SE multi tube design

[Lynxtrap]

I have some difficulty grasping the concept of impedances in output stages.
Say I want to make an SE amp, or even parallel SE, where I could use any common type of octal tube, and want to use an 8 ohm speaker.
I figure the Hammond 125ESE would be a suitable transformer.

How do I make sure the different tubes see the right load?

And also, what would be acceptable "universal" plate voltage and cathode bias resistance?
You often see something like 470 on the cathode, and I guess the possibility of using 6V6 kind of sets the limit for plate voltage.

[tubeswell]

2 x parallel tubes of the same type want to see a total load that is 1/2 of what one tube of the same type would see. 3 parallel tubes want to see 1/3 of the total load and so on.

For centre-bias Class A operation, a useful approximation for the optimal load impedance (Zout) is:

Zout = Va/(Pa/Va).

Where:

Va = plate-to-cathode voltage
Pa = the total power dissipation of the output tube(s)

Say you have a 6V6 idling at 350V with a cathode voltage of 20V (Plate-to-Cathode = 330V)

330/(12/330) = 9,075R (9k)

There's an important caveat to this.

Screen voltage is important to factor in, because that determines the peak screen current the tube will see when plate voltage drops to minimum during the signal cycle. This is where you need to plot a load line. Ideally you want the screen voltage to be as high as it can, without causing excessive screen current under this condition. As a rule of thumb, most of the time, the tube will be happy if the screen is sitting 30-50V below the plate for SE Class A centre bias with an optimum load. If you want to run the screen at the same voltage as the plate, then reduce the reflected load a bit (to somewhere between 7k5 and 8k). This will help minimise screen dissipation when you crank the amp.

[martin manning]

I have some difficulty grasping the concept of impedances in output stages.
Say I want to make an SE amp, or even parallel SE, where I could use any common type of octal tube, and want to use an 8 ohm speaker.
I figure the Hammond 125ESE would be a suitable transformer.

How do I make sure the different tubes see the right load?

And also, what would be acceptable "universal" plate voltage and cathode bias resistance?
You often see something like 470 on the cathode, and I guess the possibility of using 6V6 kind of sets the limit for plate voltage.

Start by reading through the procedure here: http://www.valvewizard.co.uk/se.html Choose an anode voltage and run the simple calcs for the tube types you are interested in, finding Zpri = Va^2/Pa, where Pa is the dissipation limit for each tube type times the number of tubes. 300-350V is probably reasonable for the anode voltage, but select a PT and use the estimated VDC you will get from it. There won't be a universal Rk, so you may want to have a switch or pot to dial it in for each tube type. Maybe have a couple of screen resistor options too.

[statorvane]

Say I want to make an SE amp, or even parallel SE, where I could use any common type of octal tube, and want to use an 8 ohm speaker.
I figure the Hammond 125ESE would be a suitable transformer.

I did something similar using two Hammond 125ESE OTs, and a Hammond 270HX PT. I wired the OT primaries in parallel and the secondaries in series. I added a switch to select between two different primary impedances, and separate outputs for either a 4 ohm or 8 Ohm cab. I have run it with Sovtek and TungSol RI 6550s, Sovtek "5881s" (a little shrill), RCA 6L6GCs, EH 6CA7s and JJ EL34s. The EH 6CA7 and Sovtek 6550 sound the best, I usually just leave the EH 6CA7s in there. Best amplifier I ever built. It isn't biased for 6V6s or KT66s.

There is an error in the schematic. V2A's amplified voltage should come after the 220K anode resistor, not before.

[Lynxtrap]

Thanks! I'll have to dig into the math.
I guess I was thinking there would be kind of a one works for all,with the right cathode resistor, the right OT and secondary impedances matched.

AFAIK there have been some amps on the market that are advertised as working with any tube. I have taken a fast glance at a few different sources.

In TUT3 O'Connor writes that his updated Champ can take any tube although with no change in output power. He doesn't go deeper into that subject.

I also have Dave Hunters book where he has a build plan for an amp called the Two Stroke. It runs two 6V6's in parallel or one of the bigger tubes in one slot. It includes an impedance switch but not much explanation.

Then there are the different AX84 projects that promise to accept all octals, but seem to be designed for either EL84 or KT88. And still no explanation as to swapping them to 6V6, Kt66 etc.

[statorvane]

Then there are the different AX84 projects that promise to accept all octals,

Check the old P1eX project/schematic at AX84. There are footnotes on the schematic suggesting appropriate cathode resistors for different power tubes. These are good starting points; the PSE power amp I built is based on two of these paralleled.

I suppose you could bias a design for twin 6V6s that could also accept larger octals, but you need to make sure the PT could handle the heater current, and you probably won''t get any more power than the 6V6s deliver.

[Stevem]

If you peer into the RCA tube manual, 300 volts on the plate of 6V6 will likely push it out of class A.

[Lynxtrap]

Thanks statorvane, I'll have a look at that schematic.

I also looked at the THD Univalve. Plate voltage switchable between 300V and 450V and should be able to take any tube, with 300R on the cathode.

Funny the manual doesn't say anything about impedance matching when swapping tubes...

[martin manning]

If you peer into the RCA tube manual, 300 volts on the plate of 6V6 will likely push it out of class A.

I don't know what you mean by this. Can you explain?

For 300V Va-k the simple math above (Zpri = (Va-k^2)/Pa says that at 12W Pa (6V6), the operating point will be centered (plate voltage wise) on a 7k5 load line when biased at 40 mA. That is very close to the op point seen on the Fender '57 Champ schematic (Va 322, Ia 36mA, Pa 11.6W), which likely has 8k Zpri.

I also looked at the THD Univalve. Plate voltage switchable between 300V and 450V and should be able to take any tube, with 300R on the cathode.
Funny the manual doesn't say anything about impedance matching when swapping tubes...

None of that means that the load lines and operating points will be optimum, or even in the best place tone-wise. Larger tubes (with more plate dissipation capability) can be run at high primary impedance and low idle current without any problem, they are just not producing all the power they are capable of. At a given plate voltage and load, output power will be nearly the same for all types, as long as there is enough drive signal. Different types will produce different sounds, and maybe that is all that is wanted.

[Lynxtrap]

None of that means that the load lines and operating points will be optimum, or even in the best place tone-wise. Larger tubes (with more plate dissipation capability) can be run at high primary impedance and low idle current without any problem, they are just not producing all the power they are capable of. At a given plate voltage and load, output power will be nearly the same for all types, as long as there is enough drive signal. Different types will produce different sounds, and maybe that is all that is wanted.

Yes, it is of course always a matter of compromise with that kind of amp.
I was mostly thinking about matching the speaker load to different tubes.
The way I understand it, different tubes would like to see different loads on the secondary side, everything else being equal?

[statorvane]

At a given plate voltage and load, output power will be nearly the same for all types, as long as there is enough drive signal. Different types will produce different sounds, and maybe that is all that is wanted.

Martin nailed it. Pretty much what I have experienced.

There was a version of P1eX with a switch for power tube cathode resistors - switched them from 220 Ohms to 600 Ohms, so you could swap in a 6V6 along with the big octals. The original P1eX schematic from Phil Rowley suggested both cathode resistor values and impedances for several octal tubes. That is a good place to start. Most OTs from Hammond support multiple secondary impedances (have several impedance ratios) so it should be possible to work out something feasible for a variety of octals.

In the end, unless you play with the anode voltages as well you will wind up where Martin left off above.

[Lynxtrap]

At a given plate voltage and load, output power will be nearly the same for all types, as long as there is enough drive signal. Different types will produce different sounds, and maybe that is all that is wanted.

Martin nailed it. Pretty much what I have experienced.

There was a version of P1eX with a switch for power tube cathode resistors - switched them from 220 Ohms to 600 Ohms, so you could swap in a 6V6 along with the big octals. The original P1eX schematic from Phil Rowley suggested both cathode resistor values and impedances for several octal tubes. That is a good place to start. Most OTs from Hammond support multiple secondary impedances (have several impedance ratios) so it should be possible to work out something feasible for a variety of octals.

In the end, unless you play with the anode voltages as well you will wind up where Martin left off above.

Thanks!
As to impedance, I looked at the specs for Hammond 125ESE (see pic below), I also found that original P1eX schematic.

So to sum it up - let's see if I got it right:

With an 8 ohm speaker, I would use the WHT secondary for tubes that like PRI 2500 (KT66, 88, 6550).
For tubes that want 5000 (6V6 and 6L6), I would use the YEL secondary. These tubes could also be used with a 16 ohm speaker on the WHT secondary.
EL34 might work with either secondary into 8 ohm.

I could make the bias variable with a pot, as shown in some of the AX84 schematics.
If I'm really getting into it, I could make anode voltage switchable as in the Uivalve, or else live with the compromise.

Does that seem right?

[statorvane]

Yes, I think you are correct. Are you going to run the 6V6s in parallel and the others as single?

[Lynxtrap]

Yes, I think you are correct. Are you going to run the 6V6s in parallel and the others as single?

Thanks! I haven't even got that far in the thought process yet.
If I could squeeze 15W out of one of the bigger tubes, there might not be much point in bothering with parallel 6V6's. But I have a possible donor chassis with two output tube sockets.

[martin manning]

If I could squeeze 15W out of one of the bigger tubes, there might not be much point in bothering with parallel 6V6's.

You could get 15W from a 6L6GC with a 5k load. The anode voltage would be just under 390V, and you would be under the 80mA continuous current limit for the Hammond 125ESE OT. However, assuming that the anode voltage doesn't change, lower wattage tubes will need increasingly higher load impedances to center the operating point, 6k for EL34, and climbing up to about 12k5 for a 6V6GT.

If you want to run constant anode voltage and load impedance for several tube types, then the output power will have to be reduced to a level that works for the lowest plate dissipation tube that will be run at that load impedance. For example, with the right anode voltage (346), it's possible to land on the two available load impedances, 5k for EL34 (25W Pa) and above, and 10k for 6V6 (12W Pa). The output power will be about 6W for 6V6GT, and about 12W for all the larger tubes.

Another output transformer option for this type of thing is Classic Tone’s SE project type, which has 5k and 8k primary. It’s current limit is lower (70mA), so the anode voltage has to be reduced to 310 (8k matches 6V6 at that Va), and output power goes down to just under 10W for the big tubes. There’s a 100mA PT that goes with it, and the price is lower than the Hammond OT.

For tubes that operate at a given load and anode voltage, the anode current at a centered operating point will be constant, and that means that for any given cathode resistor the cathode voltage will be approximately constant, and this will determine the screen voltage. What I think would be interesting is to have available a range of cathode resistor values (a string of 2W resistors), and an adjustable screen voltage (VVR type). That would allow a variable bias voltage and therefore variable distortion threshold (grid signal level at clipping) at approximately constant output power.