Sag effect, what time constant?

[bjoish]

Hi

Do anyone here have an idea what time constant is needed for a nice sounding sag effect?

My idea is to experiment with sag in my home built amp.
I have googled a lot without getting a grip of the time constant needed.

Thanks,

[guitarmike2107]

I don’t have a great technical answer for you on this, but my understanding is that you get voltage sag for different reasons in different amps. Studying some of those classic amps that do sag allot would be a good place to start.

Is it a push pull amp or a SE amp you are wanting to try it on as that effects what you can do to introduce sag?

[lowell]

Not sure a time constant figure is what you're after. I may look at it as more of a current and resistance figure. The 2 largest factors regarding sag for me are preamp grid leak values and screen grid resistance. What is your home built amp? It'd help to know .

[bjoish]

Thanks for your replies.

What I am after is experimenting with a simple supply voltage sag effect.
Basically a RC network.

The amp I am working on is an SS-amp inspired by Pass-labs Zen-amp; simple MOSFET circuits, no NFB, but in my case class AB.

[pdf64]

The germanium fuzz face type circuit I built has a noticeable sag effect, mainly I think because the bias shifts in response to the signal level.

[lowell]

This amp is class A. Is yours different? Class AB? From the schematic I might experiment with a resistor instead of a ccs.

[Steve Conner]

0.5 second? Google compressor settings for guitar: it's the same thing.

[bjoish]

This amp is class A. Is yours different? Class AB? From the schematic I might experiment with a resistor instead of a ccs.

Yes, my design ended up a 'little more complex'.
For a guitar amp a prefered class AB so I added push-pull.
Some trimming was needed for the bias and the linearity and a dc-servo for handling the dc-offset.

The result is very promising so far. The high output impedance let the characteristcs of the speaker shine thru.
I think the sound is not as harsh as ordinary ss-amps sometimes tend to sound.
My idea is to simulate a weak power supply in order to add even more character to the sound.

Has anyone here experimented with similar MOSFET circuits?

[bjoish]

Thanks for the reply!

Do you think 0.5s is representative for a tube amp with a weak power supply?

[Steve Conner]

Roughly. The time constant is the DC resistance of the PT's HT winding, together with the filter capacitance. So maybe 47 ohms and 100uF.

That's shorter than 0.5 second, but there are actually several time constants. With the pentode and tetrode tubes commonly used in guitar amps, the plate voltage only sets the maximum output power before clipping. It's screen voltage that sets the gain, so you have the time constant of the screen filter network, too. Sometimes that has a choke, so it's actually a second-order network that can oscillate and bounce, and can't be described by a single time constant. Though with the usual values of L and C it is well overdamped. Then there are the time constants of the preamp filtering, etc.

If I had to sum all of that up in a single figure, I'd say 0.5 second.

I've built a few hybrid amps. My latest one is a SE EL84 amp that can drive the speaker directly, or the output can go through a power soak to reduce it to 0.3W, or into a current booster based on BJTs, that takes it to about 30-40W. (I measured 42W clean on the bench.)

It also has high output impedance, and works pretty well. It took me years to think up a simple output stage that was linear enough, didn't depend too badly on transistor beta, had protection and needed no DC servo.

The tube part being Class-A, it doesn't sag in the low power modes. But because of bias shift in the overdriven stages, it sounds as if it does. Bias shift and sag are two separate but interacting effects.

[bjoish]

I suppose that if the mains power transformer is under rated, this would also add to to the sag.
I have read somewhere that rectifier tubes has an internal resistance of about 100 ohms...

Thanks for reminding me about the effects of the bias shift.
I think I should consider some experimenting...

[pdf64]

Thinking about it, aren't bias shift and sag both forms of 'signal level modulation of device operating conditions', 2 sides of the same coin?

[loudthud]

As a point of reference, consider the 5F6A.

The effect of sag on tone as a function of time is a complex function. The effect on clipping level is rather immediate, only one or two cycles of the line frequency. Not only is the clipping level reduced, but there is the 2x line frequency sawtooth injection on the tops of the waveform. Next comes the gain reduction from the sag at the screen grids followed shortly by the effect of sag on the phase inverter. Then finally there is the much slower gain reduction and bias shift of the preamp.

I have experimented with the slower effect on the gain of the preamp. This was simply an incandescent bulb through a resistor from the main power supply node coupled to a CdS photo cell arranged to reduce the signal slightly in the signal chain. It worked fairly well and gave a slightly compressed quality to the signal.

Attached is a single shot oscilloscope photo of a 50W four input Marshall. IIRC it was into a dummy load.