• 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.

I want to make a simple UL PP amp

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
hi all,

my goals are:

1. use the least number of tubes
2. have enough gain to drive 89dB speakers with no preamp
3. get 30W something from it
4. understand all of these :)

in order to achieve #1 and #2, i chose to make use of 6SL7. half of which will be the driver and half is the phase splitter.

i used tubecad to come up a textbook grounded cathode driver for 28.6dB gain with an unbypassed cathode resistor. the bias will be 1.5V and Vout is +45/-50 V.

now the phase splitter, with one triode i can only do a split load.

in a split load, does the Ra computation matter? normally it would matter to determine the gain of the circuit, right? however, since a split load does not have gain, how do i compute this correctly?

the 27K in the circuit was just a "guess" to find out what gain i will get which turned out to be -0.455 dB

what else should i look for? any additional readings on split load? (incidentally, there is a very good article about differential phase splitters in the Aiken Amplification website)

cheers!

ps.

B+ is 420V and B++ is 260V
 

Attachments

  • ulpp.gif
    ulpp.gif
    9.1 KB · Views: 708
To estimate gain you have to figure half of the output voltage in series with the input, cathode follower-style NFB but half. This is current feedback so the Zo at the plate increases by as much as distortion and gain are reduced, while all three figures are reduced at the cathode. Important to know when it comes to HF response, which means some sort of buffer (CF or plate follower; the latter would make it a Williamson driver topology) is preferable. You're driving pentodes so I'll hesitatingly say it won't make much difference. But for 30W output I'd rather have something like a 12AU7 triode there instead.

Um back to the cathodyne... equivalent AC circuit places the EL34 grid leaks in series, paralleled with the plate and cathode resistors in series, for the plate load.

In general you can assume you will get about as much voltage out as in, so 30V input at the grid will produce 30V at the cathode and -30V at the plate. Obviously it'll be a bit less, but I wouldn't expect more than 3/4 loss unless you really screwed up your resistor values. ;)

If you want to maximize power output, you'll have to drop near class B. More efficiency allows more power output for a given plate dissipation. (Mind the ideal load resistance in this situation!) Supposedly starts sounding like SS though...

Tim
 
hi tim,

is your choice of 12AU7 has got to do with the transconductance, plate resistance, or both? is a higher transconductance, lower plate resistance also beneficial for a split load?

> In general you can assume you will get about as much voltage out as in, so 30V input at the grid will produce 30V at the cathode and -30V at the plate. Obviously it'll be a bit less, but I wouldn't expect more than 3/4 loss unless you really screwed up your resistor values.

ok, i am shooting then for the same Vout as my driver stage without having too high a B+ for me to worry about the cathode-heater issue.
 
At 27K, you're running the 6SL7 pretty hard, and with such a small value, the swing will be limited. I'm not optimistic about this circuit. If you want to use a split-load, you might consider one of the dual triodes made for vertical oscillation/deflection in TVs; they usually have a high mu/low current triode paired with a low mu/high current triode. Use the former as the input stage, the latter as the split-load.

Have you considered a constant current diff amp instead? With a 6SL7 and 100K plate loads, you should be able to get a gain of roughly 30-35, and the current source can be a pretty simple fet or bipolar circuit. This would be even better if you could bend to accept one more tube to interpose as a driver diff amp, then take some local feedback, but I don't want to restart THAT war.

There are a bunch of details to clean up, too (you'd want input resistors on the output stage, possibly screen resistors, too, and you only need one cathode resistor if you're bypassing, etc, etc), but let's get the basic topology nailed down first.

EDIT: Looks like fscarpa nailed down that last point already while I was getting coffee :)
 
fscarpa58 said:
I think you can avoid one res and one bypass cap
at the cathode of the output tubes as follows
federico
that would be great! i'd have to double the wattage of the cathode resistor though, right?


SY said:
At 27K, you're running the 6SL7 pretty hard, and with such a small value, the swing will be limited. I'm not optimistic about this circuit.
hi sy, i think tubecad confirms what you said as my Vout is now lower than the input stage and i need to ramp up B+ to catch up.
SY said:
If you want to use a split-load, you might consider one of the dual triodes made for vertical oscillation/deflection in TVs; they usually have a high mu/low current triode paired with a low mu/high current triode. Use the former as the input stage, the latter as the split-load.
like a 6EM7 perhaps? or how about a 12AU7 as my split load? i'll keep the 6SL7 as driver?
SY said:
Have you considered a constant current diff amp instead? With a 6SL7 and 100K plate loads, you should be able to get a gain of roughly 30-35, and the current source can be a pretty simple fet or bipolar circuit. This would be even better if you could bend to accept one more tube to interpose as a driver diff amp, then take some local feedback, but I don't want to restart THAT war.
yes, i did. that will be may last option due to the CCS which needs a visit to Gary Pimm's website. :) i have nothing against feedback so that' s not a problem with me :)
SY said:
There are a bunch of details to clean up, too (you'd want input resistors on the output stage, possibly screen resistors, too, and you only need one cathode resistor if you're bypassing, etc, etc), but let's get the basic topology nailed down first.
oh ok, how about 1K grid stoppers to the EL34 grid?

many thanks to both of you!
 
A 6EM7 will work, as will a 6DR7- and many more, too. A conventional choice would be a 7247.

IME, you don't have to get fancy with a diff amp current source. A single bipolar circuit will work very well and save you the hassle of making a B- rail. One transistor, two resistors, and a diode.

Between 1 and 2.2 kohm on the EL34 grids will work fine.
 
7247 is half and half 12AX7 and 12AU7 combined :)

Yeah, cathodynes don't like low volts...you need as much headroom as you can get. It'd be a good idea to run the 6SL7 from +400V and bias it at 1/4B+. Here you'll have maximum possible output from the splitter (because it's cathode will be at 1/4B+ (100V) and plate at 3/4B+ (300V) making for 200V p-k voltage). Sy's right, an LTP would be another good choice here. If you prefer tubes ;) you can use a pentode just as well instead of a transistor CCS. You'll have a bit less headroom, but then, you still have that 400V supply..

In answer to your second post, 12AU7 is just an all-around good compromise. In this case a 6C4 would be better since you only need half (unless you want to parallel both halves).. It has reasonable current levels (um 12mA max at 100V?) and <10k Rp making it good for drivers. Low mu, and likewise Gm (rounds out the triangle: mu = Gm * Rp) but what are 'ya gonna do.
You can use a 6SN7 or equivalent (um single is 6C5 isn't it?) just as well.
You're just driving EL34s so I wouldn't bother spending the current for anything larger like a 6BX7 or vertical driver tube (6FD7 etc.).

UL still has an output impedance so I'd want more gain than a 6SL7 or 12AX7 for the first stage, so you can add some global NFB. I'd recommend a pentode front end like on my Frankenhouse.

Tim
 
I wonder

if the Allen Wright design of a PP-amp would be a good choice. What I read about it says that it is the best: state of the art"".
What tubes does he use ?
I suppose that his design is not more difficult to build or much more expensive. So why not go straight for the best solution ?

Jaap
 
Re: I wonder

SY said:
A 6EM7 will work, as will a 6DR7- and many more, too. A conventional choice would be a 7247.
now i'm not sure if i can source them locally.
SY said:
IME, you don't have to get fancy with a diff amp current source. A single bipolar circuit will work very well and save you the hassle of making a B- rail. One transistor, two resistors, and a diode.
Ok, or a tube :) but that will violate my goal, he he.
SY said:
Between 1 and 2.2 kohm on the EL34 grids will work fine. [/B]
Ok and I chose 1K.


Sch3mat1c said:
7247 is half and half 12AX7 and 12AU7 combined :)

Yeah, cathodynes don't like low volts...you need as much headroom as you can get. It'd be a good idea to run the 6SL7 from +400V and bias it at 1/4B+. Here you'll have maximum possible output from the splitter (because it's cathode will be at 1/4B+ (100V) and plate at 3/4B+ (300V) making for 200V p-k voltage). Sy's right, an LTP would be another good choice here. If you prefer tubes ;) you can use a pentode just as well instead of a transistor CCS. You'll have a bit less headroom, but then, you still have that 400V supply..

and because of this, i am bailing out of the split load for now, until i source some tubes more appropriate as sy and yourself has recommended.
Sch3mat1c said:
UL still has an output impedance so I'd want more gain than a 6SL7 or 12AX7 for the first stage, so you can add some global NFB. I'd recommend a pentode front end like on my Frankenhouse.
in my circuit below, is this still an issue?

Jaap said:
if the Allen Wright design of a PP-amp would be a good choice. What I read about it says that it is the best: state of the art"".
What tubes does he use ?
I suppose that his design is not more difficult to build or much more expensive. So why not go straight for the best solution ?
Jaap
hi jaap, as much as i'd like to but allen's circuit is triode strapped and i want more power :)

---

because of what have transpired so far, i am ditching the split load in favor of a long tail pair. i'd like to give the 6SL7 another try, as shown in the diagram below. i thought the -1.5V tail would be nice for a battery source?

i'd appreciate further comments on this circuit gentlement.

cheers!
 

Attachments

  • el34pp-ltp.gif
    el34pp-ltp.gif
    9.4 KB · Views: 508
Hi,

Why dont keep the split load phase inverter with the 6SL7 but change the resistors at anode and cahode to more conventional values, say 100k?

A long time ago, (~35 years ago) when I was very young I used a 6SL7 with 47k + 47k to drive 6146's with 150k grid leaks, it is not ideal but it worked surprisingly well, I changed the 6SL7 to a 12AX7 later but the difference was mimimal.

Check old conventional circuits for normal component values, it is quite common to have grid leaks ~2-4 times the anode resistors, in your case you have a factor of 12.

Regards Hans
 
i thought the -1.5V tail would be nice for a battery source?

That's where you least need a voltage source like that and a small value resistor. High impedance at that point is a GOOD thing and will let you get the plate resistors more equal, dropping distortion. And that's the point at which you've got a high supply rejection, too. If you're determined not to use a silicon current source (bipolar or fet), at least make that tail resistor a proper tail and take it down to a nice negative rail, like -50 or -100 V. Or you can capacitively couple the input to the amp, raise the B+ on that stage, and perch the diff amp on top of a nice big tail resistor to ground.
 
hans and frank,

yes, why not. i'd give it a shot.

sy,

with 47K tail i will need -44V and Ra2 - Ra is close with only 2.32K between them (Ra = 75K, Ra2 = 76.8K). gain dropped from 34 to 26 though. my Vout (max) is quite balanced at -71/+71.2 V. i could probably play with B- to find a better operating point of the 6SL7 probably up to 1mA @ tube so my B- will be around -92V.

thanks a lot gentlemen :)
 
my goals are:
get 30W something from it


You mention you needed something like 30watts. It would be hard to do so using this design. UL should give you around 18watts only plus or minus(I recalled with in class A, not sure for Class AB). How about you try working around direct coupling the drivers with the el34. THis would allow you to go into AB2. Might atleast push the power up more.

I find when using ltp, it is worthwhile to use high B+ to allow a better swing and linearity. You might want to use a pot between the anode resistors to balance for AC. I would go for a high B+ as well as a high B-. I used something like 300 volts for my diff stage 6922. The plate was 50 volts above the cathode, which is the right voltage to run the 6922.

For some view on how i did my direct coupling check out my website. I direct coupled a 6922 to a el34. Sounded fine and great. However i used triode mode and biased them into A. was getting around 12 watts with a 6k trans
 
hi nick,

i'm going AB and i think is the only way to get those 30W. textbook figures show at AB1 with ~420V 130mA Rk=470 Zout=6K6 i should get 37w at 1.3% THD

ok, i can push the B+/- of the LTP up to max where i won't need to lift heater voltages.

i visited your website, i'm sure i'll learn more with the information contained therein.

cheers!
 
That sounds right, you ought to be good for at least 25-30 watts with that load and B+. If you want to squeeze a few more watts out, just use so-called fixed bias. The cost and complication of a negative rail and a couple of pots is probably less than putting in a direct-coupled cathode follower- and it's fewer tubes. But really, it may not be worth it- we're quibbling over a couple of dB in max SPL.
 
he, he, frank, everytime i power up a new amplifier i use that Bose. it is 16 years old, so if the drivers blow up, i won't feel bad. :D

it is now playing with my resident Infinity Kappa 200 (made in Denmark ;) ) listening to some Carmen Cuesta, Jacintha, and Metallica. :)

mids are ok, a bit warm. bass is firm and tight, i think there's a top end roll off. i'm using really cheap parts - carbon resistors, white coffin cathode resistor, taiwan electrolytics, and the only branded one is the pair of coupling caps made by Electrocube. output transformers are custom wound.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.