Output Transformer driven hard, What does I sound Like ?

[HaroldBrooks]

Never got a clear answer on this.

I believe most of the small Valco Push pull amps I own were made to be played modestly... But once I get my hands on them, all of that goes out the window fairly quickly !

I play a variety of music, but most of it is 70's to 90's hard rock, so I usually drive the bejesus out of everything I own, and I am fairly sure I am putting an abnormal load on the output transformers of at least a few of my amps.

I've never had one that got hotter than around 125 deg F, but I have a feeling the output transformer was contributing to some type of clipping or compression to the final signal.

Anyone have a detailed explanation / experience in what driving an output transformer too far does to the signal ? And more importantly, the changes to the real world sound and response of an amp as a result of the punishment ?

Thanks for any inputs !

[Helmholtz]

Core saturation limits power transfer at low frequencies. So expect losing low end. Core saturation also stresses the tubes, as it strongly increases plate currents at low frequencies.This changes tube operating conditions and will e.g. show in increased intermodulation distortion.

[R.G.]

It's hard to drive an OT too far inside a push-pull amp that's already working. Not impossible, but rare.

To the amplifier driving it, an OT looks like a big inductor in parallel with the transformed speaker load. You can saturate this inductor by applying too much voltage across it for too long a time. It's the voltage times time integral that saturates an inductor.

In a pre-existing push-pull amp, the voltage is already fixed - it's B+ minus whatever voltage the output tube can "saturate" to. So you have to put a voltage across the OT that is one-sided for enough time to push the OT's magnetic flux into saturation territory. For guitar the lowest frequency is relatively well defined - it's 82.4Hz, or maybe a bit less if you're fond of slack tuning. You can run a bass into it, of course, or put a signal generator on it to drive it at a lower frequency.

But the preamp and phase inverter stages will have their own high pass filters - the coupling caps from stage to stage. In addition, most preamp triode stages have a bass "step" dropoff from the action of the cathode resistor and cap. So even if you run a much lower frequency into the input, it usually doesn't get to the power amp and hence the OT.

Again, it's possible, but rare.

A much more likely source of distortion from driving a P-P amp harder is what happens to the circuit. It's very likely that the bigger drive to the output tubes causes grid conduction bias shift long before you're able to do anything much to the OT.

Transformers do have their own distortion mechanisms, but just overdriving a previously working amp won't highlight that much. The tubes' operating conditions do have well-known distortion mechanisms that come into play on massive overdrive, though.

[Old Tele man]

Pushed" OT's do not handle squarewaves (grossly over-driven sinewaves) anywhere as they do for true sinusoidal waves, which is what they're designed to handle. Note that a true squarewave/pulse transformer is both mechanically & electrically different from our common audio OT's.

[SoulFetish]

Core saturation limits power transfer at low frequencies. So expect losing low end. Core saturation also stresses the tubes, as it strongly increases plate currents at low frequencies.This changes tube operating conditions and will e.g. show in increased intermodulation distortion.

Also (if I understand it correctly) when a transformer reaches saturation and the current waveform becomes heavily distorted, this affects all frequencies, not just the lower frequencies which are pushing the core into saturation.

[J M Fahey]

Basically yes, if a core becomes nonlinear, it will be nonlinear for anything going through it at the same time.
In a nutshell, expect heavy intermodulation.
Of course, that´s part of "Guitar Toob Sound"

[R.G.]

Pushed" OT's do not handle squarewaves (grossly over-driven sinewaves) anywhere as they do for true sinusoidal waves, which is what they're designed to handle. Note that a true squarewave/pulse transformer is both mechanically & electrically different from our common audio OT's.

Well, they do, but neither "normal" transformers nor pulse transformers act like we commonly think they do. Both kinds act like a series leakage inductance into a parallel combination of the primary inductance and the reflected secondary load. In the designed-for-square-wave case the input signal is better known, and the time-domain response is critical. In the audio signal (sinusoid) case, the frequency domain response is critical, and this makes the designers optimize different things.

In every transformer, what goes into the primary is split up by the primary inductance and the reflected secondary load. For OTs, you want the primary inductance to be infinity, but settle for however much iron and copper you can pay for. For squarewave signal designs, you can let the primary inductance current penalty get bigger as long as you account for it in the design.

Also (if I understand it correctly) when a transformer reaches saturation and the current waveform becomes heavily distorted, this affects all frequencies, not just the lower frequencies which are pushing the core into saturation.

Saturation is an instant-by-instant thing. If you're feeding a transformer a sine wave with enough volt-time integral to saturate it, the instant it reaches saturation, the M-field no longer changes much. It's already as high as iron can support. So the primary inductance drops, letting through much larger magnetizing currents and the secondary coupling drops to only the coupling the coils would have if there were no core there at all. Normal transformer iron doesn't saturate that abruptly, but you can think of saturation as the core blinking out of existence during saturation - it no longer concentrates the M-field and couples the coils together.

That has the result that in the secondary, output voltage no longer follows the primary volltage, so the secondary output does become heavily distorted. In the primary, current skyrockets because the primary inductance no longer limits magnetizing current. This has the effect of overloading the active components driving the OT.

Basically yes, if a core becomes nonlinear, it will be nonlinear for anything going through it at the same time.
In a nutshell, expect heavy intermodulation.
Of course, that´s part of "Guitar Toob Sound"

Yep.

I was always fascinated by magnetic amplifiers. They're worth a web search or two if you are similarly inclined.

[Helmholtz]

Pushed" OT's do not handle squarewaves (grossly over-driven sinewaves) anywhere as they do for true sinusoidal waves, which is what they're designed to handle.

Any transformer is a bandpass filter (the lower limit being caused by the increasing magnetizing current, the upper limit by leakage inductance). So even without core saturation an OT will change the shape of squarewaves depending on frequency. (I wouldn't say that OTs are only designed to handle sinewaves as they need to handle the complex (non-sinusoidal) waveshapes of instrument signals.)

Clipping increases the halfwave area (area below halfwave = voltage x time area) compared to a sinewave of same frequency and amplitude. As peak flux density is proportional to the halfwave area, the risk of saturation increases with clipping at low frequencies.

[Malcolm Irving]

When a transformer is driven into saturation by a sine wave, the wave-shape of the current going into the primary takes on a very recognisable shape. This could easily be displayed on a scope as the voltage across a small resistance in series with the primary. So far, I've never seen that characteristic wave-shape in the context of a guitar amp output transformer. (Not saying it can't exist - just I've never seen it in real life or on the web.)

My own theory is that the falling magnetising impedance, at low frequencies, loads the output tube(s) so heavily that they become incapable of producing enough voltage swing across the primary for saturation to occur.

The heavy loading on the output tubes readily causes the output tubes to distort, which gives a different distorted waveform, which we can see anytime at low frequencies.

[Helmholtz]

When a transformer is driven into saturation by a sine wave, the wave-shape of the current going into the primary takes on a very recognisable shape. This could easily be displayed on a scope as the voltage across a small resistance in series with the primary. So far, I've never seen that characteristic wave-shape in the context of a guitar amp output transformer. (Not saying it can't exist - just I've never seen it in real life or on the web.)

My own theory is that the falling magnetising inductance, at low frequencies, loads the output tube(s) so heavily that they become incapable of producing enough voltage swing across the primary for saturation to occur.

The heavy loading on the output tubes readily causes the output tubes to distort, which gives a different distorted waveform, which we can see anytime at low frequencies.

I agree in principle. You won't find total or hard saturation when the transformer is driven from something like a current source. But I tend to interprete the falling magnetizing inductance as the onset of core saturation.

[Malcolm Irving]

... the falling magnetizing inductance …

Apologies. I meant to write 'falling magnetizing impedance' - I have subsequently corrected post #9.

The reactive component of the magnetizing impedance (2.pi.f.L) falls with frequency. If saturation occurs, the inductance itself (L) falls.

The low frequency -3dB point for the OT is where the magnetizing impedance has equal magnitude to the load impedance reflected from the secondary. I think that is typically somewhere around 80Hz for a guitar OT.

The possibility of saturation depends on the time-integral of voltage across the primary, halving the frequency will double the integral (because the time to integrate over is doubled) provided the voltage stays the same. However if the frequency is halved, the magnetizing impedance is halved. At low frequencies, below the -3dB point, the magnetizing impedance is the more significant load. For an approximately constant current drive, the voltage to the primary is then approximately halved - bringing the time-integral of voltage back to square one.

[Helmholtz]

I think the second picture in post #40 shows the typical signs of beginning saturation:

https://music-electronics-forum.com/...l=1#post523817

Max voltage distortion shows in the vicinity of zero crossing as that's where the 90° out of phase and non-linear magnetizing current has its maximum.

Admittedly this is down at 40Hz. But I tried to give a general answer to the question what effects could be expected when overstressing a somewhat undersized OT.

[Malcolm Irving]

I think the second picture in post #40 shows the typical signs of beginning saturation:

https://music-electronics-forum.com/...l=1#post523817

Max voltage distortion shows in the vicinity of zero crossing as that's where the 90° out of phase and non-linear magnetizing current has its maximum.

Admittedly this is down at 40Hz. But I tried to give a general answer to the question what effects could be expected when overstressing a somewhat undersized OT.

Yes, thanks for posting that. I agree - the two 'shoulders' near the zero crossings, with one shoulder being slightly lower than the other, is exactly what I've been looking for. It seems my 'theory' does not seem to be correct in every case.

[Helmholtz]

The pictures in the link also demonstrate that the voltage distortion effect may be at least partly "masked" by NFB. But during clipping NFB loses its corrective power.

[Malcolm Irving]

Further thoughts:

I think that plot is the output voltage of the OT across a resistive dummy load. What I really need to see is a plot of input current to the primary. The magnetizing current only occurs at the input of the OT and is not directly reflected in the output. In a sense, it is only reflected to the output indirectly by loading down the drive and hence reducing the driving voltage. The two unequal height shoulders are caused (I think) by hysteresis rather than saturation.

The real ‘proof of the pudding’ would be to see an OT input current waveform with spiky current peaks at the instants in time where saturation is occurring.