V1 Input: Grid Leak and Grid Stopper

[pdf64]

...for things like blocking distortion, does the placement of the grid stopper make a difference?

If mounted away from the socket, there would then be parasitic capacitance acting on the wire between the (likely high value) grid stopper and tube socket terminal; they may act as a low pass filter, (possibly to a noticeable degree), or reduce the margin of stability (possibly - but unlikely - causing oscillation), or a bit of both

[Mike Sulzer]

That's not what I would consider significant particularly when the resonance is only affecting harmonics; you get as much variation simply by putting your speaker in a different position from yesterday or using somebody else cable.
And when you remember to include Miller capacitance (typically 180pF) the difference shrinks even more. No one can hear that; no one who spends more time playing than naval gazing, anyhow!
[ATTACH=CONFIG]44081[/ATTACH]

I like to think of this change as reducing the bandwidth to about 71% of its original value while changing which harmonics get most emphasized by the 5 db peak. When does reducing the bandwidth by about 30% right in the sensitive range of human hearing not matter? Remember, it is the harmonics that determine the sound, and even the low E open string (82.4Hz) has audible harmonics in the 3 KHz range. Of course it matters! It would matter more if it were a factor of 2 rather than sqrt(2), and if you moved the resonance up to about 5KHz a 30% change would be less noticeable since you would be at the top of the guitar system range (and maybe at the top of the range of some of us with aging ears exposed to a lifetime of noise).

[Merlinb]

I like to think of this change as reducing the bandwidth to about 71% of its original value while changing which harmonics get most emphasized by the 5 db peak. When does reducing the bandwidth by about 30% right in the sensitive range of human hearing not matter? Remember, it is the harmonics that determine the sound, and even the low E open string (82.4Hz) has audible harmonics in the 3 KHz range.

The ear is logarithmic so 30% off the top end as actually not very much (30% off the bottom end would be a different story). Like I say, you get more audible variations simply by turning your head or moving the speaker.
The low E String does not produce audible harmonics anywhere near 3k. Beyond the 5-7th harmonic they're decayed so much as to be essentially drowned by the distortion generated by the amp and speaker.
I'm not sure why you're making a big deal out of such a trivial thing. It would take someone less than a minute to try with and without the cap to decide for themselves whether it matters to them; they don't need an internet expert to help them decide this one.

[bob p]

what are you guys using for your technical drawings? i'm seeing some kick-ass layout and schematic diagrams for tube circuits.

[Mike Sulzer]

The ear is logarithmic so 30% off the top end as actually not very much (30% off the bottom end would be a different story). Like I say, you get more audible variations simply by turning your head or moving the speaker.
The low E String does not produce audible harmonics anywhere near 3k. Beyond the 5-7th harmonic they're decayed so much as to be essentially drowned by the distortion generated by the amp and speaker.
I'm not sure why you're making a big deal out of such a trivial thing. It would take someone less than a minute to try with and without the cap to decide for themselves whether it matters to them; they don't need an internet expert to help them decide this one.

http://music-electronics-forum.com/t26893/ That is a link to a discussion in which the difference in how a single coil pickup and a pair of closely spaced coils (humbucker) sample harmonics in the KHz range differently. It is investigated theoretically (like the Tillman computations), with spectral analysis, and sound files where the differences can be heard. These harmonics can be measured and heard, and it is clear that this explains one difference in how the wound guitar strings sound with humbuckers and single coils pickups. (These harmonics are out of the range of the system for the plain strings; there are also damping differences between wound and plain strings.)

If you want to say what can and cannot be heard, you have to do the work.

[SoulFetish]

Well this thread just got more interesting..

I think both Mike and Merlin make interesting points. Although, isn't the level "significance" base on a completely subjective value system in this case?
I agree with mike in that small changes in the bandwidth in which we are most sensitive can be quite dramatic in how we perceive them. Yet, I also agree with Merlin's point (if I understand post #33 correctly) that the level of importance in these changes is dependent on the listener in many cases. I would definitely agree that the position of the listener relative to the front center of the speaker is probably much more dramatic than some of the changes we're talking about. I've made the mistake of dialing in the tone controls while looking directly down at a combo amp only to go right back and change it once I walked out and heard it from the front.
I'm curious as to how some of the things I have found to be significant in shaping the sound of my guitar, would looking in simulations. For instance, I have 3 single coils in strat using 250K pots (they can almost be thought of as fix values, as I only ever roll them off if I'm playing an overly bright amplifier). I've always thought that any single capacitor value shared by the bridge and neck pickup makes one or the other sounds terrible, and was a cost/time saving decision by Fender. I found that using a .047µF in the neck tone circuit to be significant and much improved. I use .027µF for the Bridge, which is admittedly a more subtle change over .022µF. I think that equally significant is the height and tilt of the pickups in relation to the strings, which usually require a small adjustment as the strings age over some time. I'm lucky, in that my fingers don't sweat so I can get away with changing my strings maybe once a year or so if they don't break. I'm definitely curious as to how these would work out in a simulation, but I've never measured these things because I've never felt the need to convince anyone of their significance

[Gingertube]

Ceriatone Latouts are just copies (warts and all) from other designs. I doubt that any real technical nouce has been applied.

The grid stop resistor is included to supress any tendency to parasitic oscillation, to do its job the resistor body MUST be hard up against the tube socket pin.
The minimum recommended grids stop resistor for any triode is 8/gm, for a 12AX7 that means 8/1.6mA/V = 5K Ohms. As a tube ages its gm drops so a slightly higher value is recommended.
Next - the major source of noise in a hi gain amp is thermaly introduced noise from that gridstop resistor. As noise is proportional to resistance then you have a competing requirement, you want the grid stop as low in value as possible.
So Merlin suggests 10K. It will contribute x6.8 less noise than the "traditional" 68K.

Miller Capacitance at the grid will be approximately 100pF [(1 + gain) x Cga] where Cag is typically 1.7pF, add CgK and a few pF for wiring strays and we get that approx 100pF value.
The old 68K + 100pF will give you a high frequency roll off at 23kHz which provides some RF immunity.
To maintain that HF roll off with a 10K grid stop then Merlin suggests adding an additional cap from grid to ground or grid to cathode. You would need to get 680pF total for the original roll off so add a 470pF or 560pF.

The series Cap is essential for some old amps which use grid leak biasing but you would never use that in a higher gain amp anyway (grid leak biasing is VERY noisy). The DC blocking function it provides maybe handy if you are using some old pedals with leaky output caps.

Just rearrange those components in your build to have the gridstop at the tube pin socket. The grid leak (1M) can go the input socket side of the gridstop.

Cheers,
Ian

[FourT6and2]

Ceriatone Latouts are just copies (warts and all) from other designs. I doubt that any real technical nouce has been applied.

The grid stop resistor is included to supress any tendency to parasitic oscillation, to do its job the resistor body MUST be hard up against the tube socket pin.
The minimum recommended grids stop resistor for any triode is 8/gm, for a 12AX7 that means 8/1.6mA/V = 5K Ohms. As a tube ages its gm drops so a slightly higher value is recommended.
Next - the major source of noise in a hi gain amp is thermaly introduced noise from that gridstop resistor. As noise is proportional to resistance then you have a competing requirement, you want the grid stop as low in value as possible.
So Merlin suggests 10K. It will contribute x6.8 less noise than the "traditional" 68K.

Miller Capacitance at the grid will be approximately 100pF [(1 + gain) x Cga] where Cag is typically 1.7pF, add CgK and a few pF for wiring strays and we get that approx 100pF value.
The old 68K + 100pF will give you a high frequency roll off at 23kHz which provides some RF immunity.
To maintain that HF roll off with a 10K grid stop then Merlin suggests adding an additional cap from grid to ground or grid to cathode. You would need to get 680pF total for the original roll off so add a 470pF or 560pF.

The series Cap is essential for some old amps which use grid leak biasing but you would never use that in a higher gain amp anyway (grid leak biasing is VERY noisy). The DC blocking function it provides maybe handy if you are using some old pedals with leaky output caps.

Just rearrange those components in your build to have the gridstop at the tube pin socket. The grid leak (1M) can go the input socket side of the gridstop.

Cheers,
Ian

Did you see the reply I posted back on the first page? Seems like the setup in this amp is this way because the plate voltage on the stage is so low. Otherwise you get scratchy volume pot on the guitar. If I move to a more typical topology without raising the plate voltage, will I run into noise?

[Dave H]

I'm definitely curious as to how these would work out in a simulation, but I've never measured these things because I've never felt the need to convince anyone of their significance

I think it is as Mike said -

"A guitar tone control is almost purely resistive over most of the range; that is, it lowers the Q, but does not affect the resonant frequency until near the bottom"

i.e. There won't be any tone difference between 0.022u and 0.047u until the tone pot is turned well down. The plots below are for 0.022u and 0.047u with the tone set at 250k and 25k.

250k


25k

[Gingertube]

Yes - I saw the bit about preventing scratchiness from the guitar volume control. That actually makes sense (because there will be DC across the guitar volume pot and each time you move it it feeds a new DC level back into the amp) but its a bit of treating the symptom and not the disease.

I would be modifying that input stage anyway.
470K anode resistor and 820 Ohms cathode bias resistor are just not going to work well together - 820 Ohms allows too much anode current and hence too low an Anode Voltage and also excessive grid current. Drop the tube current a bit to get the anode voltage higher.
Try 2K2 or 3K for teh cathode bias resistor instead of the 820 Ohms.

Compare that to the Yeti input stage I traced here:
http://music-electronics-forum.com/t43046/
390K anode load resistor with 3K cathode bias resistor. Note I drew the voltages on the schematic trace, the anode voltage was significantly higher and there is no series cap. That makes more sense.
V1 should always be about max clean gain anyway - to get the signal level up enough to overdrive following stages.
To minimise bias disturbance from high grid current you can also lower the grid leak resistor from 1M to 470K but I really think that you should be tryong to change the operating point to get away from that high grid current area. That is get the anode voltage up, just from my own experience I find that +160V anode voltage is minimum practiable and I usually aim for +180 to +200V for high gain and 2/3 rail for super clean.

Cheers,
Ian

[FourT6and2]

Yes - I saw the bit about preventing scratchiness from the guitar volume control. That actually makes sense (because there will be DC across the guitar volume pot and each time you move it it feeds a new DC level back into the amp) but its a bit of treating the symptom and not the disease.

I would be modifying that input stage anyway.
470K anode resistor and 820 Ohms cathode bias resistor are just not going to work well together - 820 Ohms allows too much anode current and hence too low an Anode Voltage and also excessive grid current. Drop the tube current a bit to get the anode voltage higher.
Try 2K2 or 3K for teh cathode bias resistor instead of the 820 Ohms.

Compare that to the Yeti input stage I traced here:
http://music-electronics-forum.com/t43046/
390K anode load resistor with 3K cathode bias resistor. Note I drew the voltages on the schematic trace, the anode voltage was significantly higher and there is no series cap. That makes more sense.
V1 should always be about max clean gain anyway - to get the signal level up enough to overdrive following stages.
To minimise bias disturbance from high grid current you can also lower the grid leak resistor from 1M to 470K but I really think that you should be tryong to change the operating point to get away from that high grid current area. That is get the anode voltage up, just from my own experience I find that +160V anode voltage is minimum practiable and I usually aim for +180 to +200V for high gain and 2/3 rail for super clean.

Cheers,
Ian

Yeah, for my build I'm using 390K V1a plate resistor and 2K7/.68uF cathode bypass on V1a and V1b. Staying with 820R/.68uF on V2. I've also changed the 0.1uF PI output coupling caps to 0.047uF and the PI grid leak/bias splitters to 150K (from 110K). I'm also using a PT with much higher voltages. My B+ will probably be around 480v. So even with a 390K my plate voltage on V1 should be higher.

All the Yeti amps I've seen have the same input topography as the Chupacabras. If the one you mapped out was built by someone as a kit, and not by Ceriatone, that would explain why it was missing the cap...

As far as the 820/470K not working well together, I can say I rather like how the amp sounds like that. I have a Chupacabra 50 already and I dig it. Here's a clip: https://youtu.be/rKW5f5THPQY?t=30s

[SoulFetish]

I think it is as Mike said -

"A guitar tone control is almost purely resistive over most of the range; that is, it lowers the Q, but does not affect the resonant frequency until near the bottom"

i.e. There won't be any tone difference between 0.022u and 0.047u until the tone pot is turned well down. The plots below are for 0.022u and 0.047u with the tone set at 250k and 25k.

250k
[ATTACH=CONFIG]44124[/ATTACH]

25k
[ATTACH=CONFIG]44125[/ATTACH]

I LIKE this. Perhaps, I'm full of shit....
But, I like how John Linsley Hood puts it: "Listening trials are difficult to set up, and hard to purge of any inadvertent bias in the way equipment is chosen or the tests are carried out. Human beings are also notoriously prone to believe that their preconceived views will prove to be correct. The tests must therefore be carried out on a double blind basis, when neither the listening panel, nor the persons selecting one or other of the items under test, know what piece of hardware is being tested."
Could you show a schematic of the circuit as it was simulated. To be honest, this plot is really surprising to me. But, I have to consider that after rewiring my guitar a hundred times over the last 15 year, I'm could be totally wrong in how I remember it or perceive it.

Now, having said all that, I'm not totally convinced yet. This bares more investigating before I'm ready to concede. But I'm willing