Tubes and Loadlines!

[craigmillard]

Hi Guys,

Im trying to up my tube knowledge a bit at the moment and want to build a low power poweramp out of 6N6P russian tubes.

Tube Datasheet - www.lampizator.eu/UPGRADE/RU-6N6P.pdf‎

Iv been trying to figure out the math and loadline techniques for a Push Pull amp with a fixed bias.

I found this guide which is quite detailed but seems to skip over certain steps:
http://www.turneraudio.com.au/loadmatch3-pp-triodes.htm

I was hoping someone with more knowledge than me can look over what im doing and point me in the right direction:)

Anyways i have been following the guide DESIGN METHOD, PP CLASS AB1, EH6550 IN TRIODE MODE from the above website and converting it to the 6N6P datasheet.

In step 1 its pick the desired working point.. so from the first graph in the datasheet i had picked:
Ea = 240v
Ia = 17.5mA
Eg = -11v

The next step was to draw a tangent to the RA curves to calc the Ra. Ra = (300-215) / 41 = 2070 ohms
Next is to calc the RLa-a is:

RL min = 3.2 * (240v / 78ma) = 9846 ohms
so
RLa-a = 10000 ohms

Next is to draw the load lines for one half of the pushpull circuit:
1/4 RLa-a - 2000 ohms
Ea at working point = 240 / 2000 = 12mA
Ea minimum = 45v
Ea Swing Pk = 240-45 = 195v PK

Finally calculate AB1 power output using 2 x (Ea - Ea min) squared / RLa-a.  :

2 * (240 - 45) Squared / 10000 = 7.605W

I think that is all right? just want someone to confirm my maths and method really:)

So in a fixed bias power amp design such as the AX84.com 2watt:
http://www.ax84.com/static/corepoweramps/2W_PP/AX84_2W_PP_Poweramp_Schematic.pdf

I should have 240v on B+2 going into the transformers primary centertap and a bias of 17.5mA on each tube to achieve an output of 7.6W which should be around -11v.

Cheers All!!! My head is now going to explode!

Craig

[morcey2]

Hi Guys,

Im trying to up my tube knowledge a bit at the moment and want to build a low power poweramp out of 6N6P russian tubes.

Tube Datasheet - www.lampizator.eu/UPGRADE/RU-6N6P.pdf‎

Iv been trying to figure out the math and loadline techniques for a Push Pull amp with a fixed bias.

I found this guide which is quite detailed but seems to skip over certain steps:
http://www.turneraudio.com.au/loadmatch3-pp-triodes.htm

I was hoping someone with more knowledge than me can look over what im doing and point me in the right direction:)

Anyways i have been following the guide DESIGN METHOD, PP CLASS AB1, EH6550 IN TRIODE MODE from the above website and converting it to the 6N6P datasheet.

In step 1 its pick the desired working point.. so from the first graph in the datasheet i had picked:
Ea = 240v
Ia = 17.5mA
Eg = -11v

The next step was to draw a tangent to the RA curves to calc the Ra. Ra = (300-215) / 41 = 2070 ohms
Next is to calc the RLa-a is:

RL min = 3.2 * (240v / 78ma) = 9846 ohms
so
RLa-a = 10000 ohms

Next is to draw the load lines for one half of the pushpull circuit:
1/4 RLa-a - 2000 ohms

1/4 * 10000 = 2500 Ohms

Ea at working point = 240 / 2000 = 120mA

240v / 2500 Ohm = 96mA,  way over the current limit according to the datasheet.

Ea minimum = 45v
Ea Swing Pk = 240-45 = 195v PK

Finally calculate AB1 power output using 2 x (Ea - Ea min) squared / RLa-a.  :

2 * (240 - 45) Squared / 10000 = 7.605W

I think that is all right? just want someone to confirm my maths and method really:)

So in a fixed bias power amp design such as the AX84.com 2watt:
http://www.ax84.com/static/corepoweramps/2W_PP/AX84_2W_PP_Poweramp_Schematic.pdf

I should have 240v on B+2 going into the transformers primary centertap and a bias of 17.5mA on each tube to achieve an output of 7.6W which should be around -11v.

Cheers All!!! My head is now going to explode!

Craig

I think you need to pick a much higher load, like 16k or even 20k Ohm for the RLa-a, and possibly a lower quiescent voltage than 240V.  The triode characteristics graph in the datasheet leaves much to be desired in terms of doing load-lines for something like this.  It's been a couple of years since I've done loadlines, but I'm starting to remember things.   The guys over at AX84.com use ECC99s, which are very similar characteristics-wise to the 6N6P with a different pinout, into 22.5K loads and it's supposed to sound good. 

Matt

[craigmillard]

Cheers for the response Matt

Bit of a maths blunder there! lol

When ever i increase the RLa-a i cannot get the wattage up. Say for a 20000 rla-a at 200V i get a power output of around 1.2.. How can i get the wattage up close to its maximum of 4.8w per triode???

[craigmillard]

Ea at working point = 240 / 2000 = 120mA

Also does the mA matter at this point as it is only to work out the Ea minimum to calc the Ea peak?

im confused lol

[psychedelicfish]

I suggest you have a look at Merlin's Website. There's lots of guides on how to design various tube amp stages, including push pull output stages. While his push pull guide is on pentodes, most of the information applies to triodes as well, with the exception of anything to do with screen grids.

You should also take a look at Doug Hammond's Dragonfly, which uses ECC99s, which I believe are similar/identical to your 6N6Ps. Note that he's using quite a high primary impedance. If you can't find a suitable transformer (fender reverb transformers are good, IIRC) you might be able to find a mains transformer that will do the trick. If you can find a mains transformer with two 120V primaries and one 5V secondary rated at 10VA or so, that might work.

The 4.8W per triode is talking about heat, not audio output. If you've got 200V on the anode, and 20mA of anode current, your anode dissipation will be 200*0.02=4W. If you've got 300V on the anode and you're drawing 30mA, the anode dissipation will be 300*0.03=9W, which is over the limit. When I design output stages, one of the first things I do after printing off the curves (if it's not already done) is to draw a line for the max anode dissipation. I do this by making a dot above 100V, 200V 300V e.t.c. and join the dots with a smooth curve. The load lines cannot cross this curve at any point, or the valve may overheat (guitar amps frequently go over max dissipation and other ratings, but this is only done with common guitar amp valves).

What Matt was talking about with the current limit was the max cathode current. If you have 10mA current flowing into your anode, there's going to be 10mA coming out of your cathode (if we're using conventional current and talking about triodes). The max cathode current on the ECC99 datasheet is 60mA. This means that the anode current shouldn't exceed 60mA.

[craigmillard]

Ok thanks for the tips again guys! I think im getting it a bit more now, that other site i was following was confusing me more through bad english!!
I had seen merlins site before but for some reason the info didnt sink in until id messed around with the other way! I have both merlins books and wish there was a poweramp version too!!

So here we go again:

If i have a 6N6P with the following:
Ea of 240V
RLa-a of 22500

that will give a load line with max I as 240/5625 = 42mA so i have drawn from 240v to 42mA on the graph..
Then i push the line up using the same gradient so that at no point does it cross the max dissipation line to about 70-80%.
Then i take the mA level at where it cuts the 0v curve which is 38mA and the V which is 100v.
The peak V is therefore 240-100 =140v swing and the output power being 140 * 0.038 /2 = 2.66W
To get this output i will fix bias it at 15mA or -11v at idle..

How does that sound? Hopefully closer to the mark?

Cheers all

[psychedelicfish]

I found the guide you linked to very confusing. I suggest you use the Valve wizard guide I linked because it's much easier to follow. It does concern pentodes, but the article applies to triodes too, except you can ignore all the stuff about screen grids.

You've got the first pat right, in that you draw a line for 1/4 of your RLa-a. That is the class B part of the line, where only one valve is "on". You then draw the class A part of the line, where both valves are "on", which is 1/2 the RLa-a. The class A line is the one you slide up and down to get the right bias.

[PRR]

> low power poweramp... a Push Pull amp with a fixed bias.

WHY? Fix-bias is for getting more power out of given tubes. 6N6P is not a small tube. You may be able to hit a reasonable spec in Self-bias. And when-ever possible, use self-bias. It gives less trouble.

> Ea = 240v  Ia = 17.5mA

Slight err here. 240V*17.5mA is 4.2 watts. The Max Rating is 4.8W for one unit but 8W for both... 4W each when both are used hard.

I had an answer before I saw you had one worked-out.

For "best" power from triodes, run the plate voltage to the limit, swing down 2/3rd of the supply voltage. So 300V supply, swing down to 100V. Find where the zero-grid line crosses 100V, for this tube it is 37mA. Your per-plate loadline is 200V/37mA or 5,333 ohms. 20.6K plate-to-plate nominal. Your peak power is 200V*37mA or 7.4W; sine power is half that or 3.7 Watts.

With fix-bias you may set the idle current anywhere from dead-zero to the dissipation rating. Dead-zero idle bias (you can't really get there) means huge grid swings and large distortion. Too high idle means the "off" tube doesn't turn-off and is a load on the "on" tube. It is reasonable to try half the peak current, pure class A. That was 37mA so try 18mA. That makes 5.4W idle heat, too much. We can go 13mA.

If we have 37mA peak the full-load Sine current is 26mA/pair. If we are limited to 13mA per tube, the idle current is 26mA/pair. The total cathode current does not change from silent to blasting. Resistor bias will work. The grid bias for 13mA 300V is off the curves but I estimate -15V. 15V/26mA= 577 ohms (0.4W so use a 1W part).

> draw a tangent to the RA curves

What, make load equal to plate resistance? Yes, you can do that, and in push-pull the THD does not rise much. In this tube at these voltages the loadline seems to cross the Pdiss curve. This is usually quite OK for audio, plate won't melt in a half-cycle of 40Hz.

Big issue I see: the higher the transformer impedance the more it costs and the worse the treble response. That's theory; in practice today our choices are very slim. I say 30V and 20K, you say 10K 240V, and it may come down to what you can actually buy, not what your pencil tells you.    

I do not get your 7.6 Watt output number. I won't beat my head about it because either 3.7W or 7.6W will probably suit the goal of "low power". I think you could be buying parts and sort it out in the build.

I might point out that triodes are not rock-and-roll (too polite), that two 6AQ5 can be had a buck cheaper than 6N6P, can for-sure do the 8 Watts at 250V or be starved below 1 Watt at 100V. 6AQ5 attempts to be a small/cheap 6V6, and 6V6 have been used in popular guitar amps.

[psychedelicfish]

+1 on 6AQ5s, they sound great and are way cheaper than any of the popular guitar amp valves. Sockets for them can be a pain though. 6P1P is a russian near equivalent with a common 9 pin base.

[gtudoran]

6N6P is a very nice valve - 4w per section is not a small number for a small tube. I've designed an amp around this tube and 6n2p in preamp section with  concertina phase splitter (don't use it with a self-split output section because it will not give you all it can)

The schematic is this http://i479.photobucket.com/albums/rr153/gtudoran/6n6pamp-complete.jpg i've used a negative bias voltage for the grids ~8v but, i've used a trimmer so i could play with different values. With this tube you will have a very pleasant sound with a nice power section overdrive if you crank it up (on a 12" Celestion speaker with a SPL of 98db/m ... let's say it's pretty  loud)

And here you have some sound clips: https://soundcloud.com/dass101/demo-amp-analog-sound

The output transformer is a 6k to 8 and 4 ohm (nothing fancy). If you have other questions, i more then glad to help you.

Best regards,
Gabriel Tudoran
Analog Sound

[kingswayguitar]

i can't add to the technical discussion but the sound clip you posted in nice

[tubegeek]

S. Bench's Tube pages:
http://www.jacmusic.com/techcorner/SBENCH-PAGES/sbench101/

In the index area, note the "Technical Articles" section. There is a 4-part series on loadlines. Bench is the benchmark for internet tube info. It's certainly a benchmark to which Merlin has risen, but the Bench series is also a good way to learn.

[craigmillard]

WOW! Cheers Guys thats a ton of information!:)

As this is all really just a learning exercise for me i want to get it right!

Thanks for the Info PRR that is great! I know triodes arnt very rock and roll but wanted to give this one a go:) Good tip on the 6p1p though psychedelicfish will give them a go next! I find it easier to source russian tubes from europe than alot of the other ones.. More reliable sources etc..

Tubegeek that link is very informative will have another read over the weekend!:)

I had another go with Merlins way using a 22500 transformer and got the following:

If i have a 6N6P with the following:
Ea of 240V
RLa-a of 22500

that will give a load line with max I as 240/5625 = 42mA so i have drawn from 240v to 42mA on the graph - the class B load line..
Then i calc a 1/2 load line for class a =21mA and have drawn from 240v to 21mA.
Then i push the class a line up using the same gradient so that at no point does it cross the max dissipation line to about 70-80%.
The load line is then the comp of both of these..
So i take the mA level at where it cuts the 0v curve which is 27.5mA and the V which is 80v.
The peak V is therefore 240-80 =160v swing and the output power being 160 * 0.0275 /2 = 2.2W
To get this output i will fix bias it at 14mA or -11.5v at idle..

Hopefully that is closer to the mark.. At no point is it above the max dissipation..

?

[psychedelicfish]

That looks much better

[PRR]

> output power being 160 * 0.0275 /2 = 2.2W

Consider 300V B+, 22K CT load, 200V 36mA or 3.6W out, all below the 4W Pdiss line.



The idle-point could be near 300V 12mA/tube, 24mA/pair. With full Sine excitation the peak current of 36mA is 24mA average, so resistor-bias will work. About 15V bias, so the final spec is 315V B+.

However I -do- favor NOT running TV Tuner tubes at FULL rating continuously. In TV use, "any" watchable signal will hit the AGC and turn-down the current in the tuner. While these tubes will survive a night after the station goes off the air, they may not give 1,000-5,000 hours of satisfaction *at* 300V 4W/plate, while they may run 10,000 hours at a turn-down (watchable signal) bias nearer 2W/plate. A 240V scheme is more conservative. (OTOH, how many hours do you really want to run this amp? If it is an experiment that turns out dull, 100 hours may be excessive. OTOH if it turns out wonderful and you play it constantly, tube cost of $18/1000hrs or 2 cents an hour may be unimportant.)