Subtle circuit difference, massive difference in tone

[strelok]

So I finally got around to ordering a different OT for my 6v6 Marshall. Its the one I built a while back with a push/pull pot for 1959/2203 modes. I'd replaced the Hammond 125E in my 6v6 express with one from Allen Amps that was a bit beefier and liked the results. Tamed the highs a bit and added some low end and fatness. Since I'd used the 125E in this amp as well I figured it might benefit from a larger OT too. The amp had always suffered from way too much high end, it was ok with a les paul and the cheesy hammer shop guitar, but god help you if you plugged in a strat, even with humbuckers. Which really interfered with my ambitions for it, basically I wanted something like my '68 Metro 100 watt, but much smaller and more manageable, to do the EVH thing with, among other things. So I was somewhat disappointed when the transformer didn't help with the brightness much.

I started going over the amp with a fine tooth comb, trying to figure out what the hell was different. Triple checked all the pot values, wiring etc. Nothing. Hooked up the scope, dummy load, and signal generator to see if there were any HF oscillations, again nothing. Not that I really expecting anything, this has been by far the most stable/quiet amp I've ever built. Despite its excessive brightness, you can dime all the controls even in 2203 mode and its nothing but tube hiss, no buzz or hum whatsoever.

So what the hell was going on? I opened up my metro last night and the only thing I could see that was different was the 470k/500pf R/C network was after the volume control like in a 1959, and not before like in a 2203. I should mention that I converted the metro to a 2203 preamp after hearing rockstah's experiments with it. Apparently I'd left the R/C network in the original 1959 configuration so I wouldn't have to pull up the board or something. Anyways this amp had it before, mostly out of convenience for the layout and not having room for it on the pot/switch as you can see in the pic.

I'd never really given it much thought, one way or the other, but once I did it hit me like a truck almost instantly. The kind of thing were you just held some incorrect assumptions about something taken for granted as being trivially simple or inconsequential. Years go by, new knowledge is acquired and an issue surrounding those old assumptions comes up. Then you have one of those moments of just god dammit I feel like such an idiot for not thinking this through sooner I want to bang my head against the wall.

Basically I'd assumed that in 1959 configuration the RC network was acting as a high pass filter, same as it would be in 2203 configuration. Which doesn't make any sense as there's no resitive load on the other side of it. I knew the resistor in combination with the miller capacitance acted as low pass, but I just never put the two together. What you end up getting is compensated LP network, kinda similar to what is used in scope probes to help flatten out the response. You can see what I mean in the following links:

1959:
http://www.falstad.com/afilter/circuitj ... 25+1+-1%0A

2203:
http://www.falstad.com/afilter/circuitj ... 25+1+-1%0A

You'll have to adjust the dB scale as it doesn't save in the link. You can click and drag away the 500p bypass cap in the 1959 or the 470k resistor in the 2203 version and see the effect that is has. In both cases the attenuation maxes out at about 3db or so, but thats a difference of ~7db for some high frequencies between the two. May not sound like a lot, but this early in the signal path it makes a massive difference. So just as a test I removed the 500p cap, amp landed squarely on its feet. No more excessive brightness, has the growl and lower mids you'd expect from a Marshall, and just generally sounds awesome and much more like my Metro. Will eventually move the 470k/500p after the pot, might be too much low end a result, will just have to see, but I think have a new favorite amp.

Lesson learned. I'm a dumbass, but hopefully if you made it this far in my ramblings it might save you some time.

[Reeltarded]

Rockin! Dood, just say something. I am happy that you are happy and both of those amps are seriously two favorites. It looks wonderful inside!

[Colossal]

Strelok,

I always enjoy your posts, especially those about your single ended trainwreck adventures. Great looking layout.

[strelok]

Thanks!! I'm quite pleased with it now. For a while there I thought it was just gonna be an amp that didn't work well with my strat, not so anymore.

It's nice to hear you enjoy my ramblings, I keep trying to get myself to write shorter posts and just get to the point, yet it always ends in a wall of text That SE express was a fun experiment, though I ultimately decided to change it over to 6v6's in P-P. For whatever reason the pre-amp sounded fine but when you started pushing the output section into distortion by cranking the master it just got so harsh and bitey. Lack of a PI probably didn't help either. Still doesn't sound as good as a real express but its pretty close.

Next step will be to get the RC network moved to after the pot, should be fun trying to finagle it in between the wiper and the switch. May have to add a terminal strip for it. Might experiment with the first coupling cap (.022u now) and the outputs (.047u). Then I might stick the 500p compensation cap back in for the 2203 mode (currently all the bright caps have been removed).

Also for whatever reason this amp has a TON of gain. Way more than a normal 1959. I'm sure the reason just yet. 3-4 on the gain is ACDC, 7 is well into VH and 10 is just overly compressed madness with a tweed deluxe esque sag. I wonder if the PI is driving the output tubes too hard? It does get some weird waveforms on the output grids when pushed really hard, though 6v6's should require about the same drive as an EL34. Might try adjusting the PS dropping resistor for the PI. If that doesn't help I'm also thinking of adding a PPIMV, if only to save my ears just a bit. VVR gets too muddy/compressed by the time any serious volume reduction occurs.

Here's a pic of the full chassis before the VVR got put in:

[Reeltarded]

though I ultimately decided to change it over to 6v6's in P-P. For whatever reason the pre-amp sounded fine but when you started pushing the output section into distortion by cranking the master it just got so harsh and bitey. Lack of a PI probably didn't help either. Still doesn't sound as good as a real express but its pretty close.

I'm not going to say that SE amps make better bass rigs and sound weird when driven, in general.

I have said that before!

(hides from the rock that Dave is about to throw)

[Colossal]

(hides from the rock that Dave is about to throw)

...you watch your ass...

harumph.jpg

[strelok]

Lol.

I think the main problem with mine was the plate voltage was too high and the OT from Edcor was probably intended more for Hifi than anything, but it was the only one I could find that wasn't super expensive. Not a lot of people building SE guitar amps in the 25 watt range.

SE bass amp might be cool for certain sounds. Obviously would only be good for practice/studio use but might give a unique sound. Would need a meaty transformer though. If you're looking to build one I just so happen to have an OT laying around.

[Reeltarded]

LOL this take-down writes itself..

People dreaming up 10,000 watt SE amps with substation sized transformer arrays might include tag member 3610..

https://ampgarage.com/forum/memberlist. ... ile&u=3610

I really should contribute more to the development of a fine and decent culture of amp building. (but joking around is really fun.. too.. it's really fun too)

[Colossal]

[strelok]

You can get tubes for that here:

https://www.greenstoneusa.com/tetrodes-vacuum-tubes/

[Reeltarded]

lol

#EL84s #hardrowhoe

[strelok]

You'd only need about 2,000 of them

At which point you wouldn't even the need the OT because the output impedance would be low enough to drive a speaker directly lmao.

Edit: I couldn't help but run some numbers

First the easy one, filament current: 1520A, which gives a power draw of 9576 watts.

Plate voltage is probably the only remotely reasonable figure in this hypothetical monstrosity, 350v would do fine. Since the output impedance is so low you wouldn't be loosing any voltage swing to ratio reduction.

Plate current on the other hand would be about 50 amps, assuming a 4 ohm load.

Screen grid current we'll just say for the sake of simplicity about 10ma, or 20A total.

Now the fun part. The driver circuit. An EL84 has a Cg-k of about 10pf, and I think a gain of about 20. That gives a miller capacitance of 200pf. That gives a total capacitance of .4uF. If we use a standard 5.6k grid resistor that's a parallel resistance of 2.8 ohms. It typically takes about 15 volts to run the gamut on an el84. If we were to try to put that voltage onto that capacitance the instantaneous inrush current would be just over 5 amps. Now we won't be switching obviously but we also would need to take into account the additional capacitance added by all the wire runs that would connect to all these tubes. A few dozen more EL84's or a dozen or two KT88's in a parallel follower configuration would probably do.

Now if we laid all those tubes out on a square grid, each one is about an inch wide, so lets say 2 inches between mounting centers, we'd need a chassis that was roughly 8ft x 8ft to fit just those 2000 tubes on them.

You'd also need 60-80 250w speakers to handle the power, the ultimate series parallel challenge to get the impedance right.

Only way something like this would get built is if you locked Colin Furze and Photoinduction in the same room with a pound of meth.

[Tony Bones]

I can't speak to the fantasy 10000W SE amp, but a friend of mine did once build a 50W per channel stereo SE amp with Eimac 304TL output tubes. I don't recall how many individual transformers there were, but I do remember eztimating that it weighed 400 pounds.

[maint_tech]

"You can get tubes for that here:

https://www.greenstoneusa.com/tetrodes-vacuum-tubes/"

LOL, I used to work on radio gear that used a pair of 4CX1500's to get a clean 1KW of RF out. I always wondered what it would take to make a PA out of that box! Beryllium oxide insulators, nasty stuff.

[strelok]

I love the look of those big power triodes. They sound really nice as well. 400 pounds sounds about right lol. I'd love to build one around something like that one day, but in P-P. I think the SET guys are nuts lol. I've never seen anyone go to such lengths just to get a few watts of power. We had one come into the shop a while back, it was a stereo 22 WPC unit. Or so they claimed, best I could get it to do was about 15 watts before clipping. It sounded really quite good, except it just didn't have the power to make bass at a reasonable volume, so you need some really efficient speakers, the ones in the shop are 97dB, which is pretty high, but you really need something in the 103+ range. It weighed probably 150-200 pounds. For 15 watts. Transformers were massive. But what will really get you are the chokes. Since you don't have the common mode hum cancellation of a push pull output, you really need a heavily filtered power supply. I found that out with the SE Express. It took two 50uF caps and a 3H 200ma choke as a pi filter to get the hum down to acceptable levels. This thing had a choke the size of the PT. Which itself was about 8-10 inches high and maybe 6-8 inches in diameter (each one was encased in a potted chrome plated cylinder).

Triodes are amazing for Hifi. They can have linearity that's hard to beat. But you don't need to build a several hundred pound behemoth to get exceptional sound out of them. Now if only we could convince the SET community of that, we'd be in the money.