Whistling causes

[Paul1453]

I've been trying out some simple two tube circuits on my breadboard.  I'm finding whistling issues with many different circuits/tubes.

I started with a simple EF86 6L6 circuit that whistled horribly.  As I roll up the voltage the whistling begins, usually starting around 75V.
With this circuit I added some capacitance on the 6L6 cathode and that seemed to help some, whistling started around 120V then.

Tried a simple 12AX7 6L6 circuit to see if the EF86 was the problem.  Nope.  This circuit whistles too.

So then I tried a simple 12AX7 EL84 circuit.  More whistling.
On this circuit I tried swapping in a 12AT7,12AU7, and my Hammond 7247.  Some of these whistle less.
With my volume control at 0, I can push the voltage up to 300V.  Then when I roll on the volume the whistling resumes.
The whistling occurs sooner with the tubes with the most gain.  With the 12AU7 at 250V I can get to about 7 on the volume before the whistling begins.

One thing I found that I did not expect was that all these circuits will make reasonable amounts of output with voltages as low as 50V and no whistling. 

Do you guys have any suggestions I can play with in these circuits?  I'm trying to learn how these changes effect my circuits.
I'd like to build up my understanding of these whistling issues with a few simple 2 tube circuits on my breadboard.

Thanks in advance for your input/suggestions.   

[John]

It could be picking up RF. Is your input jacked grounded just like it would be in an amp?

[Paul1453]

Yes.  I even put some metal shields on the tubes with wires to ground.

Another odd thing I noticed, was that when I was using my metal can 6L6 the whistling was reduced when I touched the metal can.
When I wrapped a bare metal wire around the can and connected it to ground it did not reduce the whistling.
That made me think I was adding some capacitance by touching the can.   

I couldn't touch the can for long, metal 6L6s get quite HOT.   

[uki]

Yes.  I even put some metal shields on the tubes with wires to ground.

Another odd thing I noticed, was that when I was using my metal can 6L6 the whistling was reduced when I touched the metal can.
When I wrapped a bare metal wire around the can and connected it to ground it did not reduce the whistling.
That made me think I was adding some capacitance by touching the can.   

What about follow some known circuits schematics for the testing, maybe you are missing something in the circuit, that is why it is whistling, don't know, look from another point of view? Think from a different angle?!

I couldn't touch the can for long, metal 6L6s get quite HOT.   

That must hurt OUCH !!!   

[HotBluePlates]

I've been trying out ... circuits on my breadboard.  I'm finding whistling issues with many different circuits/tubes. ...

Tubes could be microphonic, but the most likely problem is your wiring for the breadboarded circuit. We don't know because we can't see what you have.

Keep all wiring as short and direct as possible. Keep outputs away from inputs (for both individual tubes and for circuit sections/whole amp). Keep grid wires as short as possible; plate wires can be longer if needed. Keep output tube plate wires away from lower-level preamp circuits. Assume anything high-impedance (over couple-hundred-kΩ) will pick up radiated signal out of the air (which includes high a.c. voltage- and/or current-carrying wires)

[Paul1453]

I've been trying out ... circuits on my breadboard.  I'm finding whistling issues with many different circuits/tubes. ...

Tubes could be microphonic, but the most likely problem is your wiring for the breadboarded circuit. We don't know because we can't see what you have.

Keep all wiring as short and direct as possible. Keep outputs away from inputs (for both individual tubes and for circuit sections/whole amp). Keep grid wires as short as possible; plate wires can be longer if needed. Keep output tube plate wires away from lower-level preamp circuits. Assume anything high-impedance (over couple-hundred-kΩ) will pick up radiated signal out of the air (which includes high a.c. voltage- and/or current-carrying wires)

Yes, I know you are correct with your breadboard wiring assessment, HBP. 

I will work on shortening things, but are there any other avenues for me to explore?
I am seeking to learn what certain mods do to my circuits.

I have noticed that more gain = more whistling sooner.
I am currently investigating how I can add a NFB loop from my OT to my circuits to see how that affects the whistling issue.   
Maybe that's not the direction I need to go, but will help satisfy my curiosity about the NFB loops affect.

Any suggestions about adding or changing capacitance values, or other gems of tube wisdom would be greatly appreciated.   

[Willabe]

I will work on shortening things, but are there any other avenues for me to explore?

Can you post a few pictures of your BB?

I have noticed that more gain = more whistling sooner.

Sure, more gain = amplifying what ever the problem is faster.

Look at what HBP said again;   

Keep all wiring as short and direct as possible. Keep outputs away from inputs (for both individual tubes and for circuit sections/whole amp). Keep grid wires as short as possible; plate wires can be longer if needed. Keep output tube plate wires away from lower-level preamp circuits. Assume anything high-impedance (over couple-hundred-kΩ) will pick up radiated signal out of the air (which includes high a.c. voltage- and/or current-carrying wires)

[HotBluePlates]

I will work on shortening things, but are there any other avenues for me to explore?
I am seeking to learn what certain mods do to my circuits.

You really want to start from a perfectly-functioning platform first. Then when something changes after a modification, you will know the change was due to the modification. Otherwise, it could be a change due to changing how the original malfunction is behaving (i.e., too many variables).



Guy talks to his mechanic: "I want to build a race car."
Mechanic points out: "But a wheel fell off your car. You need to replace that wheel first."
Undeterred, Guy says: "Yeah, I'll get around to replacing that wheel. In the meantime, how much faster will my car go with changed gear ratios?"
Mechanic shakes his head: "Not faster, much slower... The car will plow into the pavement as soon as you step on the gas. Now about that new wheel..."

[Paul1453]

Well, I did get a simple 12AX7 6AQ5 circuit to work on my breadboard with no whistling or other issues.

I did rearrange and shorten some things this time.
Input and signal wires on the left, HV and output wires on the right.

That 6AQ5 is quickly becoming my favorite low watt output tube!   
This little thing Rocks with only the 5" 4-ohm speaker from a crappy 10W SS amp.

Now comes the really challenging part.
Getting the new PT, OT, tubes, and layout to fit properly in that crappy 10W SS amp housing.
Preliminary investigation shows it should all fit with some customization work, but as always the devil is in the details.   

[Willabe]

Well, I did get a simple 12AX7 6AQ5 circuit to work on my breadboard with no whistling or other issues.

All right. 

[PRR]

> "But a wheel fell off your car."

Spaghetti hay-wire on a HIGH-gain audio amplifier is more like oil leaking into the air, water leaking into the fuel, and the kids' soda oozing out of the radio.

Wire layout and exposure should be like a home water/sewer system. More-so, because pipes are kinda leak-free and treble always leaks all over. Keep the dirty stuff away from the clean stuff. Keep as short as you can for minimum seepage.

> Input and signal wires on the left, HV and output wires on the right.

Good. Sometimes insufficient. But with well and water lines on the east and poop-lines on the west of the house, I don't get much cross-contamination. And if I do, it is easier to find.

[Paul1453]

This one is a really simple circuit.  15 components besides the tubes as setup on my breadboard.

I use my Bench PS, and have found extra filter caps in the layout to be unnecessary.
Bench PS is B+, 33K 1W resistor makes B1, no E caps needed.

This one never whistled or hummed at all.  Maybe I'll post a pic later tonight.

Of course it didn't work 100% on 1st try.  Do they ever?  Usually not for me.

1st power up I was trying to use one of my little Philco OTs.  It was the one, (I got 2 from a junk gear purchase), that I smoked with the 6L6 circuit.
I didn't know how much these OTs could take and I was having whistling issues with the 6L6 circuit from about 75V anyway.
Once I got that circuit to stop squealing at around 200 V, I noticed a small puff of smoke escape from the OT.  I instantly cut the DC supply.
Pulled OT and let her sit a day hoping she might still be OK.  Checked R measurements of smokey OT vs it's twin and it appeared OK.  Same-Same.
Tried to use it in this 6AQ5 circuit 1st, to verify it survived.  Started PS at 0 and tried to give her juice.  Amps started rising before DC, should be other way around.
I knew I had killed smokey right then.  I put it's twin in and all was well.

So like with the B+, I always start with the gain pot at 0.  Power up to 250Vdc like data sheet states, and ignore schemo's V reading to start.  Around 30mA draw, cool.
Try the guitar, sounds great, clean and quiet.  OK, give her some gain.  Uh Oooh.  Something's not right.  Roll on gain, sound gets scratchy, current starts ramping up?
After quite a bit of messing with the 1000pf cap, switching in other values etc, I finally found the real problem.  Ground issues on the pot.

I am using a terminal block with a wire tying all the terminals together that is tied to my Bench PS Gnd.  Well that wire apparently broke mid block giving my pot and input jack intermittent poor Gnd.
After I found that no more issues.  I had been just clipping in different values for that 1000pf cap.  After Gnd fix I find I don't really hear any difference with that cap in or out or any value in between.

Can someone enlighten me on what that cap is supposed to be doing, is it really necessary, and if so what value should it really be?  Thanks!

[HotBluePlates]

I use my Bench PS, and have found extra filter caps in the layout to be unnecessary.
Bench PS is B+, 33K 1W resistor makes B1, no E caps needed.

You didn't use a filter cap after the 33kΩ (20uF in the schematic)? That half your whistling problem right there...

The series resistors and cap to ground, along with providing successively more-filtered d.c. for the preamp, decoupling gain stages from each other. If you omit those, there is feedback through the power supply from later stages to earlier stages. The feedback we care about is positive feedback from 3rd stage to 1st stage, etc.

Long story short, copy the power supply of your breadboarded circuit from the first filter cap onward.

Can someone enlighten me on what that cap is supposed to be doing, is it really necessary, and if so what value should it really be?  Thanks!

If you have it connected properly, it makes the sound brighter for any volume setting less than full-up.

[Paul1453]

I use my Bench PS, and have found extra filter caps in the layout to be unnecessary.
Bench PS is B+, 33K 1W resistor makes B1, no E caps needed.

You didn't use a filter cap after the 33kΩ (20uF in the schematic)? That half your whistling problem right there...

The series resistors and cap to ground, along with providing successively more-filtered d.c. for the preamp, decoupling gain stages from each other. If you omit those, there is feedback through the power supply from later stages to earlier stages. The feedback we care about is positive feedback from 3rd stage to 1st stage, etc.

Long story short, copy the power supply of your breadboarded circuit from the first filter cap onward.

Can someone enlighten me on what that cap is supposed to be doing, is it really necessary, and if so what value should it really be?  Thanks!

If you have it connected properly, it makes the sound brighter for any volume setting less than full-up.

Thanks for the info on that 1000pf cap.   

Please explain more on why I need those PS filter caps? 
My Bench PS is putting out well filtered DC.  Dare I say exceedingly well filtered DC.
No caps in there now and no HUM or Whistling at all with this 6AQ5 circuit.  0, nada, zilch 

I used to put them in before, and somehow it seemed to make HUM much worse.   
Along with making my breadboard setup much more cluttered.



OK, so thinking about what you are telling me.   

From my experience, the E caps in any PS are there to remove/reduce AC ripple from your rectified DC.
I never heard that they were also there to prevent signal feedback from one level of voltage to another.
Live and learn, I guess.  So if that is true, then I have a question.
If I have a normal amp that has whistling issues,
could I possibly eliminate or reduce that problem by putting say a 600V .047 regular signal cap in parallel with the E cap for that stage?
Is that what you are telling me?  Because I've never seen that done before, but it seems it just might work if what you are telling me is true.

[Paul1453]

Can you post a few pictures of your BB?

If you promise not to make fun of me.   

It still looks quite hectic, but this one is working great.

So from my Bench PS on the left I have my 6.3VAC to a terminal board with 3 outputs.
Next to it I have my Gnd and HV terminal board, with 1/4 100 ohm R's tied to Gnd from AC board.
So my white wire on far right of Bench PS is HV or B+.   It feeds my Philco OT and pins 5,6 on 6AQ5.
I then have a 1W 33K R feed a B+1 terminal strip.  It feeds pins 1,6 of V1 with 2 different Rs or voltages.
Last terminal strip is my GND block tied to the Bench PS GND.
Input feeds little black board for easy connections and changes.  Input is as far away from HV and OT as it can get.
The rest is just basically connecting up the components as called for on the schematic.

I know it doesn't look all that great, but it is working perfectly for this circuit.   

[Paul1453]

Can you post a few pictures of your BB?

If you promise not to make fun of me.   

It still looks quite hectic, but this one is working great.

So from my Bench PS on the left I have my 6.3VAC to a terminal board with 3 outputs.
Next to it I have my Gnd and HV terminal board, with 1/4 100 ohm R's tied to Gnd from AC board.
So my white wire on far right of Bench PS is HV or B+.   It feeds my Philco OT and pins 5,6 on 6AQ5.
I then have a 1W 33K R feed a B+1 terminal strip.  It feeds pins 1,6 of V1 with 2 different Rs or voltages.
Last terminal strip is my GND block tied to the Bench PS GND.
Input feeds little black board for easy connections and changes.  Input is as far away from HV and OT as it can get.
The rest is just basically connecting up the components as called for on the schematic.

I know it doesn't look all that great, but it is working perfectly for this circuit.   

Oh, 1 part I forgot to mention is the OT is feeding the 5" 4 ohm speaker in that black box.
That is the crappy 10W SS amp housing that I want to try to put this circuit in.   

[HotBluePlates]

Let me start by adding emphasis the the piece of my post which should have gotten the most attention:
 

I use my Bench PS, and have found extra filter caps in the layout to be unnecessary.
Bench PS is B+, 33K 1W resistor makes B1, no E caps needed.

You didn't use a filter cap after the 33kΩ (20uF in the schematic)? That half your whistling problem right there...

The series resistors and cap to ground, along with providing successively more-filtered d.c. for the preamp, decouple gain stages from each other. If you omit those, there is feedback through the power supply from later stages to earlier stages. ...

Everybody focuses on the cap-to-ground, with an eye towards power supply filtering. However look at what follows the first cap-to-ground at the rectifier cathode: a series resistor (or sometimes a choke) followed by another cap to ground.

Valve Wizard points out under "Smoothing Filters" that these R-C filters do 3 things (emphasis added):
1: Smooth/filter out residual ripple voltage;
2: Bypass/provide a local energy supply for sudden current demands;
3: Decouple/isolate each amplifier stage from the rest.

The power supply might be your source of d.c., but each tube stage has a current drawn from the power supply which varies when a signal is applied. So you now have a.c. traveling over the + and - wires from your bench power supply. A series decoupling resistor is inserted to form a relatively high impedance path back to the ultimate power supply + terminal, while the local filter cap (from perhaps the power-supply end of a 12AX7 plate resistor to the ground side of that 12AX7's cathode resistor) supplies/bypasses the short-term a.c. demands of that 12AX7.

Having a decoupling resistor and cap-to-ground for each gain stage (plus another decoupling filter between output tube plate and screen power supply nodes) keeps a later-stage a.c. from creeping into an earlier stage through the power supply. The series resistor also gives a convenient place to tweak (generally downward) the supply voltage for each stage moving towards the input of the amp.

The ideal is each stage gets its own decoupling resistor & filter; however, two successive gain stages (or 2 plate-loaded stages plus a cathode follower) are often hooked to a single filter cap as the builder notes that the signal a.c. of 2 inverting stages are opposite polarity and don't pose much risk of coupling through the power supply.

If you go into the Library of Information past the first handful of links to "Links to technical and amp service information" there is a download link for "Radiotron designers handbook 4th edition". Pages 535-538 present information about power supply decoupling.

... My Bench PS is putting out well filtered DC.  Dare I say exceedingly well filtered DC.
No caps in there now and no HUM or Whistling at all with this 6AQ5 circuit.  0, nada, zilch  ...

Your bench supply will filtered and regulated. However, no supply has zero impedance from d.c. to light, under all loading conditions (resistive, reactive, load current, line voltage, etc). It is the common power supply impedance across which signal currents interact to cause feedback through the power supply. Hence why you should still include the decoupling filters, to isolate individual stages rather than for improved filtering.

[Willabe]

That's not so bad. 

But, the wires after the plate coupling caps are now grid wires, they need to be as short as possible and/or shielded. Grid wires act like an antenna and pick up stuff from the air. That alone can cause whistling, oscillation and other weird problems.   

I take it the pot is for volume? You have that pot far away from the tube and those grid wire runes to the pot and out of the pot to the next tubes grid are pretty long.

Move the pot closer to the preamp tube and/or use shielded wire.

[HotBluePlates]

... I used to put them in before, and somehow it seemed to make HUM much worse. 
Along with making my breadboard setup much more cluttered. ...

You also found issues with your grounds previously, so all bets off there.

How you arrange the grounds has an impact, both for supply decoupling and hum. For the circuit schematic you posted, pin 3 of the 12AX7 should be grounded at the same point as the ground end of pin 8's 3.9kΩ, and to the ground end of the 20uF filter cap for the 210v source. Likewise, the 210v-side of pin 1's 100kΩ, pin 6's 270kΩ and the + side of the 20uF cap should all be connected directly together.

Similarly, the "285v source" 40uF filter cap should have direct connection to the other 40uF filter cap's - and to the ground end of the 6AQ5's 560Ω cathode resistor.

Those 2 "ground nodes" should then be connected together (as they're both Ground), but best practice is each filter cap connects as directly as possible to the specific circuit-chunk it feeds.

... No caps in there now and no HUM or Whistling at all with this 6AQ5 circuit. 0, nada, zilch  ...

Feedback due to decoupling issues, much like feedback due to the field around current-carrying wires induces signal voltages in high impedance circuits, is never black-n-white, guaranteed-to-happen. And if you've dealt with parasitic oscillation before, you know it can happen at sub-sonic, low-audio, super-sonic or RF. Therefore it can be hard to know it's happening, except the "circuit just doesn't perform quite right." But that's why there are design and layout best practices. My 4-yr old has never jammed anything into any outlet, but I still put the plastic plugs into my wall outlets to make sure he doesn't try...

... If I have a normal amp that has whistling issues, could I possibly eliminate or reduce that problem by putting say a 600V .047 regular signal cap in parallel with the E cap for that stage? ...

No, that would really help because it's no different than the cap having a part-percent more capacitance.

You should strive to execute your breadboard layout physically the same as if it were a final build in an amp. Yes, you'll have terminal blocks, etc for component swaps. However the physical ordering of input-to-output, grounding, lead dress, etc should be reasonably like you'd expect to arrange it in a finished build. That's because all the same gremlin-potential exists on your breadboard as well as in a finished amp build.

[Paul1453]

Thanks for your detailed explanations and directions to further resources, HBP. 

My main purpose with setting this up on the breadboard was to see how it sounded,
and if there were any slight modifications I wanted to make before trying to cram the whole thing into that little amp housing.
I'm not sure I'm going to bother with that 1000 pf cap now, 1 less part and thing to go wrong.  LOL
In my mind, I can see everything needed fitting in with proper placement.
Everything will be much shorter and tighter in there.

Now to work on the real layout and tube placement.
PS on one side, input/components/tubes on the other, OT fitting somewhere in the middle hopefully.   

[Paul1453]

... If I have a normal amp that has whistling issues, could I possibly eliminate or reduce that problem by putting say a 600V .047 regular signal cap in parallel with the E cap for that stage? ...

No, that would really help because it's no different than the cap having a part-percent more capacitance.

Are you sure about that?

The way I am thinking about it is:
The E cap gives back some voltage when it drops to smooth out the DC.
The signal cap would still provide a parallel path to Gnd for other unwanted signals to follow, maybe?
Maybe not, that's just the way I was thinking it could work.   

[HotBluePlates]

No, that would really help because it's no different than the cap having a part-percent more capacitance.

Are you sure about that?

The way I am thinking about it is:
The E cap gives back some voltage when it drops to smooth out the DC.
The signal cap would still provide a parallel path to Gnd for other unwanted signals to follow, maybe?

You still remember the formula for capacitive reactance from your tech-time, right? Said at a high-level, reactance gets smaller for any given frequency as capacitance gets bigger, or if frequency goes higher.

In a tube amp power supply, you might increase a 20uF filter cap to 40uF with only a slight perceived change in performance. So would a change from 20uF to 20.047uF really make a difference?

There are reasons audiophile folks want a small cap (0.047uF) to bypass a big filter cap (40uF), but that is to make up for possible imperfections in the big cap, and the performance improvement is often debatable.

The signal cap would still provide a parallel path to Gnd for other unwanted signals to follow, maybe?

A lot of people run into a mental block by thinking "Ground" is some blackhole for unwanted stuff to go down & disappear. It's really just "reference point" or "circuit common".

Back to the decoupling resistor & filter cap. Your schematic has a "210v source" with a series 33kΩ resistor and a 20uF cap to ground. What happens to 1kHz or 100Hz signals applied to your 12AX7 stage, with respect to these components?

The 1kHz signal applied to the 12AX7 grid causes a 1kHz plate current, which is pulled from the 210v source through the plate resistors. Bad stuff can happen if these signal current mix with & are influenced by the current pulled by the 6AQ5 from the power supply due to that same 1kHz signal. We'd like to keep the 12AX7's 1kHz a.c. decoupled from that other stuff to avoid feedback due to a common power supply impedance (as the references talk about).

For 1kHz signal, the 20uF cap looks like:
X_c = 1/(2*∏*f*C)
= 1/(2*∏*1,000Hz*0.000020F)
= 7.96Ω

So that 1kHz signal current sees a 33kΩ path back to the power supply through the series decoupling resistor and a ~8Ω path through the 20uF filter cap back to the 12AX7's cathode resistor. It takes the path of least resistance (or reactance), so that signal current doesn't interact with the bigger currents pulled by the 6AQ5.

For the 100Hz signal case, the 20uF cap looks like:
X_c = 1/(2*∏*f*C)
= 1/(2*∏*100Hz*0.000020F)
= 79.6Ω

Still, the ~80Ω path through the cap is followed rather than the 33kΩ path to the power supply, and undesirable current interaction.

If you use a 0.047uF cap to ground but no series resistor between this node and your bench power supply, the 100Hz signal sees:
X_c = 1/(2*∏*f*C)
= 1/(2*∏*100Hz*0.000000047F)
= 33.8kΩ

So the 100Hz signal current sees ~34kΩ to ground and back to the tube's cathode resistor, while there's only ~0.2Ω of resistance in the wire and back to the bench supply. Interaction with the 6AQ5 signal current is assured, which is fine only as long as the power supply presents a true zero-ohm source impedance. There isn't a regulator or power supply built which can guarantee to do that under every possible condition.

As you get beyond the very basics of amp building & design, you'll find that every part has 2 or 3 different functions it performs. Often people focus only on a single function that they think is most-important. It's easy to overlook some of those other functions, which are equally (or more) important.

[Paul1453]

Thanks again HBP!

Most of the Army training I received kept us far away from mathematical equations.  At most, basic Ohms law stuff.

I was thinking shortly after I wrote that question, "Even if some of that bad signal did go to Gnd.  If I put my scope probe on the other side of the cap I'm still gonna see that unwanted signal".  I got the no mysterious black hole ground thing.
 
It was always much harder for me to figure out all the places signal is really going and what it is doing by looking at a schematic as opposed to simple voltage or current.  Most of the time when we were unsure if signal was present somewhere we would just use the oscope to see if we could see signal there.  All circuits were fully designed by the engineers to avoid problems like feedback.  And any type of circuit modification was strictly prohibited.

I guess I'm out there swimming with the signal sharks now.  I got to learn how not to chum the water in my swimming area.