One-Bottle Preamp

[Learner]

[quote author="CJ"]The 417A is an entirely different animal than most audio preamp tubes. It was designed as an IF tube in a microwave reciever. (70 mHZ) It uses a frame grid instaad of the usual posts, which means you can get the grid a lot closer to the cathode, which is why the Gm is so high. [/quote]

Any other similar tube you can suggest that is cheap to get hold of? I wonder what's the difference between this tube and beam tubes like 12A6 and 6AG7 which both have hi Gm as well, I wonder what is the reason that I can not get a good result when I bias the tube with higher current???? :sad:

 

[Learner]

[quote author="PRR"]
I'm sure learner's troubles are due to not keeping plate voltage around half of supply voltage, or not watching the grid bias. Many-mA operation of that tube has little advantage for audio voltage gain, but it sure is possible (see the H-P 200AB which has a similar pentode working at fairly high current, or many oscilloscopes).[/quote]

Is there any particular reason that the plate voltage should be at half the supply voltage?

I use 1K ohm cathode resistor as my grid bias, since when I place a negative voltage at the grid it spikes the 1khz sine input signal. It is not common to see cathode resistor bigger than 1K but may be I should try it to see how it differs, after all its the gain ,noise and distortion at output that counts.....

 

[pstamler]

[quote author="Learner"]hey CJ,
What exactly is the supply voltage?? 100V or 200V? and your bias current is 11mA!?? :shock: hmmm...... I would love to see the output on the CRO with a signal running through it, also the noise figure with closed circuit.

What is the gain of this tube?? I would not think it will be too high if the current is at 11mA.[/quote]

His gain was 10x, or 20dB, about the same as a 12AU7 or 6SN7. His supply voltage was 150 for 11mA.

Same case at the output, I have a resistor connected in series with the output coupling cap. This resistor kills both noise and gain, so its a compromise. Without it and with no FB for the tube, I get a sine wave like sort of noise at the output and its loud. I forgot to measure its frequency.....

Oscillation. Too long leads, too much junk around the tube allowing capacitative coupling from output to input.

How did you work out the bias, using load curve? Interested in replacing the plate resistor with a 500K ohm pot then run a 1khz signal through the tube, probe the plate coupling cap and tweak the plate resistor pot. Watch it on the CRO and see what value of the pot you arrive at to get optimum gain and minimum noise, you might want to add another 500K pot for FB as well, since it will help to flatten the noise you'll see it on the CRO!

That 500k pot instead of a plate resistor, and all the other high-impedance stuff around the tube, is causing the "noise" which is really oscillation. Read the earlier posts -- this looks like it's a pretty low-distortion tube.

I don't have cathode decoupling cap on any of my tubes because I found that it added noise, may be I didn't wait long enough until the noise settle down a bit after turning on the circuit. I might try that again.....

I also tried all different coupling caps and could hear NO difference, the only thing I find is that the biasing point is absolute critical to gain, noise and distortion which to me pretty much defines the sound. There is a optimum region for the value of plate R and FB R and screen R(in case of the Pentode), where by combining these 3 variables you can achieve maximum gain with minimal noise in comparison to every other combination of the 3 variables. I only tweak for max gain and min noise because there isn't much I can do to alter the distortion plus its really minimal, can not really see it on the CRO also to me it defines the personality of the tube as well as the characteristic of the sound.

The "noise" you're getting is oscillation. The actual noise generated by a tube circuit comes from:

1) Johnson noise (thermal) generated by the source impedance
2) ditto generated by the cathode resistor if it's unbypassed
3) ditto generated by the plate resistor in parallel with the load impedance, all divided by the square of the stage's gain
4) The noise generated by the tube itself; in a triode this is approximately 2.5/gm, where the gm (transconductance) is expressed in mho -- excuse me, Siemenses.
5) Hum from various sources

Really, try rigging up a really tight layout with a by-the-book plate and cathode resistor, and see what you get. No feedback resistor. I think you'll be surprised and pleased by the results.

Peace,
Paul

 

[Guest]

I believe Steve Bench did an amp with parallel 417 a few years ago. He ended up using a seperate grid stopper on each grid and just tying them all together at the input. What did you do with your 4 grid pins?

 

[pstamler]

[quote author="Learner"][quote author="PRR"]
I'm sure learner's troubles are due to not keeping plate voltage around half of supply voltage, or not watching the grid bias. Many-mA operation of that tube has little advantage for audio voltage gain, but it sure is possible (see the H-P 200AB which has a similar pentode working at fairly high current, or many oscilloscopes).[/quote]

Is there any particular reason that the plate voltage should be at half the supply voltage?[/quote]

By having the plate sit at about half of the supply voltage, you maximize the possible voltage and current swing. Do this thought-experiment: imagine the supply voltage at 250V, and the plate at 249V. Clearly the plate can swing plenty in a negative direction, but it can only go 1V in a positive direction. Similarly, if the plate is sitting too low, it will cheerfully go higher, but can't go lower without running into problems with grid current. Setting about halfway down minimizes both problems.

Peace,
Paul

 

[PRR]

> Is there any particular reason that the plate voltage should be at half the supply voltage?

ABOUT half.

For the reason pstamler said.

"Half" is very approximate. 1/3rd, 2/3rd, much the same.

If you need maximum undistorted output, some reasonable approximations suggest biasing the plate to about 0.65 of supply voltage. This is a good starting point for power-amp drivers for output triodes or pentodes with local feedback (ultralinear or cathode winding).

If you want maximum gain into a very high impedance load, use a very high-value plate resistor and bias down to about 25% of supply voltage. You see this in direct-coupled volt-amps driving cathodynes: the large plate resistor drop maximizes gain, and a cathodyne likes its input near 20%-25% of supply voltage.

With typical-value plate resistors, 20% of supply voltage is suspiciously low; 70% is suspiciously high. Until you find some reason to do otherwise, aim for 1/3rd-2/3rd.

With a Triode, if you picked the plate resistor appropriately (more than twice the book-value plate resistance), then fiddling the cathode resistor until plate is 1/3-2/3 of B+ will give a workable operating point every time.

> It is not common to see cathode resistor bigger than 1K

I see it all the time. Not so much in data-sheets (which are boast-sheets about the tube's hardest work) but in the suggested Resistance-Coupled Amplifier charts.

Your SK7s should bias-up very much like SJ7s, use this chart: http://www.triodeel.com/rca_c20.gif

Note that a pentode can make voltage gain over 100. You keep finding hiss and hairballs, and fooling with feedback and in/out resistors: clearly you don't need big gain. Wire the pentode as triode: triodes really are easier to work with, and were pretty universal in audio design. Use chart http://www.triodeel.com/rca_c11.gif

 

[PRR]

> Will this work?

Only if you don't have anything bigger.

Actually: it looks "too-good", except it isn't clearly made for unbalanced DC. The 4K4 side is center-tapped, suggesting push-pull design. The "60mA" label may apply to push-pull, meaning "heat"; we would like a rating for DC magetization. A 60mA push-pull E-I core transformer may not like more than 12mA unbalanced DC.

But try it. It may be fine. It sure has enough output taps. 600 is your everyday line output. If you want to run a hot input up into distortion, without melting your line input, use the 60 tap. You can try it for loudspeakers at the 4 tap.

> why a 10 uF bypass cap has such a profound effect

Below 300Hz, unbypassed; above 500Hz, bypassed. The gain will rise 5db to 7dB in that area. This is not unlike the core EQ used on lead guitar to get the basic bright sound: it punches the midrange up. I call that an effect, and can probably be done more flexibly in an EQ than by mucking an amplifier.

Try no-cap and with >>100uFd. These two conditions will be audio-flat, but different gain and distortion.

> Isn't this a fairly low output impedance circuit, even without a transformer?

Tube is around 2K, so with the 4K you see about 1.3K at the output. "Low" compared to typical 10K board inputs and modern gear. "High-ish" for driving true 600 ohm loads or multiple 10K loads without interaction. With the transformer, voltage is less but it is at 200 ohms, loading does not matter.

> 20 ma thru a 4.7k plate resistor seems like it would pull a freight...

It will just about make 11V peak in a 600 ohm load, +20dBm. And to get there, you have to flap the 417 to extreme current, several-percent distortion.

To do the same +20dBm with a 6K:600 transformer, you only need +/-6mA variations around the 20mA standing current. Distortion is less.

 

[CJ]

Thomas, I used just one lead for the grid connection. I figured the 4 pin grid setup was for 70 mHz operation.

I hooked up that WE output transformer and it sounded prettty darn good. Lost a little high end, but the sound was smoother. I bet a good transformer would really make this a great sounding preamp. Maybe an A/B switch for tranformer or cap coupled would be slick. That way, if you wanted more detail....

I tried using different value cathode resistors. This changes the grid bias. Lowering the resistor lowers the bias, which increases the plate currrent a bit.
Amplification rises with a decrease in the cathode resistor. Amplification stays about the same with a decrease in plate voltage, so why not run this thing down around 150 B+ for less noise, as headroom is not rerally a problem in the first stage of a mic pre.

I tried a 500uF bypass cap again, and like the results. I am thinking a rotary switch with different value bypass caps might be the way to go for a cool tone control.

I have another 417A, so I think I will build another channel and bolt this thing up.

Yes, the DuKane is about a 1:14, my mistake. 150:30K ohms It has a center tapped pri, but there is no need to use it, as this thing has more than enough gain with a 1:14 transformer.

So this thing likes:

150 vdc B+
4.7 K - 2 W. Plate resistor
65 Ohm - 2 W Cathode resistor
1.5 uF - 400 vdc coupling cap
500 uF - 16 V bypass cap
And possibly an output transaformer to be named later.

Thanks to everyone for the help, and Eric Barbour for the idea.

cj

 

[Learner]

[quote author="pstamler"]

Really, try rigging up a really tight layout with a by-the-book plate and cathode resistor, and see what you get. No feedback resistor. I think you'll be surprised and pleased by the results.

Peace,
Paul[/quote]

Hi Paul,
I have taken your advise and decided to give it another try, after all the method has been applied by engineers for decades without a glitch and they can't all be wrong can they? :wink:

I have rebiased the 6SN7 with great result using the loading curve method, eventhough it is much easier to do than pentodes but I'll apply the method with all my other tube and see how it goes. The problem that I was having was that I used to probe the plate when its unloaded and I got a huge sine wave like curve, however as soon as I connected it to my power amp it disappeared and the noise was down to <50mV. I never bothered to connect the output to the power amp due to prior experiences of being blasted/shocked by the noise :shock: :roll: , which lead me to biasing on the CRO using FB and make sure that the output noise was less than 50mV with a open circuit output. That's the reason why it has led to me such a low biasing point of current, I am glad that I have taken a right step toward doing this correctly.

A BIG THANK YOU AND TO PRR SO MUCH for your patience and persistence in trying to show me the right way, donno what I'll do without you guys!

What else can I say, a HUGE RESPECT out to you EE ol'timers.
:thumb: :guinness: :guinness:

 

[Learner]

[quote author="CJ"]
I hooked up that WE output transformer and it sounded prettty darn good. Lost a little high end, but the sound was smoother. I bet a good transformer would really make this a great sounding preamp.
cj[/quote]

Hey CJ,
Great to know that you are pleased with the result, just wanna say don't mind me about my STUPID biasing thoughts as I still need a lot of practice on it. I just feel like an idiot going around in loops..... :roll:

 

[NewYorkDave]

(EDIT)

 

[Svart]

ok so i'm such a noob when it comes to valves.. how would you control the volume? would you put a pot between the trafo and the tube?

 

[NewYorkDave]

The variable gain, along with the switchable pad, give a pretty wide range of control. But if the output is still too hot with the gain set to minimum, a better way to control the level would be with switchable pads on the input side of the transformer, since if the signal's that hot in the first place it's probably overdriving the transformer. I don't like the idea of putting a pot between the transformer and the tube grid since--besides the fact that it will do nothing to avoid transformer saturation--it also changes the capacitive loading on the transformer secondary as you turn the pot up and down. That's because you're adding variable resistance in series with the input capacitance of the tube.

 

[Svart]

gotcha

 

[Svart]

also.. since i am still getting ready to order the parts for this.. I would really only need a 1:1 PSU trafo and rectifier for the 150V correct? How cruicial is the ripple on the B+?



I am looking for the 417A. found some but they are kinda high priced.. what is a "normal" price?
:thumb:

 

[NewYorkDave]

Svart, are you asking about my preamp or the other one? This thread has become a bit confusing now.

 

[Svart]

I was asking about CJ's version but I'll be building your version as well!

as I am totally new to valves i'll be having many more dumb questions soon..

Also, Learner, CJ and everyone else.. I am interested in this "tingly" sound you are getting.. why does that happen? is it pleasant? something to be used?

EDIT: also looking back over everything, CJ what are you using for heater voltage? DC, AC?

also, what is a decent modern trafo for these applications.. of course something both available and affordable.. :green:

 

[Svart]

still wondering..

 

[CJ]

No transformers required for that 417 circuit.

Dig around some tube sites for cheap WeCo's.

Hmmmm, maybe a tube source Meta, but people are reluctanct to post their secret stash sources.

Ringing was resonace of the glass/elements. Not a big deal if tube is shock mounted.

 

[pstamler]

[quote author="NewYorkDave"]The variable gain, along with the switchable pad, give a pretty wide range of control. But if the output is still too hot with the gain set to minimum, a better way to control the level would be with switchable pads on the input side of the transformer, since if the signal's that hot in the first place it's probably overdriving the transformer. I don't like the idea of putting a pot between the transformer and the tube grid since--besides the fact that it will do nothing to avoid transformer saturation--it also changes the capacitive loading on the transformer secondary as you turn the pot up and down. That's because you're adding variable resistance in series with the input capacitance of the tube.[/quote]
Not to mention playing hob with the noise figure. Typically you want to load a transformer with 10x the nominal secondary impedance. So let's say it's a 15k secondary...loaded with a 150k pot. With the pot up (effectively out of circuit) the tube sees a source impedance of about 13.6k, with Johnson noise to match (~2.1uV in a 20kHz bandwidth). Turn the pot down 5.2dB and suddenly the source impedance is about 41.25k, with Johnson noise of ~3.7uV, about 4.9dB worse.

Peace,
Paul