6SK7 tube amp circuit

[Learner]

Hi guys,
here are the pics of my circuit....

Output with no FB, input is 2V p-p. output is 5V increment on the CRO roughly 15V p-p.
This is the cleanest signal I can get without feedback.

Output with FB, for some reason even the input is hairy. Must be contaminated from some where???

Noise with no FB, at 1V increment on the CRO.

Noise with FB, at 50mV increment on the CRO.

I suspect the noise is from my PSU also because of the type of the tube that picks this up, since I don't seem to have this problem with 6SN7 tubes..... sigh..... :?

I also get slightly different result from my first 6SK7 circuit, will post some pics shortly.

 

[pstamler]

Unless I miss my guess, your tube is oscillating.

Peace,
Paul

 

[Learner]

Hi P,
What can I do to get rid of this osillation???

Thanks! :thumb:

 

[bcarso]

Do you have the 100 ohm stopper R in series with the grid as recommended by Paul S.? If it's there try making it a bit bigger (2-300)---you have a pretty relaxed layout there.

You could also be getting some external noise pickup with things strung around like that and no shielding/ground planes etc.

 

[Learner]

Ok, back with some more pics.

Here is the update of the circuit...., have placed a 100K potentiometer in place of the stopper R.(between the grid and the grid shunting R, right Paul & B???)

The output is at 5V increment. With the "added" stopper R, it seems to have stopped the oscillation. Eventhough the pot is completely shorted out :shock: :shock:

The stopper R looks like a great tool for even harmonic distortion control??? :shock: :shock:
That is at close to 100K ohms, using the potentiometer.

Finally the noise....at 1V increment.

As you can see, the noise is close to 2V !!!!!!!

sigh...... :sad:

Tried various FB value with a 1 meg pot(with green wires) as seen in the pic, it doesn't help with noise but just attenuate the output.

bah!

 

[PRR]

That's not "noise". It has a beat, a regular repetition. Stabilize the trace and find its frequency.

> potentiometer in place of the stopper R

No. You have TOO MANY FLYING WIRES already, each of them transmitting and receiving signal to each other. A grid stopper is SMALL, FIXED, and VERY CLOSE to the grid pin.

 

[bcarso]

That last waveform is probably noise pickup. Noise is random---that waveform is highly periodic. Do you have some strong interference source in your lab? Fluorescents maybe nearby?

 

[Learner]

my PSU has a fan and I believe that is the most likely source of that frequency, I will try to enlarge the trace to find the frequency.

 

[analag]

Post the schematic for this setup you have running now. I bet you that tube is no good. I bet you if you change it your problem will disappear. Never build a circuit around one sample of a certain part (6SK7), always have a couple on hand. Sometimes a little experience will kick theory's ass in a real world scenario.

Analag

 

[Learner]

[quote author="analag"]Post the schematic for this setup you have running now. I bet you that tube is no good. I bet you if you change it your problem will disappear. Never build a circuit around one sample of a certain part (6SK7), always have a couple on hand.[/quote]

I am playing around with 3 tubes at the moment will try more tubes to see if this is consistant.

Sometimes a little experience will kick theory's ass in a real world scenario.

Analag

ha.....that's the impression I got, theory and analysis is only good to use as a point to start tweakin. The result of the circuit in the real world is often somewhat different, since analysis and theoretical prediction does not include every single factor that exits in the real world condition which'll contribute to the outcome.

There are always "new" problems that'll arise during constrcution, simply because of fact that you did not know it was going to be a problem until it is a problem. :sad:

 

[analag]

Connect that sucker like a triode and see what happens, easier to isolated the bad part that way. Do not use a cathode bypass cap yet. I've had designs where the electrolytic made the circuit go crazy....as recently as last month. Start with the simplest configuration (triode) then work your way up into full pentode operation. Check the heater circuit too, use DC. No AC near the control grid, move the board around and see if the noise level fluctuates, use a 1k as the grid leak resistor if that makes it go quiet then it's a layout issue. Most of all KNOW that you can fix it and that you WILL fix it. Hang in there, this is a great learning experience.

Analag

 

[Learner]

[quote author="analag"] Most of all KNOW that you can fix it and that you WILL fix it. Hang in there, this is a great learning experience.

Analag[/quote]

HEHEHE...... :green:

I don't give up that easily, in fact I have barely just begun!!!! :shock:

I have done this before but just not properly because I had to use full feedback to get it to work, so I've decided to get it right and do it the "correct" way; follow the data sheet plot the load line, calculate the bias .....the whole 9 yards......

Anyway, 2 hours of tweakin could've done the job but it took me nearly a whole week :roll:

Here is the full working bias setting.....

Here is the circuit again, all I did was added another 1 meg variable resistor at the plate and tweak away..... should of done this on the first day!!! sigh...

Here is the output, input = 0.4V p-p(0.2V increment). Output is 12V p-p(2V increment), gain of 30 or 29.5dB whatever....

Finally the noise!~ At 50mV per step increment , the noise swing just within 1 step.

You might be able to tweak it a little bit more but I am pretty certain not much better before it gets much worse.

This is one helleva lesson I won't forget, tried to follow the data sheet thinking that I can bias the tube at around 5 mA to 8.5mA and get at least like 5 mA peak swing. Went from 3 mA up to 13 mA back and forth stuck with the gorilla hair ball signal for days :roll:

Finally decided to abandon the whole theory/prediction method and just stick a 1 meg pot in and start tweakin, never would of thought that I would settled the bias at 0.6mA!!! :shock: :shock:

As the current increase, the noise just gets astronomical and I was thinking of using this in parallel with 3 more tubes to drive some low Z load.....

Looks like its only good for voltage amplifier, with that tiny amount of current.

Case closed, will move onto next tube biasing won't fall for the data sheet plot curve again! . :razz:


edited for schematic drawing error...

 

[vibrolax]

Paralleling tubes to obtain lower output impedance is effective when the plate resistance is pretty low to begin with, i.e. <2K ohms. The plate resistance of a pentode is typically several hundred K ohms to a few megohms. To drive a low impedance load, we either use an impedance matching transformer, and/or add an additional tube (or MOSFET) wired as a cathode follower.

-Jon

 

[Learner]

[quote author="vibrolax"] To drive a low impedance load, we either use an impedance matching transformer, and/or add an additional tube (or MOSFET) wired as a cathode follower.

-Jon[/quote]

I am thinking of using a MOSFET as a cathode follower but have not been able to find a schematic of that type of circuit, or may be just a fat BJT or a darlington transistor got any suggestions?

Or may be something like this but making a high voltage dual regulated PSU for this could be bit of a hassle....

or may be something like this....

 

[BYacey]

Nice scope. Is that a V1065 or V665? I have the 1065 and it was some of the best money I ever spent on test gear.

 

[PRR]

> settled the bias at 0.6mA

You aint running 0.6mA from 250V through that new 410K resistor. 0.6mA*410K= 246V, leaving 250V-246V= 4V to run the circuit. Yes, you can get a faint signal through a tube dropping 4V, but you show 12V p-p signal! Something don't add up.

> the gorilla hair ball signal

Of course. The breadboard looks like a hairball. As we hinted several times, your hay-wire breadboard and stray capacitance ensures that it will want to oscillate at some high frequency. Especially with the high power gain of a good RF tube like 6SK7 working at 5mA. By brute-force cutting current to 0.6mA, what you did was starve the tube so bad it won't oscillate (or pass signal) at high frequency.

You also don't get gain of 30 with that unnecessary 119K feedback resistor and the 1Meg pot at 50% as the schematic says. I assume you had the pot full-up so it saw the low-impedance signal generator. OK, but the 10 ohm cathode resistor ensures that even small inputs will drive the grid positive and suck current. With a low-Z source, that can happen. With a high-Z source, it will clip madly.

GET RID OF ALL YOUR POTS AND HAY-WIRING!!! Lose the 410K. Lose the 119K. Use 100K plate resistor, often a good trial value. Screen grid resistor can be 470K, screen grid cap has to be like 0.1uFd for audio. Use 2K7 cathode resistor, no bypass for now. Have a >10uFd cap on the B+, within 6 inches of the tube. Keep everything tight! It should all fit in less than 2"x2"!

In this condition: plate current will be around 1mA (let's leave higher current and higher gain for until you master neat stable no-hairball breadboarding). Gain will be about 20, we can fiddle that later. Maximum output into an o'scope should be near 100V p-p, maximum input near 5V p-p. THD at 1V level is about 0.1%.

 

[analag]

This is the schematic I suggested earlier in the post, it is quite similar to what you propose. Minus the cathode bypass and it's what you got, pretty much. Maybe our friend just like to do his own thing, I think that's a very strong possibility.

Analag

 

[Learner]

[quote author="PRR"]> settled the bias at 0.6mA

You aint running 0.6mA from 250V through that new 410K resistor. 0.6mA*410K= 246V, leaving 250V-246V= 4V to run the circuit. Yes, you can get a faint signal through a tube dropping 4V, but you show 12V p-p signal! Something don't add up. [/quote]

Just measured the voltage at the plate, it comes to 19.2V. ??

the 1Meg pot at 50% as the schematic says. I assume you had the pot full-up so it saw the low-impedance signal generator. OK, but the 10 ohm cathode resistor ensures that even small inputs will drive the grid positive and suck current. With a low-Z source, that can happen. With a high-Z source, it will clip madly.

Right, the pot is on full. Sorry about that.

GET RID OF ALL YOUR POTS AND HAY-WIRING!!! Lose the 410K. Lose the 119K. Use 100K plate resistor, often a good trial value. Screen grid resistor can be 470K, screen grid cap has to be like 0.1uFd for audio. Use 2K7 cathode resistor, no bypass for now. Have a >10uFd cap on the B+, within 6 inches of the tube. Keep everything tight! It should all fit in less than 2"x2"!

In this condition: plate current will be around 1mA (let's leave higher current and higher gain for until you master neat stable no-hairball breadboarding). Gain will be about 20, we can fiddle that later. Maximum output into an o'scope should be near 100V p-p, maximum input near 5V p-p. THD at 1V level is about 0.1%.

I realized the error in the first schematic where I connected the 1 meg pot in series with plate and screen resistor, here is the corrected schematic( I didn't think anyone would be interested) thanks for pulling me up on that one PRR! :thumb:

Got 20V p-p output with 0.4V p-p input, gain of 50. Plate voltage is 35V and screen voltage is 16V.

 

[Learner]

[quote author="analag"]

This is the schematic I suggested earlier in the post, it is quite similar to what you propose. Minus the cathode bypass and it's what you got, pretty much. [/quote]

Hi Analag,
Sorry about the delay, I have tested the value in your sim, biased at 2.45mA with input of 0.4V I got 15V output with FB resistor of 61K. The difference between the 2 bias setting I have is that the noise is a bit higher, I think its because of the bias current is higher too.

The simulators result works pretty well comparing to plotting load lines on the curve chart.

Maybe our friend just like to do his own thing, I think that's a very strong possibility.
Analag

I do like to do my own thing, however I love suggestions/advise equally as much simply because it provides more option/alternatives for me to compare. :grin: :thumb:

 

[pstamler]

[quote author="Learner"]Sorry about the delay, I have tested the value in your sim, biased at 2.45mA with input of 0.4V I got 15V output with FB resistor of 61K. The difference between the 2 bias setting I have is that the noise is a bit higher, I think its because of the bias current is higher too.[/quote]

No, it's because the value of the cathode resistor is higher, and it generates more Johnson noise. Depending on your application, this may or may not make any difference.

The problem with your circuit, aside from its susceptibility to oscillation, is that you're using a teeny-tiny cathode resistor, and it's pulling enough current to haul the plate down to 35V. Ideally the plate should sit someplace about halfway between B+ and zero; in this circuit that would be about 125V. Not an absolute hard-and-fast rule, but a good starting place.

Try doing what PRR said; use his values, get rid of the feedback resustor and pots, and put in a plate stopper (100R to 1k) right next to the grid. Do this on the bench, not in a simulator. Find the clipping point; it will be way higher than what you have now. And with careful wiring (short leads, grid stopper) you won't have oscillation problems.

Peace,
Paul