Voltage drive for 6550 (Mullard/HK Citation V circuit)

[GordonW]

OK. I'm having a wild-hair moment. Thinking of using a Citation V circuit, for 6550s.

The setup I'm thinking of, would run the 6550s in pentode mode, at 600v plate, 300v screen. This alignment seems to want a grid bias voltage of about -35v or so (to get around 100ma idle current for a pair of tubes).

The Citation V uses 6L6s (specifically, 7581s), at about 430v on plate and screen. Grid bias voltage is specified at -42v.

Now... AFAICT, grid bias voltage is directly related to necessary AC drive voltage on the grid. In essence, you need to be able to drive the grid to 0v in the "positive", and to double the bias voltage (negative) the other way. If you can do that with low distortion, you can basically run the output tubes "wide open".

Well... since the 6L6/7581s "need" +- 42v drive (-42v grid bias)... and the 6550s, at 600v/300v only "need" +-35v drive (-35v grid bias)... wouldn't the existing Citation V driver circuit be able to provide plenty of "swing", for 6550s used in this manner? IOW... the 6550 would actually have slightly MORE "effective gain" than the 6L6, in this alignment?

I can see how this might work- as pentodes do seen to usually exhibit more gain, with lower screen voltages...

I like this idea (if it will work!) for its sheer simplicity. A simple Mullard-circuit driving 6550s?! That could be sweet!

Regards,
Gordon.

 

[audiodon]

Using Citation iron?
What CIT V power supply changes would be required? Would the stock one have enough reserve for this hot-rodded circuit?
If not, how would you beef it up? You'd keep the solid-state rectification, correct?

 

[GordonW]

I'm using a pair of Bogen M60A chassis (PTs and OPTs). 4.2K primary outputs.

Decided that UL operation wasn't practical (the tertiary windings aren't coupled well enough to the primary)... so, pentode mode was the next choice. Hence, using the Citation V circuit, instead of the Deuce...

Regards,
Gordon.

 

[wildone]

I am sure you an work this out. I too am working off of the V for my latest project.

Oh Ya! You are the one to suggest this great setup. I have been working on getting this all worked out and this was just the ticket I was looking for.

Your the Man!
WO

 

[Kegger]

Yep Gordon that will work just fine an push the 6550s as far as they will go an then some.

Even the one Dave did for the HF60's was fine there, but he did need more input since he
went with a 20mu input tube, but still worked just fine there.

I've done similar with 30mu tubes for input and 20mu tubes for the LTP to drive from EL34
right on through KT88 in UL. (but I used less feedback an kept 1v input)

But yah that circuit is just fine driving 6550 or KT88..
(I still like the idea of replacing the 12BY7 with 6EJ7)

 

[dcgillespie]

I haven't got the circuit of a V in front of me, but the only thing I would mention, is to make sure the design of the V can drive an output stage that has no more than a 50K DC grid return resistance for each output tube. That was the crux of the problem I was dealing with in my modified HF-60 design, since the original Eico design employed 180K resistors in that location. Reducing those values required raising the value of the coupling caps, but even then, the original phase inverter stage became woefully inadequate to provide adequate low distortion drive. By raising the values of the resistors associated with the inverter stage, AND raising the effective B+ it operates from (now ~ 550 vdc with the addition of the negative supply for the tail), the stage was then able to deliver nearly 80 volts peak drive PER SIDE, with very low distortion into the 47K grid return resistors used in the modification for the 6550s. That result contributes greatly to the overall success of that modification.

From memory, doesn't the V use a 6CG7 in the inverter section? It's basically a lower dissipation 6SN7, so if that is the tube used, and it can't deliver the necessary drive into the appropriate load as designed, you could always use the value and voltage levels I used to achieve it. Or you could use the same values without the negative supply, but use a CC device instead to generate a similar current flow if you're not opposed to a little sand in the circuit.

Either way, I have no doubt that your wild hair moment will produce great results. Happy Thanksgiving to all!

Dave

PS: Keg is the absolute king of knowledge when it comes to the capabilities of the different triodes available for different applications. If he is recommending a different tube to use in either stage of the V's front end, I would seriously consider it!

 

[Kegger]

Thanks for the confidence Dave.

Yes you could triode strap the 6EJ7 an use as that for input, or leave it as pentode.

-----Obviously Gordon has more power supply voltage available----- But..

Dave here is the "modified" Citation V schematic we have been working on in another
thread, your comments are more then welcome, (yes for a 6550 want to lower grids).
(I don't think you need to go as low as 50k, I've used 100k grid res total on KT88's) ?

 

[BinaryMike]

I don't think you need to go as low as 50k, I've used 100k grid res total on KT88's ?

RCA's databook gives 50K for the upper limit with fixed bias on 6550s, but I've observed a similarly cavalier attitude in more than one relevant discussion. So I feel compelled to ask: Under what conditions is it acceptable to violate this specification? Is it ever acceptable in amp designs intended for production?

 

[Kegger]

I haven't worked with as many 6550 as I have KT88, and the KT88 spec is 100K.
(with many amps run well over 150k there without issue on old stock tubes)

My assumption for being able stretch that is 2 fold, first "quality" tube samples, an
second how far your punishing the tubes to where they attempt to draw some grid
current at a decent clip, so my thinking is if the KT88 can take it, and you are not
really running the crap out of the 6550 tubes that 100K should be acceptable.

But I'm all for hearing other info on the subject as well..

 

[GordonW]

I could always go with something like a 5687, if we REALLY wanted to use some low-resistance plate and cathode resistors in the driver/phase inverter stage.

I think I am going to model the 5687 under some relatively hefty current/low impedance drive conditions, in an LTP... I got plenty of those. I also have a good supply of 12BY7s...

With a 5687, and maybe 10K plate resistors or such... the 50K grid-bias resistor may not matter much. I'd rather be safe than sorry, as far as bias control (i.e, maintaining grid voltage no matter what) is concerned...

Also, the 5687 has lower Cgp than the 6CG7. That can't hurt...

Regards,
Gordon.

 

[dcgillespie]

Sounds like a neat project underway!

Regarding the allowable dc resistance in the grid circuit (and assuming a quality tube), it all depends entirely on how much quiescent current/dissipation a tube is subjected to in a given design. For example, when I published my EFB modification for the small Dynaco amplifiers, I was asked why I didn't lower the grid number one resistance for the output tubes in those amplifiers, as the original value (470K, which I retained) is higher than recommended when 6BQ5 tubes are operated in fixed bias mode. However, because the EFB modification lowers the dissipation levels of the output tubes to no more than 75% of the Design Center rating for those tubes (with EFB, the tubes dissipate < 9 plate watts each, versus nearly 12 for the original design), the fact that the resistors are over spec is a non issue. Of the many, many now who have converted their amplifiers to EFB with the stock Dynaco boards and components, not one has complained of unstable output tube operation.

The bottom line is that the spec for allowable dc grid resistance for any given tube is based on the tube operating at max dissipation. As the tube is increasing operated below max ratings, then some leeway can be had with the spec.

For 6550s, in production amps that typically run lower quiescent currents, they can get away with higher than spec values -- but of course, this is usually at the expense of distortion as well. In high fidelity applications where the quiescent current it typically run higher, I would not go over 100K for these tubes, and try to keep it at spec if at all possible.

Dave

 

[BinaryMike]

For 6550s, in production amps that typically run lower quiescent currents, they can get away with higher than spec values -- but of course, this is usually at the expense of distortion as well.

Does the higher distortion result from the grid going more positive due to increased ion bombardment on anode current peaks?

 

[PakProtector]

Does the higher distortion result from the grid going more positive due to increased ion bombardment on anode current peaks?

No, it is because the designers of said circuits are picking the idle point for maximum power delivery, not maximum fidelity delivery. The type 813 is even worse, with a 30k grid circuit maximum. In practice stepping around that with a grid choke works well.
cheers,
Douglas

 

[GordonW]

I did some modeling in TubeCad, comparing the 6CG7/6SN7 to the 5687.

Looks like a good candidate.

The 6CG7, under the conditions of the original Citation V, has an output impedance of about 5K or so.

I took the 5687, increased plate voltage a bit (to about 460v- because I could- I have up to 600v available!), increased the current (to 6ma per tube section), and reduced the plate resistor to 11K/12K (top/bottom) and the cathode resistor to 12K. Requires about 20v more grid voltage (which would also just require increasing the plate voltage to the 12BY7, which is just as easy)... but you wind up with an output impedance of less than 2K in the phase inverter! That should work better with 50K grid bias resistors, as the original did with the 6CG7 and 120K bias resistors.

I also increased voltage swing a bit. More drop across the anode resistors, and such. Probably about a 10% increase from the stock Citation V values.

Anyone know anything about the 5687, that would be a factor as to why I should NOT use it, in this application?

Regards,
Gordon.

 

[Kegger]

I can't see anything wrong with that setup Gordon.
(when I get a chance I'll look it over more closely)

 

[GordonW]

Oh, BTW: I modeled it with a 2K bias balance pot, instead of a 5K, and 7.5K resistors to the cathodes of the output tubes, instead of 10K/15K. That should keep the bias range about right. I'm only running one channel per amp, so the rest of each bias supply should work (about twice the load per channel, but half as many channels per bias supply). Naturally, I'll have to tweak the bias supply for the Bogen circuit (tapped from the primary OUTSIDE the transformer, with a voltage divider and separate rectifier)...

Regards,
Gordon.