Abnormal Current in SE EL84 on first power up

[JoeTele]

Ok, so I replaced the wood board with vulcanized fiberboard and and wired the EF86 screen to the right side of the EL84 cathode. First the good news:

The voltages look good! Just over 20 on the EF86 plate (plan calls for 20) and 30 on the cathode (2v over spec). I'll ultimately have to tweak a couple other dropping resistors now that I have voltage data from real world currents but all in all I'm stoked.

The bad news: Once I get up to around 90-100 volts on the variac, the OT starts to sound like Ant-Man's weedwacker, giving off a small scale motor like sound whose frequency changes as I up the variac, and when I probe the power rail with my meter. It's hard to tell, but it sounds like maybe the EF86 is buzzing too, though it could just be their proximity. There is usually a delay from the time I hit the triggering voltage on the variac and the onset of the buzz, though I suppose that could just be the amp warming up. Thing is, I measured almost 8v AC across the OT secondaries, which up to now have only been connected to dummy loads. Could my early high current overloads have done in my OT? Also, when I turn the tone and volume knobs the frequency changes. It's audible enough that you could almost make music with them, and again this with no speaker. I've attached a picture of the OT secondaries on my oscilloscope, which was set to 1ms per div. This starts at the EF86 plate. Any ideas?

Joe

[JoeTele]

Oops. Here is the aforementioned picture.

[JoeTele]

Edit: That ac is also on the grid of the EF86. Grounding the grid makes the buzz go away. I'm using shielded cable, grounded at the input jacks. I actually use 2 rca jacks even though it's a mono amp and connect both to the same volume pot terminal, which is the input in this circuit. This so I'm not always needing stereo to mono converters, and I've measured losses across such adaptors. Also, my filament winding is attached to the hot end of the cathode, rather than ground in case that's worth knowing.

Thanks!

Joe

[JoeTele]

I think I may have got it! I switched the the OT primary leads (whose orientation I never had an authoritative source on to begin with) and that seems to have done the trick. No more buzz, no more voltage across the OT secondary. I'm wondering if the proximity of the OT to the (as yet unshielded) EF86 was causing the oscillating transformer to induce a signal into the tube? I'm going out on a bit of a limb with that one, and would love any insights but I think I'll start testing with signal conditions and for noise in the speaker.

Joe

[JazzGuitarGimp]

From the schematic, it looks like the design has global negative feedback. So maybe it was just wired backwards in the beginning?

[martin manning]

I think I may have got it! I switched the the OT primary leads (whose orientation I never had an authoritative source on to begin with) and that seems to have done the trick.

Well sure, if the phasing of the OT is reversed the global negative feedback becomes positive feedback, and the whole thing becomes an oscillator. I would keep the dropping resistors as shown in the schematic and only adjust R15 to get the specified 290 volts at the EL84 plate.

For turning stereo into mono, I would not connect the stereo outputs directly together; the output stage of the stereo device may not like that. It would be better to include a series resistor from each channel, say 470 ohms to 1k.

[JoeTele]

Thanks for the stereo/mono tip! I pretty much changed all of the power resistors to accommodate the 5AR4 I'm using in place of the EZ80. Now that I have near-correct current levels going through the amp, I'm going to remeasure everything and tweak my resistor values to get things to spec. Most of what I have was based on using the Duncan PSUD which got me pretty close (3-5 volts off on first filter cap, EF86 plate, and EL84 cathode) but I don't think I ever had a good estimate of the EF86 current.

[martin manning]

If you are 3-5V off at the first filter, I'd say you are done. It might be 5V off in the opposite direction tomorrow due to line voltage variation.

[JoeTele]

Ok, so I've done some tweaking of resistor values. My main motivation for doing so was that despite being only a few volts over on the first filter cap, I was not getting anywhere near the effective drop on the plan for this amp, which stated 310v at first filter cap but 290 on the EL84 plate. I was only dropping 5-6v through the OT primary so my few volts over at the cap meant about 310v on the plate. So, I changed my first dropping resistors to get the plate right around 290. The EL84 was dissipating over 16w at that point, so I proceed in working to drop the EF86 plate from approx 25w down closer to the stated 20. Of course, as I got there the voltage on the DC coupled EL84 started going up, pretty much keeping me just shy of 16w at the EL84 plate (a bit less given the calculation was based on cathode voltage and didn't take into account the current out the screen), which was back up to 300v. I calculated that if I add enough series resistance with the plate only to bring it down to 290v, I'd be just over 15w dissipation (minus the screen factor). That said, I'm at 10v over spec on the plate, 2v over on the cathode, right on at 290 on the screen, and at 21.5v (as opposed to the stated 20) on the EF86 plate. Good enough to start testing with signal conditions?

Also, do the bigger 5w power resistors run cooler than the 2w metal film, or just handle more heat? I've only got 1.4w through the two-in-series 2w first dropping resistors, but the head and smell are still a bit unnerving.

Thanks!

Joe

[martin manning]

I see, so with the first filter voltage very close to the published value, the EL84 still had too much current. Which resistors have been changed, and what are the new values? As you have found, there are a lot of interactions that combine to set the DC operating points for the tubes.

It's typical to select resistors rated at twice the actual power that will be dissipated. For a given resistance and current, the power dissipated (watts) is always the same, but since resistors with higher wattage ratings will be physically larger, the surface temperature required to dissipate that power will be lower.

[JoeTele]

Ok, so I did one final step down of the first dropping resistors. I've got about 490 ohms off the 5AR4 (two paralleled 2w/1k which each measures a little shy of spec) for about 298v at the first filter cap, and 292v (290 stated) at the EL84 plate, 280 on the screen (290 stated) and just shy of 30 on the cathode (28 stated) and now just under 21 on the EF86 plate (20 stated). I recalculated my EL84 plate dissipation, taking off 5% current for the screen, and am floating right around 14.5 watts. As it happens, my EH EL84 is rated for 14 watts. I know I could increase my cathode resistance a bit to bring it down further, but for class A operation, I wonder if the 14.5 is ok for this tube?

Though I'm at 1.4 watts through those very hot 1k dropping resistors (4 watts with the 2 in parallel) I think I'm going to replace them with bigger power resistors.

[martin manning]

Are the other resistors in the power supply (3k9, 390k) and the EF86 plate load (1M) still at the schematic values?

As long as you don't get red-plating you might be okay. To reduce its dissipation you could try lowering R9 to see what effect it has, like parallel another 22k across it. Increasing the EL84's cathode resistor is another possibility, be aware that the EF86's operating point will be affected too.

[JoeTele]

Plate load is the same. 3k9 on the schematic is 2k2 in my actual circuit and and the 330k in front of the EF86 plate resistor is 680k on my build.

Joe

[JoeTele]

Listening to her now through the bass reflex enclosure/Fostex full range 4" 93db driver I built to go with it. Outstanding. Will post a full review with pics as some people on a different thread about it earlier back were interested. Before I actually listened to it, I ran the o-scope through the stages to make sure all looked good, then ran it to clipping on the dummy load, did the math, and landed right at 3.5 watts. The volume is perfect for my apartment. Will monitor heat conditions for a while. The PT is plenty warm but I don't think at all unreasonable for a class A amp. I can clamp my hand on it for way longer than I care to before it even approaches uncomfortable.

Thanks again for all the help. I really learned a lot on this build!!

Joe

[martin manning]

The 2k2 substituted for 3k9 is probably not a good idea since it will increase the EL84 screen voltage and plate current. I don't think forcing the voltages to match the published values is necessarily a good idea either. The cathode bias on the EL84 and feedback to the EF86 screen should set the operating points within reason unless the tubes are way off nominal spec. Are these new production tubes or NOS?