Thursday, June 14, 2012
Ultralinear Guitar Amp (1): Tubes - 6L6GC/6BG6
I am putting together a guitar amp, and hoping for about 40 watts RMS.
I chose 6L6GC as the robust tube of choice (I hate EL34s, especially newer unreliable crap), and am aware of bad tube-eating designs of the past.
I am limited by what I have on hand:
(1) a good Ultralinear 100w Output tranny (5K Zprimary). This will allow me to expand power later if I can beef up the powersupply (see below).
(2) a nice 300-0-300 HV power tranny, with a 5 and 6.3v winding. This is underpowered, but I hope to squeeze more out of it by leaving the 5v winding unused, and not overloading the 6.3v winding. I tested these windings with loads, and found:
600v c.t. @ 160 mA (from a load test)
5v rect. heater @ 2A
6.3v main heater @ 4A (before sag)
This doesn't match any current Hammond, but looks like a mix. Its a no-name brand from the 60s or 70s (surplus), possibly JAN.
When I use solid-state diodes, I can wire it full-wave with a 10H choke and a stacked cap giving 100uF (4x 100uF series/parallel 450v), and it looks like I can count on 520 volts under a real load (projecting 4x 6L6 at 40 mA each, push/pull).
(3) 520-540 volts is just a little too much to put on the tubes (even good 6L6GCs), so I'm also opting for self-biasing, so I can drop about 40-50 volts across the cathode resistors). This leaves me at the edge of the design-center max for the 6L6GC, but the low idle current (40mA/tube) will help prevent too much dissipation.
(4) I intend actually to use 6BG6s, which can take 700 volts anyway (topcap plates) without worries. But in a pinch I should be able to put in a set of 2 or 4 6L6GCs made to take 500 volts. I'm going to rig something special to protect the plate caps and people's fingers, and to hold the tubes in regardless of orientation. Also I will probably float the mounting so that the tubes have some shock-insulation, to improve microphonics and protect the tubes from banging.
(5) Ultralinear Topology: I intend to take advantage also of the Ultralinear tappings, so that I can avoid building two extra powersupplies (screen and bias). This should allow lower idle current without undue distortion.
(6) Four Output Tubes: The idea here is to lighten the current load on each tube, and also drop the internal output impedance of the power stage.
(7) The splitter/ driver circuit is still up in the air. I haven't chosen a topology for this, and wish to hear what others think from their personal experience (comparisons) especially in regard to keeping the output stage balanced as tubes age, with low maintenance, and a good match to provide current to drive the output stage grids, in case they draw current.
(8) Looking at the basic loadline, I find there is a small 'danger' zone at the lower-right, where the voltage across the 6L6s could climb above 500 volts (the design max) as the tube shuts off, if the input signal were driven hard enough to push the stage into class B temporarily... I think I can get a good 40 watts without too high an input voltage, but in any case the 6BG6s will be able to take the overdrive/classB. Its only the 6L6CG scenario that worries me. Any comments would be appreciated. But with self-bias, there is much more headroom, even with the 6L6s.
(9) The 'guitar-amp' sound will be mostly derived from manipulating the input section, so I would really like to hear what people have to say about good preamp designs re: guitar sustain and harmonic content.
(10) I'd like to have a simple but effective Parametric EQ, something like the EQ in a BOSS MT-2 pedal (bass, mid/Q, treble) only modded for a better choice of frequency ranges. I could use a hand on that, again, I want to use tubes for this (maybe an 12AT7 or two).
Obviously I can do this in stages, as I get the results I need at each step, and move on to the next addition/mod.
Right now, I have on the table the following issues,
and could use some advice:
(a) Grid Stoppers: (1.5k - 5.6k ohm range?) I'd like advice on these, I'm not worried about losing a bit of signal, but would rather have stability and immunity from RF...
(b) Screen Limiters: (450 - 680 ojm range?) I read a discussion somewhere on here about guitar amps, and the problems with EL34s etc., along with typical values and preferred. But any input or advice would be good, especially as apparently screen current will significantly affect sound as well as tube-life.
(c) Grid-Leak / Input: (100k - 470k? ) with self-bias, I should be able to tolerate values a little higher, but I am more inclined to want the self-bias on the grids to work well, rather than 'maximize' input Z here. Any advice?
(d) Cathode Bias: (500 - 1.5k ohm ?) I found experimentally that to get a good 48-50 volts on the grids and keep the current down to near 40 mA per tube, I needed at least 1.3k cathodes (with Svetlana 6L6GCs).
But the issue doesn't stop there, because I could use one large common cathode resistor per tube-pair (e.g., 650 ohm). However, Hafler found that a 10 ohm common resistor was apparently the right amount of 'common resistor' to minimize intermodulation distortion (presumably the rest of each cathode resistance was separate). Any leads on this would be welcome.
Also whether or not to bypass, as it seems I would trade off lower output impedance (which may be unnecessary with 4 tubes) against the sound degradation that say 47uF electrolytics would introduce....please tell me what you think!
Other issues I need to settle are:
(e) Input isolating caps for input each side (.22uF? 600 v?) of power stage. Since its not a hi-fi, perhaps polystyrene or something not too expensive could go here?
(f) RF inhibitors and stabilizer additions:
Here I'm thinking of an RC series between screen and plate on each side, an RC series across the primary of the output xformer, and also a 100 ohm resistor across the Secondary in case a speaker is unplugged or goes open-circuit.
Any other safety features would be also welcome.
I am going to place an internal fuse between the HV B+ and the output xformer primary c.t., but what else can I do to protect tubes and xformer from spikes due to failure or user-error?