Ampeg SVT CL ring modulation distortion at high volumes

[nigel-ma]

I have an Ampeg SVT CL which produces a ring modulation sound at high volumes.

The situation at present is that it has a fresh set of power tubes (6 x 6550) which are matched with 5mA of each other at 72mA per side (3 x 24mA nominal each), and the bias setting LEDs are both green. The pre-amp tubes (2 x 12AX7) have been swapped out, plus the buffer/phase splitter 12AX7. The two 12AU7 driver tubes have not been replaced since I don't have any to hand, however I did try substituting a pair of ECC81s. The power supply capacitors have also been replaced.

Looking at the supply rails on the power supply I have one anomaly according to the schematic I have (07S419-XX Revision B dated 7-28-00). The output from the 15k 5W resistor (R54/C17) is nominally +470V, but I am reading +540V. The output from the 47k 1W resistor (R58/C20) is specified at +345V and I am reading +410V. Strangely enough, the other voltage from the output of the 47k 5W resistor (R55/C19), also sourced from R54/C17 is specified at +343V and is within specification. All other supply voltages are close to specification (+660V, +380V, +365V, -180V, and +/-15V). The resistors (R54/R55/R58) measure within specification.

At lower volumes the +660V supply has less than 2V p/p sawtooth ripple, and the amplifier produces a clean sound. However, at higher volumes the ring modulated sound occurs. With the signal generator at 100Hz, (50Hz supply in the U.K.), the amplitude modulation pointed to the power supply full-wave rectification, and when the +660V was traced on the oscilloscope the ripple increases to about 15V p/p. As might be expected, this is the reason for the ring modulated sound at the output. The diodes and capacitors have been swapped out.

The 10 ohm 5W cathode resistors for the 6550s are within specification, as are the 220 ohm screen grid resistors, and 470k control grid stopper resistors.

The anomalous voltage (+410V instead of the specified +345V) feeds the 12AX7 buffer/phase splitter, resulting in +250V on pin 1 (anode plate) of the 12AX7, and also +310V on pin 6. I have swapped out this 12AX7 and this anomaly remains.

I have pretty much checked 90% of the amplifier, apart from the speaker output transformer, but I am stuck in regards to this anomalous voltage on the buffer/phase splitter supply, and I wonder if that is overdriving the amplifier leading it to overstress the capabilities of the power supply and ultimately causing the ring modulated sound at high volumes. The amplifier was playing cleanly at high volume, but one hour into a gig the ring modulated sound suddenly appeared.

Please would anybody comment as to where I should look next, or how to proceed from here?

[Jazz P Bass]

What does this 'ring modulated' thingy look like on a scope?
I know you stated that the power supply caps are new but the 2 Vac to start off seems a tad high (and then it climbs!)
You may have a duff cap.

[Steve Conner]

I don't have the schematic, but I'd expect there is some sort of regulator that's supposed to be regulating (and removing ripple from) the voltage at R55/C19, and it has failed.

Another possibility is that one driver tube has died, so the output stage is now unbalanced and can't reject ripple. But in that case the amp would be seriously down on power.

The ripple voltage seems reasonable for the raw supply rail of a high-powered amp.

[nigel-ma]

Thankyou for your replies Jazz P Bass and Steve Conner. I apologise for not replying sooner, my computer has been having problems and my last post disappeared into the internet "ether" after I clicked on the "submit" button.

The schematics I refer to can be found on these forums via the following link:

http://music-electronics-forum.com/t...55/#post138355

I am trying to help out a friend, so I am not in the workshop at present. I asked him to photograph the oscilloscope traces but unfortunately he told me they were to blurred and the camera flash unit caused reflections. I will attempt to obtain some photographs myself when I am next there on Monday and report back. At full volume the overdriven signal is difficult to describe, so a photograph is worth a thousand words. Before maximum volume the sine-wave is pure and as the volume increases the artefacts of the power supply ripple can be seen affecting the sine-wave, although this is to be expected. As the volume increases the sine wave becomes slightly compressed, but at full volume I don't think the SVT bass amplifier is supposed to produce the distorted waveshape, nor sound the way it does. I am open to advice on what can be determined from observing the output on the oscilloscope if I manage to post any photographs. It certainly sounds bad to my ears.

I understood a ring modulated sound was best known as a sound effect, such as the "Dalek" voice from "Doctor Who" if you know of that. It is the sum and difference of two frequencies, and sounds like a dissonant distortion. The two frequencies in this case being the 100Hz signal generator and the power supply ripple (full rectified U.K. 50Hz). The 2V p/p ripple doesn't seem much on +660V to me, but I lack experience of what is "normal" for these things. I certainly cannot hear any ring modulation at lower volume levels. However, when the volume is at maximum the ripple increases to about 15V p/p and I would expect to hear a ring modulated sound since that level of ripple seems high by any account for audio electronics. Apparently this is typical of failing power supply electrolytic capacitors. The capacitors have been exchanged for two new sets and the problem remains. I suspect at this point the distorted ring modulation sound is a symptom of another cause, but that is only a guess. I am wondering if the amplifier is being pushed beyond its' design limit, and asking far too much of the power supply, at which point it is perfectly reasonable for the power supply regulation to be completely insufficient.

I remain concerned about the schematic +345V supply being measured at +410V. I believe this only feeds the 12AX7 buffer/phase splitter just after the power amp input. I thought the increased supply voltage may be overdriving the output tubes. I forgot to take a measurement when looking at the output on the 'scope, I will do that next time. The output transformer is set for 2-ohms, driving a 2-ohm 400W dummy load (four 8-ohm 100W power resistors in parallel).

Other things I forgot to mention in my first post were that the pre-amp output was sent to another power amp, and the sound was clean and clear at maximum volume. This removed the pre-amp from suspicion.

I have also lifted one leg of all the capacitors surrounding the 12AX7 buffer/phase splitter and measured for d.c. leakage, there is no obvious problem there. The capacitors are C6, C7, C8 and C9. That would seem to eliminate the +410V supply, which should be +345V, from producing a d.c. offset for the surrounding circuitry. I don't think the increased supply voltage would affect the gain of the phase splitter.

I think when I am next at the bench I will check for continuity between the power supply smoothing capacitors and the tube sockets. I wonder if there are any opens/shorts in those paths.

In regards to the anomalous supply readings, it defies electonics laws that the if the voltage at R54/C17 is incorrect, then the voltage at R55/C19 is correct when the voltage at R58/C20 is incorrect; unless of course the actual circuit differes from the schematic, so that is what I need to search for. Something is not drawing sufficient current down paths R54 and R58. This indicates perhaps an open circuit somewhere.

The power tubes bias correctly with both the LED bias monitoring circuit, and by measuring the cathod currents with bias probes. The power tubes have been swapped for fresh units, although it is true that a valve can fail at any time. However, monitoring the bias currents, and the bias monitoring LEDs do not suggest any problems there.

I wonder if the ring modulation sound and power supply ripple are a distraction, and the real problem is caused early in the circuit by whatever is causing the power supply to be higher than specified (+540V at R54, and +410V at R58).

I will report back when I have further information.

[nigel-ma]

The bassist commented that he had unintentionally run the amplifier with the speaker output transformer set at 4-ohms, but into a couple of 4-ohm cabinets in parallel for a load of 2-ohms. To eliminate the possibility of speaker output transformer damage d.c. resistances were measured across the primary and secondary windings, between windings, and between wndings and ground. Plate to plate resistance was 58-ohms, split 30-ohms one side of the centre tap and 28-ohm the other side. resistances betwwen windings, and between windings and ground were in excess of the meter capability at greater than 20M-ohms. The speaker output transformer seems fine.


I also confirmed the power tube output board control and screen resistor values were good at 220-ohm and 47k-ohm respectively.

I revisited the power supply measurements and confirmed my earlier readings. Unfortunately I was unable to obtain photographs of any oscilloscope traces with the camera available. That is something that needs to be solved for future debugging I think. The 660V anode plate supply to the 6550 power tubes has a normal sawtooth wave ripple of 2V p/p under idle conditions, and 15V p/p at full volume when demonstrating the distorted ring modulated sound. When looking at the speaker output, the meter read approximately 35V a.c. into a 2-ohm load for a very clipped waveform. I don't imagine a bass amplifier should produce such a clipped output, perhaps a guitar amplifier distortion channel, but not an SVT. For a square wave this represents over 600W of power. The 400W dummy load gets very hot, very quickly. I believe this is why the power supply drops from 660V to 600V and the ripple increases from 2V p/p to 15V p/p, thereby producing the ring modulated sound on the output. I would expect the power supply to be unable to regulate when attempting in excess of 600W output. I think this is a symptom of a problem elsewhere.

Still concerned about the +345V supply being +410V, whilst the +341V supply derived from the same source was correct, I thought that path must be taking more current than expected. On removing J3 IDC connector to the pre-amp board, a signal injected to the power amp now becomes clean. I think this must now be the area of my next investigation.