Specific grid stopper question

[optimationman]

Not being a tube amp expert, I turn to the more qualified DIY members. I have read several threads concerning grid stopper resistor types and values needed. There is a divided camp as to the need and the valued necessary if used. Here's what I would like to have help with. My amps sound "bright" to me(my opinion, It's not the mutiple sources or speaker systems, cable etc).

The driver/gain stage is a ECL86, with a 68k(IRC MF I think) grid stopper on the triode section, directly connected to the pentoide section. This is then capacitor coupled to the output stage consisting of EL34's operating in PP @ 645v/30ma with 100 ohm(MF) G1 stoppers. G2 is connected to 390v through a choke,in series through 4 tungsten lights, then to a 39 ohm(CC) resistor. G3 is grounded.
The amps are Lab Instrumentation refugees rescued from sure destruction(scrap heap). The spec sheet claims full power up to 40khz and strong operation up to 100khz.. I have checked many other circuits and 1k grid stoppers on G1 is common. There is lots of info concerning types of resistors, circuit configurations, inline components that make brighter sound, I would like to limit this discussion to real life grid stopper experience, and leave the engineering for another day.

My question;
Would raising the value of the stoppers tone down the amps? Some would say it is counter productive, but would it tone them down and what value would be a good starting point.
I am interested in the sound I enjoy, not technical correctness!

 

[JonSnell Electronic]

"Grid stoppers", used on the control grid only usually 10k Carbon Film, stop the valve from taking off at RF and don't usually affect lower frequency audio from say 50kHZ and lower.
The purpose of the grid feed resistors on the screen grid is to reduce the current and therefore the voltage to that grid. We don't want the screen grids to have a higher voltage than the anodes or the grids will glow like a light bulb and the valve may be damaged in time.
Increasing the stoppers to say 1M would reduce the current available to drive the grids and possibly produce a slightly non linear effect.

If the amplifier is too bright, add some negative feedback at higher frequencies.

 

[SY]

The most important thing about grid stoppers is location- the body should be as close to the grid pin as possible. 100R is on the low side- I'd use at least 1k-4k7. For pentodes, it can be quit a bit higher. Resistor type is secondary- the traditional choice in power amps is carbon comp.

 

[optimationman]

Thanks for the quick info.
-----------------------------

grid feed resistors on the screen grid is to reduce the current and therefore the voltage to that grid.

I presume that is what the function of those bulbs must be, in an excessive power condition, Thanks for that.
---------------------
If the amplifier is too bright, add some negative feedback at higher frequencies.

There is global neg feedback and additional negative feedback from the EL34 plates, according to the manual. This later feedback takes care of the oscillation above 100khz, again, according to the manual. Is there a way to increase it?

-------------------------
100R is on the low side- I'd use at least 1k-4k7. For pentodes, it can be quit a bit higher.

Access to the stoppers is easy enough as they are located upon the circuit board that the sockets are mounted. Do you think it would tone the amp down? What about the used of a single stopper on the driver pentoide(one) versus 8 on the el34s?

 

[DF96]

If an amplifier has to work at higher frequencies than the usual audio range then the grid stoppers may be smaller than usual. The risk of parasitic oscillation then has to be tamed by careful layout, such as following RF-style standards. That may be why the EL34 stoppers are only 100R. Alternatively, it may be that the output stage operates in Class AB2 so grid current requires that stoppers be small.

A 68k stopper (if true) is being used as a tone control, or perhaps a dominate pole loop stabiliser, in conjunction with Miller capacitance.

Some people claim that stoppers should be carbon composition. I assume they are worried about the inductance of a spiral cut resistor like carbon film or metal film. Irrelevant, in my opinion. Normal resistors are mainly resistive up to around UHF frequencies, so fine for suppressing valve parasitics. The inductance does no harm; it could even be an advantage.

Screen grid resistors reduce the screen voltage, and so the current too. Low values (1K or less) are really just stoppers.

If an amplifier is 'bright', then it may be faulty (loop compensation capacitor gone open circuit?) or it may just need tone controls.

 

[SY]

A 68k stopper (if true) is being used as a tone control, or perhaps a dominate pole loop stabiliser, in conjunction with Miller capacitance.

If a pentode, not likely so. Large values like that are often used to reduce recovery time after overload- with the low Cin of a pentode, the pole is generally several octaves above the output tranformer's. The large stopper reduces the maximum current drawn from the phase splitter or driver at clipping, reducing blocking. The Red Light District uses this old trick (it has the output stage driven by a cathodyne) and the bench results confirm the efficacy.

 

[optimationman]

Lets assume the amps were designed for high freq. The 68k stopper is there and on the ecc83 section of the ecl86. I thought that this was a normal grid stopper value for the triode(?).

The amps were built in 1963, and I have update many parts. No updating had been done in the compensation loop if that pertains to the plate feedback circuit. It consists of a parallel 470k/22pf, after which a series resistor set of 10k and 50k to ground with a 1uf film/foil between them to ground. Then the circuit continues to the phase inverter/driver input thru and additional parallel 470k/6pf. I will check the film/foil cap for fault, the rest looks fine.

 

[SY]

For a triode with gain, 68k is indeed a tone control as Dave speculated.

 

[optimationman]

The manual states the the circuit for the ecl86 is arranged so that the gain of the ecl86 "is very nearly equal to the amplification factor of the triode section of the tube" so yes that is were the gain is located. Ok, what kind of tone control? Would there be a better value? Could you speculate as to the effect this would have on the amps response?

 

[SY]

It would roll off the treble. If you calculate the input capacitance of the triode (which will be roughly Cgk + gain*Cag), then you can see that this capacitance forms a first order lowpass filter with the sum of the grid stopper resistance and the source impedance of the driving stage.

If that's greek to you, then get yourself a copy of "Valve Amplifiers" and start reading.

 

[optimationman]

I am confused. Yes it is greek to me. I will try to find out some additional information concerning this issue.
I have done additional research concerning this particular amp. It is an early version of the Wolcott amp, which I believe is well respected. That version also has this 68k resistor on the triode section. Would this be used to reduce or as you say roll of the treble, that is ultrasonic freq not desired?

 

[optimationman]

Checked the film/foil caps on both sides, both amps. Appear ok. I am still wondering about the value of grid resistor that might tone down the amps to my liking. At least a starting point. Will this have the desired result?

 

[DF96]

My guess is that the 68k is being used to set a dominant pole to stabilise the feedback loop. Doing it this way (as in a typical opamp) may be better than just adding a capacitive load to an anode somewhere. In this case you can't just change it; small changes won't affect the frequency response very much and big changes could destabilise the loop.

The best way to implement tone controls is to add tone controls to a line stage. Second best would be to find a triode grid which is before the loop and increase the grid stopper there.

 

[optimationman]

Thanks for the insight;
small changes won't affect the frequency response very much and big changes could destabilise the loop.

Well, I guess I will leave this circuit as it will not take me to positive territory in my effort to tone the amps down.

-----------------
I would like to avoid the tone control route. I like the source directly connected to the amp, without any preamp. I guess the amp has a built in passive pre in the form of a 250k pot.
-----------------

Second best would be to find a triode grid which is before the loop and increase the grid stopper there.

The first stage of the amp is what is describe as a cathode follower(ecc83) with an emitter follower(set of 2n3053, one for each triode section). There is no stopper on this input tube. Is this the point where a stopper may help? The anoide is connected to a dedicated regulated PS. Could you suggest a grid stopper value to start with for experimentation if this would be appropriate?

 

[optimationman]

Let me think about this a second. The input to the grid of the ecc83 tube has a 250k pot connected? Adding a stopped would just add to this? Sorry, I am not sure where this is going. I answered my own question, here is not the place for a grid stopper.

I guess I am back to the EL34 stoppers. At 100 ohms, these could be increased as recommended to at least 1-5k with no appreciable audio freq change. Is there agreement on this?

 

[DF96]

Adding a stopper to a cathode follower will not change the frequency response as a CF has no Miller effect.

The EL34 stoppers may be low for two reasons I outlined in an earlier post: good ultrasonic performance, Class AB2 operation. If so, increasing them could either restrict the frequency response (which may be what you want) or could introduce distortion on louder signals (not what you want?). Before you change a circuit, first you have to understand it.

 

[Miles Prower]

My question;
Would raising the value of the stoppers tone down the amps? Some would say it is counter productive, but would it tone them down and what value would be a good starting point.
I am interested in the sound I enjoy, not technical correctness!

Not if these are inside a NFB loop. Any changes made to the open loop characteristic are neutralized by the NFB. That's one of the reasons we use it: to make performance more independent of open loop characteristics.

If that amp sounds excessively bright, it may be due to the characteristics of the finals. Some audio finals make more higher order harmonics than others, and so sound excessively bright or aggressive. Not much you can do about that without an extensive redesign of the whole thing.

 

[dgta]

Yes, you need to get back to basics. What exactly is "bright" and what is the problem (if any)? What are you trying to "tone down"?

Measure the frequency response and see if there is an anomaly there and what exactly it is. If the frequency response is flat within reason, look at the distortion spectrum, like Miles said it could be odd harmonics you're subjectively reacting to. That's a totally different issue.

 

[optimationman]

I have done many 20-20k sweeps while looking on the scope across the speaker terminals. Amplitude is flat.

When I say bright, it is to my taste levels. The dyna have a soft dome tweeter which should give a somewhat less aggressive sound, but for me, still more than I like. I have first reflection absorber(not diffusion) walls and ceiling, full carpet etc. I have changed sources, but still to bright.

I changed the 100 ohm EL34 grid stoppers to 100 ohm CF which has taking me the direction I want to go. A little less sparkle, but still with plenty of detail.

One additional component issue. The output transformer has 1000pf/330 ohm 5w strung across the primary. There are five terminal, 1-5. There are sets of cap/resistors between 1/2 and 4/5. Why?

Will try a CF in place of the MF on the gain stage ECL86 soon.

 

[DF96]

Those are snubbers across the OPT primary. They stop an OPT HF resonance from turning the output stage into a power oscillator. A common feature of UL.