6-70 Push Pull Tube Monoblocks

ÖogliiBlocks 6-70 Triple Chassis Tube Monoblock Amplifiers

The ÖogliiBlocks started as a project to satisfy my desire to build my own reference level mid-power tube amplifiers after a few years of noodling around with tube preamp designs.

The original design (which I call Mark I, Mod 0) was directly lifted from Maple Tree Audio Design's Super Stealth + push pull EL34 monoblocks. These are simple little amps that use 2 EL34s per channel with 6SJ7 input stage and 6SL7 driver/phase splitter. I chose this as my jumping off point for a few reasons: I wanted a push pull design using EL34s in ultralinear, I didn't want to build a bog standard Mullard 5-20 clone, and I wanted something that used oddball input tubes rather than the boring usual suspects. Octals are always fun and have their own character so the fact the SS+ used the unusual 6SJ7/6SL7 combo I decided this was the way to go, if only to buck the norm. Bonus was that using unusual tubes means getting good NOS examples is way cheaper than if you use "popular" options. Oh, and Maple Tree is a Canadian company so a wee bit of nationalism is at play.

There was one key requirement the SS+ did not fulfill: I wanted at least 50W/ch to have a decent amount of power to drive my Magnepan LRSs. But this isn't that hard to satisfy - just parallel more tubes for more power! Ok so it's a little more complicated than that but that's essentially how this started.

Mod 0 was simply a Super Stealth with 4 output tubes per channel. No other mods except adjusting a couple of grid leak resistors, increasing the coupling cap size for better bass response, and running the input tubes off DC heaters. And running a 3.3K primary Edcor output transformer vs the 6.6K spec'd for a pair of EL34s. This worked well, sounded good, and made about 50W/ch at clipping, a great start.

Mod 1 was the first evolution. The topology was left the same but AB-Qi autobias modules replaced the cathode bias Mod 0, and the power supply was redesigned to use better Edcor power transformers with bias voltage taps. Hammond transformers SUCK. Their power transformers are buzzy and sag more than they should, while their OPTs are only suitable for guitar amps. Edcor is basically the same price and leagues ahead in all respects. Sorry not sorry.

These changes netted a lot more refinement to the sound and a bit more power, now around 60w/ch at clipping.

Next came Mod 2, which is where this became my own design rather than a hot rodded clone of the SS+. The input 6SJ7 pentode was rewired into a triode to reduce open loop gain to around 40db. Negative feedback was then reduced from the -30db (!) of the SS+ to -14db, netting 29db closed loop gain. Unnecessary high frequency filters were removed as the design was perfectly stable without them. The phase split circuit was altered and I added AC balance controls. Driver stage voltages were increased to increase headroom. Bias and plate load values in the input/driver stage were optimized for minimum distortion once I got my hands on a HP 8903B audio analyzer. Final feedback tuning was set to stabilize a 10KHZ square wave at 10W with a slight overshoot and no ringing.

My final tube selection is Sylvania 6SJ7GT glass envelope JAN VT116A inputs, RCA 5691 drivers, and Gold Lion KT77 outputs. 2nd place choices are RCA 5693 inputs, Tung Sol 6SU7GTY drivers, and Svetlana EL34s.

Power now peaked at 70W/ch in ultralinear and 35W/ch in triode. I ended up removing the triode option altogether as I never used it and I prefer the sound in UL. 1% distortion is reached at 65W, with 0.1-0.2% (THD+N) pretty much anywhere below that. Frequency response at 1W into 8 ohms is within 0.5db between 20hz and 42khz, with -3db point at 63khz.

To finalize my work and clean up the prototypical bodges I rebuilt everything into what I call the Mark III revision. These use Hirata Tango FW-100-5 output transformers, and replaced the first cap in the power supply with a monstrous 200uf/800V TDK film capacitor followed by a 1.5H Triad C17X choke. Otherwise everything is pretty much as before, but cleaned up and wired in far tidier fashion.

With the Tango transformers distortion is lower but so is peak power - they are 5K primary impedance, so you trade off a little Class AB peak power in favour of more Class A operating region and lower distortion below the peak (lowest I measured was 0.05% THD+N in the 5-10W range). Estimated Class A is around 10-15W of that, about double what you'd get with the 3.3K OPTs. Frequency response is identical. 

OPT impedance matching is a bit of a lost art. If in doubt and by default you should choose the value that matches the anode to anode impedance of your output tubes (RL - A to A on the tube datasheet for push pull circuits. That's how I originally came to 3.3K for parallel push-pull EL34s). But this value usually errs on the side of low impedance to boost peak power capability - the number that looks best on paper. Increasing the impedance over this will reduce the peak but give you far more Class A operation in a Class AB circuit. Also you can play with the secondary tap matching to further adjust - put an 8 ohm load on the 4 ohm tap and you effectively have doubled  the primary impedance the tubes see. You should usually do this when hooking up speakers if possible, use a lower impedance tap than your speakers are nominally rated at. Flip side is you shouldn't use a higher impedance tap than your speakers (you'll make more peak power but with more distortion, and the tubes are run harder with signal swings exceeding the dissipation limits).

I wouldn't say the Tangos are dramatically better than the Edcors. That's not a slight against Tango, that's a credit to how good Edcor is for the money. I wouldn't bother spending the extra for the fancy Japanese iron, the Edcors perform superbly and sound damn near as good.

Specs:

Loosely based on Mullard "5-20" circuit 

Push-Pull Ultralinear Class AB1

1x 6SJ7 voltage gain, 1x 6SL7 driver/phase splitter, 4x EL34/KT77 output (per channel)

29db gain

-14db NFB

0.7V input sensitivity

65w/ch into 8 ohms at 1% distortion THD+N, both channels driven

Frequency response -1db 20hz-42khz. -3db 20hz-63khz.

Signal to noise ratio 1W output 85db