Stage 1&2 oscillation

[dennyg]

I usually hang out on AX84 but thought I'd get to know the folks over here with my first post.  I celebrated my 2nd anniversary of amp building by tearing down my first build, High Octane SEL from AX84, and rebuilt using ideas and techniques I've learned since.  I fired it up and has that annoying high-end fizz with gain full on that I've suffered in several high gain builds (e.g. the AX84 Uber) that has usually been traced down to parasitic oscillations of some sort, mainly because I started using terminal strips/gothic ring layouts which I found to be more challenging (and temperamental!)   But this build uses an eyelet board and I didn't change the component layout, only the wiring.  One diagnostic tool I've used for oscillations is to open up the input jack with a 1/4 plug, no cable, turn the gain up while monitoring with a scope, and look for an oscillation build up.  Doing that with this build I'm getting feedback between stage 2 and input stage.  Oscillates with gain pot at 3 o'clock.  Disconnected the coupling cap from stage 2 to 3 to eliminate downstream broadcasting.  Used a battery for heater supply on input valve to eliminate noisy heater supply.  The power supply is well filtered, flatline on scope at the preamp B+'s, use UF diodes and CLC reservoir/filter.  I'm using high quality shielded cable from input jack, 10K grid stopper.  Can't eliminate it.  Any ideas?  Will gladly post a gut shot. 
Denny

[sluckey]

Did you rebuild as a Hi Octane? Did you put a resistor on the grid of the input tube?

[dennyg]

basic HO SEL circuit with KT88; added footswitchable clean second channel (feed stage 2 into single tone control to separate MV; also switches out stage 2 bypass cap for lower gain).  On the rebuild, I rewired such that most wires are now soldered on top vs underneath to make easier to mod/service.  I don't like the original orientation of the pre-amp sockets and may change that for first pre-amp tube so I can do more traditional wiring layout - Pic attached. 

[shooter]

these are some of the things I've learned here and tried with success;

250pF to about .001uF across plate R, or from plate to cathode.
"breaking" a shared ground between effected tubes n inserting a "new node"
decreasing gain per stage.

once the osc is tamed I usually have to go back and "adjust" tone some to make up for changes made to eliminate osc.

 

[terminalgs]

basic HO SEL circuit with KT88; added footswitchable clean second channel (feed stage 2 into single tone control to separate MV;

do you have a schematic with mods added?

[dennyg]

thanks guys - I re-oriented the tube socket 90deg such that the heater wires are approaching from the chassis wall - more traditional separation of wires for each stage too - and that cleared up the oscillations.  I've still got some heater buzz which I think is coming from the DC switching supply.   Humdinger doesn't kill it so I'll try adding some .1uf bypass caps. I was going to ground the neg side of the DC to chassis but I read somewhere that having a ground on either side of a bridge rectifier could short the diodes (the other side being the humdinger).  Any thoughts on that or other methods to clean up the heater supply other than resorting to a separate tranny? 

I've got a hand-drawn schematic for the clean channel mod - I'll shoot a pic and upload later. 

[terminalgs]

I've still got some heater buzz which I think is coming from the DC switching supply.   Humdinger doesn't kill it so I'll try adding some .1uf bypass caps. I was going to ground the neg side of the DC to chassis but I read somewhere that having a ground on either side of a bridge rectifier could short the diodes (the other side being the humdinger).  Any thoughts on that or other methods to clean up the heater supply other than resorting to a separate tranny? 

whatever voltage source you are using to heat the heaters should have a reference to the same ground that the cathode of the tube has, either directly or via a ground lift.


You might sketch your heater circuit with values and snap a pic of it for us as well.


You mentioned using a battery to supply heater voltage in testing, have you tried it since you eliminated the parasitic oscillations ?


If the heater hum increases as you turn up the volume, it's likely that the hum is being introduced before the volume pot.  If the hum is the same regardless of volume setting, its most likely to be post-volume control,  perhaps at the power tube.

[jojokeo]

Intrigued by the socket rotation helping the noise issue. Please see attachment. Also, what waveform or description can you give for seeing the oscillations you mention on your scope?

[dennyg]

Intrigued by the socket rotation helping the noise issue. Please see attachment. Also, what waveform or description can you give for seeing the oscillations you mention on your scope?

Yes, socket rotated per your question - pic attached.  note that I used a different socket and avoided drilling new holes, so could make the case that previous socket may have had loose lugs causing oscillation, but I do think a more conservative wiring layout allowed by the rotation was the remedy.  note in original pic that wires were crossing the heater wire which was a first build error.  The waveform of the oscillation was typical high freq (5K or so?) sine wave, which I could watch build as turned up the gain until around 3pm where it would burst into oscillation. 

[dennyg]

whatever voltage source you are using to heat the heaters should have a reference to the same ground that the cathode of the tube has, either directly or via a ground lift.


You might sketch your heater circuit with values and snap a pic of it for us as well.


You mentioned using a battery to supply heater voltage in testing, have you tried it since you eliminated the parasitic oscillations ?


If the heater hum increases as you turn up the volume, it's likely that the hum is being introduced before the volume pot.  If the hum is the same regardless of volume setting, its most likely to be post-volume control,  perhaps at the power tube.

Here's what I've done since yesterday:
* replaced humdinger with 100R's referenced to KT88 cathode; also disconnected the separate elevation circuit
* disconnected the DC switching supply - made no difference in the hash so reconnected (at lamp)
* the heater circuit is now pretty much the standard: secondaries from 270EX PT to lamp, heater wires (20gauge) tightly twisted run along chassis wall to KT88, 100R's to cathode, 22g twisted along chassis wall to preamp sockets
* there is no 60 or 120hz hum in the amp, only buzz, which is only audible with first gain (stage 1->2) up;  hash is characteristic switching transients but puzzling that humdinger had no effect on buzz in any position, although it did effect hum when turned either side of center. 

My other 2 builds are 12V DC but also ultra high gain designs.  The SEL is only medium gain and i am determined to solve this issue and create a quiet 6V AC heater amp!

BTW am loving this forum and the prompt helpful responses!

[dennyg]

I've got some new info regarding the buzz:
* Used a battery to power stage 1 heaters - made NO difference in the buzz - which is consistent with why the humdinger had no effect on buzz when adjusting the pot - so ruling out heater noise as the source (is this valid deduction?)
* moving the scope probe over the rectifying diodes produces the identical spikes as seen when probing each stage - inverted as appropriate per stage
* deduction is that the diodes are radiating crap into the preamp.  I'm using UF5408's. 

any suggested solutions? i saw a power supply schem on ax84 that used two N4007's in series on each leg, with .01caps in parallel. Is this 'snubbing' the diodes?  is this worth a try?

[PRR]

What is your wiring from PT HV and rectifier to first filter cap?

Show us a diagram, picture, or both.

The PT-rect-1stCap loop has BIG spikes. If the rest of the amp is connected unfortunately, the spikes are injected ALL over.

[shooter]

.01caps in parallel. Is this 'snubbing' the diodes?  is this worth a try?

Not sure on the diodes, but the .01uF are common way to clean up B+ in digital stuff.  I've started adding them in my builds, 1 at each preamp tap, usually on the turret board.  I used 4 across a bridge (1 for each diode) although I didn't have a problem to start with.

[dennyg]

What is your wiring from PT HV and rectifier to first filter cap?

Show us a diagram, picture, or both.

The PT-rect-1stCap loop has BIG spikes. If the rest of the amp is connected unfortunately, the spikes are injected ALL over.

Pretty much standard first ps stage:  PT-> UF5408's -> 100uf reservoir
I just replaced 1K to the next filter with hammond choke (see the flying black wires in pic which should be tightened up) and added a 100R balancing between first and second 100uf caps on ground side.  My inclination is to rebuild the first stage on a term strip tucked along back wall.  Also I used the larger 100uf caps just because i had them although most supplies seem to use smaller (33-47uf) caps - would larger caps have an effect on switching spikes? 

[shooter]

would larger caps have an effect on switching spikes? 

by the time you get to v1 tap it's normally low mV's ripple, spikes are high freq transients that need something small, hence the .01's peppered around digital boards.  I've been hanging mine on the board near where the tap is.  Are you using individual diodes?, if so just tack a cap in parallel and see what happens.

[dennyg]

Well I woke up early on black friday and took advantage of the wife being out shopping to get in some amp time.  Made a bit of progress on the hash/fizz issue but still needs more work.  I've reduced the hiss/buzz  a bit but still see the spikes when scoping stage 1 & 2 coupling caps, and still have fizz with pre-amp gains dimed.  Very prominent at low master volumes, less so as MV pushed and at higher volumes I don't hear fizz but high end seems a bit harsh.  I'm not sure if the spikes are causing the fizz, or if it's parasitic ultrasonic oscillations trashing the signal (nothing shows up on the scope), or both.  A couple more diagnostic bits:
* With guitar plugged in (humbuckers), I get some buzz and faint hum.  Touch the strings, goes away, although fizz remains when playing.
* With gain 1 dimed using channel 2 (i.e. stage 2 direct to one-knob tone control to second MV), the slight distortion is harsh so I'm staying focused on first two stages for now
* With input grounded, gain 2 (from stage 2 to 3) dimed and MV dimed, no hiss, buzz or hum.  Turning up gain 1 (from stage 1 to 2) brings on the hiss.

What I tried this morning, and results:
* strapped .01 caps across rect diodes.  No change in hash  nor hiss, nor fizz when played.  Removed caps.
* inserted a 47uf reservoir cap, left the 100uf cap in place for additional smoothing (added 100R between 47 & 100 caps).  assumption is that 47uf cap might reduce the spikes due to lower top-off required vs larger cap.  This did reduce the hiss somewhat but not sure if due to smaller reservoir or simply having an additional smoothing cap. 
* added .01 bypass caps to stage 1/2 heaters. No change in spikes nor hiss nor fizz.  I didn't expect a change since the humdinger didn't eliminate the spikes, but left caps in place since couldn't hurt.
* I used a .001cap probe-to-ground to bleed off high freq along the signal chain from input thru stage 4 (CF), while playing.  The effect seemed pretty much the same at all points - i.e. attenuated the spikes (but still visible), reduced the harshness on high-end, but can still hear muffled excessive fizz underneath. 

Still scratching my head.  Updated gut shot attached. 
I have some time over the weekend and would love to try some new ideas from you fellers

[dennyg]

Added basic DC heater circuit for the pre-amp tubes and I'm fizz free!!
The signal spikes were so darn text book switching transients coupled via heater that I figured what the heck I'll try DC. 
My conclusion to why I couldn't tame it with a humdinger is that these were common mode coupled. 
Is that valid? 

Thanks to all who chimed in with suggestions. 
Denny

[jojokeo]

Added basic DC heater circuit for the pre-amp tubes and I'm fizz free!!
The signal spikes were so darn text book switching transients coupled via heater that I figured what the heck I'll try DC. 

Kind of weird & interesting that hooking up your battery earlier (Used a battery to power stage 1 heaters - made NO difference in the buzz...) didn't work?

[silverfox]

This is an interesting concept: The rectifier transients are induced into the heater windings. If so, what size cap on the heater windings would filter that out without shorting the heater winding?

silverfox.

[dennyg]

JoJo - I wondered about the battery DC diagnostic too - so I reconnected the battery and while connecting it I realized potential problem with the test.  I was using long leads (14") with alligator clips on each end.  Sure enough, I saw noise on the signal when scoped.  The leads were antennas!  I shortened the leads to 6" and soldered them directly to socket.  No noise. 
Thanks for challenging me on that - otherwise I probably wouldn't have redone the DC test, nor learned yet another lesson the hard way!

[jojokeo]

It's a good thing for everyone to learn one more thing to file away by your experience and you clarified it nicely. But still wondering what that fizz sounded like compared to hiss and buzz? Thinking of the sound of a soda poured in a glass?

[dennyg]

I may not be using the proper descriptors but to me hiss is like white noise (like the old 5150's), buzz is more 'electrical' sounding noise - like if you hold your guitar near a computer.  To me hiss and buzz problems have generally been unrelated to guitar plugged in (i.e. present w input grounded).  I use fizz to describe excessive high end distortion on a signal, which in my limited experience has been caused by parasitics of some sort.  And I've never been able to see the 'fizz' on a scope, and deduce that ultrasonics are trashing the signal.  Signal looks same before and after fizz remedied.