Lots of craziness (weird noises and osc) happening in point to point 5W EL84 Amp

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
I figured with 2 tubes (a 12AX7 and an EL84), it'd be easy enough to just wire it point to point (physically from jack to tube to pot to next tube to OT). I'm getting some crazy oscillations with this thing. I can see it on the scope even after the first gain stage on the 12AX7. It's in a metal enclosure (with the lid off of course when probing). I have all the preamp grounds at one point (at the input jack) and the EL84's ground going to the first cap on the power supply (after the choke).

Curious if the reason for the oscillation is because there are parts and some wires 'floating' in the chassis. I did not use shielded cable but just ran the wires to tube grids right along the chassis floor. If I turn the volume up and down it sounds like an old sci fi film (wobbly frequency sweeps and almost sounds like an old radio being tuned).

I know how to follow a schematic, I know how to shape the tones I want if I need to make changes, I know close to NOTHING about dealing with interference so any advice would be appreciated! Hum isn't a problem by the way. Hum wise, it's pretty quiet.
 
Last edited:
Sure here's a quick and dirty edit to what I currently have.
 

Attachments

  • 5Watt.jpg
    5Watt.jpg
    167.7 KB · Views: 157
I have connections running to and from pots and directly to tube sockets. Layout wise with the chassis, this was originally a crate v5 am and it worked quietly so I know the tranny/tube spacial relationship is ok. This was originally on a 2 layer PCB with one layer being ground. Now it's wired free floating with the parts hard wired to the tube sockets and pots/jacks.
 
Bypass every PSU electrolytic with a film cap. Value not critical, say 220nF. This is usually necessary to prevent oscillation in my guitar amps with multiple 12AX7 gain stages. In HIFI I use at least one big film cap in the PSU, preferably the last before regulation.

Electrolytics don't really shunt radio frequencies.

If this is a HIFI amp not an instrument amp, why use so much gain? Why two volume controls?
 
This originally had one volume control but I liked having the two so I could balance how much it distorted (for guitar) between the 12AX7 and the EL84.

I notice however that if it's too loud, it sounds like something's out of bias (fizzy bad distortion) and also that I get much more oscillation.
 
Administrator
Joined 2004
Paid Member
I'd add a grid stopper resistor 1K or so right at the grid of the second triode and the EL84 particularly since there is a run of some distance probably to the grid of the output tube. (High transconductance types are very prone to oscillation)

Is the 470 ohm resistor right at the screen grid of your EL84 - if not move it.

Post a picture or two if you can, it should help to address layout concerns.

And finally :cop: since this is clearly a guitar amplifier I will move it per forum policy to the Instruments & Amps forum. You'll get lots of good advice both on the topology you've implemented and your specific stability issue.
 
I don't see any obvious major problem with the schematic. Grounding must be all done in one place, not part at the input jack and part over at the power supply. I'd use a separate ground return wire for each section of circuitry. The whole circuit should have only one connection to the chassis. That means the input and output jacks should be insulated from the chassis, unless one of those are the one connection to the chassis for all circuit grounds. The earth wire of the AC line cord (green) should also connect to the chassis, but in a different place (ideally right where it enters the chassis).

Power supply bypass caps of roughly 0.1uF/600V+ are a good idea. The noisy pot could be that there is DC on the pot, which suggests that the cap feeding the pot is leaky or shorted. I'd put at least 1K R's on all grids (right at the tube pin). Going significantly higher in value than that can form a significant low pass filter but can also add hiss noise. Metal film R's have the lowest noise.

When the amp is first turned on, and none of the tubes have started conducting yet, the B+ will be higher than you might realize. Some caps may have had their max voltage spec exceeded.

I'm not sure what the 10 ohm and 0.1uF at the output are for. You never want to run a transformer coupled output without a load (the tranny turns into a spark coil and can blow up the tube and/or itself). The nice thing about the switching output jack is that you can wire it to switch in a 10 ohm 10 watt resistor in place of the speaker, when the speaker is unplugged. This doesn't guarantee safety if a loud signal is present, but is a good idea anyway.

Single ended no feedback circuits are among the best sounding IMO. Personally I'd have added a tone control section after the 2nd triode, and before the master volume control.
 
Last edited:
Disabled Account
Joined 2010
A few things I would try,

Disconnect C23 from ground just to try it.

Put 10K resistors in series on the input (grid) to 2nd tube and power tube right on the valve socket.

The heater resistor CT what value are the resistors? (2X 220 Ohm would do it)
what's the PSU circuit like?

If the above has no effect move all the grounds to the negative of the first cap on the PSU.
Make sure the input ground is isolated from the chassis and only grounded at the same point.

Regards
M. Gregg
 
Last edited:
One thing to keep in mind with any amp circuit is that you want to keep any high impedance nodes, and/or input circuits, as far away from output or power supply wiring as is practical, to minimize the probability of the circuit oscillating. Also, keep all leads as short as is practical. Especially with high impedance circuitry (grid circuits in this case), parts should be as close to tube pins as is practical. Stray capacitance and inductance is not a good thing.

Also, put a passive Rf filter at the input (maybe a 10kohm R in series, and a 330p ceramic or mica cap to gnd., which will give you a -3dB point at 24kHZ if the source Z (guitar pickup) is about 10kohms - which it is on a strat and I think most guitars).

Also a 0.01uF 3kV ceramic cap across the AC input, or better yet from each of the hot and cold AC connections to the earth gnd which connects to the chassis.

A guitar makes a nice antenna for Rf, and any audio amplifier is highly likely to act as an AM detector for those frequencies, which will result in audio band noise and distortion (not the good kind), especially if you're in a place that uses light dimmers, or any switch mode electronics.
 
Last edited:
Does turning the volume down turn the oscillation down or otherwise affect it? Does pulling the preamp tube kill the oscillation? Poke around with a wooden chopstick while it is doing this, does moving anything affect the volume or tone or frequency of the oscillation?
 
The Zobel won't hurt anything, but is generally only needed when the amp is using negatory feedback. If negative feedback gets shifted in time (phase) by the reactance of the load (speaker), then it may cause the amp circuit to oscillate spuriously or worse.

For hum problems, all gnd connections need to come back to a "star center" gnd post (bolt is my rec). The exceptions are the AC cord green "earth" wire (tie it to the chassis in a different place), and the wiring between the power tranny secondary ground connection (often the center tap) and the large filter caps of the power supply. You want to connect that secondary gnd connection to the filter caps separately, and then run a wire from there to the "star center" location for the entire rest of the circuit. The idea is to keep those high current surges away from the rest of the gnd. connections. The circuit ground must only tie to the chassis in one place, so any I/O connectors should be floating.

The early guitar amps didn't do this right; there were ground loops from non-floating 1/4 phone jacks, and no Rf filtering anywhere. They hummed, squeaked, sputtered and didn't need to.

For noise issues, use all metal film R's in the early stages and anywhere the R value will be over a few kohms. Those who think metal films sound metalic are fools. The hiss that carbon films or comps make have a psychological effect that makes some people think they sound more or less metalic. The only other noise of any real significance is from Rf being demodulated (AM detected), or from the tubes themselves - some tubes are significantly noisier than others, even of the same tube number.
 
There is a spec for heater to cathode max voltage, which many designs ignored. It's usually around 100 volts. It's important that the heaters are somehow referenced to ground, so they can be reliably referenced to the cathode. In the case of a follower topology, the cathode will be sitting up around 1/3 to 1/2 of B+, which can cause this spec to be exceeded.

I make a power supply filter cap bleeder resistor that is split giving me a roughly 90VDC tap, which I tie to one side of the heater supply, with an AC shorting cap off that tap to gnd. If you're running the filaments on AC, use at least 18AWG single strand wires, and twist them fairly tightly so the electromagnetic field they emit largely cancels itself out. Keep that twisted pair as far as you can from any high impedance wiring.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.