• WARNING: Tube/Valve amplifiers use potentially LETHAL HIGH VOLTAGES.
    Building, troubleshooting and testing of these amplifiers should only be
    performed by someone who is thoroughly familiar with
    the safety precautions around high voltages.

Oscillation in tube amps

Any preamp, and Any line stage should have a resistor from the coupling cap to ground.

Otherwise, if you power the tube and B+, and after that, you connect the output to the next stage . . . you will apply the full voltage of the output tube plate to the next device. Ouch!
Only when that RC output circuit has charged up the capacitor, should you connect the output to the next stage.

All generalizations have exceptions.
 
Thanks for that - I put in a 1M resistor at the output of the coupling caps.

I have a second 6C4C preamp which I have in my system now with the same circuit and component values. It started off OK but then started tapping continuously. I put it back on the bench and added the 1M to the outputs and moved the plate chokes 70mm away from anything else. So far so good.......... With this preamp the filament supply and the PSU are both external.

With the first slow-tapping preamp the filament supply is onboard with the signal circuit. The PSU is external. That's going back on the bench.

I never had this problem with other valves. Is there something about using a 2a3/6C4C right at the input? I know it's unusual. Any reason why? It sounds great to me but is it prone to oscillation or is it just a case of getting the layout right?
 
Are your B+ capacitor, and your cathode bypass capacitor located close to the plate choke and cathode respectively.
And do the other ends of those capacitors connect to each other and at the local ground (no long wires permitted).

If those capacitors are on the power supply, and only connected to the external ground, it might cause the ticking.
An earlier post told about bursted RF oscillations.
A wire, especially a long wire has 'significant' inductance.
Inductance and capacitance is a resonator (often at very high frequency RF). Add a vacuum tube to that, and you might have a bursted oscillator.

Your schematic does not tell us how long the 12uF cap and 30uF cap wires are. It does not tell us how long the wires from the tube to the plate choke, the self bias resistor and cap, and the grid circuit.
One end of the grid stopper resistor needs to be connected directly to the tube socket tab (no long wire from that end of the grid stopper resistor).
Your plate choke has distributed capacitance, it can potentially resonate with a long lead from the plate to the choke.

Sometimes the worst happens:
Build an amplifier stage, get an oscillator.
Build an oscillator, get an amplifier.

Happy Troubleshooting.
 
Last edited:

Rod Coleman

Member
Paid Member
2004-07-25 6:55 pm
UK
> One end of the grid stopper resistor needs to be connected directly to the tube socket tab (no long wire from that end of the grid stopper resistor).

Good practice, but in the case of some of these old DHTs, the internal wiring of the triode is long, unshielded, and runs parallel to the anode, and somewhat blunts the effect of good stopper positioning. The low gm keeps the risk low, usually though. But short anode wiring, routed well away from the grid wiring is always worthwhile.
 
> One end of the grid stopper resistor needs to be connected directly to the tube socket tab (no long wire from that end of the grid stopper resistor).
Good practice, but in the case of some of these old DHTs, the internal wiring of the triode is long, unshielded, and runs parallel to the anode, and somewhat blunts the effect of good stopper positioning. The low gm keeps the risk low, usually though. But short anode wiring, routed well away from the grid wiring is always worthwhile.

As far as I am concerned all grid stoppers are wired directly to the socket tab. But I'm wondering if extra measures need to be taken when using a large tube like the 6C4C/2a3 as an input tube. For example enclosing the whole preamp or putting some sort of shield around the tube, like a perforated metal tube?
 

Rod Coleman

Member
Paid Member
2004-07-25 6:55 pm
UK
Power DHTs with 3-6mA/V like the 2A3 should not usually need a grid stopper, provided the wiring is done reasonably well. I have never needed them for 300Bs - and I can measure burst oscillation up to 1GHz here.

You can experiment with screening. Does that pulsing change its "bpm' if you bring a hand near the valve?

If you tune an AM radio to a weak station,and bring it near the valve, what kind of noise does it make on the radio?
 
Before spending lots of time chasing down a burst high frequency oscillation, switch off ALL LED based lighting in the house and either turn off or separate in distance all burst mode RF devices from the amp and scope under test. I spent two days chasing a burst of oscillation only to find that it was still on my scope with the amp switched OFF. It vanished however when the amp was unplugged. It turned out to be my LED overhead shop lights. They also raise THD measurements about 0.1 to 0.2% when on.

Some SMPS devices go into a low frequency burst mode when lightly loaded. They can make a "heartbeat" sound in a nearby amp. Phone, tablet and laptop chargers without a device connected can do this. Many laptop chargers will go into burst mode when the laptop is off, and charging is finished. So will some lightly loaded Meanwell or other SMPS's. They do this to be compliant with standby power consumption regulations.

If you use a similar SMPS in your amp design to power the tube heaters, make sure that its metal case is connected to amp chassis ground. It took me a day to find a bad clip lead on a bench top prototype. This looked like a constant 30 KHz haze over the amp's output which changed when you moved stuff or got your hand near a tube, just like a real oscillation would. The clue that told me that it was NOT oscillation is that only the amplitude changed, not the frequency.

Other known offenders are cell phones, WiFi devices and wireless routers, some bluetooth devices, and even my wife's Apple Watch! We have an old Sony TV upstairs that emits a constant RF signal at 146 MHz when TURNED OFF. It goes away when switched on. I found this one while hunting down ham radio reception interference.
 

Gnobuddy

Member
2016-03-01 4:10 pm
...short anode wiring, routed well away from the grid wiring is always worthwhile.
I agree. And yet, in many valve guitar amps, there are relatively long transformer leads connecting the output tube anodes to the output transformer! Basically, it amounts to connecting a short antenna to the largest signal voltage in the entire amplifier. Voltage swings of hundreds of volts, and several inches of wire to radiate EM waves from.

The few OTs I've used have all come with flying leads, encouraging this practice, as nobody wants to clip those leads short and "disfigure" an expensive transformer in the process.

Why isn't it standard practice to use shielded wire to connect output anodes to the output transformer? If the TX manufacturer chooses to provide flying leads, why don't they use coax cable leads on the primary winding?

-Gnobuddy
 
I agree. And yet, in many valve guitar amps, there are relatively long transformer leads connecting the output tube anodes to the output transformer! Basically, it amounts to connecting a short antenna to the largest signal voltage in the entire amplifier. Voltage swings of hundreds of volts, and several inches of wire to radiate EM waves from. -Gnobuddy

This seems to say that an external power supply is a really bad idea. Or at the very least, the cable connecting it should be short and shielded. Yes?
 

Rod Coleman

Member
Paid Member
2004-07-25 6:55 pm
UK
Andy, this is the OT wiring, rather than the PT.

PT wires should also be short - but that only means that the rectifiers & capacitors of a DC supply should be close to the PT. It's OK (within reason) to run the DC lines to another chassis.

> And yet, in many valve guitar amps, there are relatively long transformer leads connecting the output tube anodes to the output transformer!

The better designs run the wires away from grid wiring, and along the chassis but many Guitar amps' wiring is atrocious. Anyone who has had the misfortune to look inside the "Rivera" Fenders will vouch for it!
 

Rod Coleman

Member
Paid Member
2004-07-25 6:55 pm
UK
> switch off ALL LED based lighting in the house

Diodes in translucent packages (DO-35 diodes, many types of LEDs etc.) can generate photoelectric signals if exposed to PWM-driven or otherwise flickery LED lighting too. Opening the amp housing to debug might reveal this feature! Masking the diodes might be worthwhile.
 

Gnobuddy

Member
2016-03-01 4:10 pm
This seems to say that an external power supply is a really bad idea. Or at the very least, the cable connecting it should be short and shielded. Yes?
Let's say we're talking about an external DC power supply, and that it's well designed, so that it supplies clean DC, without a lot of AC ripple on the output.

If there is a properly-sized filter capacitor inside the amplifier enclosure, and also one inside the power supply, then the current from the external power supply to the amplifier is mostly direct current (DC), with only relatively small amounts of AC mixed in. Direct current doesn't radiate electromagnetic interference. Only alternating (or time-varying) current does.

What little AC there is, will be at pretty low frequencies - basically, relatively slow fluctuations in the current needed to keep the amplifier's big filter cap charged up. If the filter cap is rather marginal, only big enough to power the amp for, say, one-tenth of a second, then it only needs to be topped up ten times a second, very roughly speaking. That means fluctuations in power supply current are at frequencies below 10 Hz.

With a properly sized (bigger) filter cap in the amplifier, able to power the amplifier for longer, these frequencies will drop even more.

Low frequencies tend to radiate less than high frequencies (more accurately, they won't radiate well unless you have a really long wire run, because of the long wavelengths of lower frequency radio waves).

So in this situation (external DC power supply with good filtering, connected to an amp with adequate internal filter caps), there shouldn't be a lot of radiation to deal with from the power supply cable.

Meantime, the transformer inside the power supply tends to radiate EM waves, and so does the wiring around it. Even in an old-fashioned 60 Hz / 50 Hz power supply, the currents through the rectifier diodes have very "spiky" waveforms that contain frequencies far above 60 Hz - not only all through the audio band, but well into the LW and AM radio frequency ranges as well.

So it makes sense to keep the power supply relatively far away from sensitive input circuitry, by making it external. You keep the noisiest bits (power transformer and diodes inside the power supply) far away. The power cable itself is much less of a problem than the bits inside the power supply enclosure.

What about switching power supplies? They operate at tens or hundreds of kilohertz these days, and some of those high frequencies do leak onto the DC output power wires. Even though many have quite small amounts of (AC) ripple - 100 mV of ripple is typical for many small switchers - this is at the switching frequency, so its high enough to radiate significantly from a few feet of wire. Which is probably the major reason why almost every small switching power supply we see uses shielded cable from its enclosure to its power plug, just as you suggested.

Because the frequencies are higher, they tend to radiate better from the little power transformer, PCB, and switching MOSFETS inside the power supply. So it often makes very good sense to put the switcher far away from any audio equipment it's powering.

For instance, I use a 1Spot 9V switching power supply to power my guitar effects pedals. If I get my guitar too close to the 1Spot, there will be lots of buzz and hum as the guitar pickups respond to all the electromagnetic interference radiated by the power supply. But as long as the guitar is at least several feet from the 1Spot, there are no audible hum problems.

In my earlier post, I was talking about valve/tube guitar amps, in which I often see many inches of unshielded wire running from the output valves (power tubes) to the output transformer. Those output valves may be quite willing to oscillate at frequencies as high as a megahertz or more, and the voltage swings are huge (hundreds of volts at full output). So this is a really bad place to connect many inches of unshielded wire - it's too easy to turn your tube amp into an accidental RF transmitter! It only takes a tiny amount of the radiated output leaking back into the amplifiers input to make this happen.

I've owned three Fender guitar amps - a Super Champ XD, a Blues Junior, and a Princeton Reverb reissue. All three seemed to have left the factory only marginally stable - they were right on the edge of RF oscillation. The SCXD and PRRI would burst into RF oscillation if I simply placed my old Zoom G3 multi-FX pedal on the ground near the amp and connected the two. The Blues Junior would burst into RF oscillation if you just changed the output valves(!), unless you dressed the wires running from PCB to valve socket just right.

I wasn't the only one who had these sorts of problems with these amps. Bill M. (RIP) briefly offered a better output transformer for the SCXD, but some customers found that installing it would make their SCXD oscillate wildly. Bill stopped offering these for sale.

Part of the reason why the SCXD was so sensitive to the output transformer was because it had just about the worst-designed PCB I have ever seen. I have never seen worse grounding or PCB trace routing in an audio amplifier.

But that's another story!

-Gnobuddy
 

Gnobuddy

Member
2016-03-01 4:10 pm
Diodes in translucent packages (DO-35 diodes, many types of LEDs etc.) can generate photoelectric signals <snip>
You brought back childhood memories. :)

I remember carefully filing the top off a metal-can NPN silicon transistor as a boy, in order to convert it into a photo-transistor. In the time and place where I grew up, there was no electronics store that sold actual phototransistors, so this was the only way I could get my eager paws on one.

I don't remember all the things I did with that phototransistor, but I do remember one. I magnetized a piece of one of my dad's disposable razor blades, glued it to a little cardboard disc, and balanced it on a pin to make a compass. Then I wound a little coil of copper wire placed around the compass "needle" to turn it into a galvanometer. Finally, I connected the DIY phototransistor to the coil, and made a cutout in the cardboard disk to block and unblock the light to the phototransistor at the proper positions, so that the phototransistor would switch the current through the coil on and off, depending on the position of the needle.

Et voila, an electric motor! You needed a bright light above it, and you had to turn it the right way (coil parallel to the magnet) to get it to self-start. But, that done, it would start and run. Slowly, and with no usable output power at all, but just the fact that it ran was a huge thrill for me.

I remember expanding on that idea later, with a new cardboard rotor cut to have multiple "teeth", and several small nails glued to it, nail heads out, tips pointing in. The single galvanometer coil was replaced by an electromagnet made by winding copper wire onto a larger nail. The phototransistor was mounted under the disc, facing up, so that the teeth would block and unblock light as they moved over it.

With a lot more power pulses per rotation (one pulse per nail), this motor would spin faster and more vigorously, once you got the phototransistor positioned just right.

-Gnobuddy
 

Gnobuddy

Member
2016-03-01 4:10 pm
Most all of those transmissions are at frequencies most ordinary toobz do not even see,
The GHz stuff may be too high frequency to be picked up by "toobz".

But lots of switching power supplies still operate in the range from audio frequencies, through old-fashioned long wave radio. As a for-instance, I just bought some little modules that run on 5V DC, and spit out symmetrical +/- 15 V DC. The spec sheet says they operate at between 70 kHz - 140 kHz (variable frequency, depending on load).

Another interesting factor is that many of us bought lots of $1 tubes, no longer popular, because they were originally designed to operate at RF, not audio. My little 6AK5 pentodes, for instance, are "Useful at frequencies up to 400 MC" according to the RCA datasheet.

-Gnobuddy
 
  • Like
Reactions: 1 user
The GHz stuff may be too high frequency to be picked up by "toobz".

But lots of switching power supplies still operate in the range from audio frequencies, through old-fashioned long wave radio. As a for-instance, I just bought some little modules that run on 5V DC, and spit out symmetrical +/- 15 V DC. The spec sheet says they operate at between 70 kHz - 140 kHz (variable frequency, depending on load).

Another interesting factor is that many of us bought lots of $1 tubes, no longer popular, because they were originally designed to operate at RF, not audio. My little 6AK5 pentodes, for instance, are "Useful at frequencies up to 400 MC" according to the RCA datasheet.

-Gnobuddy
And lots of little switch mode power supplies operate in cycle skipping or "hickup" mode where they send bursts of 2Mhz (in more recent devices. I use a lot of little 1-2Mhz Buck convertors in my day job) but at a rate of several hundred Hz.. it's these burst like the afformentioned TDMA noise (which is at 217Hz) that get turned into audible crap.
 

wg_ski

Member
2007-10-10 5:21 pm
Bursts of RF gated on and off at 217 Hz would be quite audible. Just like bursts of RF gated on and off at 120 Hz from a ringing power transformer can produce annoying ”bizzzzzzzzzz”. Anything can pick up this kind of RF interference - it’s best to try to eliminate it. Transistor amps also have an annoying tendency to produce RF rectification - they like to demodulate AM radio stations. All it takes to do that are nonlinear capacitors, which is what transistors are. Tube amps are a lot less susceptible to that particular garbage.