What is Gain Structure?

Joined 2002
Paid Member
Great write. Thank you.
Although we are concerned here with playback on commercial recordings where mastering has already taken care of high peaks, VU meter readings is much lower from Peak meter dbs. With classic and opera works, Peak meter readings are between 12db and 18db higher than VU meter dbs .
I’ve seen up to 24db difference.

Therefore, the 6db overhead that is mentioned, it should be applied to Peak meter indicators. If the instrument is a VU meter, the overhead has to be a conservative 20db.

That the meters are implemented in the software, it doesn’t mean that they are designed to respond to the signal's impulsive content instantaneously.
Even when one is monitoring through the digital bit depth meter (dbs rel dbFS ) , he has to know the “ballistics” of that particular instrument. Not every SW supporting documentation makes things clear on this.



  • VU vs Peak.JPG
    VU vs Peak.JPG
    130.8 KB · Views: 185
Last edited:
Great article. Explanation is spot on. I've worked as a professional recording and mix-engineer, both live and in the studio. Maintaining a good, healthy and clean signal throughout a complex system, can be a challenge. Especially when you add devices that add frequency-specific amplification (or attenuation), like EQ's. Compressors in the chain can increase noise problems dramatically!

The only point of criticism I have about this article, is the mention of the DCX2496 digital active cross-over processor. It's said to be meant for "the pro-audio world". Let me rectify that.

The DCX2496 is a Behringer device. This device may offer a buch of features that are quite interesting for hifi-enthousiasts who want to expand their systems (multi-amp processing, for example), in terms of sound quality is certainly isn't the best choice. Behringer, although it has a good reputation of delivering "bang for buck", is a brand not frequently used or loved in the higher ranges of the pro-audio world. It simply doesn't deliver the sound quality required in most hi-end studio's or hi-end PA systems. This has to do with durability, S/N ratio, linearity and accuracy of the AD/DA conversion, linearity of the input and output stages, etc. Since most well-developed home-audio systems sound as good as some studio systems, it would be a shame to expose them to a component that sounds not as good as it should.

If you want a device with similar functionality, look at names like DBX, BSS minidrive or omnidrive, Dolby or Electro Voice. Really. These can be had on the used market for decent prices. Still more expensive than a new Behringer, but your ears will love you for it. Quality comes at a price - that's a universal truth.

Let me add that since Behringer acquired the Midas brand, they have incorporated a bunch of Midas technology into their own designs. That can only be a good thing. Take a look at the X32 mixing desk, for example. Tons of features and good sound at a competitive price. At the moment, however, there is no Behringer active cross-over on the market with Midas technology in it.
Last edited:
Antoinel: "I advocate the design doctrine of Nelson Pass which is: "keep it simple". A CD player has a volume-controlled headphone amp. It is a high fidelity mini power amp ~15-20mW (rms) into 30 Ohm loads. The CD player's volume control may be manual (e.g. Marantz), or motorized via a remote controller (Sony). The resultant output then goes directly to the input of the power amp. This approach is a high efficiency and least-noise gain structure."

Adason: "while it is feasible, to put headphone cd output straight to the amp, it usually does not sound very good. The reason is that the best cd players do not have headphone output at all, those who have have it for some sort of monitoring, not for real listening. Well, I tried it, it just does not compare with dedicated high quality headphone amp fed by line signal. The line level out can be high quality, but headphone amp inside cd player is usually of mediocre quality. Most of the time one OPA. Rarely discrete. Never tube headphone output. If yes, please point me to such cd player."

I agree that connecting a headphone output direcly to the line-input of a (pre-)amp is not a very good idea. You might get away with it on some pro-audio gear (with low input impedance - 600Ohms or so), but not with most cosumer gear. The reason for this is output-to-input impedance matching. A 30 Ohm output will not drive a 47+ kOhm input very efficiently, resulting in suboptimal sound. Usually thin sounding low end and, brittle highs and no real 3D-scale in the sonic image. Not to mention added distortion because of the simple headphone amp, that was not designed for critical listening sessions.
Joined 2004
Paid Member
James - thanks! Glad you like it. I hope that it will give users unfamiliar with the concept something to thing about and a basis for improving signal flow.

Quite right that it gets even more complicated when filters and compressors are added. That can take some real juggling. Most home users have no idea how complex live sound can get.

I agree and disagree on the DCX2496. Yes, in stock form it ain't pretty sounding, but if you search the forums here you will find that there is a quite a bit of modifying going on. Mostly it's the analog sections of the DCX that suck, followed by the power supply and maybe the clock. But the A/D, DSP and D/A are quite good, really I have a DBX Driverack 480 here and it uses many of the same AKM chips. It just has more in/outs and a much better analog section than the Behringer.

Have you used any of the Bi-Amp processors? We use them now at work and they are very nice. We replaced the DBX with a Nexia processor and so far are happy with it. Sounds good, very flexible. Mostly made for installed systems, but that's how we use it.
Great article! I've been re-examining my hifi setup (all DIY tube amplification with store-bought turntable, SACD player and FM tuner). My power amp has very low sensitivity, by home audio standards. It takes about 4V peak input to get to clipping.

I used to use a line preamp with about 12X gain (about 22dB) and output impedance of about 2k ohms. The grid bias on the input was about 4V, able to accommodate the 2.83V peak output from a CD player without distress. At one point, I put the 100k volume control on the output of the preamp, right after the output coupling cap. The wiper went to the RCA output jack. Max output impedance would then be 25k, but power amp's input impedance is about 10x that, at 220k (all of this stuff is vacuum tube).

This idea was criticized for having the preamp run at full tilt into the output level control, which means more distortion from the preamp.

My question is, how is putting the volume control on the output of the preamp different than putting it on the input of the power amp? What are the pros, cons? Or are you recommending using a preamp volume control *and* a power amp input level control, to balance the gain structure to maximize signal to noise ration for your particular setup?

Sorry for changing the subject.

Joined 2004
Paid Member
A good question and not off topic, I think.

IMO, putting the volume pot at the output of the preamp or at the input of the power amp are almost the same thing - except for cable impedance. Don't forget that the last stage (whatever it is) of the preamp has to drive the cables and the input stage of the power amp. Having a low impedance/high current output stage is usually helpful if cables are long.

Pro audio gear usually has both input and output volume controls, as well as buffers on the ins and outs. That because we don't know what over pieces of gear will be connected to it today or tomorrow. Home stereos are often much easier

In most cases I think that 12dB is plenty for a home preamp, but it might not be enough for your amp with its low sensitivity. 12X amplification does seem a bit much, tho, unless your sources are low output. About 1/2 that ought to drive your amp into clipping even from something like a 0.75V phono preamp.

I would say that if your interconnect cables are short and of good quality, then there won't be much practical difference in putting the volume pot at the output of the preamp or input of the power amp. Just be sure your preamp isn't clipping or driven into other distortion with high level sources, like a CD player.
I used to use a line preamp with about 12X gain (about 22dB) and output impedance of about 2k ohms. The grid bias on the input was about 4V, able to accommodate the 2.83V peak output from a CD player without distress.
the 2.83Vpk (=2Vac) needs 2times gain (+6dB) to drive a 4Vac sensitivity power amp.
12times (+22dB) gain pre-amp is adding too much noise that has to be thrown away again at the vol pot.

Short cables between the vol pot and the power amp are limited by the capacitance of the cables.
I'd suggest less than 1m of typical 50pF/m cable is probably OK in addition to the other capacitances that exist.
You could look at a unity gain Power Buffer.
That would remove the amplifier gain from your total system gain.
Typical power amplifiers have a gain somewhere in the range +20dB to +35dB
Omitting that amp gain gets the system gain down to levels that can drive a very efficient speaker.

I think PASS has shown a Power Buffer. An Italian amp designer (Cuffoli ?) has a Buffer. and ESP mentions it.
Hi Guys!

Many thanks for the help!

Andrea Ciuffoli (thanks)

I went to his site and found his paper on his
"Power Follower 99" a solid state class A amp 30wps at 6ohms

He also says

"The presented topology, has NO gain (actually it looses something) and should be preceded by voltage gain stage, with an output swing not lower than 10Vrms and Rout < 1000Ohm. "

This sounds perfect! Can this zero gain amp be implemented DIY or is it too esoteric and should be purchased from a person like Ciuffoli?
Ciuffoli Power Follower(Buffer) has very few parts and could be done as hardwired.

A gain of 1times is equal to +0dB = -0dB
A gain of 0.99times is equal to -0.09dB

If you preceded the Power Follower with a very low gain pre-amp stage like switchable between +0dB, +6dB and +12dB, you could probably drive the output to maximum power with most of your sources.
Possibly BA 3 pre, or Juma, or many others.
Hi Andrew

Many thanks for your help. I understand conceptually what I need - I just lack any technical ability to build. Are there people that will build these custom for a fee?

If you had a generous budget which of the pre-amps you listed would you recommend (given a zero gain amp)? (should it be dual mono?)

On the zero gain DIY amp - is there a Honey Badger equivalent of a zero gain design out there that can be used?

Thanks again
Now the issue is where to attenuate. Doing it at the power amp imput is NOT the ideal solution, the best option would be to adjust the gain of the amp itself (like Ayre and ASR do).
The ideal gain structure would be
2-2,5V from digital source -> input buffer -> adjustable VAS from like 0,1x to 20x (to provide for less efficient speakers) -> unity current gain.

And after 6 years nobody made a DIY circuit that realizes this... :eek:
Addenda: if we do NOT employ GNFB, stability shouldn't be an issue.
i have a question

Dear Bro, I used tda 7294 ic shan board for channels. What is the appropriate input volt used for best result in this board? How many ampere is should be use for best result? pls reply. Is it convenient volt transformer 24 0 24 10amp, 28 0 28 8amp, 30 0 30 8amp? pls plss reply. I wait for your response. plss comment
It is nice to know that something I vaguely think from DIY experience is right, from words of the person with knowledge. Each gain stage adds its character and distortion and I feel relieved that I am using a sensitive speaker that does not need high gain.

Thank you for the article!