Take Five Minutes And Help the Tube Noobs -First Tube Stuff

:grin: What should I do with this thing?

I've repaired tube amps but never have designed before. This is going to be my first testbed for tube stuff. It's an old TV set I pulled off of the curb in front of someone's house before the garbage collector got there. :guinness: It has what seems to be a raucusly large power supply, lots of tubes and sockets, and oodles of giant wattage resistors and caps. I think the picture tube was blown, cuz everything else seems to work fine.


I'm just learning tube theory, using FET & bipolar (not manic depression) theory as a springboard. I know transistor theory pretty decently, excluding current sinks, on which I haven't found a good resource for study yet.

Where do I go from here? I'm interested in class A preamplifiers almost exclusively. Mostly mics & microphone signal chains. I also have waaaay too many audio transformers... :wink:

I have just drawn up my first basic tube preamp, and I'll probably be able to direct couple it if I want to.



I know there's enough info on this board to keep me going, and I'm almost at information overload every day on the Microphone META META... (Bowing in the direction of PRR and Gyraf)


I found a few links to work from:
http://www.nmr.mgh.harvard.edu/~reese/RC10/rc10.pdf

http://www.tpub.com/content/neets/14178/

http://boozhoundlabs.com/howto/

R1 is way too big for a cathode-biased tube.

C1 is way bigger than it needs to be.

What's the purpose of the 16MHz trap on the output?

6K is a rather low load for that circuit to drive.

The circuit doesn't make any sense, actually. But this falls in line with what you said about "just learning tube theory."

I don't think you're at the point of making a design yet. Start by studying known, working designs, find out what makes them tick, build copies if it helps you to grasp how they work. Build "example circuits" out of old textbooks or RCA tube manuals. Throwing semi-random combinations of components into SPICE is not designing, and it does not bring understanding. Put in the time to learn.

I just replaced the FET with a tube, trying to take into consideration optimum operating current for the tube. The design was my primary sketch for my microphone head preamplifier. R1 has fluctuated from 500k to 50Meg in my simulations. I just thought R1 would determine the loading on the previous stage, and was nearly arbitrary as long as it was referenced to ground.

What effect does erring on the high side with coupling capacitors have? I just always throw something too high in there, and if I am going to actually produce the circuit I might do the calculation...

The trap, when simulated with this circuit, actually cuts at about 100k -3db. That's what my Circuitmaker simulation says anyway... I just thought it had to do with some internal capacitance in parallel...

The 6k load is my generic standard load. I use it for almost everything, assuming that the next stage is going to be built either for a 600R input or 150R input. This standardized control element has helped me to compare designs. Will it significantly affect the Circuitmaker simulation?

Thanks...

[quote author="northsiderap"]I just thought R1 would determine the loading on the previous stage, and was nearly arbitrary as long as it was referenced to ground.[/quote]

I think in this circuit you may or may not run into trouble with a grid leak resistor that high... But do a search of the forum to see why too high a resistor might cause trouble. I'm pretty sure it's been covered before.

[quote author="northsiderap"]What effect does erring on the high side with coupling capacitors have? [/quote]

You might get some low frequency instability. Also, a 200uF cap would usually be an electrolytic, which would be to the detriment of the circuit's sound quality. A 1uF polyester or polypropylene cap will work fine.

[quote author="northsiderap"]The trap, when simulated with this circuit, actually cuts at about 100k -3db. That's what my Circuitmaker simulation says anyway...[/quote]

Looks to me like that trap would work at some MHz frequency... What's the point of that, anyway???

[quote author="northsiderap"]The 6k load is my generic standard load. I use it for almost everything, assuming that the next stage is going to be built either for a 600R input or 150R input.[/quote]

Well, a 12AU7 plate isn't a 600 Ohm source, now is it? It's actually on the order of about ten times that (say 5K to 10K).

[quote author="northsiderap"]Will it significantly affect the Circuitmaker simulation?[/quote]

If it doesn't, the tube model isn't right.

Here's what I would do for starters:

-For simplicity, get rid of R5.
-Make C1=1uF.
-Make R1=1M.
-Raise the supply voltage.
-Raise R3.
-Make C3=220uF.
-Make C2=4.7uF - 10uF.
-Get rid of that resonant trap.
-Make R4=100K or so.

Good luck!

Peace,
Al.

May I suggest that a few long evenings with Bruce Rozenblit's "Beginner's Guide to Tube Audio Design" would pay large dividends?

Peace,
Paul

[quote author="pstamler"]May I suggest that a few long evenings with Bruce Rozenblit's "Beginner's Guide to Tube Audio Design" would pay large dividends?[/quote]

Definitely! And since you seem to know the basics, Morgan Jones' Valve Amplifiers would do you even more good.

And I'm sure RDH4 goes without saying...

Peace,
Al.

[quote author="alk509"][quote author="pstamler"]May I suggest that a few long evenings with Bruce Rozenblit's "Beginner's Guide to Tube Audio Design" would pay large dividends?[/quote]

Definitely! And since you seem to know the basics, Morgan Jones' Valve Amplifiers would do you even more good
[/quote]

Maybe Valve Amplifiers would be useful AFTER BGTTAD but I dunno... I had the Rozenblit book a few years ago, it disappeared while I was away so I bought the Jones book but the math is so effing hard I just don't get anything out of it at all, nothing. With the Rozenblit book I felt like I at least had a chance -Morgan Jones may not be for beginners at all if they are as math impaired as me.

Kiira

[quote author="kiira"]I bought the Jones book but the math is so effing hard I just don't get anything out of it at all, nothing.[/quote]

Kiira,

Jones' book has more math in it than Rozenblit's, but it's really not that hard at all - definitely not any harder than what's on BGTTAD, just basic algebra/arithmetic. You can even skip over the math and still get what the man is trying to say.

If you're "mathematically impaired" (or "dumb", as my high school math teacher would call me!) read it over and over again and try to get it in in small bites. You'll gain an understanding of things that BGTTAD won't give you. Believe me: if I can do it, anyone can.

Stick with it, it's worth it!

Peace,
Al.

It's SO helpful to conquer the math anxiety. Take baby steps by calculating more often. A little bit of basic algebra is so helpful, and unfortunately even that is being taken over by machines now, for the lazy.

The best way if possible is to sit with someone and work through progressively more complex problems, while describing the thought process as you work them so that the misconceptions can be caught and corrected. A classroom situation is a whole different vibe with all kinds of anxiety-provoking stuff---yuk!

Every tube has a MAXimum grid resistor, often 1 Meg on small tubes.

Don't go higher without special care.

Don't go lower without special reason: you will load the previous stage, you may need a bigger coupling cap (caps cost MONEY). At some point, you need to switch from plastic-film to electrolytic caps, which leak too much to be used with typical grid resistors.

Basic triode voltage amp:

Pick a tube.

Look up its data for an operating point similar to what you plan to use (yes, you need some experience). Note that the book values often show-off the extremes of the tube, we often run audio voltage amps at much lower currents than book values.

Pick a plate resistor that is 2 or 3 times higher than the plate resistance. For 12AU7 working at high current, Rp is ~6K so pencil 12K to 18K, say 15K.

Pick a next-stage input impedance 2 to 5 times higher than the plate resistor. That suggests 30K to 75K.

Problem: if the load should be 30K to 75K, we are far-far from driving 600Ω or even 10K loads. On the other hand, if we drive another grid, it is probably 1 Meg. We can load high, but it suggests that a high-current 12AU7 is "too good" for this work. Working 12AU7 at a lower current is cooler and makes power buzz filtering cheaper. Or switching to 12AX7 gives more voltage gain per bottle and per watt of heater power.

12AX7 at 100V, 0.5mA has plate resistance of 80K. 2 or 3 times higher gives plate resistor of say 200K. 0.5mA flowing in 200K is 100V, plus 100V across the tube and 1V cathode bias, gives 201V power supply. And only 0.5mA supply current needed, which means 10-20 times less filter-costs than a high-current 12AU7. Taking the next-stage input as 5 times the plate resistor suggests 1 Meg, which happens to be the maximum grid-leak resistance of many small tubes.

If you want to drive a 600Ω load directly, why are you modeling with 6K??? Use the right value. Don't pretend all loads or all amps are the same: sometimes we do drive heavy loads, and sometimes we aren't.

Working backward: for 600Ω load, we pick a plate resistor 2 to 5 times lower, or 300-100Ω. We pick a tube with plate resistance 2 to 3 times lower than that: 150-33Ω. We in trouble: the only such tubes are monsters. We should also check the power output: when audio-heads say 600Ω we imply +18dBm-+30dBm or 50mW-1000mW. Since a resistance-coupled amplifier won't be even 5% efficient, we need 20 times that much plate dissipation, 1 Watt to 20 Watts. Or we look at peak current: we want 20mA-60mA peak in 600Ω, tube peak current needs to be higher. This is big-bottle territory.

Don't heroically re-invent wheels. The Resistance-Coupled Amplifier table in tube manuals will give you working values. If you are driving grids, these answers are good. There is some fashion for running tubes at higher current, perhaps to swamp stray capacitance. Anything wildly outside the R-C tables is dubious.

Ahhh...

My biggest no-no I think sofar was neglecting to even concieve that there could possibly be a plate resistance. I take it this is similar to emitter or drain resistance?

Being a self-taught newb, I often get into books where a few wise words would make everything clear up in an instant... This might be one of those times.

Yes, it is hard to find a book that is suited to the individual's learning needs. Most books I read are either the kind that tells me that that grid resistor was indeed a nearly arbitrary number picked mostly to load the previous stage, or the kind that you can't find a picture of a tube in and introduces a new theoretical equation and variable to solve for in every paragraph.

I have (tried) to read both... I think MOST people would prefer a happy medium... I like PRRs style. Giving me equations, but actually referring to practical functions and parts of the tube.

Thanks for being understanding, and I think I'll be able to revise this after looking into a few things...

Wiggle the voltage on the plate a little. How much current did it take from your wiggler to make that change? How much of that is due to the load resistor? Account for that and then you can calculate the plate resistance in that vicinity. It is different a little for each operating point but it isn't jumping all over like some pathological function. The mathematical notation for what you are wanting is intimidating, with backwards 6's and so forth, but it's conceptually just asking what a small change in one variable does to another when everything else is held constant.

If all the change in current for that change in voltage were given by ohm's law and the value of the plate resistor, then the plate resistance would be ~infinite.

The concept of a perfect current source/sink is a useful one here, so it would be good to get that under your belt at some point.

There are also some good reads online.
I have found Steve Bench's pages to be pretty good, and though he's illustrating power amps in this link, the idea of a "load line" is pretty handy.
http://members.aol.com/sbench101/#Theory
As always, Yah, what PRR said. If designs lifted wholesale from the RCA manuals are good enough for Fender amps, they're a good starting point for us all.

You'll have to pardon bcarso. He's my nephew, who is mentally ill. He has been "studying" tube theory with me, trying to learn. I let him try anyway. Please don't make fun of him.

You weren't supposed to tell!!!

> neglecting to even conceive that there could possibly be a plate resistance.

It is right there in the tube data sheet. RTFM before you even think about designing.

No, a triode's plate resistance is not much like that of an FET, BJT, or pentode. It is a lot like a resistor, and even more like a vacuum tube diode. In fact tube textbooks always drill you with diode theory before you ever find a grid. The grid is just a control lever, the cathode-plate diode does all the heavy work.

> PRRs style. Giving me equations

You may have confused me with someone else. I rarely do more than a Add or a Times here. In secret, I've never wrassled with more than 12 terms in an equation, and that was an all-summer chore. I had to scratch a lot of hair to figure what a "backward 6" was; I don't do differential equations.

I also think that perfect current sinks are a dubious concept in triode work. Vital to pentode and BJT understanding, but triodes in practical applications are just warped resistors. Long-tail pairs may be the exception, but most new-age designers grossly over-estimate the value of a stiff tail. A triode longtail actually gives best CMRR with a specific resistor tail. Triodes ain't pentodes or BJTs.

"I also think that perfect current sinks are a dubious concept in triode work."

I was going to say just a big R with a big voltage as the plate load, then thought maybe a big-assed choke. But since the poster (known now as Uncle Northside forevermore I guess) had mentioned not getting to current sinks in his studies as yet, I thought I could make the hint.

[quote author="PRR"]> neglecting to even conceive that there could possibly be a plate resistance.

It is right there in the tube data sheet. RTFM before you even think about designing.
[/quote]

What if the data sheet doesn't give the plate resitence then? I am trying to design a line stage from this tube (got a pile of them cheap)
http://www.2tough.com/~kiira/rec/1626.pdf

Is there enough data to compute the plate resistence? Without the plate resister value I don't know how to draw a line for the operating point and at least get a rough idea of where to start.

PRR wrote:

Basic triode voltage amp:

Pick a tube.

Look up its data for an operating point similar to what you plan to use (yes, you need some experience). Note that the book values often show-off the extremes of the tube, we often run audio voltage amps at much lower currents than book values.

Pick a plate resistor that is 2 or 3 times higher than the plate resistance. For 12AU7 working at high current, Rp is ~6K so pencil 12K to 18K, say 15K.

Pick a next-stage input impedance 2 to 5 times higher than the plate resistor. That suggests 30K to 75K.

Problem: if the load should be 30K to 75K, we are far-far from driving 600? or even 10K loads. On the other hand, if we drive another grid, it is probably 1 Meg. We can load high, but it suggests that a high-current 12AU7 is "too good" for this work. Working 12AU7 at a lower current is cooler and makes power buzz filtering cheaper. Or switching to 12AX7 gives more voltage gain per bottle and per watt of heater power.

Click to expand...

I wish there were a book of explanations about designing like this so I could understand the math from the practical understanding and not have to try and do it the other way around.

KIira

I'm guessing that one can ascertain plate resistance via the slope on the graph of relationship between current and voltage on the plate per given grid volts... Correct me if I'm wrong, and I probably am here...

Your 1626 tube has at grid = 0v:

100ma at 200v = 20R plate resistance?
and 50ma at 125v =6R?

That seems off to me.. I almost know it's wrong...

Or would you create the slope based on plate current and volts responding to grid bias/operating signal voltage? Based on previous posts I am trying to assume that the plate resistance is just the resistance of the "diode" between the anode and cathode... Similar to a PN junction, but I know the properties of the two have to be completely different...

Sorry, haven't gotten to any books or links yet. I have a band coming into the studio today and it's been busy... :guinness: :sam: :guinness: :sam:

[quote author="northsiderap"]100ma at 200v = 20R plate resistance?
and 50ma at 125v =6R?

That seems off to me.. I almost know it's wrong...[/quote]

It is. The plate resistance is a dynamic parameter... In other words you calculate it as the change in plate voltage over the change in plate current. That's where those "backward 6's" come into play.

Looking at the curves, at -18V grid:

~250V plate, rp is around 2K.
~200V plate, rp is around 4K.

That's a 2:1 ratio! Which is why it's important to know where to bias a tube.

This 1626 is some kind of power tube, I think - WAY overkill for a voltage amp.

Peace,
Al.