|mikepukmel||4/15/2018 7:27 AM|
|JTM45 cascading gain mod|
There was a great, and short, thread that is pretty old, about this topic, Tom had posted some instructions on a hard wired version of this mod. http://music-electronics-forum.com/t2205/ A couple of open questions came up when I started going through the details. Please excuse my poor drawing (done in MS Paint and not quite complete)
On new builds the suggestions are fairly unanimous to use shielded wiring from the inputs to V1, and move the 68k grid stops right to the tube socket pin.
This cascading "mod" is to route the wire coming off the normal channel volume pot, after the 270k resistor, back to pin 7 (input) on the bright channel, and disconnect the wire coming from the input jack.
If I moved the grid stops to the pin on the tube, then should this wire go through the grid stop, or around it?
Is it OK only to disconnect one half the input for this tube, or do I need to shut off both inputs? I.e. would something horrible happen if the cascaded gain was turned "on", and someone put a jumper from the live input going to pin 7 to the other input jack pair going to pin 1?
I recall lots of bits and comments, e.g. Blankenship, talking about people who added gain stages and oscillated their amps to death (its was a comment he made about all repairs he does bringing the amps back to stock/factory). Since higher gain, is more likely to lead to unwanted oscillations, what can I do to reduce that chance? V2 has no grid stop on its input, would adding one there help to reduce this chance? I did plan on putting 5.6 or 6.8K grid stops on the output tubes, where the original-original schematic had none.
Mike The Tinkerer
|Chuck H||4/16/2018 5:06 PM|
|Oscillations happen for all sorts of reasons. There is no blanket fix or additional single component to insure against it. Sometimes you need a grid stop resistor, sometimes not and sometimes something more drastic like a treble bleed or local feedback circuit. More to the point...|
Will you be employing a master volume? Without one you will not be able to enjoy the extra gain unless the amp is at full volume. That being the case, the JTM45 can produce a lot of overdrive at that level anyway. So the channel stack is pretty much for uber gain tones at full volume without a master volume included. And further more...
If you ARE going to include a master volume, which type you choose could (should) affect your choices when building or planning the rest of the circuit. So that's information we don't have yet.
There are a lot of options for a channel stack mod. I personally don't like clunky mods that require unplugging from one jack, into another and then you have to NOT use anything else in another jack or you have to plug a dummy jack or open cord into yet another jack or two to make adjustments. I personally prefer a switch. And the best way to do that without adding a toggle to the front of the amp is with a push/pull pot. When the pot is pulled the channels are stacked. Volume 1 controls gain and the channel 2 jacks are out of the circuit. When the pot is in you have the stock amp. You can jumper channels and do all the normal Marshall stuff with it. Simple. JM2C
|mikepukmel||4/17/2018 5:23 AM|
Someone said to put the master vol before the phase inverter like the 2204 does, but that will make the tone stack less effective. Wish I had the technical expertise to quantify that. I only know how much my guitar signal changes character when going from half to full volume, so I understand descriptively some of the effect.
There's a wild master volume mod, with a dual gang pot, really complicated thing, I think that is to combat some of the tone stack issues.
I'll look into push pull pots, thanks for the tip.
The master vol mod, one I found [ATTACH]48517[/ATTACH] I understand the ganged pot, each pot between the output of the phase inverter, and the output tube, but not quite sure about how the bias is affected. They have 2 2.2M resistors between the bias feed and the output to pin 5's, but the Marshall schematic has 2 220K resistors there. There's some kind of voltage divider thing going on that I don't understand. Anyway, I see lots of photos of builds on various amp build sites, with a ganged pot on the back of the chassis, that looks like this mod.
Ive written too much already, but is this the 'best' mv mod for this amp?
|Dave H||4/17/2018 8:42 AM|
The bias isn't affected. There's no voltage divider. The 2M2 resistors are there for insurance against loss of bias when the pot wiper contact fails open circuit.
|mikepukmel||4/17/2018 5:32 PM|
|FOund one thread here: |
Rob has a nice schematic showing this part:
Eek, this can't be right, can it? With the pot wipers all the way 'down', the bias will go up (larger negative value), The bias will change as the pots are rotated. I looked through the various arguments for/against the few mods, and they all seem to have a significant drawback. If the pot blows up, then the bias goes through only a 1.8 meg resistor (instead of stock 220k). Won't that still be a small enough bias (small negative value) to cause the tubes to redplate?
|Chuck H||4/17/2018 6:56 PM|
EDIT: I thought to take another few minutes and explain, further, why the PPIMV in question works. It's simple really. The AC signal can find it's way to ground via the bias filter caps. The DC bias is not loaded by the capacitors because capacitors do not allow DC to pass. It is not strictly resistances that are affecting circuit parameters. There is also division, current and impedance to consider. Keep these things in mind and you may actually admire the circuit rather than questioning it.
|mikepukmel||4/17/2018 8:15 PM|
|How cool is that, thanks Chuck! |
Dayyumm, I got the circuits 1 & 2 book, even in E format, just have to read it and do the exercises.
thanks for the explanation.
|mikepukmel||4/17/2018 8:27 PM|
|OK now I feel a little on the stupid side. Maybe more than a little. |
Bias is potential. No current. They put a potential between the grid and cathode. (It is grid to cathode?) and just keep the potential, with no DC current flowing. That small a DC potential can't force charge to jump from the grid to the cathode or vice versa. So, put whatever resistors in the path, the potential is still the same.
So, how did the amp builders back in the day, then, come up with the 2 x 220k resistors that go between the bias feed, and the junction where the signal goes to the output tube grids? Why not 20 meg or 1k? What is special about this value?
|Chuck H||4/18/2018 7:45 AM|
|The 220k grid load resistor value is a nod to the limitations of the tubes construction and what how it manages the chore of amplification. This is beyond my knowledge, but there is a limit to the allowable resistance between the tubes grid and ground. This figure is listed in the data sheets. So that sets the upper limit of the resistance. The lower limit simply affects output from the PI. Since we want a good, strong signal from the PI a value is typically chosen that is near the max resistance. Some designs do go lower. Going lower, obviously, reduces signal (but not bias voltage). This is why the PPIMV works. It reduces the load resistance of the signal while increasing the series resistance. A voltage divider.|
FWIW the TW type PPIMV can be arranged in the other direction. With the output of the PI on the wipers. I can't say how it changes the function with respect to tone, but circuit wise it looks like it may increase the amount of signal attenuation, at least partly, via local NFB more than the typical arrangement. And because it operates more as a blunt load to the signal, rather than a voltage divider, it probably also adjusts much faster at the bottom of it's range. But I know it works pretty well anyway because guys that have bias vary tremolo have done it this way. The PPIMV could interfere with the trem function otherwise. This arrangement also has the advantage of delivering the bias voltage via the outer lugs on the pots which is more reliable than the wiper. I would feel comfortable omitting the 1.8M or 2.2M safety resistors with this arrangement.
Also, even with the typical arrangement that has the bias voltage delivered via the pot wiper... Why do we use the safety resistors? You can observe countless amp designs that deliver bias voltage via a pot wiper in bias vary trem amps. We don't actually read much about failures in these circuits so why are we worried about it with the master volume? Well...
Because a master volume is likely to get a lot more use than a trem. And there are two pots in the circuit so the possibility of a failure is doubled. Not to mention the typically lower quality and greater fragility of modern panel pots compared to the old CTS products and such. FWIW the last time I built a bias vary trem into a design I included a safety resistor.
|pdf64||4/18/2018 12:43 PM|
See p1 of http://www.mif.pg.gda.pl/homepages/f...93/6/6L6GC.pdf
Under 'maximum ratings' there's 'grid-number 1 circuit resistance', noted as 0.1 Megohm in pentode fixed bias and 0.5 Megohm in pentode cathode bias.
Obviously most fixed bias 6L6 amps exceed this; tube life may be extended if we were more fussy about it.
The reason for the limit is that although it's effectively infinite resistance, g1 may pick up the odd electron resulting in a slight negative current via the 220k grid leaks and go slightly negative WRT the applied bias voltage. Or rather worse, it may release the odd electron (especially if the tube is dissipating heavily), resulting in a slight positive current and so go slightly positive with respect to the applied bias.
If the grid leak was 20M rather than 220k, that slight positive current would create 10x more positive voltage at g1 than would a 220k grid leak. The risk is that would cause the tube to run hotter, more electrons would be released from g1, creating more positive voltage and pretty soon there's a thermal runaway effect and the tube redplates.
If the grid leak was 1k, then no regular phase splitter could push much signal into such a low load. Rather mini power amp would be needed to drive 1k grid leaks.
Compared to the power triodes of previous era, the great benefit that the power pentodes we know and love brought was that they were fine with high value grid leaks and so were easy to drive and allowed them to produce high power output in class AB1, greatly simplifying amp design.
|mikepukmel||4/20/2018 7:58 PM|
|Wow, thanks, I will re-read this thread a few times, hope more sinks in.|
|Chuck H||4/21/2018 9:37 AM|
|Just for clarity, here's a redraw of the PPIMV with the pots flipped and the safety resistors removed. This is how I proposed it for use with bias wiggle trem some years ago and it's reported to work well.|
|mikepukmel||4/22/2018 7:42 AM|
would you also recommend use of the 1.8M safety resistors?
|Chuck H||4/22/2018 9:19 AM|
|My hunch is that the standard wiring for the Trainwreck/LarMar type PPIMV works better. I was just being clear about the circuit I described. So for the standard wiring, yes on the safety resistors, because the bias relies on the pot wipers it's better to include the safety resistors.|
The original grid load/bias feed resistors are 220k. The 250k pot acting as a grid load has the safety resistor parallel. So 250k parallel to 1.8M would be a grid load of 219.5k in the stock amp, master full up setting. I've commonly seen 2.2M used also. So that would be a grid load of 224.5k. Either value safety resistor gets the circuit within stock spec tolerance for the original grid load AND has the benefit of keeping the tubes from melting in the event of a wiper failure.