Preamp supply sag

[EFK]

Has anyone come up with a way to modulate the voltage supply to the preamp in a semi-automated manner via the audio signal? Obviously if the preamp is drawing it's B+ supply down the chain from the power amp supply, the supply voltage is going to bounce up and down as is typical to most amps as the power amp draws current. However, how about in a situation where the preamp is supplied from a different or very stiff source? The current draw of the preamp (LTPI through input stages) is at @ 8mA total, and doesn't really change one bit from idle to full bore with a signal injected. This is a later 60s plexi preamp, so it has a lot of capacitance which further keeps it very stiff. I may try to lower the capacitance somewhat, maybe to JTM45 level or even lower, but it may let some increased hum back in which I don't really want and frankly I doubt it will really affect the sag - or I should say, lack of sag, as the current draw just for the preamp really is not changing. What I need to do is figure out a way to cause the preamp supply to drop 20-30 V or so when the amp is being driven to emulate what will happen in a typical plexi circuit when the power amp gets cooking and the B+ sags because right now it is not sagging and the preamp is just too stiff. I suppose it could be done using a 2W pot as a variable resistor, maybe 10K or so, but this seems to typically be frowned upon and it would have to be adjusted manually. So I was thinking is there some way to use the variable audio signal as a "guide" and implement some way of causing the B+ to increase/decrease while tracking the audio signal? Or some other method? I guess I'm looking for something like one of the popular VVR controls, but tied only into the preamp supply and somehow running in an self-adjusting manner. Is this making sense?

[EFK]

I found this over on the AX84 board from just a couple of months ago, looks like someone was thinking along the same lines. I know nothing about mosfets though so I guess I'm going to have to read up to figure out if this could be implemented:

"A crude overkill sort of scenario I had in mind was to take a mosfet like a IRF820 (or a valve stage), power it from the same power cap as the most critical drive stages in the preamp, and bias it just at cutoff. Feed it from the signal somewhere, but don't take any output from it. Then at idle it does nothing but as the signal increases it amplifies half the wave and starts to draw current, causing the B+ to sag and the valve stage on the same supply capacitor to lose headroom. "

Is this something that would work?

[teemuk]

So I was thinking is there some way to use the variable audio signal as a "guide" and implement some way of causing the B+ to increase/decrease while tracking the audio signal?

I've seen two methods:

1. Use a regulated supply. But instead of using "fixed" DC voltage reference for the regulator modulate the DC reference with signal coming from an envelope follower. Input to envelope follower is the signal you wish to "track" for simulated sag characteristics. Make the circuit operate so that higher amplitude output from the envelope follower decreases regulator's output voltage. (i.e. you may wish to invert the envelope follower output).
This is the scheme used, for example, by Maven Peal, Fryette, Line 6 and Quilter. Maven Peal and Fryette have patented their circuitry.

2. Add something that draws more current than the low current preamp stages as is. Roland's older Blues Cube series amps, for example, had a little "power amp" in the preamp section running to a dummy load. The operation is pretty self explanatory: The additional amp circuit driving a moderately low-ohmic load draws more current than the typical preamp stages. Since both are fed from same supply the current draw sags the voltage.

[pdf64]

Bear in mind that the power tube screen grids tend to sag more than the plates, and in terms of the effect on things, it may be more accurate for the pre-amp HT sag to model the screen grid sag rather than (power tube) plate.

[teemuk]

A typical supply will also have a "ripple" component in it because the DC voltage is created by means of rectifying and filtering a sinusoidal signal. The magnitude of this ripple component will increase dynamically with the current draw since the capacitance of the filter circuit can not be charged as effectively due to increased current draw. Basically, the greater the sag, the greater the magnitude of ripple.

Like sagging, this ripple component will likewise modulate clipping thresholds, superimposing itself to the clipped signal. The audibility and "musicality" of this effect likely varies a great deal with each individual. Some hear obnoxious ghost notes due to superimposed ripple, some hear practically no effect from ripple modulation (that would be me) and some people sense that an essential ingredient of "tube-yness" is lost if there is only supply voltage sag but no associated ripple modulation.

[EFK]

Bear in mind that the power tube screen grids tend to sag more than the plates, and in terms of the effect on things, it may be more accurate for the pre-amp HT sag to model the screen grid sag rather than (power tube) plate.

I believe you're correct. Just insofar as what I'm trying to do, however, I'm going by memory based upon the last time I had a Marshall here which was a early 70s metal panel I sold about a year ago. I know the supply voltage at the input stage feed was sagging about 20-30 V just going off what I recall when I measured it w/ a 1 kHz input signal and the amp 'cranked.'

2. Add something that draws more current than the low current preamp stages as is. Roland's older Blues Cube series amps, for example, had a little "power amp" in the preamp section running to a dummy load. The operation is pretty self explanatory: The additional amp circuit driving a moderately low-ohmic load draws more current than the typical preamp stages. Since both are fed from same supply the current draw sags the voltage.

This is very interesting because I was thinking about a way to maybe string a pentode down at that end to do something like this, however a SS design would certainly be more discrete I suspect. Will have to look into this.

Tried to read up on the Maven Peal design but I see it's patented.

What I've been fiddling with here is a lower watt plexi type amp as I sure don't need 50W plus. I've got it down to 10W at full-on output into an inefficient speaker (single greenback), which for a true fixed bias pp power amp sounds awesome and a hugely noticeable volume reduction. I'm running the power amp at 240V plates/230 screens but I maintain a 500V supply to the LTPI/preamp which is filtered and dropped and then puts the pi/pre at typical 'black flag JTM' (50W) or later 60s preamp voltages. But, because the preamp supply is not being pulled down by the power amp when driven, it's just too stiff. For cleans of course it's awesome but when driven it's just a bit too much sledgehammer to the face.

I may try to just lower the supply voltage via a simple/fixed resistive method and see how far I can go before it begins to break up too early to sound right but I thought I'd investigate a more complex solution first. Maybe this is backwards!

Should probably try some capacitance changes too. Right now the supply for the pi/preamp is at 32 uF/15K/32uF, at which point it feeds the ltpi, then the standard 10K/32uF for the cathode follower then another 10K/32uF for V1. I measure only 9 mV of AC at the pi node and the amp is darn near dead silent insofar as hum.

Thanks guys.

[EFK]

Hmmmmm, very good point. And I actually like those ghost notes! The majority of my work is with trashy little amps like old Silvertones and Valcos, so I *expect* to hear them.

[tedmich]

Tried to read up on the Maven Peal design but I see it's patented.

That usually makes it much easier to read up on! A patent is just a license to sue anyone who makes too much money on an idea you claim, little people are largely exempt.

The European Patent office is a better place to get nice pdf files of granted patents IME, while applications are best searched at the clunky USPTO site.

[printer2]

I believe you're correct. Just insofar as what I'm trying to do, however, I'm going by memory based upon the last time I had a Marshall here which was a early 70s metal panel I sold about a year ago. I know the supply voltage at the input stage feed was sagging about 20-30 V just going off what I recall when I measured it w/ a 1 kHz input signal and the amp 'cranked.'



This is very interesting because I was thinking about a way to maybe string a pentode down at that end to do something like this, however a SS design would certainly be more discrete I suspect. Will have to look into this.

Tried to read up on the Maven Peal design but I see it's patented.

What I've been fiddling with here is a lower watt plexi type amp as I sure don't need 50W plus. I've got it down to 10W at full-on output into an inefficient speaker (single greenback), which for a true fixed bias pp power amp sounds awesome and a hugely noticeable volume reduction. I'm running the power amp at 240V plates/230 screens but I maintain a 500V supply to the LTPI/preamp which is filtered and dropped and then puts the pi/pre at typical 'black flag JTM' (50W) or later 60s preamp voltages. But, because the preamp supply is not being pulled down by the power amp when driven, it's just too stiff. For cleans of course it's awesome but when driven it's just a bit too much sledgehammer to the face.

I may try to just lower the supply voltage via a simple/fixed resistive method and see how far I can go before it begins to break up too early to sound right but I thought I'd investigate a more complex solution first. Maybe this is backwards!

Should probably try some capacitance changes too. Right now the supply for the pi/preamp is at 32 uF/15K/32uF, at which point it feeds the ltpi, then the standard 10K/32uF for the cathode follower then another 10K/32uF for V1. I measure only 9 mV of AC at the pi node and the amp is darn near dead silent insofar as hum.

Thanks guys.

What output tubes are you using? I am thinking of doing a Plexi with 6AQ5's as outputs.

[ThermionicScott]

What I've been fiddling with here is a lower watt plexi type amp as I sure don't need 50W plus. I've got it down to 10W at full-on output into an inefficient speaker (single greenback), which for a true fixed bias pp power amp sounds awesome and a hugely noticeable volume reduction. I'm running the power amp at 240V plates/230 screens but I maintain a 500V supply to the LTPI/preamp which is filtered and dropped and then puts the pi/pre at typical 'black flag JTM' (50W) or later 60s preamp voltages. But, because the preamp supply is not being pulled down by the power amp when driven, it's just too stiff. For cleans of course it's awesome but when driven it's just a bit too much sledgehammer to the face.

Are you sure that it's a lack of preamp voltage sag that's causing the stiff sound?

[Roberto]

Hi,

Based on the assumption that the sag is the right solution against the stiffness of the preamp, my solution is:

- the piloting system has to be the piloting system of a compressor: analyse the output swing;
- the piloted system can directly be the power supply: rectify the ac coming from the secondary of the PT and go to a first cap, then to a variable voltage divider, then to the cap that supplies V2 of your preamp (or even better to V3, including the PI, and then to the preamp). The "variable voltage divider" will be a 1k resisor in series with the supply, plus a power transistor (piloted by the output swing of the preamp) and then a 1k resistor in series with the supply (the two resistors and the transistor form like a T on the schematic).

This way you lower the voltage on the supply each time you increase the output signal swing (because you reduce the ratio of the voltage divider by decreasing the Resistance between collector and emitter of the transistor), obtaining the effect that you want. I'd put a resistor in series with the transistor, in order to reduce it's effectiveness.

Pay attention to isolate the output of your preamp from the B+ of the transistor you are piloting.

[loudthud]

Are you sure that it's a lack of preamp voltage sag that's causing the stiff sound?

This (the stiffness) is generally recognized in amp setups that utilize a separate (perhaps rack mount) preamp where the preamp power supplies are rock steady. Not all players like the effect, especially super high gain metal players. It's more suited to Tweed, blackface, JTM45 and 5F6A style amps, not Triple Rectifier or SLO types.

[Roberto]

I "solved" this issue once in a WDW setup by supplying the preamp from the last node of the dry PA.
But it's something I can do for my rack, I won't feel safe to do it for someone else.

[ThermionicScott]

This (the stiffness) is generally recognized in amp setups that utilize a separate (perhaps rack mount) preamp where the preamp power supplies are rock steady. Not all players like the effect, especially super high gain metal players. It's more suited to Tweed, blackface, JTM45 and 5F6A style amps, not Triple Rectifier or SLO types.

Shh! I wanted to get EFK to walk us through his thought process.

[EFK]

Some more good stuff. Roberto, thank you - interesting approach and I will look into this. I've not ever done anything with transistors but your explanation makes sense and sounds theoretically like it would do what I want it to do.

For anyone still interested, I set up the PS here in a roundabout way to get the voltages I wanted, and after some more experimentation I'm beginning to wonder if it may be contributing to the over-stiffness in some way that I don;t quite understand because when I take measurements at idle vs. under load, odd things happen which I am not used to seeing.

I am running a 380 vdc, CT power transformer into a 6CA4 tube. The rectifier here is acting as the forward portion of a bridge I believe, then I am running two reversed diodes off the secondaries to ground (anodes to ground). This is giving me the rectified high (490-500 vdc) preamp voltage. I run this into a pair of stacked can caps, each a dual 32 uf. I run the PT center tap to the mid point of the totem setup, and so I get around 245-250 vdc there. This is where I'm supplying the plates and screens only. Bias is fixed, cap coupled off a 55 VAC tap on the PT.

When I hit the amp with load, there is only a 9 vdc voltage drop from idle to "dimed." This is at the power tubes. The preamp does not sag at all, not one bit. I had the bright idea to work with a sag resistor between the tube rec and the top of the totem pole, but this is where it's getting weird. I went up to 400 ohms but the whole thing hardly sags at all. ??? the only thing that appears to be sagging is the preamp voltages taken from the top cap, and they are only dropping a minuscule fixed amount with no other change under load. I would think that despite the fact that the power amp supply is being pulled from the mid point of the totem pole, it should still pull down the whole chain with a big honking sag resistor right off the rectifier, but this is not the case.

This setup is a first for me, so maybe something is going on of which I'm not aware?