Sustain - Grid Blockin or Anode Clipping

[lowell]

What do you all find gives an amp good overdrive sustain? My thoughts are good impedance matching between stages is very important here. TUT says center biasing is relevant for clean sustain. I'm also thinking anode clipping might produce better sustain than grid blocking. Want to pick everyone's brains.

[Merlinb]

I'm also thinking anode clipping might produce better sustain than grid blocking. Want to pick everyone's brains.

Grid blockin (blocking distortion) is always bad, and anode clipping doesn't mean anything I'm aware of. Do you mean grid-current limiting versus cut-off clipping?

[lowell]

Anode clipping is when the "anode clips" and the available plate supply is exceeded. It's my understanding that grid blocking is not ALWAYS bad if that's something you're going for, especially in preamp stages. Correct me if I'm wrong there.

[TransLucid]

I believe the best approach is to use several stages in series, none of which are overdriven too much. Individual level controls can be used in front of each stage to allow proper drive. This way the distortion is mainly even-order, and builds up gradually.

[R.G.]

Anode clipping is when the "anode clips" and the available plate supply is exceeded. It's my understanding that grid blocking is not ALWAYS bad if that's something you're going for, especially in preamp stages. Correct me if I'm wrong there.

I think you're wrong here. At least as I understand the words, I think so.

Grid blocking means (to me, at least) when you put so much signal into a capacitor coupled grid that you pull the grid positive with respect to the cathode. This pulls electrons onto the grid side of the capacitor that can only escape through the grid leak resistor. In effect, the grid rectifies the positive going signal peaks into the coupling cap, shifting the grid more negative, and hence toward cutoff. It can bias the grid all the way off, and it only comes back when the signal gets smaller and the grid leak resistor dumps the excess off-bias.

But that's just my understanding.

[lowell]

So grid blocking and grid clipping are 2 different things? I thought when the input signal positive cycle exceeds the cathode voltage, the remainder is clipped. Is that grid blocking or just grid clipping?

[Old Tele man]

So grid blocking and grid clipping are 2 different things?

...yes, grid-blocking is like a 'transitory' excursion into -2 tube operation, where the control grid goes 'momentarily' positive on a signal peak, clips, and continues to clip less and less while the capacitor "charge" is being reduced (discharged). It's an RC-exponential "backing off" of control grid conduction due to bias shift.

...see page 7 for "grid clipping" and page 46 for "grid blocking" of Blencowe's book, and especially Fig. 2.12 (pg 46) for an excellent illustration of "what's happening" during grid blocking.

[R.G.]

So grid blocking and grid clipping are 2 different things? I thought when the input signal positive cycle exceeds the cathode voltage, the remainder is clipped. Is that grid blocking or just grid clipping?

Again, it depends on the definition.

Grids change from near-infinite input impedance when they're below the cathode voltage to a small impedance when they're above it. For the 12AX7, assuming I'm remembering correctly from the last time I looked it up, the grid drops somewhat nonlinearly to about 5K ohms as it passes the cathode voltage going positive.

A positive-going grid voltage is trying to make the tube conduct more. Since a tube cannot "saturate" in the sense that semiconductors can, a low impedance signal driving a grid positive can in fact continue increasing plate current for some voltages with the grid above the cathode. The thing that causes clipping on the positive-going grid signal for direct-coupled signals is the ratio of the source impedance to the grid impedance. A 12AX7 plate resistance is something around 67K as a rule of thumb (again if my memory is still OK) so a 67K impedance can drive a near-infinite impedance just fine. But it balks at trying to pull a 5K impedance any more positive.

So the grid gets clamped by the ratio of the impedances between the signal and it's own input impedance. For the nominal case of a 12AX7 plate driving a 12AX7 grid and ignoring other side issues, the grid simply can't be pulled very positive by a plate because the plate can't provide enough current to do it, and you get pretty flat clipping. Biasing a triode grid to the B+ supply was a trick used in some switching circuits; they'd use a high resistance between grid and B+. Any negative going signal pulled it out of "clamp", and the tube was said to be "running in clamp".

It's not well known, but a 12AX7 clips more softly if you drive its grid from a low impedance. This allows the signal to pull the grid mildly positive on positive peaks, and hence pull the plate further down (i.e., higher current). In tubes in good condition, there is usually enough cathode emission to supply this additional current. When you drive it with a big enough signal to shut the plate current off, it's truly off, and plate clipping on positive going plate/ negative going grid happens. The grid is always high impedance here, and the tube simply cuts off.

This is the basis of running output stages in Class AB2. A low impedance driver is direct-coupled to the output grids and can wring more current out of the output tubes by pulling the grids positive. You get more swing, and hence more power out of a pair of tubes and a given B+.

But this won't work if the grid is capacitively coupled. If it's capacitively coupled, the grid conducts, but the current has nowhere to go but to build up on the coupling cap, so it adds to the negative bias already there when the signal swings back down. The grid conducts one way at about 5K putting charge on the coupling cap, but the nominally 1M grid leak resistor is what pulls it off the cap. So too-big signals on a capacitively coupled tube grid wind up causing a bias shift into cutoff, which leaks away slowly. To me, this is grid blocking.

I guess I'd say that the DC coupled thing is grid clipping, as it does indeed clip the signal by suddenly changing the grid impedance; and the AC coupled thing results in grid blocking, as the grid bias is shifted and blocks signals from passing for a bit.

[Merlinb]

So grid blocking and grid clipping are 2 different things?

Just to chime in with the others, yes they are two different things.

Grid clipping (which I prefer to call 'grid-current limiting' is when the grid is driven positive and starts to conduct.

If, when the grid starts to conduct, the coupling cap is able to charge up, when the input signal swings negative again the voltage across coupling cap will "drag" the grid voltage really far negative, so the valve is stuck in cut-off; it is 'blocked'. This is blocking distortion, and is never good for audio because the valve won't amplify anything until the cap has discharged.

What you call "anode clipping" is what most people call 'cut-off'. It is when you drive the grid so far negative that the valve turns off.


One thing which promotes sustain rather than hard-fuzz clipping is avoiding the use of cathode-bypass capacitors. Unfortunately this means you sacrifice gain, so it's a bit of a balancing act.

[tubeswell]

From what I can gather, typical sustainy sounding hi-gain amps use several stages that are alternately biased so that one stage goes easily into grid current limiting and the next stage goes into plate clipping and the next goes back into grid current limiting and so on. Look at the SLO100 pre-amp circuit, or the PV5150.

(Incidentally - I modded my PV classic 30 to emulate the first four stages of a SLO100 about this time last year, and got something that did sort-of-sustainy. There some threads about it here.

http://music-electronics-forum.com/t9416/

http://music-electronics-forum.com/t10001/ )

[defaced]

The other trademark of some high gain amps is a cold clipping stage toward the end of the of the cascade. I've seen values as low as 9K (Partially bypassed, VHT) and as high as 39K (SLO, Recto), not that I've really looked all that hard. I'll be the first to say I don't fully understand why this is done, but I'm pretty sure it's to clip at the cathode.

[TransLucid]

Grid distortion

There's a nice discussion of this on the Aiken amps site...

[wizard333]

One thing which promotes sustain rather than hard-fuzz clipping is avoiding the use of cathode-bypass capacitors. Unfortunately this means you sacrifice gain, so it's a bit of a balancing act.

Why? If there is some technical reason for this, I can't think of what it is. Further, no matter what the theory is, I just do not hear it in practice.

There are a lot of things that affect sustain, primary among them 1) The guitar, 2) Midrange in your tone, which will cause the guitar to want to feedback musically at a lower volume (I dont mean screaming feedback; a lot of this can happen when the signal is 'clean').

Unless you really f up your circuit somehow in a way that causes the signal to clamp, whats going on in the amp is going to be very secondary to those 2things.

As far as CKs causing fiz vs not having them, or somehow hurting sustain, I just don't see why. If you're using a small CK its going to boost the high frequencies more which, if combined with high gain, can lead to more of a 'fiz' perception. They also increase overall gain, which can lead to the perception of fiz if you push it too far.

All things being equal though in terms of gain levels with vs without them....I just haven't heard that be the case.

[Merlinb]

Why? If there is some technical reason for this, I can't think of what it is. Further, no matter what the theory is, I just do not hear it in practice.

The reason is that an unbypassed cathode resistor introduces series current feedback which increases the input impedance of the valve.

In essence, if you do use a cathode bypass cap then when the grid starts to conduct it does so fairly suddenly, and the grid voltage will be clamped. You basically get hard clipping (not as hard with an SS diode of course, but hard in valve terms). This can be useful for a more gritty, aggressive overdrive tone, but too many stages like this can easily lead to a fuzzy tone, the kind I associate with modern metal players. I'm not a metal player, so this doesn't suit me.

If you remove the cap then when the grid starts to conduct the grid current has to flow in the cathode resistor, so a feedback voltage develops which opposes the grid current. The signal does not get clamped so soon; there is a more compressive transition into clipping, and a more rounded overdrive tone (which is the kind I like). Of course, you get less gain, so you typically have to cascade more of these stages together to get the same level of distortion as the above case.

I suppose you are right that the sustain can be the same either way, but the timbre of the overdrive is different is what I really meant.

[wizard333]

I guess the affect on timbre depends on how much gain you are pushing?

What I found in my designs is that when I don't have at least a small CK, the more even frequency response tends to want to fight the pick attack in a noticeable and unpleasant way. There is too much "thud" and not enough "zip" on the very front edge of the attack envelope.

Any suggestions on how that could be alliviated in a no CK design?

Still not sure I'm hearing any sustain difference in practice since I've experimented with the no-CK vs CK thing and nothing jumped out at me except the pick attack thud from the no-CK thing, but if I could get around that I might play with it more.

I'm not a high gain nor a metal tone player, but I do like my amps to be usable for everything from Larry Carlton type tones (more where my personal tastes lay) to Metallica type tones and everything in between, so anyone that plugs into it can get something out of it they like.

If I was making them just for myself, I'd do it different, but I'm shooting for versatility.