Oscillation WE417A

[regal]

I know this has been covered before and have grid/plate/cathode stoppers, but what about the heater pins?

Ferrite beads a possibility? Having trouble finding a good part number for this option?

I don't have a sope that going to 200mhz to say for sure if there is an oscillation but the power output sems to drop off prematurely (i.e. ditortion jumps fairly quickly as the output ac is increased.

Also would a $300 DSO usb "scope" be a useful tool for validating oscillatin free operation?

 

[lakeside]

Yes, the heater on the 417A/5842 can be a source for problems, try bypassing each heater pin with a cap to ground, ~200pF.

 

[regal]

I did this, only had 470pf ceramics, went close as possible to the pin. My rudimentary distortion vs power testing showed an amazing improvement.

Prior the amp went into a sharp clipping going from .4V to .5V output across a 600 ohm load and now no clipping up over 1V.

I am really surprised to see such a dramatic difference. I usually listen around .3V and even here the sound is more natural.

Only issue is I had to take a shortcut and introduced a tiny groundloop since my dc heat is lifted, bypassing on the last return heater pin is a challenge as I have the heater supply in a separate chassis.

I guess with this tube one should connect the resistor divider ground lift directly on the tube's heater pin.

 

[DavesNotHere]

I take it your tying to get rid of a hum. a 20MHZ scope will see your power/noise issues 200Mhz is a little overkill. but lets talk about the nature of the heater winding. not all tubes' heater windings are built the same, but they cause anomalies within the tube. The most common is the imposition of the heater's AC on the elements of the tube. In AC heater world, it was common to "hum balance" the heater winding by placing a 120-250 ohm potentiometer across the heater windings and connecting the wiper of the potentiometer to B+ ground. The Hum Balance Control was born. The reason why this works is the heater winding is electro statically at a lower potential and in the case of common dual triodes (at, au, ax, sn,) and common power tubes (L6,V6,A7)the adjacent heaters are counter wound and thus thier opposing magnetic fields cancel any effect on the electron beam inside the tube and thus causes a focusing effect between anode and cathode. In circuits that use those signal tube listed above, it was cheaper to build the amplifier by substitution of the hum balance control with a resistor network of (commonly 120 ohm 1 W resistors and the center tap is reference to ground. Now on some circuits (cascode, SRPP, Akido) where two tube sections are placed one on top of each other the heater ground reference is tied to 1/4 of the B+ via a voltage divider network. this is commonly used to satisfy the cathode to heater voltage constraints.

 

[regal]

No audible hum actually, just mesureable, the original issue was power robbing RF oscillation (these tube are very prone to it). The fix suggested by Lakeside was to bypass the heater pins with small caps to ground which worked great, (the other pins were well stoppered.)

It is DC heat @ 12.6V for the two WE417's in series (LM1085) which I do have lifted about 20V' via resistor divider from B+. However the issue is the DC heater regulator is in a separate chassis, and connected to the divider there. So when I added the rf bypass cap's to the tubes heater pin I created a small ground loop.

Its really insignificant and inaudible but I can measure it and being a perfectionist I guess I will move the divider to the amp chassis, so I can maintain a proper star ground. Building headphone amps you have to be part RF-engineer as the drivers are extremely sensitive.

 

[DavesNotHere]

Now lets talk about DC. (Sorry fourm timed me out) DC heating has been used typically in instramentation, computer, and in professional audio/video and back in those days, car audio. DC powering like AC, has its own faults too, but if applied properly, in my opinion, it wins out for a much more stable and lower noise floor, and wider power bandwidth gain to distortion product. there are two ways you can attack the differential noise unbalance: electrostatic referencing: two power supplies use, they are never connected directly (even by chassis touching) to each other. The same voltage divider reference (hum balance circuit) happens in DC as was in AC. But the DC power supply's ground must be isolated to the b+ ground. the voltage divder tap (or wiper if it is a pot ) is referenced to the B+ ground or voltage divisor (in the case of cascode) either direct or capacitive coupling. Usually, if the heater's voltage divider is referenced to a voltage, it is capacitive coupled. If your circuit requires certain active rfi snubbing, both B+ ground and DC heater ground should be only connected to the main's nutral (wall outlet) or field ground (earth ground) by isolation capacitors (ex:.01) but not at the time unless you are battling noise on the signal ground issues and injecting differential noise. They should never be conected directly to them. Electromagnetic noise injection this form of DC heater use as to eliminate noise in previous high gain stages. the heater(s) from the previous stage is part of the final tube's cathode circuit. The heater's constant current demand is taken advantage of by the output tube by making a dynamic constant current source. Any noise generated by the B+ supply or the output tube's heater circuit (Any AC circuit or DC electrostatic referencing) is imposed back on the previous tube elements and its hum balance product (the wiper's position on the classic circuit) is derived from the capacitive coupled hum balance for the output tube's heater supply against the B+ voltage via capacitor with a resistor voltage divider network or a voltage divider network made from two capacitors in series. the voltage reference (center tap ) represents the amount power noise injecting into the output power tube which flows back to and removes itself plus noise in the previous stage electromagnetically via the heater windings in the cathode. DC power heater supply in all cases require them to be low noise and low hum. Battery chargers, AC wall wart style DC supplies, and laptop supplies have too much noise as all of these expecting the appliance you are plugging them into regulate and condition the DC voltage. Desktop computer supply, industrial switching supply, and industrial linear regulator are good sources for clean power. In all cases, they must have a capacity of 2-3 times more than the tube load demands. If you use a transformer and build or use a suitable winding off of the plate transformer, it must accommodate the current demands of the rectification, regulation and current demands of both the power supply circuit and the heaters. for others who need to know about more.....

 

[DavesNotHere]

rf noise causes positive feedback. sometimes tis is caused by the hash noise or tube noise imposing on the grid. high gain tubes like this one are all prone to them. grid snubbing caps (5pf-22pf) place between them and a signal ground (either B+ or B+ ground or hum balance reference) and plate snubbing (same application but on the plate). In pushpull output, you put the snubbing cap across the output transformer. in balance line, across signal positive and signal negative plates. there is a certain degree of noise that is in rf region of the tubes by nature. typically 1-10% of the amplified signal. this is caused by the electrons hitting the plate. reducing the gain of the stage helps lowering it and lowering the heater voltage can help too (but that depends on the tube and how high the input impedance and what gain is needed).

 

[regal]

rf noise causes positive feedback. sometimes tis is caused by the hash noise or tube noise imposing on the grid. high gain tubes like this one are all prone to them. grid snubbing caps (5pf-22pf) place between them and a signal ground (either B+ or B+ ground or hum balance reference) and plate snubbing (same application but on the plate). In pushpull output, you put the snubbing cap across the output transformer. in balance line, across signal positive and signal negative plates. there is a certain degree of noise that is in rf region of the tubes by nature. typically 1-10% of the amplified signal. this is caused by the electrons hitting the plate. reducing the gain of the stage helps lowering it and lowering the heater voltage can help too (but that depends on the tube and how high the input impedance and what gain is needed).

Yes but the tough one is snubbing the heater pins, they were the culprits picking up RF like an antenea with the long wirings to the LM1084 DC powersupply, parasitic power loss. I had well stoppered all the other pins.