New Marshall JTM-45 clone build making noise on high gain.

[Alex R]

Building from scratch, making mistakes and working out why, is a great way to learn. Like Steve implies, probably that's how the classic guitar amps were designed anyway. Oops, that doesn't work, let's try it this way.

[Steve Conner]

"Ooh, that's sounding great now, quick, ship it before it explodes."

A time honoured design flow that's still used by some boutique builders even today. Not to mention Mullard with the 5-20.

[PfeifferElectronics]

Thanks guys, I feel better now. I am learning a lot from this build and from this forum. I really appreciate everyone's help. I couldn't do it without you people.

Latest news on this build: I replaced the two 470 ohm grid resistors with two 3K resistors. This changed the bias circuit so that I can't get down to 30 - 40 mV on pin 8 anymore. So I increased the bias resistor to 110k (was 78k). Now my bias voltage is too low at 12mV. I'm going to split the difference and put something around 90k in there and see what happens.

[Alex R]

hey pop the PI out when you are setting the bias. If the thing is oscillating the current draw will be all over the place.

[PfeifferElectronics]

not sure if I misunderstood, but replacing the two 470 ohm resistors that go from the power tube grids to the choke has only made things much worse. I replaced them with 3K and now it oscillates all the time.

[Alex R]

Ah no those are the screen grid resistors, pin 4, they are fine at 470R. The suggestion concerned resistors on the grid that the bias and signal are applied to, pin 5.

[MWJB]

Steve O Connor wrote,"For me this is the real voodoo of tube amps. A good designer makes a circuit that is tolerant of layout. The performance of the circuit is determined by the components you can see, and the "hidden circuit" only has minor effects." Well, I have to disgree with this, the circuit doesn't become a tangible device until it is physically laid out & wired up. Even when using a proven circuit & board layout, I have seen guys snatch defeat from the jaws of victory, simply by not paying attention to wire routing...by that I literally mean an amp that has the same circuit as the one next to it, same parts, but one works & the other...well, it makes a "noise" when a signal is fed in, but you couldn't gig with it - which by my standards identifies it as malfunctioning. Just look at amps like the RI Marshal 1962 & the PRRI, both share the circuit with their (usually) functioning ancestors (which admittedly can still be improved upon, Fender & Marshall certainly worked to standard thet they could "get away" with, but at the same time, they had a dedicated test shop to revive basket cases...a facility that the novice home builder does not have), but both suffer with sonic artefacts related to the physical layout of that circuit, rather than component values.

The guy who designs the circuit isn't usually the guy who builds it, the designer designs something that should work, the guy who builds it can make the difference between functioning & malfunctioning. If you need oversized grid stoppers, who is at fault....the designer who specified a value that doesn't eliminate oscillations in one amp that he has never even seen (but does in countless amps with the same circuit), or the guy who put an amp together only to find that his specific amp is unstable with the specified value? The circuit should function well enough with 1.5K or 2.2K grid stoppers (if it was perfectly laid out it can get by with none), so to a point I agree that trying some reasonable sized grid stoppers is the next logical step, rather than rewiring the amp, but if you need values of 10K or over it's time to reasses the layout. In the OPs case...I wish him luck, but given the big picture, you could be trying to steal tiny margins here & there to no avail for weeks, it's probably best to bite the bullet and put in a few hours work revising the layout now & hopefully have a working amp by next week.

[PfeifferElectronics]

FIXED!

Wanted to give everyone an update. The amp is working great now! I installed 5.6k grid stoppers on the power tubes and shortened the leads on them from the circuit board.

I'm running plates at 494V with the bias set at 33mV. Does this sound ok for KT66s? I think I might need to lower that some.

Thanks for everyone's help. This build was meant to be a learning experience and it definately was. Next build I'll be paying a lot more attention to layout. And now I know what a grid stopper resistor is and what it is used for. Cool!

[pdf64]

Rather, or as well as, being due to less than ideal lead dress, bear in mind greekie's experience reported in post #10, to the effect that modern KT66 are more prone to oscillation. Pete.