me for buttin' in but need to ask a question regarding the 88's lower screen voltage as it's listed in the DS. ... G2 voltage and think he left it up near the plate voltage? ... Is the lower voltage listed in the DS because of a hi-fi thing(?) and it's perfectly fine to not worry or address it being lowered and/or regulated for a geetar amp?

PRR answered the question with all the relevant dots, so I'll just connect them for you.

Screen voltage is mainly about 2 issues: peak plate current and idle bias.

You see in the left two graphs on the Svetlana data sheet that plate current for all grid curves increases if the screen voltage is increased.  For output stage design, you care first about power output, and are looking to select plate voltage (and implied plate voltage swing) along with peak plate current (and implied plate current swing) to get your target power output.  The OT primary impedance is selected as the impedance which allows the right balance of plate current & plate voltage swings (and also by availability).
  -  Example:  I'm looking for about 20w of output power, and my tube & supply voltage are such I can allow about 400v of peak plate voltage swing while leaving enough voltage across the tube to conduct.  400Vpeak = ~283Vrms -> 283Vrms²/4kΩ = ~20Wrms, which might be an available OT impedance.
  -  283Vrms/4kΩ = ~71mA of plate current -> 71mA * 1.414 = ~100mA peak plate current (above idle in a SE amp)
  -  If everything looks good up to this point, I'd be looking for what screen voltage supports my idle current plus another 100mA.  I'd find this looking at the triode curves where the grid is at 0v; plate is tied to the screen in triode-mode, so the "plate voltage" where the 0v gridline touches my required plate current is equal to the screen voltage I need to allow that much peak plate voltage.

The other issue is idle bias.  Assuming an otherwise-happy design, lower screen voltage in pentodes leads to "higher gain".  Low screen voltage lets G1 bias be smaller for the same idle plate current, implying a smaller input signal slams the tube as far as it can go.  In the Svetlana data sheet example, if the screen were 400v instead of 225, G1 bias would have to be more like -45v instead of -17v for the same idle current.  That implies a peak input voltage of 45v rather than ~17v peak for the same plate voltage/current activity, and a lower gain.

The Svetlana does not give the load impedance, a critical detail. ...

I think they had ~1.7kΩ in mind.

Since Max Signal Plate Current is only ~21% over idle, I'm assuming that's an RMS plate current.  19w/(105mA²) = 1723Ω

1.7kΩ would mean plate voltage swing is only 150mA * 1.7kΩ = 255v peak, so plate voltage could have been lower.  Or maybe they were thinking a B+ of 450v (with the usual trick of bridge rectifier, CT to midpoint of stack filter caps, giving an easy 225v supply point), cathode bias, and B+ voltage which sags a bit under the full load.  87mA * 450v is only ~39w, and less once you subtract bias due to the cathode resistor.

The whole design could be different (plate voltage near screen) with a lower supply voltage and a little higher load impedance.

[shooter]

Nice dot drawings, have you ever seen a Seurat drawing?
while I wait 2weeks for oils to dry;

Re-biased the 88 to 80% MaxPlate
Vp  295  Vk 23.4  I 121mA  Rk 194  (G2 not factored, yet)

Checked pre gain;   (1Vacpp in)
voltage gain 6.8
Changed grid leak on KT from 100k to 220k
Bumped gain to 7.8
Installed Ck on the 6SN7, (that I forgot)
Gain 13.8, just like the data sheet predicted!

The 88 is now putting out clean 19Vac pp, or 6.8Vac rms, (according to scope #’s), but I need 2.64Vac into the pre to get that.

To convert pp to rms with math, I thought it is pp * .707 = rms,  which yields 13.4Vac, which is twice what the scope says???  Any insight there.

Next,
measure signal at G1 of kt and see how much  loss from V1 plate to V2 grid.
Re-read about DC coupled CF, since my brain says there is “some” gain boost from that config.

[sluckey]

V_rms = .707V_peak    (That's peak, not peak to peak)

V_pp = 2V_peak

[shooter]

That's peak

AH!
so the other peak don't count

so what is the "best?" way to keep apples from becoming oranges, keep everything pp?

I do have to re-check the 19Vpp since that seems way high at the speaker 19^2/10 yields 36W! when the tube is idling at 32.8W

[sluckey]

so what is the "best?" way to keep apples from becoming oranges, keep everything pp?

Why bother? Nobody cares what the peak to peak power rating if an amp is. Least not since the '60s when Silvertone , Monkey Ward, and various other consumer electronics retailers were trying to pull the wool over us dumb consumers by advertising their products as 85 watts (peak to peak) power for a 2x6L6 amp. Hey, Fender and Ampeg could only get 30 to 40 watts from their 2x6L6 amps! Most of us were too ignorant to know that 85W_pp was the same as 30W_rms. That's a lot like false advertising! Thank goodness the honest manufacturers (and maybe some lawyers) figured out a fair rating system.

So forget trying to keep everything pp unless you're only gonna talk about signal voltages measured with a scope. If you're gonna talk power as relates to audio, then please use RMS. And use your rms dmm (not your scope) to measure the voltage at your speaker. Makes the math simple. If you insist on using the scope to measure voltage at the speaker with the intent to calculate power remember the relationships between RMS, peak, and peak to peak.

I do have to re-check the 19Vpp since that seems way high at the speaker 19^2/10 yields 36W!

See, even you are believing the slick advertising lie! 19V_pp = 6.7V_rms and that makes only 4.5W_rms into a 10Ω load.

[shooter]

even you are believing the slick advertising lie!

I did eat a couple special browniestoday but I could care less about anything but the design aspect and speaking in the proper terms for personal consumption, not commercial

19Vpp = 6.7Vrms

I have the scope set up to display true rms, so it reads 6.7rms, and 19pp and thats where the apples n oranges got messed up when I did the math both ways!.

If you're gonna talk power as relates to audio

I also want to talk grid drive, which is usually expressed Max Peak, so *.707 and I still get apples?

thanks for the skoolin!

[sluckey]

I have the scope set up to display true rms, so it reads 6.7rms, and 19pp

Ah, one of those fancy scopes. I thought you were making all measurements in pp and using a calculator to convert (like I have to do).

[PRR]

You put 33 Watts into the tube, and you get 5 Watts out?? Where it gets bent?

Something is quite wrong. With ideal parts you get 50% efficiency, or 16W out. With tubes the wave gets bent by 40% efficiency, 13 Watts out. You got half what you should.

What is your load impedance?

[shooter]

You put 33 Watts into the tube, and you get 5 Watts out?? Where it gets bent?

Idle current - no sig is 32watts.
In, well out of 1stage pre at plate is 13.8v with 1v in,

I'm using a 5K:8 edcor but using a 10ohm 250W DL

It gets bent when I apply 2.64Vpp to the input jack, so today i'll follow the sig frm V1 plate to V2 G1 and get those numbers, also if the HV probe works, 88 plate.

one of those fancy scopes

When I got outsourced an downsized, I forgot to drop off some equipment

[PRR]

> 5K:8 edcor ... 10ohm 250W DL

So you have 6.25K at the primary.

A quick-stab guess at "happy" impedance for tube working Vp 295  I 121mA is 294V/0.12A= 2,400 Ohms

Your load is almost 2.5X bigger than you need to suck out the power that tube has available for you.

> if the HV probe works

Tack 1Meg and 1K from plate to ground. You get milliVolts, multiply by 1000 and the truth is in Volts. 295mV or 0.295V observed is 295V at plate.

[shooter]

A quick-stab guess at "happy" impedance

Thanks, original "design" was for 400v but when I did my self imposed troubleshooting I changed to lower B+ PT, halving the output load, soon. (might have a 3K:8 in the barn)

I did do a run-through from input to 88 plate.  I did use pp since my "understanding" is up to this point I'm concerned with voltage gain, once I pass through the OT I want to be concerned with audio power - rms. no?

clean sin at 800hz
1Vac pp in = 13.4Vac pp at 88s grid = 168Vac pp at 88s plate
2Vac pp in = 28.4Vac pp    "     "      = 364Vac pp   "      "

I do want to "spec out" with both PT, 300v and 400v just because,

[HotBluePlates]

A quick-stab guess at "happy" impedance

Thanks, original "design" was for 400v but when I did my self imposed troubleshooting I changed to lower B+ PT, halving the output load, soon. (might have a 3K:8 in the barn) ...

Or get another one of those 10Ω resistors and put it in parallel with the one you have.

[shooter]

only have 1, but I do have 2 - 2ohms, and 2 .5ohms.
So I put a 4ohm load on
all readings pp, except at speaker (DL) that is rms

in    G1 - kt88  plate kt88  DL
1v     14.6v         142v       1.92v
2v     28.4v         274v       3.77v
3.36  47.6v         414v       6.10v
3.36 is the max I can get from sig-gen.  signal clean throughout.
math says 9.3W audio.  I'm hoping - shooting for 12 - 14w clean.

each tubes gain seems data-sheet correct 14.1 6SN7  8.7 KT88
plate at 292  cathode at 23.5  Rk 194ohms  Pdis 32.5W idle
looks like I need to add a small gain stage from V1b

[PRR]

> add a small gain

Change 6SN7 to 6SL7.

It will "work" on the same resistors, just to prove you like more gain; but best to find 6SL7 resistance-coupled amplifier suggestion chart.

[shooter]

I updated the working schematic in post 29.  I tested the pre section, got what I expected, gain of the un-bypassed stage 7.2, bypassed stage 13.1.  I can now get 100Vac pp out of the pre clean.  problem now is when I add the kt88 there is a 21vac 7.4hz modulation 
6sn7 by itself fine, kt88 driven with sig-gen fine, put them together and walla gremlins on steroids!

I'll try splitting the 2 halves of V1 with a new R&C on B+.
bypass the 2nd gain stage and throttle back the gain with a voltage divider.