807pp suggestions and guidance please

[aardvarkash10]

Miles - your help in here!

I recently took delivery of an octet of 807s. All are NOS JAN RCA. Enough acronyms.

I have lusted after a mid-output tube amp for a while. I've built a trio of 12ax7/EL84 SE units, a 6j5 buffer, and a 6CM5 SE spud. I'm now overconfident and recognisably underskilled.

Did I mention I intend to wind the OPTs myself?

OK, so here are the issues for me. I greatly admire Miles' 807pp design and understand the concepts if not the detail of the schematic. However, my leaning is toward some thing a little less complex - preferably three stages. I'm thinking a pentode preamp, cathodyne triode splitter/driver to the 807 outputs.

Source is entirely digital - no vinyl to have to equalise.

I have a clutch of 6j5 triodes, 6sj7 pentodes, a mix of the usual console-derived novals. THe EL36's don't belong in this project, but if I let you know I have a few, who knows what may be suggested - voltage regulator pass tubes perhaps?

Think monoblocks folks, and separate power supplies to each.

Music is bass heavy, a mix of rocks, dance/house, punk, moving toward blues jazz and light opera. OK, so I'm eclectic.

I figure AB is a given and looking for around 35W - the spec sheet says 500V anode, 300V screens, -27V grid, 8k a-a load.

SUggestions? Sanity Checks? Christmas greetings?

[Shoog]

Consider building something along the lines of this.

The positive rail can be derived from a valve full wave rectifier, the negative rail can be derived from a silicon full wave rectifier. Both rails can come off the same 320-0-320V transformer.

Expect output to be in the region of 25W of class A Pentode output. This will sound better than almost anything else you could build. The design is a simplification of Gary Pimms Tabor amplifier.

The only serious issue to face is that the CCS's under the outputs can see in excess of 300V over them, which requires a special kind of design. I used TV deflection transistors configured with a darlington driver - consider this a single darlington pair, this is then configured as a ring of two with a high hfe transistor in control. Once I worked this out they have presented no issues over many on off cycles.


Shoog

[Klimon]

As a side-note on 'simpler' 807 PP designs, what would Shoog or anyone else for that matter think of this circuit: http://tinpan.fortunecity.com/saints...mer/807-pp.gif

To a beginner like myself it looks charming in it's simplicity, uses some rarely seen feedback methods... and at last provides a way to make use of all those Hammond opt secondaries

Cheers - Simon

PS I see the link doesn't work; googling fi primer 807 should get you there

[Shoog]

Looks like a nice neat little design. Should sound good.
Still, for little more effort I think there are technical reasons why my design will sound better.


Shoog

[Klimon]

Thanks for the seal of approval; the Fi Primer amp has several practical feats that make it tempting for me to build - though I don't doubt your design or Pimm's probably has the edge.

Simon

[aardvarkash10]

807 cathodes in the design that klimon linked? THey appear to be connected to the opt secondaries and nothng else! How can this be???

[aardvarkash10]

I had left the transformer-splitter option off my list but had been considering it. It seems to give a number of advantages, not the least being it splits the signal at the very front end and leaves the active units to amplify and drive - the things they do best. I'm not interested in inter-stage transformer splitters - I just can't justify the iron to get the required bandwidth....

[Shoog]

In my design I use a mains toroidal at the front end to do the splitting. In order to preserve bandwidth it is essential to step down the signal. Since the 6AU6 has plenty of gain, I use a 4:1+1. This is a nice cheap option that has worked well with no issues to speak of.

Shoog

[Miles Prower]

Originally posted by Klimon
As a side-note on 'simpler' 807 PP designs, what would Shoog or anyone else for that matter think of this circuit: http://tinpan.fortunecity.com/saints...mer/807-pp.gif

In a word: YUCK!.

In the first place, a 6SL7 doesn't have the moxie to properly drive the grids of an 807. Though not exactly a difficult load, the grid circuit has more Cgk + Cmiller + Cstray than a 6SL7 can handle. The 6SL7 was intended for useage as a high gain small signal triode. As such, it has eeeeeeeeeeeeeenormous static and dynamic plate resistances. It'll work at sub-milliamp plate currents. This making possible the use of huge plate loads without requiring insane rail voltages. What the 6SL7 likes to see is a Hi-Z, Lo-C load, and an 807 ain't it. Furthermore, when the finals clip, the 6SL7 will simply roll over and die. Even before clipping, you're gonna have a slew problem at the high end. Fixed bias and RC coupling is bad news. An overdrive transient that turns on the GK parasitic diode will accumulate an additional negative voltage at the grid (how RF amps and oscillators derive most, if not all, of their operating bias). That will put the finals into a more nonlinear part of the plate characteristic. Even if you don't hear an actual clip, you will notice the resulting sonic degradation.

Secondly, a triode IS NOT a constant current source. If you're going to use an LTP/Schmidt/differential phase splitter, you need a CCS in the tail. That means a small signal pentode, or better still, cascoded BJTs.

If you're gonna go to the expense and bother of building your own, then build something worth your time and effort. This PoS isn't it.

If you want "simple" than go with Shoog's design. 6AU6s look pretty good so far as linearity goes. Looks like he's got a good Q-Point. Since it's a differential, that'll improve the THD situation. The active tail loads on the 807s will mean no Class AB, crank it to 11, but you can live without those extra few watts.

The only suggestions I would make is to stiffen up that screen supply with a voltage divider instead of a series dropping resistor, and adding the LR plate stoppers. 807s tend to want to make RF.

[Shoog]

Thanks for the excellent analysis Miles, Many things I missed.

With regard to stiffening the 807 screen, after consulting with Gary Pimm on it, I dropped the screen filtering altogether. This is because with a smoothed screen supply the amp tends to hum more as it amplifies the supply ripple at the plates. I use minimal supply filtering overall (8uf 5H choke and then 50uf). I reference the screens to the raw +B. As we have discussed before the plate to plate feedback linearises everything a treat and I have an almost hum free amp to boot. Output impedance is sub triode strapped.
Also on the screen supplies to the 6AU6 the pot has been replaced with two 10K resistor, one for each screen. Again gary advised that this would lower distortion by introducing a bit of degradation. In the Tabor original Gary uses the pot to balance current in the finals. I deliberately wanted to avoid that and that is why I put the CCS's in the tails of each of the 807's so that they wouldn't drift at all. I am certain there is a slight sonic penalty, but sufficiently small to be outweighed by the advantage of not having the chore of bias balancing. This is critical in my design because I have used 120+120V : 6V mains toroidals as outputs. These can tolerate no more than a couple of milliamps of imbalance. Response is good down to 20hz and present to 10hz.

Shoog