New Build - Nona Rose KT66 amp

[WRC34]

Just finished a new build which I named 'Nona Rose'. It uses a Classictone 40-18095 PT along with a Thordarson 24S79 OT and a 4H 130mA Philco brand choke (used). 5AR4 rectifier, KT66 output tubes. The preamp is where I had fun - it's got two 6SF5GT tubes going into a 6SL7 phase inverter. The 6SF5 is like an octal version of 1/2 a 12AX7, single triode. It was very noisy at first so I resorted to using shielded wire and attaching the shield to the plate of the first tube (leaving the other end unconnected). This energizes the wire so I wrapped it in shrink wrap to be safe. It worked very well at taming the crazy hiss I was getting. I didn't expect the amp to be so vulnerable to noise only having two gain stages but it was. Through some very brief testing (I mean like a few notes and a couple of chords - we live in an apartment) I settled on a 918 ohm cathode resistor w/25uf bypass cap on input tube, and an unbypassed 5.6K cathode resistor on the second 6SF5GT. Will get a chance later this week to see what it can really do, cant wait.

I used the 312.5-0-312.5 secondary HV tap on the PT and got about 400 VDC on the plates. However, I experimented with various size resistors going to the first leg of the bias pot and found that the mA current available to the tube seemed to have a direct relation to the plate voltage. For example, with too low a resistor the KT66s were pulling waaay too much current but their plate voltage was also lower, as low as 375 VDC with an 22K resistor, and conversely the plate voltage went up to about 412 VDC with an 82K resistor (but the tubes could only get around 23mA of current, not enough) I finally installed a 56K which s stock for the JTM45 style bias circuit and voila - 400 VDC on the plates with 43mA easily dialed up on the bias pot. I was aware that all things inside a tube amp's circuit affected each other, but I didn't think it was so dramatic. Good stuff.

Oh, the speaker output jacks are wired for an 8 ohm load to 3.8K primary. This OT also has the option of 3.3K primary which could be mismatched for 6.6K, might be worth trying at some point.

Thanks to everyone here for helping me get to a place where I could build an amp like this.

[Stevem]

The hotter ( higher current) you idle the outputs at, the more you tax the PT and the lower you power supply voltage (V+) will be!
If the amp is not set up and biased for class A then the louder you play it the more the V+ will sag / drop.

[WRC34]

Stevem, thanks for the explanation. This amp is definitely not set up and biased for class A. It's set up for fixed bias and with KT66s 70% with idle plate voltage at 400 should be about 43 or 44 mA.

When you say "the louder you play it the more the V+ will sag/drop" this means that the current will increase resulting in plate voltage falling due to the demand?

Here's the big question - I know everyone has a difference of opinion on this and that's fine, but let's just take the garden variety fixed bias aim of 70% max plate dissipation, which is what I'm using to get my 43/44mA target (KT66 max dissipation being 25 watts, 70% is 17.5, divided by idle plate voltage)
SO, I'm setting the mA current at 70% when using the idle voltage, does this mean that as I crank up the volume, even though the mA draw and plate voltage will change, that I will at most be operation at 70% dissipation when cranked? Because obviously with the volume all the way down I'm not dissipating ANY watts, yet this is the situation (vol down) that I'm using to plug the 70% numbers into. Or am I wrong? I know this is a complex issue and the accurate technical answer may be difficult to properly describe, but I'm looking for an easily understood translation in working man's English!

[xtian]

Even at idle, your power tubes ARE dissipating energy as heat. This is what the max. dissipation rating in the datasheets is about.

This is different than the work being performed the speaker, which, when silent, is zero.

Your tubes will dissipate ever more energy the louder you play--the power supply will not sag enough to keep dissipation at your 70% mark. In fact they will dissipate well more than 100% of their rated maximum for short periods safely as you blast away.

[pdf64]

Putting out a full power square wave into a resistive load, the power 'wasted' as plate dissipation is minimal, almost all plate current gets delivered to the load.
That's the principle used for class D digital amps, to make them so efficient, ie the output devices are either full on or full off.
Although the above should be ok for the plate, its screen grid will take a beating.

[WRC34]

Great, this is making sense to me now. xtian, of course you are right, there must be dissipation with everything turned down otherwise the tubes wouldn't get hot and while they aren't sizzling, they certainly need a moment to cool down before touching them after being powered up to adjust bias.

I don't suspect I'll approach anything close to Class D but maybe it will be smart to dial back the bias to between 60-70% before I document the results of "blasting away" so that with the amp cranked I'm not going too far above 100% dissipation for any significant period of time.

Will report back shortly!

[WRC34]

Upon first test the amp sounded muddy and woofy, not at all what I was going for.

A couple of days later I was at a friend's house who has a fully set up work bench and a home studio. He had a band in the studio all day and let me use his tools to do some swapping of components. I didn't have much time but I swapped out the 25uf cathode bypass cap on V1 for a 3uf Solen fast cap, went back to a 1.6K cathode resistor on V2 (unbypassed), removed the cap going from the volume pot to V2's grid (.0047 value) and used it from V2 to the input grid of the phase inverter (originally it had .022)

It's got much more bite and sounds better, but is a bit noisy and still needs work. I'm thinking of re-installing the .0047 between the volume pot and V2 grid and installing a .01uf cap to the PI input grid. The cathode bypass caps on the first two tubes seems to have the most effect though, so I'm going to mess with those more.

[WRC34]

Have been working on this and it's sounding good. First triode now has 2.2K cathode resistor now with switchable .68uf or 250uf bypass cap, second has 1.6K cathode resistor with fixed 25uf cap. Coupling caps are all .01 now into PI, and from there to grids of output tubes are .033

Two of the three HV secondary windings are now accessible via a large DPDT switch on the front panel next to the power switch. This switch cannot be safely used while the amp is in operation. The amp must be switched off entirely, then the HV switch can be thrown, THEN the amp can be turned back on and biased appropriately.

Adding the HV switch on the front sort of necessitated a way to bias the amp more easily than having to remove the chassis from the box, etc. So i added external bias points and an external bias pot. I used a 34K 1w resistor and a 50K linear standard alpha pot mounted between the output tubes. I had a Fender type knob so boom it's on there.

Plate voltage with the 312.5-0-312.5 winding is around 404/405 VDC, and on the 345-0-345 it is 40 volts higher, 444/445 VDC. Screen voltage is one volt below plate voltage in all scenarios.

Lastly I swapped the choke for one from a failed AC/30 build, it's bigger and rated higher (10H, 110mA, 225 ohm). I was chasing trying to get screen voltage lower, but it turns out I may not have needed to do this. Either way I'm much happier with the results so far and will post a video w/sound clips soon.