Randomly, one of my monoblocks gives off bursts of similar to radio noise (ssshhhsh hash). At first, it started to appear rarely a year ago, for no apparent reason, a few times in a month maybe. The turning off and on of the amplifier or reconnecting the RCA jack made it go away. Back then, I heavily wondered about the problem, disassembled the amplifier, checked soldering joints, cleaned tube socket and pins, measured voltages, frequency response and looked at waveforms. There were not apparent HF oscillations at the output.
The first and second grid have 470R grid stoppers soldered directly to the socket. The tube is loaded with an IT.
One day, while tapping with a plastic rod around the amplifier, I found out the culprit was the EL802 driver tube itself. When tapped, it began exhibiting these burst of hash noise randomly. Each hit changes the frequency and amplitude of the bursts. If struck multiple times, it would go away for a while.
I'd really like to know if my EL802 is faulty or the problem lies somewhere else. To be honest, I'm a bit greedy and I don't want to throw it away if I'm not convinced the tube itself is the problem. Could it be a mechanical problem?
Could the problem lie in the fact I push the tubes close to the max. dissipation power rated by their datasheets? It gives 6W for the plate and 2,5W on G2. I push them to 6,5W.
The first and second grid have 470R grid stoppers soldered directly to the socket. The tube is loaded with an IT.
One day, while tapping with a plastic rod around the amplifier, I found out the culprit was the EL802 driver tube itself. When tapped, it began exhibiting these burst of hash noise randomly. Each hit changes the frequency and amplitude of the bursts. If struck multiple times, it would go away for a while.
I'd really like to know if my EL802 is faulty or the problem lies somewhere else. To be honest, I'm a bit greedy and I don't want to throw it away if I'm not convinced the tube itself is the problem. Could it be a mechanical problem?
Could the problem lie in the fact I push the tubes close to the max. dissipation power rated by their datasheets? It gives 6W for the plate and 2,5W on G2. I push them to 6,5W.
Last edited:
Have you swapped the tubes between amplifiers to confirm that it's the tube,
and not the socket? If so, have you cleaned the tube pins?
Yes, I did. I forgot mentioning the fact I swapped the tube to the other monoblock and so did the noises.
I also did clean the tube pins including the socket.
Then about the only remaining remedy would be to try resoldering the tube pins.
You could try a ventilation fan to see if it happens less often when cooler.
Last edited:
I was curious to find out what could be the reason of the tube getting noisy? Could it be a faulty weld making poor contact? Or some kind of mechanical oscillations?
Why does the noise go off when I unplug and plug the RCA jack again? Could it be some kind of electrical oscillation?
The noise appearing is like a sputtering hiss.. There's a low-level hiss in the background, and randomly, bursts of crackle appear.
Why does the noise go off when I unplug and plug the RCA jack again? Could it be some kind of electrical oscillation?
The noise appearing is like a sputtering hiss.. There's a low-level hiss in the background, and randomly, bursts of crackle appear.
Last edited:
I have been chasing similar problems for years in my D3A/GM70 fully transformer coupled amps. I always suspected the tube (D3A) and the problem generally occurred only in one of my two amps, sounds exactly as you described and also confusingly followed at least some tubes. There was more than one tube that would do it, in fact many more than one tube, always a tube that measured amongst the best. Some tubes would be fine for the first hours of use and develop symptoms after a few hours, others did it immediately, and a dwindling number in my stock didn't do it.
Years of pulling out my hair I identified that in my design tubes above a certain transconductance threshold oscillated at a frequency beyond the range of either of my scopes, consequently could not see it.
Recently I fixed it, I replaced the grid stoppers with two in series, halving capacitance, doubling resistance and inductance. (The same resistance value by itself did not solve the problem)
I also increased the value of the screen stopper resistor.
I've tried beads and they didn't work, apparently not lossy enough at the frequencies where instability was occurring. Rearranged grounds to reduce inductance, etc.
Now the amplifiers show no sensitivity at all to the tube plugged into the driver stage socket.
Years of pulling out my hair I identified that in my design tubes above a certain transconductance threshold oscillated at a frequency beyond the range of either of my scopes, consequently could not see it.
Recently I fixed it, I replaced the grid stoppers with two in series, halving capacitance, doubling resistance and inductance. (The same resistance value by itself did not solve the problem)
I also increased the value of the screen stopper resistor.
I've tried beads and they didn't work, apparently not lossy enough at the frequencies where instability was occurring. Rearranged grounds to reduce inductance, etc.
Now the amplifiers show no sensitivity at all to the tube plugged into the driver stage socket.
As Kevin said, these very high transconductance (it is high both on g1 and g2) tubes are very sensitive to spurious oscillation due to layout and design. Initial aging often makes that a tube appears to work fine for a couple of hours, only to burst into oscillation after. If you tap on the tube (don't do that, will you?), you momentarily change its geometry, which might kill the oscillation or trigger it. If your amp is "on the edge", oscillation appears very erratically, even though the individual components are fine.
I have a uTracer, and I have used Kitagawa RI-4-10-2 beads in its socket wiring loops. With these beads in place, I can test firecrackers like the E810F, 12GN7, EL802 and other high transconductance tubes, both new and used ones, without a trace of oscillation. The trick is to use more than one bead, distributed unevenly, because their placement is most effective when they are placed in a current antinode (i.e. the place where the current is maximum) of the standing wave. The location of nodes and antinotes is frequency dependent. If you place just one bead, there is still a multitude of frequencies which aren't sufficiently suppressed, so you still have a good chance of oscillation. A good starting point is to use two beads, one as close to the grid socket pin as you can get it, and the other one about halfway the wire.
To make a long story short: don't discard your tubes, the behavior you are seeing is rather normal for them. Can you post a schematic or (perhaps even better) a picture of your build?
It would be nice to study old color TV schematics to see how the beast was controlled in this application. This is the intended application for the EL802/12GN7/12HG7 family of tubes(*), they are the final video (luminance) amp driving the CRT.
Edit: here's one:
http://www.philipstv.org.uk/blog/wp-content/uploads/2012/03/img047.jpg
(*) Please tell everyone that these tubes are yucky icky TV tubes. This keeps them affordable for the more adventurous/knowledgeable souls who want to realize their performance potential, while the audiophile community keeps steering away from them
I have a uTracer, and I have used Kitagawa RI-4-10-2 beads in its socket wiring loops. With these beads in place, I can test firecrackers like the E810F, 12GN7, EL802 and other high transconductance tubes, both new and used ones, without a trace of oscillation. The trick is to use more than one bead, distributed unevenly, because their placement is most effective when they are placed in a current antinode (i.e. the place where the current is maximum) of the standing wave. The location of nodes and antinotes is frequency dependent. If you place just one bead, there is still a multitude of frequencies which aren't sufficiently suppressed, so you still have a good chance of oscillation. A good starting point is to use two beads, one as close to the grid socket pin as you can get it, and the other one about halfway the wire.
To make a long story short: don't discard your tubes, the behavior you are seeing is rather normal for them. Can you post a schematic or (perhaps even better) a picture of your build?
It would be nice to study old color TV schematics to see how the beast was controlled in this application. This is the intended application for the EL802/12GN7/12HG7 family of tubes(*), they are the final video (luminance) amp driving the CRT.
Edit: here's one:
http://www.philipstv.org.uk/blog/wp-content/uploads/2012/03/img047.jpg
(*) Please tell everyone that these tubes are yucky icky TV tubes. This keeps them affordable for the more adventurous/knowledgeable souls who want to realize their performance potential, while the audiophile community keeps steering away from them
Last edited:
Excellent and priceless experiences of yours, thank you for sharing.
This will put me on the quest to put further efforts into taming this probable oscillation.
I remember this started happening after changing my g1 resistor from carbon comp to metal film. I had to, because my CC resistor burned due to an accident and I didn't have a spare.
At first place, I'm thinking of putting a wirewound g1 series resistor that should imitate a lossy choke, like kevinkr said.
This is the schematic.
Here's a photo that slightly shows the build around the driver stage. I circled it in green in the top right corner. The long wires that might cause trouble are the decoupling capacitor -> ITT -> anode - cathode connections.
* I have both of EL802 and EL183, these are excellent sleeper tubes. Low Rp, high gain and excellent sound. They're still cheap I and hope it remains so. Use them while they still are.
P.S. We also have the uTracer at work, although it's kind of broken lately. What are the signs of occurring oscillations? I measured quite a bit of high gm tubes and got perfectly looking traces.
This will put me on the quest to put further efforts into taming this probable oscillation.
I remember this started happening after changing my g1 resistor from carbon comp to metal film. I had to, because my CC resistor burned due to an accident and I didn't have a spare.
At first place, I'm thinking of putting a wirewound g1 series resistor that should imitate a lossy choke, like kevinkr said.
This is the schematic.
An externally hosted image should be here but it was not working when we last tested it.
Here's a photo that slightly shows the build around the driver stage. I circled it in green in the top right corner. The long wires that might cause trouble are the decoupling capacitor -> ITT -> anode - cathode connections.
An externally hosted image should be here but it was not working when we last tested it.
* I have both of EL802 and EL183, these are excellent sleeper tubes. Low Rp, high gain and excellent sound. They're still cheap I and hope it remains so. Use them while they still are.
P.S. We also have the uTracer at work, although it's kind of broken lately. What are the signs of occurring oscillations? I measured quite a bit of high gm tubes and got perfectly looking traces.
Last edited:
Let's discuss this further. A wirewound resistor simulates a longer length of wire. Although this contributes to RF picking, it also gives more inductance that will put the cutting point of the resulting RLC filter at a lower frequency. It's not unusual to see a resistor with a a few turns of wire wound around them in radio practices.
Of course, the parasitic capacitance of a wirewound resistor is also not something to neglect. Maybe it might be the culprit?
What if the resistor is screened using a piece of foil around?
Of course, the parasitic capacitance of a wirewound resistor is also not something to neglect. Maybe it might be the culprit?
What if the resistor is screened using a piece of foil around?
Especially wirewound should not be used as grid stoppers... small metal film resistors might work, but carbon based resistors are best as grid stoppers.
How is the driver tube biased? Although I can't enlarge the pictures, I don't see any biasing provision of the driver. Or do I... Is there a battery in series with the signal grid? If so, it is an nonadjustable fixed bias which diminishes as the battery ages. These steep tubes should never be used with fixed bias (let alone a battery), because small bias fluctuations and tube-to-tube variations make the actual bias point vary all over the place.
With the utracer, the first signs of oscillations are offbeat measurement points. In more severe cases, the MCU might even crash. But that only happens when the tube goes completely bananas. An oscilloscope of sufficient bandwidth should give you a definitive answer.
How is the driver tube biased? Although I can't enlarge the pictures, I don't see any biasing provision of the driver. Or do I... Is there a battery in series with the signal grid? If so, it is an nonadjustable fixed bias which diminishes as the battery ages. These steep tubes should never be used with fixed bias (let alone a battery), because small bias fluctuations and tube-to-tube variations make the actual bias point vary all over the place.
With the utracer, the first signs of oscillations are offbeat measurement points. In more severe cases, the MCU might even crash. But that only happens when the tube goes completely bananas. An oscilloscope of sufficient bandwidth should give you a definitive answer.
This is a parallel RC arrangement, while a WW resistor has its inductive component as a series component. WW resistors have parasitic capacity, which together with their inbuilt inductance, makes them act like a parallel resonant circuit, which becomes capacitive above its resonance frequency.
The parallel arrangement in radio circuits becomes resistive above the RL lowpass cutoff point. The resistor is a low-inductance (i.e. a carbon comp or carbon film) type. Even in multi-kW transmitters, this resistor is can be made quite small, because it carries almost no current.
Battery bias and unadjustable fixed bias will force you to select tubes for operating point, in the long run this will prove to be more trouble than it's worth.
I now use either some variant of cathode bias that is designed to swamp the operating point variations caused by large variations in transconductance or fixed bias that I can adjust as needed for the specific tube.
I now use either some variant of cathode bias that is designed to swamp the operating point variations caused by large variations in transconductance or fixed bias that I can adjust as needed for the specific tube.
High transconductance tubes are often used with a positive voltage bias on the grid (w.r.t ground) and a large cathode resistor between cathode and ground, in order to keep their operating current more constant over the life of the tube. Otherwise, even "traditional" cathode bias may give excessive bias current variability.
Then going back to a carbon composition resistor will be my first try.
Yes, there is a battery in series with the grid for bias. It has a long shelf life. I have the practice to check it's voltage and it hasn't changed since the moment I bought it (1.6V) AgO.
My practice is to carefully match tubes using the uTracer before inserting them to avoid the tube-to-tube variations. The only other issue is the varying power supply voltage.
Yes, WW resistors have parasitic capacity that is more significant in non-inductive ones, with an Ayrton-Perry or bifiliar kind of winding.
These are the direct links to the images, if you can't zoom then from the forum's interface.
https://postimg.org/image/j6f4khtsn/
https://postimg.org/image/3yz5051xz/
Yes, there is a battery in series with the grid for bias. It has a long shelf life. I have the practice to check it's voltage and it hasn't changed since the moment I bought it (1.6V) AgO.
My practice is to carefully match tubes using the uTracer before inserting them to avoid the tube-to-tube variations. The only other issue is the varying power supply voltage.
Yes, WW resistors have parasitic capacity that is more significant in non-inductive ones, with an Ayrton-Perry or bifiliar kind of winding.
These are the direct links to the images, if you can't zoom then from the forum's interface.
https://postimg.org/image/j6f4khtsn/
https://postimg.org/image/3yz5051xz/
Varying power supply voltage is an issue with fixed bias triodes, as well as pentodes with unregulated screens. The supply voltage varies along with mains voltage. I don't know how "tight" your mains voltage is, but this amp needs a redesign in several points if it is to provide worry-free operation.
You are right about the mains variation, although it pretty stable at the place I live. I measured a variation of ~6mA of the plate current of the driver tube, which is not a small value, so I made sure the IT can cope with the higher current to avoid core saturation.
The issue with cathode bias is the huge effect on the sound of cathode bypass capacitors. I preferred to use a battery to make a good sounding and cheap solution.
Or I could use keep the battery and use voltage stabilization.
The issue with cathode bias is the huge effect on the sound of cathode bypass capacitors. I preferred to use a battery to make a good sounding and cheap solution.
Or I could use keep the battery and use voltage stabilization.
For these high transconductance valves you may want to think about using a surface mount type resistor soldered directly to the socket or even cutting the lug down right near the pin and mounting it there.
Getting a 6E5P stable took me some time. Ferrite beads in the cathode and surface mount grid stoppers did the job. They are so physically small that inductance is minimal.
I have also found that modern reputable manufacturers 1/4W and 1/8W metal films have less inductance than the standard AB 1/4W 1K carbon comps I had previously been using. YMMV.
Cheers
Matt
Getting a 6E5P stable took me some time. Ferrite beads in the cathode and surface mount grid stoppers did the job. They are so physically small that inductance is minimal.
I have also found that modern reputable manufacturers 1/4W and 1/8W metal films have less inductance than the standard AB 1/4W 1K carbon comps I had previously been using. YMMV.
Cheers
Matt
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.