EL84 pp blocking distortion

[anson]

the voltage swings are 42% of the plate swings. I'm not sure about the implications of being 2v higher at idle, but I suspect it's normal (lower dc resistance from taps)

That's right. Sorry about my misleading information. should have been clearer.

[imaradiostar]

Let's double check a few things.

The primary impedance of an output is measured plate-to-plate. You're using 7600 ohms into an 8 ohm load, right?

7600ohms/8=950.

sqrt(950)=30.8 so I would expect a 30.8:1 voltage ratio. Something has gone awry here. How are you measuring each of these voltages? You can measure turns ratio pretty simple by connecting a wal-wart or variac to the unloaded transformer and measuring the plate-to-plate primary voltage and the secondary voltage.

With a 16:1 ratio you're really using a 2k plate load- quite a bit below what a pair of EL84's are capable of driving.

As for the rms/sine/peak thing- it can be very confusing to communicate but let's make sure we're on the same page.

There are multiple ways to express voltage. You can't have "peak rms." RMS is a measurement based on the energy delivered to a load. A 1 volt rms sine wave delivers the same energy to a load as 1 volt DC. It takes a sine wave that measures 2.8 volts from top to bottom to deliver this amount of energy. If you measured from zero crossing to the top of the sine wave it would be 1.4 volts. Same from zero crossing to the bottom. This is where the term "peak" comes from.

As it applies to undistorted sine waves, 1 volt rms = 1.414 volts peak = 2.828 volts peak to peak.

jamie

[ValveMan]

Hi Bruce,
A question on the power supply: running 2 12AU7's (2 gain stages & LTP) and 2 EL84's - what current draw would you plan for?
I may have underestimated - using a 5Y3GT & a Hammond 272BX 300-0-300 @ 115mA
The 12AU7's pull ~ 12mA between them leaving ~100mA for the EL84's (I thought that'd be plenty for PP & UL mode, but now I'm wondering . . .)
Cheers

[BiBi]

The 115mA is current before it's rectified. Here's a helpful source to determine the current in DC-- go to page 11: http://www.hammondmfg.com/pdf/5C08.pdf
I had a hard time figuring that one out too!

[ValveMan]

Still can't get over 160V swing on each side. I did fix the blocking problem though - used the zener trick on the grid - also across the cathode to lock the bias at 14V.

At 12Vpk-pk (sine wave - not RMS), there's 160Vpk-pk on each side. The current draw (for the 84's) goes from 60mA to 78mA from the total draw of 90mA for the whole circuit (shunting the current at max signal before clipping, I measure 95mA AC pk-pk). From previous tests, I've been able to pull 120mA from the power supply so there's obviously some headroom there.

I know the test setup doesn't represent actual music, but I'm trying to prevent clipping on loud (particularly low-end) phrases. I could roll off the bass, but I don't want to give up the low-end response if I don't have to.

Ideas?

[imaradiostar]

I think you could sub in a different output transformer- one from a known good 2xEL84, 6v6 or 6aq5 amp. If the numbers you gave before are correct then you are looking at a substantial impedance mismatch. An EL84 can't "handle the weight," so to speak- it can't sink the current and needs to see a plate-to-plate impedance in the 6 to 9k range. A transformer substitution from a known good amp (or an off-the-shelf part like this) will eliminate a possible source of error.

jamie

[imaradiostar]

To add to that- an El84 is a pretty efficient valve- you should see the plates swinging from close to ground (maybe 30 volts) up to almost twice B+. Assuming a 330 volt plate voltage that'd be 600 volts available swing at each plate. If the tubes were seeing an 8k:8 reflected load (a common and conservative value for an EL84) the turns ratio of the output would be around 31.6 to one so we'd expect to see about 2*19 volts at the secondary- primary winding is two equal halves so we have to multiply the impedance ratio by two. This means 38 volts peak-to-peak (sine) at the secondary. 38/2.8=13.57 volts RMS at the 8 ohm load. Vsq*Z=W so theoretical max possible dead clean wattage out would be around 23 watts. The real world values would probably be between 15 and 19 watts.

jamie

[pdf64]

I agree, the issue raised in post #32 re OT primary impedance needs to be addressed.

[ValveMan]

Many thanks guys - OT seemed like the likely problem area, based on all the evidence so far, but I'm still struggling . . .
I'm using Hammond 1650F (7600 p-p impedance, CT, 25W rated), and I've got 6 of them on the shelf - all the same. I've double checked voltage, impedance & turns ratios (turns = 32:1).

I dropped in a new 272DX which gives an extra 25mA over the 115 I could get from the 272BX, dropped the bias back to 25mA per tube, switched to pentode (using 1k5 on screens) and I still top out at ~160V pk-pk each side with a B+ of 324V
I have the schematic cleaned up, so I'll convert to jpg & post later tonight
Cheers