• WARNING: Tube/Valve amplifiers use potentially LETHAL HIGH VOLTAGES.
    Building, troubleshooting and testing of these amplifiers should only be
    performed by someone who is thoroughly familiar with
    the safety precautions around high voltages.

6F6G Amplifier

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
How does it look? I have the Z565's, PA774, and the tubes. I'm also considering using a single 6EM7 per channel, the high gain half direct-coupled to a concertina formed by the low gain/low Ra half, Williamson style.

An externally hosted image should be here but it was not working when we last tested it.

An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.


I've got 8 Russian 6F6G types on hand, but am prepared to try new production 6V6's as well..
 
sorenj07 said:
How does it look?

Looks like a good start.

First off, the 6SL7 LTP phase splitter needs a longer tail in order to work right, preferably a CCS. Here, you don't have much of a tail at all, just the AC impedance of that diode. I also have doubts as to how well this type can drive the grids of a final, with its higher Ci + Cmiller + Cstray. I'd install a 6CG7/6FQ7/6SN7 (I'd suggest a 12AU7 for this, but 12AU7s aren't any cheaper, and sometimes even more than the more linear types) cathode follower grid drivers. If you don't want to make more holes for sockets, then a MOSFET source follower. That way, you can direct couple the cathode followers and also avoid that blocking problem that overdrive tends to cause when capacitor coupling to a fixed bias stage. You already have a negative rail, so that's NBD to add.
 
First off, the 6SL7 LTP phase splitter needs a longer tail in order to work right, preferably a CCS. Here, you don't have much of a tail at all, just the AC impedance of that diode.

the diode is reverse biased while in operation.
it seems it is there to prevent cathodes initially 150 V negative relative to grids at startup, which would be bad.
so actually it has a 150K tail.

agree with rest though; cf drivers would work nice here
 
I do have two extra EH 6SN7GT's that I could throw into this. I also have some badass 5687's but a) they're noval and b) probably overkill. My one concern is that it's a bit bulkier at 8 octals in one chassis for MAYBE 12-13W/ch...

None of them should be getting too hot though (6F6's are only running at 12W) so close spacing combined with ventilation holes should be good. Depending on financial constraints I'm going to try and CAD-design a custom stainless chassis.

One last question though: How do you design CF's into an amp? Should I just peruse a few schematics that use them? Got any good examples? How much extra current would the 6SN7 CF's take? I'm using 162mA quiescent as this amp stands out of the PA774's 180mA. It's a lot closer to choke input, which should free up some extra juice, but if the '774 already runs hot, I don't want to push things too much. Might have to go monoblock with a 150mA+ Hammond per chassis :smash:
 
driver stages 6f6 tubes

I would always look at what has been done

You will find the 12ax7 the tube used in acres of push-pull amps with the old split phase circuit used mostly.

A 6sl7 will provide plenty of drive to 6f6 tubes. These tubes do-not need a lot of drive as they are higher impedance pentodes, lower grid current demands.


If you forget the diode, use a 2.2K cathode resistor, and the neg feedback has to go to the input triode, a basic circuit will do the job fine.

Part of the Magic sound of older tube circuits is the variables in the circuit so no real need to negate everything.

I look at circuits of vintage gear to see how tubes were used before in gear that was sold and had good sonic character.

You can go to the library and get copies of SAMS photofacts, or look on the NET at schematics.

Also 6f6, 6k6, 6y6, 41, 42 and others are kind of lost tubes that have very good sonics and should be explored again.


Look at this link to see samples of some of my recent restortations and special audio projects.

http://photobucket.com/albums/b121/SalesBoy/

George
;)
 
sorenj07 said:
One last question though: How do you design CF's into an amp?

What is the Q-point bias for the 6K6s? Once you have the Vgk for these, then you know what the static cathode voltage is. Add enough Vpp at the plate for headroom, and enough standing current to provide the charging current for the Ci + Cmiller + Cstray, and draw a loadline.

Should I just peruse a few schematics that use them? Got any good examples? How much extra current would the 6SN7 CF's take?

Two examples:

Vixen Main Schemo

Le Renard Main Schemo

6K6s shouldn't present a particularly difficult load, so the required current for the cathode followers shouldn't be over 5.0mA. So far as the Vixen is concerned, the 6SL7 LTP didn't have anywhere near the drive to charge up the 807 grid capacitance for a good slew rate at the higher frequencies, and it sure didn't have enough to deal with transients that drive the grids positive. This would make for some nasty sonic degradation, even if an overt clip weren't actually heard. The cathode follower presents a much more agreeable load since there is no Miller capacitance, the Cgk is "reverse bootstrapped" to a smaller effective value, leaving only the Crt as the only capacitance that draws the lion's share of the current at 30KHz.

The situation with the Le Renard design is somewhat different. Here, the 6BQ7s would seem to have the current sourcing capability, as the gm is much higher than for 6SL7s. However, these are cascoded, and that makes for a higher than normal Zo. The grids of the 6BQ6s would drag it down too much, negating the frequency performance of the cascode. Again, the cathode follower driver represents a Hi-Z load that doesn't ruin the performance.

In both cases, having current sourcing capability greatly improves overdrive behaviour since the finals can transparantly slip a bit into A2 instead of clipping at the grid.
 
Thanks for the tips! the 6F6's (not 6K6's) are biased at around -20.6V 40mA. Unfortunately I couldn't find any ultra-linear curves for 6F6 tubes, so I based my calculations off some UL curves for 6V6 types. Here is my load-line.

An externally hosted image should be here but it was not working when we last tested it.


I'll spend some time figuring out what you said about CF's. Maybe Morgan Jones can help as well...

As a side note: Do you know who I could ask to plot some UL curves if I send them a few of my extra Russian 6F6S tubes? It'd be very useful for everyone, I think.
 
sorenj07 said:
I'll spend some time figuring out what you said about CF's. Maybe Morgan Jones can help as well...

This isn't all that difficult to figure out. Given a Po= 17W, and a Rl= 8K (P-2-P) you can figure:

V= sqrt(17 * 8E3)= 368.8Vrms (P-2-P)
V= (368.8 / 2) * sqrt(2)= 260.8Vp (per phase)

Av= 260.8 / 22= 11.6

For the 6K6, Crt= 0.5pF, so Cmiller= 0.5 * (1 + 11.6)= 6.3pF

Cgk= 5.5pF, so

Ci= 5.5 + 6.3= 11.8pF, guesstimate Cstray to be: 30pF, and

Ct= 11.8 + 30= 41.8pF

At 30KHz:

Xc= (2pi * 30E3 * 41.8E-12)^-1= 126K9

I= V / Xc= 22 / 126.9E3= 0.17mA

By the "Rule of Five", the driver plate current should be 0.85mA or better. Given that you have a -150Vdc rail already, you're good to go there. The plate of the cathode follower can be set to +150Vdc for headroom.

As a side note: Do you know who I could ask to plot some UL curves if I send them a few of my extra Russian 6F6S tubes? It'd be very useful for everyone, I think.

Yves just might have some already.
 

Attachments

  • 6sn7-loadline_i.gif
    6sn7-loadline_i.gif
    73 KB · Views: 1,021
Finally resurrecting this project. I've redesigned the 6SL7 LTP, feeding 2.5mA into the pair. By the rule of five, 1.25mA per 6F6 ought to be OK, right? In any case I still have enough heater current to spot an extra 6SN7 as a CF and I'm beginning to think it's worth it.

One question I have is the 150V that seems to feed the CF's in both schematics. Where should it come from, and how much plate current will each channel's 6SN7 need? I've got about 310V to work with but an extra 20mA total, say, will drag it down to 300V or so.

The two reference schematics also use different means of biasing the CF's. They both have capacitors to ground in common. What is their purpose/what is going on?

What is the resistor/capacitor from each output tube's plate to the CF input grids? Some form of feedback?

Also, what is the inductor/resistor network connected to each output tube's plate in both schematics?

Thanks a bunch. I really want to turn this amp into a reality.

P.S. Yves says he's super busy at the moment, and his curve tracer isn't in operation...
 
Ex-Moderator
Joined 2004
I'll try to answer some of your questions but Miles will correct me if I'm off the mark, I'm sure.

One question I have is the 150V that seems to feed the CF's in both schematics. Where should it come from, and how much plate current will each channel's 6SN7 need? I've got about 310V to work with but an extra 20mA total, say, will drag it down to 300V or so.

There are several ways of providing the 150v. It doesn't have to be exact and you need about 3mA per cathode follower, so to feed both channels you need 12mA. A simple restive dropper with a decoupling capacitor from your B+ should do the job. I make that 12.5k at 1.8 watts dissipation, so a 12k 7W wire-wound resistor should be OK.

The two reference schematics also use different means of biasing the CF's. They both have capacitors to ground in common. What is their purpose/what is going on?

The bias mechanisms in both schematics appear the same to me. The capacitors you refer to provide an AC path to ground that is independent of the settings of the bias network.

Note also the position of the 'safety' resistors in the bias networks (R17 & R18 in the Vixen, R18 & R19 in the Reynard), to provide a negative bias back-up in case the pots ever go open-circuit. I've seen other schematics in which the resistors are taken to the other end of the bias pots, or even to ground, and that's dangerous!

What is the resistor/capacitor from each output tube's plate to the CF input grids? Some form of feedback?

Yes, the local NFB loop in the Vixen helps to bring down the OP impedance of the OP tubes so that the global NFB loop has less work to do. This aids stability. Don't forget, the Vixen is not a UL design and so it has a very much higher OP impedance and needs more NFB to improve damping.

Also, what is the inductor/resistor network connected to each output tube's plate in both schematics?

Those networks help to prevent HF oscillations. The 807 may need it but the 6F6 probably doesn't.
 
sorenj07 said:
One question I have is the 150V that seems to feed the CF's in both schematics. Where should it come from, and how much plate current will each channel's 6SN7 need? I've got about 310V to work with but an extra 20mA total, say, will drag it down to 300V or so.

Since the voltage regulation there doesn't need to be anything special, I just used a voltage divider.

Vixen Main PS

Developing that really isn't such a big deal. Just give the divider resistor an idle current some 20% of the load current, and add that to the total current for calculating the divider resistors. The plate current for the 6SN7 grid drivers doesn't need to be much more than 3.0mA or so. That can still take care of the high frequency slewing, allow for improved overdrive behaviour, and not overburden the power supply.

Those networks help to prevent HF oscillations. The 807 may need it but the 6F6 probably doesn't.

That's what they are: plate stoppers. Being rather high gm VTs, the 807 and 6BQ6 can definitely use these to prevent HF oscillations. A 6F6 doesn't have so high a gm, and probably won't, but keep it in mind just in case you do get RF oscillations.

Other than that, Ray did a good job of explaining all the rest.
 
Thanks for all the explanations! I think I'm beginning to wrap my head around CF's in general. For example, that the grid of the CF needs to be biased more negatively than the cathode, so if you have the cathode feeding negative voltage to the following tube, the grid needs to be EVEN MORE negative...

One thing I don't like about the schematics above is the lack of ability (as far as I can tell) to adjust bias between the two output tubes. Is this for any particular reason? My 6F6's aren't matched.
 
sorenj07 said:
Thanks for all the explanations! I think I'm beginning to wrap my head around CF's in general. For example, that the grid of the CF needs to be biased more negatively than the cathode, so if you have the cathode feeding negative voltage to the following tube, the grid needs to be EVEN MORE negative...

Yes

One thing I don't like about the schematics above is the lack of ability (as far as I can tell) to adjust bias between the two output tubes. Is this for any particular reason? My 6F6's aren't matched.

Huh? Each final includes a 10R current sense resistor between the cathode and DC ground. Simply monitor the voltage across that resistor, adjust the bias for equal voltages, and they're balanced. There's really no need for any special current tracking as the currents are quite stable in operation. I've been running both designs for months now with no meaningful difference in no-signal plate current. You have to keep an eye on it for the first few hours after new finals go into service until they settle down, but that's basically it.
 
Ex-Moderator
Joined 2004
sorenj07,

In the Vixen, the pot R21 is used to adjust the bias of the upper 807 until the voltage across R27 shows that the tube's cathode current is at the required level. The same thing applies to R22 and R28 for the lower 807.

If it should happen that the range of bias adjustment is not what you need, you may have to change the values of R19/R20 or R23/R24 until it suits your needs. I had to do this once when changing from one brand of EL34s to another.

This is the best way IMHO of arranging individual grid bias adjustment. You just have to be careful to set each pot at the negative end (right-hand side in the schematic) of its travel before making the adjustments. It's normal to recheck and make any necessary correction to the bias settings after about an hour or two of running, as the tubes settle down.
 
All right. I've got the PSU mostly lined up, chassis, and some of the parts including a smaller choke for the CF's and 6SL7 LTP.

An externally hosted image should be here but it was not working when we last tested it.


The layout is a bit tight but I favor a more compact amp and am confident that I can wire it to be hum-free. It's funny to see a PA774 I scraped clean and repainted dwarfed by a relatively mundane Hammond 193J choke. I'm thinking of painting it red-orange to match.

I love the look of those Russian 6F6S bottles. And of course the price is rising as we speak with each mention, but hey, I've got 8 already :). I'm running them nice and easy at ~12W dissipation in UL so hopefully they'll last a long time. Either that or I can put in more boring, regular 6V6's.

I guess you could call this an all-Russian amp as it uses a 6N9S, 6N8S and 6F6S. I'll probably use Russian PIO caps in the signal as well...
 
Hey, I was just wondering if anyone had the values for the CF section of the Vixen. I'm shooting for a range of between -10 and -30V on the CF cathodes/6F6 grids.

I'm also considering replacing the 6SL7 with a 12AX7 and the 6SN7 with a 7AU7 (I have a few) for space concerns. Aside from tweaking the LTP, can I leave the CF values unchanged if I make this switch
 
^^^^

Here are the component values for the grid drivers:

R15-16: 1M
R17-18: 10M / 0.25W
R19-20: 2M5 (10M/4 (parallel))
R21-22: 100K; 24mm; Linear
R23-24: 250K
R25-26: 100K / 3W

All 0.5W metal film except where noted. 7AU7s will work, but you might have to adjust the bias resistors to tailor the voltages to your needs. Cathode follower service isn't really that demanding.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.