Help understanding tube preamp output levels

[armstrom]

I'm considering using a tube preamp in front of a high-gain (Av of 20-30x) solid state power amp (more specifically, a non-inverted LM3886). I know for sure that pretty much any tube preamp I choose will generate voltage levels far too high for the 1V RMS input requirements of my power amp. My plan is to add a simple voltage divider to bring the voltage in line and then add some additional soft clipping diodes to guard against any spikes above "normal" signals making their way though.

I'm still trying to learn how to calculate the gain of tube stages so I'm totally out of my element. Couple that with a inter-stage tone control circuit and I have no idea what kind of output voltages to expect from a simple tube preamp. Hopefully some one here has had a chance to put a scope on the output of a similar preamp and can give me some indication of what to expect.

Here's the preamp circuit I plan to use:
http://www.moosapotamus.net/IDEAS/F2B/alembic.htm

Given a "normal" guitar signal peaking around 200mv-400mv what kind of attenuation is required on the output of the preamp to bring the average voltage in the 2.8V Peak-to-Peak (+/- 1.4V) range required for my power amp?

Would I be better served by changing the tube to a lower gain member of the family like a 12AU7 or 12AV7? If so, what changes would be required to the circuit? Presumably the plate resistor would need to change and probably the cathode resistor.

Finally, am I totally wasting my time trying to go down this route? I know full well that simply having a tube preamp will not give me "that tube sound" achieved by all tube amplifiers. What I'm after is a less "sterile" tone. Something with some warmth and character. My goal is not to have the preamp produce distortion either. There are plenty of pedals out there to do that.

Thanks in advance for any information and advice.

-Matt Armstrong

[tubeswell]

Hi Matt

FWIW I found the valve wizard article on triode gain stages helpful for my understanding about pre-amp gain stages, with equations for calculating gain, transconductance and input and output impedance.

A straight voltage divider would work but it would be a bit crude, because you would lose current through the series resistor as well, and that might not be optimal to feed your power amp.

Apart from using a cathode follower stage or a small transformer at the output side of your preamp (i.e.; to lower the voltage but keep the current higher), you could look at the Peavey C30 or Delta Blues FX loop send (which uses a small transistor to drives the FX loop). I know that drives a SS power amp nicely (because I have tried it). You get the pre-amp tube distortion and compression, and you could have an elaborate preamp with multiple channels and stages. You would need a low voltage power rail, as well as a bias supply, for the transistor.

[GibsonLover]

Hi Matt

I'm sure you've Googled the F2B and know who uses it, so it is a perfectly good choice for a preamp. I've seen tubes added for output buffers and balanced outs for recording. It has plent of gain, but still in the order of several volts. I would replace the output resistor with the same value pot and proceed to turn it up until you get the level you want. Either leave it in as a Master (maybe on the back panel), or measure the value and replace the pot with an appropriate resistive divider. The sound will always be subjective, depending on what you are driving, cable length, etc., but you gotta start somewhere. My point is that at least you will be able to see if it will work with your amp.

[armstrom]

Thanks so much for your replies!

Would a voltage divider lose any tonal quality though? I'm not too concerned about the loss of current because the preamp will immediately feed into an opamp buffer. Here's how the signal chain will go:

Guitar->Pedals->Alembic F2B preamp->Buffered DSP reverb->Master Volume->LM3886 power amp.

The reverb module I plan to use only allows a maximum of 2.8V peak-to-peak (1V RMS) so as long as I condition the signal properly for the reverb unit it will be fine for my power amp. I can always play around with the gain of the power amp to get it into the range I need, I just can't go below an Av of 20x or the chip becomes unstable.

The output of the tube preamp will run directly into a voltage divider (10:1 maybe???) for simplicity I will probably use a trim pot to allow me to "dial in" the signal attenuation using my scope. The output of the buffer will be "soft-clipped" to +/- ~1.4V using diodes and fed to the rest of the signal chain. I'll draw up a schematic to illustrate this.

Regarding the use of a transformer... I have to admit that my knowledge of audio output transformers is a bit limited. I'm assuming such a transformer will lower the voltage just like a power transformer? Does the same ratio apply? For example, radio shack sells a little 1KOhm to 8Ohm output transformer... does this mean it will attenuate the voltage by 1/125? (10V on the primary becomes 0.08V on the secondary) If so, I would have to be careful in selecting a transformer otherwise I could lose all my gain Could I just use a power transformer backwards? Something like a small 120V/12V tranny would give me a 10:1 attenuation if run backwards...

Sorry for the barrage of questions... I'm a quick study but some of this information is hard to dig up.

[armstrom]

Here is a rough schematic of what I'm thinking. The voltage divider pot in the drawing would replace the 1M output resistor in the Alembic F2B schematic I posted earlier. This will likely be a trimmer or a pot mounted inside the chassis to avoid accidental adjustment resulting in smoke

I left off all the virtual ground/biasing stuff for the opamps since we all know how that works as well as some component values (AC coupling caps).

Does this seems reasonable? The soft clipping diodes should give me a voltage limit of about +/- 1.4V thereby limiting the total voltage swing to 2.8V P2P.

Edit: Oops.. just saw a pretty fundamental flaw in the design. My summing amp will attenuate the signal by half if the reverb is all the way down. Can someone suggest a better mixing circuit? Would it be sufficient to simply splice the output of the DSP buffer in with the dry signal and convert the final opamp into a non-inverted buffer?

[Wakculloch]

I don't understand why your trying to limit the input to your power amp. Once it hits the rails it'll square off, and as long as you've over adequate heat sinking it can go no farther. If you limit it before the power amp your just lowering the headroom and presenting the power amp with a pre-clipped signal. Seems a bit pointless.
Of more concern is the impedance mismatch which could start to cut treble,so as has been mentioned earlier in the thread, an cathode follower/ impedance buffer/ driver stage is more important.
Hanging a mosfet off the last valve is a proven way to go.

[armstrom]

I don't understand why your trying to limit the input to your power amp. Once it hits the rails it'll square off, and as long as you've over adequate heat sinking it can go no farther. If you limit it before the power amp your just lowering the headroom and presenting the power amp with a pre-clipped signal. Seems a bit pointless.
Of more concern is the impedance mismatch which could start to cut treble,so as has been mentioned earlier in the thread, an cathode follower/ impedance buffer/ driver stage is more important.
Hanging a mosfet off the last valve is a proven way to go.

There are a few reasons why I'm trying to limit the voltage to the power amp:

• I'm using a digital reverb board that has NO voltage protection. It's digital, so first you get nasty digital clipping, then you fry the board. Not fun.
• The power amp I'm using has a sensitivity of around 700mV (gain of 33x and +/- 30V rails) so unless I reduce the voltage somehow basically any tube preamp will clip all the time.
• I want to be able to set the volume fairly high and still allow a wide dynamic range from the guitar. If you play a loud chord it shouldn't suddenly jump into nasty opamp clipping sound. Adding the clipping diodes will help to smooth out the high voltage transients and keep the power amp from clipping. As you know, unlike tube amps you really don't want to clip a solid state power amp.

Regarding buffering the preamp... do you think my solution of using an opamp buffer will suffice? In a unity gain, non-inverted configuration the opamp has an input impedance close to infinity (many MOhms) and an output impedance of a few tens of ohms.

[Steve Conner]

I've done most of the combinations of tube, op-amp, JFET, power amp, preamp, digital and spring reverb, and many of them sound pretty good to me. A Fender Twin might be clean, but a 200 watt MOSFET power amp is really clean. Some would say it's too clean and needs help from our nine-legged friends.

The voltage divider method is totally valid, and necessary. The "headroom lowering" mentioned by a previous poster is exactly what you are trying to achieve. Imagine going into a solid-state PA amp with a treble booster and driving it into clipping: it would sound like treading on popcorn. But hitting tubes with a treble booster sounds great. The trick in hybrids is to make sure the tubes break up before the transistors do.

You're running the tubes hard to compress and dirty the sound in a desirable way, lopping off peaks that would make the solid-state finals clip, then throwing away the excess output voltage that you don't need, to drive a solid-state power amp running clean. Ideally you have a master volume here, to essentially control the amount of tube dirt.

I've tried voltage dividers as well as transformers (a mic matching transformer run backwards) and the divider probably colours the sound less than a transformer. Any "tonal degradation" will be treble loss, curable by a bright cap. So I'd say avoid transformers unless you want the coloration (bass distortion and reduced low end) of a transformer.

[Wakculloch]

I think you've got into a bit of a circular argument with yourself.
The preamp will only clip the following stages when it hits max output. I dont know what the wattage of the power chip is, but it will only clip when its rated output is exceeded. This will probably be loud. If it starts to clip turn it down a bit.
It won't clip all the time, it will clip when you turn it up enough to achieve this.
Diode clipping good, power amp clipping bad, is not a rule that holds true.

[Steve Conner]

Well, I've tried diode clippers that sound rotten, and the old Maplin MOSFET power module clips so gracefully that you don't really even notice.

But the general rule that I believe always holds, is that any tube distortion sounds better than any solid-state distortion for musical instrument applications.

Another general rule that I believe always holds, is that everyone's idea of a "clean" guitar sound actually contains 10dB of treble boost followed by several dB of soft tube clipping, which corresponds to about 10% distortion on pick attacks. You can't emulate that with just diodes and op-amps.

[armstrom]

I think you've got into a bit of a circular argument with yourself.
The preamp will only clip the following stages when it hits max output. I dont know what the wattage of the power chip is, but it will only clip when its rated output is exceeded. This will probably be loud. If it starts to clip turn it down a bit.
It won't clip all the time, it will clip when you turn it up enough to achieve this.
Diode clipping good, power amp clipping bad, is not a rule that holds true.

Have you heard an LM3886 clip? It's not a nice sound. My power amp is running at full output with about 750mV RMS input signal (gain is set to 33x and my supply voltage is +/-30V) Similarly, my reverb board will fry at anything above ~1.5V RMS input signal. I realize I can just leave everything as-is, use a 1M Master volume pot and call it a day. However, I wan't to build a robust amp. With no other over-voltage protection in place it will be possible to destroy the reverb DSP chip (more specifically, the A/D converter on the input) just by cranking up the master volume control. I doubt higher voltage from the preamp will damage the power amp chip, but it sure won't sound good. Stacking diodes will clip more smoothly than letting the power amp clip (and also serve to provide floor/ceiling voltages for my reverb chip). I'm not sure wh

[Wakculloch]

I'm trying not to enter pedantic territory here, but the preamp you're using doesnt have rediculous amounts of gain if i've understood the schematic properly, lots of gain being lost in the tone stack.
I cant see a situation where your overdriving the following stages until your well up on the volume control.
If this is the chip they stick in the cheapo marshalls and such i have heard it but didn't give it much thought.
Most valve power amps input sensitivity is between 0.5 and 1V rms, some lower.

[armstrom]

Ok, now we're getting into the information I was looking for (Hence the title of my first post). If the maximum output I can expect from this circuit (given a reasonable maximum guitar signal level) even with the volume all the way up is within my input range of ~1VRMS, then I can just add some diode clipping to protect against "rogue" high voltages frying my DSP. Also keep in mind that I'm in no way trying to emulate the sound of an overdriven tube amp. I'm simply trying to leverage the "warm" tone characteristics of a tube preamp (compared with more "sterile" sounds from op-amps). The clipping diodes are there more as protection against over voltage than for adding distortion. Under normal operating conditions the circuit should behave the same with or without the diodes in place. However, if the volume is cranked, and the signal starts to get out of a safe range then I need something to protect my digital circuitry. Ultimately, I will build the circuit and use my scope to determine the maximum output voltage before adding it into the mix. I just wanted to get some information ahead of time to see what I'm in for .

As for the power amp chip, I don't know of any Marshall amp that has used it (though I'm not up on all the latest designs). The chip is actually designed for audiophile quality Hi-Fi amps. The design I'm using is called a "gainclone" which is a clone of an outlandishly expensive Hi-Fi amp called a "Gaincard" They were charging thousands for an amp based on a chip costing $6... Even though this is a hi-fi amp chip it has been used successfully in a number of DIY guitar amps. Here's the datasheet if anyone's interested: http://www.national.com/ds.cgi/LM/LM3886.pdf

[Steve Conner]

I think the Marshalls use a TDA-something-or-other. Enzo knows, since he has to replace blown ones all the time. But it's the same kind of chip-amp deal as the LM3886. They're just commodity blocks of power, and I doubt anyone could tell any of them apart in an ABX.

FWIW, I've accidentally overdriven the heck out of one of those digital reverb boards with a cathode follower running off 250V. It sounded atrocious, but there didn't seem to be any permanent harm done.

A diode clipper in front of the reverb will certainly be a good idea, since it'll probably clip a lot more cleanly than the ADC itself. It has digital anti-alias filters that can overflow and mash the sound beyond recognition.

There are all sorts of soft clipper circuits (using low voltage zeners, LEDs, germanium diodes, high voltage diodes, etc) that would maybe sound better than plain diodes. I think Craig Anderton even suggested using a clipper in front of spring reverb, for some reason.

I have a thread on this forum about my attempts to fit one of these boards to a tube amp.

[armstrom]

I found your posts regarding the reverb board. I'm using the same board (You can get it in the US direct from the manufacturer, profusion just resells them under a different name).
Do you think your dynamic range issues were related to trying to integrate the board into a tube amp? My current thinking is that since the reverb board has about the same input sensitivity as my power amp, it should be a fairly decent match. (I hope!).

What kind of mixing circuit did you use to recombine the dry and wet signals?

If you have some free time and don't mind putting up with newbie questions I would be interested in taking this conversation to email as you've brought up some interesting options I may want to consider further. My address is armstrom at gmail.com