Power trans design and tube draw

[jojokeo]

I am wanting an amp to use any type of 8 pin power tube there is in a low watt single ended design. My criteria isn't necessary to get maximum highest output power out of each, only moderate levels are fine since it's intention is to have some fun tube sampling. I could use a multi-tap OT to connect an 8 ohm spkr to the 16 ohm tap to lower the Pri-Z for tubes like an EL34 and other types would be fine w/ a 5K-primary. (if this assumption is wrong please let me know)

If I were only to use one or two 12a_7s to go with this then I would need what/??? minimum for a power and output tranny? Also, is my understanding below okay?
1) I suppose I will need to accomodate enough heater current for the highest user being the 6550/KT88,
therefore I would need 2 Amps min for this.
2) If I'm not shooting for max output power does it matter if I use something small like a 240v @ 55mA or a higher Voltage but small current 330v @ 120mA tranny w/out worry of it heating up and/or blowing from the larger tubes? (in other words, won't the tubes only draw what they can get at the expense of output power?)
3) Depending on which PT tranny I use (or another if suggested to do so), would I simply use a 12w SE OT or would larger be better/safer?
Thanks in advance for your time guys,
Keo

[PRR]

> My criteria isn't necessary to get maximum highest output power out of each,

Then your aim is the LEAST tube.

Yes, a double-fat tube can carry double current and half load impedance for double the the power... but that's not your goal. (And double-power is not "WOW!")

If you limit to "common octal power tubes", then you build like an older Gibson 6V6, 250V 45mA 5K load or 280V 40mA 7K load.

Most of the octal audio power tubes working at 6V6 power may use the same bias resistor as 6V6.

Pin 1 is a question. In DIY you can hard-tie pin 1 to pin 8 cathode. 6V6 6L6G don't have pin 1. EL34 wants pin 1 tied to something, and cathode is normal. True 6550 has pin 1 to metal base: don't use a base-claw, and be aware you will get a 20V shock off the base. The outstanding joker is 7027: it has pin 1 tied to Screen Grid! In brief experiments, I found that a 470K 1/2W resistor 1 to 8 would hold EL34's G3 near cathode voltage yet allow the 7027's internal strap to pull up to 400V and the resistor just ran hot.

> some fun tube sampling.

Then just use a Champ/Deluxe PT. It will carry 0.3A+0.3A+0.45A+0.45A or 1.5A all night long for decades. It will surely carry 0.3A+1.6A for hours. When you get up for more beer, feel the PT. Hot is normal. Sizzz-YOW!!! is too hot.

[jojokeo]

Thanks for the heads up on the 7027, maybe I'll exclude that one.

To confirm
1) Design not to exceed the smallest tube's voltage rating, power dissipation, and try to match it's load impedance.

2) I need to have enough heater current to make the largest power tube happy and the PT's secondary mA output rating isn't significant because the tube will only use what it can get and suffer a loss in output power as a result (while not stressing the PT)? The only way to stress or blow a PT in this situation is if the heater's rating is exceeded?

3) Size the OT according to the highest output power anticipated w/ a primary impedance that gets as close as possible for the smallest tube to get it's full performace and for the larger tubes it will be less significant because they'll adjust down easier and aren't going to be at full performance anyways?

[PRR]

> try to match it's load impedance.

Tubes don't have "a load impedance".

You pick it by selecting the voltage and current.

There are limits.

The maximum useful load impedance is limited by plate voltage or transformer winding troubles. We "could" run 6V6 at 1000V 10mA 100K load.... but a 100K winding won't have good bandwidth, will cost a lot, and 1,000V is so far beyond what 6V6 was made for that we would have trouble right away.

The minimum useful load is set by the plate resistance _below_ the knee. You want to stay well above this impedance to get power out of the tube. When plate and screen voltages are similar, a good trick is to learn the tube's "Triode plate resistance" and double it. This suggests >4K for 6V6 and >2K for 6L6.

Most of the higher-voltage (exclude 110V kitchen radio tubes) Audio Power tubes were designed to handle 2K-5K loads for the big bottles, 4K-7K for the small bottles. Voltages over 300V-400V tend to cost more, windings of 10K and up cost more. The tube makers were no fools.

Getting to specifics: you have 6550 which can make 20 Watts output, 6L6 generally taken as 10W out with best conditions, and 6V6 at 5 Watts. If your goal is "MAXIMUM POWER!!", then you don't fool with tube swaps. You design for the 6550's 20W condition and only use 6550/KT88.

But even 5 Watts is pretty loud. And this is not intended to be a Big Stadium amp. This is a tube-roller. You design for the LEAST tube in your lot. Probably 6V6. 6L6 6L6GC 5881 EL34 6550 KT88 KT66 will all work in a "6V6" hole with little or no change. And with essentially 6V6 power output. Oh, the 6550 might get 10% more Watts, or the difference in EL34 curvature could give a different Watts number at the "same THD %", but such differences are small.

So aim near 250V 45mA 5K load or 280V 40mA 7K load. This will be 5 or 6 Watts output. The standard Champ Replacement OT is a safe bet.

As for heaters.... The Deluxe replacement is specced (IIRC) for 1.5A heater. One 12AX7 and one 6550 will exceed this by 27%. With some reasonable assumptions, your "6.3V" will be 0.1V low and the PT will be 0.3 Watts hotter than the designer intended. That's not Good Practice; however it isn't much of an overload and as an experimental platform it will work for years.

> for the larger tubes ... they'll adjust down...

I need to bring power to my garage. I do the math, it says I need 2,345circmil copper to carry the current. #10 wire is 2,800circmil and will work. However I find some #4 wire with 34,567circmil of copper already hiding in the ground to the garage. There is no electronic problem using a bigger wire (or tube) for a smaller load! There may be an economic problem: #4 wire is 7 times the price of #10! And in an ideal world, 6L6 would be almost twice the price of 6V6 (the historic 6L6/6V6 ratio is about 1.5). So if you have to BUY the wire or tube, you aim small. But if you already have it in the ground (or when tubes are as cheap as they are today, relative to cost of beer or gas), then just use it.

[jojokeo]

I got it  I've not "designed down" w/ tubes really and it's different than usual for me. I needed to confirm my suspicions and I really like the beer analogy.  My garage fridge is stacked up nicely w/ a lot of varieties waiting to be sampled. Mahalo

[Merlin]

Since you're one of the few people exploring the idea of non-maximum-power tone amps, you might be interested in this.

It's a push-pull output stage which you can plug a wide range of tubes into, in any combination too! So you could have a 6V6 with a KT88, or an EL34 with a 5881, whatever. Because of the cross-coupled cathode biasing, the two tubes will always bias to within a couple of milliamps of each other, with about 12W dissipation, and the output power will always be about 20W.The OT impedance should be suitable for a 6V6, say 6.6k.

Although I know the biasing system works, I have not tested it for tone. Just though't I'd throw it out there to give you some ideas. Each tube should have a cathode bypass cap, which I forgot to add to the diagram, duh...

[labb]

Check out the Single Ended Lead with external bias over on AX84 site.

[jojokeo]

Since you're one of the few people exploring the idea of non-maximum-power tone amps, you might be interested in this.

   The first thing I thought of after reading that is what Ted Weber said when he was designing his neo speakers and how efficient they were but didn't have good tone at first. How speaker manufacturers were hyped about this fact at first in their designs trying to maximize this and that they seemed to be in an efficiency contest. Good speakers should = good tone w/ whatever magnet material they had in them. He had to make them less efficient or rather limited their efficiency to get the good tone he was after. A few other manufacturers did the same and yet others he said, "were clueless."

Thank you for the schematic and feedback. Since I'm unfamiliar w/ the design, would the bypass caps go across both of the cathode resistors on each tube from ground to the tube's cathodes? Also, there must be a reason for the large capacitor at the B+/screen/CT tranny point?

[FYL]

Although I know the biasing system works

It does: it's a Blumlein garter. Yet another design from one of the true greats.

Each tube should have a cathode bypass cap

Yup. From cathode to ground, across the two resistors

[FYL]

Also, there must be a reason for the large capacitor at the B+/screen/CT tranny point?

It's a DC supply symbol.
 

[Merlin]

Also, there must be a reason for the large capacitor at the B+/screen/CT tranny point?

I just used the battery symbol to indicate the B+. In reality you would have a standard power supply, with the screen being fed from an RC or LC filter after the main reservoir capacitor.

BTW, as FYL said, it's often called a Garter Circuit among hifi people. But as far as I know Blumlein didn't come up with it; that appears to be a figment of John Broskie's imagination.

[FYL]

But as far as I know Blumlein didn't come up with it; that appears to be a figment of John Broskie's imagination.

Hmmm. Looks a lot like other Blumlein designs from the mid 30s and everybody calls it Blumlein garter since Broskie brought it back into focus a few years ago.

[Merlin]

Hmmm. Looks a lot like other Blumlein designs from the mid 30s

Heh, "looks like", but I bet you can't lay your hands on a single Blumlein circuit that actually uses it. Believe, me, I've tried!

[jojokeo]

Can any type of pi be used to feed this or is one suggested over another? The bias seems to very cold by the values but what would it be actually in operation? I don't know what to think about it. The bypass caps must help this issue out and why they're needed?

[Merlin]

Can any type of pi be used to feed this or is one suggested over another? The bias seems to very cold by the values but what would it be actually in operation? I don't know what to think about it. The bypass caps must help this issue out and why they're needed?

The resistors look bigger than normal because the cross coupling puts positive voltage on the grid and, to cut a long story short, they need to be big to get the bias point in the right place every time. As far as the big bottles are concerned, yeah the bias is cold, but that's what you wanted in the first place- an amp with the tone of big bottles, but without the massive dissipation. Think of it as an amp with power scaling permanently set low.

Use whatever PI you want, just treat it like any other push pull amp.