The Twins!

...Yes, plural.

Okay, so about a month ago I managed to pick up these two, erm, beauties:

Someone order 235W?

On top we have a Fender PA100 from ~1975, and below a Twin Reverb from 1981.

Yes, that's right, the dreaded 135W ultralinear Twin! I was really happy to get this one, even though it's in somewhat rough shape cosmetically, because it's a vintage amp that doesn't have much value (certainly not compared to the Twins even 5 years older) so without worrying about ruining a museum-grade amp I can tweak it into being a great player. On the whole I'm not thrilled with calling an amp that was made the same year that I was born "vintage" but that's how it goes.

The PA100 is also exciting - these were sold as PA heads, of course, but in reality they're Twin Reverbs, minus the vibrato, plus four preamps. All four channels are bridgeable, and while the input impedance and tone stacks aren't too great for guitar currently, those are both easy fixes. I haven't done too much with this one yet as I've been spending my time on the UL Twin.

So, without further ado, here's what I've been working with:

Yikes.

Yeah, it's a real mess in there. At least one other person has been in this amp sometime in the past 33 years - there's the occasional carbon film resistor, and an odd red wire used for the master volume ground connection. The caps, of course, were original:

emphasis on "were"

So, here's where the fun begins, and I haven't even gotten to the schematic yet. The plate voltage is hovering around 510V-520V depending on the day. That's the plate voltage loaded. We're over 530V unloaded. That's a big stupid problem with this doghouse layout! I don't have quite enough room to put caps in series at each node (though I'm going to be looking for small 350V or 300V/47uF caps to try this in the future) so I'm stuck using the only 600V rated electrolytics around. I got these ones from Weber, though I'm sure they're the same as the ones you can get from RadioDaze or other vendors, just with Weber's wrapping on them. Pretty sure the manufacturer is MIEC as they're the only company I can find putting out caps with that voltage rating.

Now, I don't care the slightest bit about mojo, and everything I can find about these caps indicates that they're reliable (I haven't had any problems so far) but down the line I'm going to either look into making a new board with totem-pole caps for each node, or knocking the unloaded plate voltage down under 500. Probably the latter.

Alright, so what's in this beast?

Nice try, CBS.

Okay, so in the interest of brevity I'm not going to go through this part-by-part and list what is designed "poorly" or even "not Blackface-acceptable," and instead I'm just going to post the current as-modded schematic:

Not that much red for a huge improvement in tone.

Now.... now, we bust out the list. Bullets!

• Horrible pull boost and all associated wiring removed. As far as the components for the pull boost, I removed the 12k resistor and left the other components on the board. 
• Screen grid stoppers raised to 2K/5W in hopes of extending tube life. Also introduces distortion a little earlier. 
• Bridged the 2 channels, so now reverb and trem are on both. 
• Removed the bright cap across the master volume. Seriously? How could that have ever been a good idea?
• Removed the death cap. 
• Lowered the output coupling caps to 22n. With the 68k grid leak, this raises the -3dB point to a hypothetically uncomfortable 70 Hz (105 Hz with the 47k) but the bass response is still quite bone-crushing, which is a refreshing change from the overwhelming mud it was with the 100n couplers. 
• Changed the pull-boost MV to a global NFB disable switch. Well, it's normally open, so when you pull the switch it engages the global NFB loop, which maintains the "pull to sound worse" functionality of the original. And sure, a 1M feedback resistor isn't "completely" open-loop, but it's awfully close and this way the switch doesn't pop.

It's pretty close to done, I gotta say. There are some other general maintenance things I've done - replaced the fiber shoulder and flat washers on the 8 ohm jack, and temporarily installed a pair of Peavey Scorpions because the Utahs it came with sound significantly worse.

Oh hey, I should mention the output jacks. This is weird.

• Plugged into the main jack, and nothing in the extension jack, we get a 4 ohm out.
• You can then plug an additional 4 ohm cab into the extension jack. The switching jack puts the external cab and the internal speakers in series, and connects them to the 8 ohm tap on the OT. 
• You can instead pull the main jack, and plug an 8 ohm cab into the extension jack. 

Me, I think they overthought it, I probably would've just gone with a 4 ohm jack and an 8 ohm jack. Decently clever I suppose.

I should mention the poor, abused screen grids in this amp. "Ultralinear" is a configuration for beam tetrodes and pentodes where the screens are connected to taps on the output transformer instead of having their own DC supply. From a sonic perspective, this is a form of local NFB, and now the screen dissipation can also contribute to the output of the amp. You combine this with the much-higher plate voltage, and that's where CBS came up with the 135W claim. (Really ~100W before any clipping, at least according to my scope. Still, this is more than the ~80W clean you really get from a typical "100W" 4x6L6GC amp.)

There are rumors that this amp was designed around a special 6L6GC that CBS got Sylvania to make with significantly more rugged screen grids, but I haven't been able to confirm that yet.

The problem with an ultralinear output section in this amp is the ridiculously high screen grid voltage - roughly 1-2V over the plate voltage, so roughly 60-70V over the 6L6GC's spec for a maximum screen grid voltage. This by itself isn't a huge problem - dissipation kills tubes (rather, electrodes inside them), not voltage. The problem is that the screen grid dissipation in the original schematic gets nuts pretty fast. One person on MEF measured the screen dissipation at roughly 20W per tube (I'll link when I find the post again) which is just a weee bit over the 5W the screen grid is rated at.

In a typical (pentode-connected) output section, the trick is to use a larger screen grid resistor - increased current draw drops the screen grid voltage enough that the resulting dissipation is safe. This is known as "sliding screen" operation, and is a pretty awesome trick. Typically 1K is large enough to accomplish this. The poster at MEF indicated that he needed to use 4K7 screen stoppers to get the dissipation under safe limits, and with stoppers that large the output and tone suffered.

So, we're trying 2K for now. Tubes seem happy enough, the output is plenty, and the added distortion (rather, distortion occurring at lower volume) is welcome.

With regards to "blackfacing" and the future mods I have planned:

The "bias balance" system is poor, and I'm going to change it to a "bias balance and adjust" two-pot system. Considering how beat up these tubes get, I'm not particularly interested in running them any hotter right now. I forget the numbers, but they're on the 50-55% range of plate dissipation right now, and the output looks fine on the scope so this is kind of a low priority.

The phase inverter typically gets the brunt of the "blackfacing" parts substitution. Honestly I see nothing wrong with the phase inverter as it is - it's better designed for a 12AT7, delivers some massive output voltage swing, and is well balanced. I may check the biasing, but those lower plate resistors (47k vs the 82k/100k "traditional" values) are really helpful for driving those 68k grid leaks - which are helpful for keeping the tubes from going into thermal runaway, and minimizing blocking distortion. I could write an essay here, but this is one section where CBS actually got something right.

The tremolo is ticking, so I'll try some of the typical fixes for that.

The reverb driver is biased somewhat uncomfortably warm, so I may try the blackface spec for that.

And someday, someday, I'll look into getting the B+ down. The amp sounds great now, so I have no interest in converting it to a typical pentode-connected output, and I haven't been eating tubes so this really isn't a high priority.

There are some cosmetic issues to address (Scrubbing Bubbles works great on Tolex!) but a full restoration would be ridiculously cost-prohibitive, so for the forseeable future this monster is going to be a little rough around the edges but louder and more reliable than the average apocalypse.

So for now, this one's good. I'll be starting in on the PA100 soon, and I've got to get going on that Voxy build. And updating more frequently. Bah!