AudioCircle
Other Stuff => Archived Manufacturer Circles => Music Reference => Topic started by: Roger A. Modjeski on 1 Jul 2007, 10:15 pm
-
Biasing of Tubes: Part 1.
If you and I went back in the “Way Back Machine” * and visited an engineer or a radio repairman (called “radioman” in the 1920's-40) and started a conversation on bias we might not get past the first sentence with this competent fellow. If we said “I adjust the bias on my amplifier to be 50 mV or 300 mV or any number under 20 Volts he would look at us very strangely. For him “bias” is typically –20 to –100 volts. No millivolt bias in 1929. He would call your bias zero and wonder how you kept your tubes and amp from burning up.
Positive reading low voltage bias came into the amplifier world in the 1955 with the birth of Dynaco. See: http://en.wikipedia.org/wiki/Dynaco for more information on that company. My dear friend, Bruce DePalma (brother of Brian) worked for David Hafler in the early years. We had many interesting conversations over gin and tonics. He told me some humorous stories about David which I will relate if there is interest.
I had worked on many a Stereo 70 by then and I knew why 1.56 Volts was the chosen numbers because I had built several meters by that time. Anyone who has built and calibrated a Heathkit VTVM or most any other meter, will recognize that number, it’s the voltage of a “fresh flashlight cell”. In the age before accurate digital meters that was the best “voltage standard” one could get. In that day the most sensitive scale on common meters was 1.5V with just enough swing at the top to allow 1.56. Some meters even had a little mark up there.
DePlama told me an interesting thing I would never have known otherwise. It turns out that a common cathode resistor of around that value also reduced the distortion appreciably. DePalma was 16 when he started working for Hafler. Another early starter.
I had better warn you now. I go off the subject sometimes to get a few details in, my intent is to fill in the gaps and history for the younger readers who may have never seen an analog meter. You know, the kind with the needle that swings up and down and all those numbers on the scales. I will say here that analog meters are still the best for many audio applications and the good ones are quite accurate. One thing to know about them is the 1/3 scale rule. The rule is you should not read below 1/3 scale. Most meters had a 1,3,10,30 series so you could range down if below 1/3 scale. If the needle barely moves, forget it. Analog meter accuracy was specified as a percentage of full range. So if you have a meter that is 5% accuracy and you are on the 5 volt scale you can be .25 volts off. Add .25 volts to 1.56 and you will be 16% high, or low if the meter is off in the other direction. Bias does not have to be dead on, but 16% is a bit much, especially if on the high side. The ST-70 tubes are already running at full dissipation, and 16% over would shorten their life considerably.
David Hafler undoubtedly knew all this. Not all meters had a 1.5 volt scale. Some had 3 or 5. The manual told you to get a fresh flashlight cell, see where it read on your meter and just adjust the bias to the same spot on the meter scale.
Now what does all this have to do with our visit back to the old radioman and why don’t you find numbers like this on a tube data sheet that he would be looking at? Look up an EL-34 data sheet and you will find typical bias is around –35 volts (note that’s a negative voltage). What’s even more confusing, when you BIASET the Dynaco you read a positive 1.56 V (all with respect to the chassis which is typically ground). By now there is great confusion in the conversation with the radioman. BTW, I have found that many students have trouble getting a feel for voltages, both magnitude and sign. I play this little game: 1 volt = $1 and you know that $1.56 is not the same as $35. You also know that if you balance your check book and you have -$35 you are going to get a note from your bank. If you have +$35 you can at least get dinner. To a technical person voltages are as real as money and reporting them should get the same care as your bank takes reporting the status of your account.
There is one other reason the radioman thought in terms of bias voltage and not plate current. The tubes he was dealing with, types 71 and 45, were all low Gm tubes (1.6 to 2.5 mA per volt) so variations in the tube parameters or the setting of the bias voltage made little difference in the plate current. When the 2A3 was introduced in RCA manual RC-11 (1933) there was no mention of bias matching though they did mention, rather proudly, the high Gm of 5.25 ma/volt. By the publication of RC-12 (1934) they had inserted a very important paragraph, undoubtedly based on things learned from applying the tube. It’s a long paragraph but the most telling sentence (paraphrased) is this: “The very high value of transconductance makes the 2A3 somewhat critical to grid-bias voltage since a very small bias change produces a very large change in plate current” In push-pull they strongly urged that the tubes had individual bias pots and the ability to read the plate current and make it the same in both tubes. Please note that both plate current and cathode current are the same in triodes and both terms are used interchangeably. In pentodes the difference is usually no more than 5% due to the screen grid current, but that’s another story. Note also that today’s common pentodes; EL-34, KT-88 etc. have Gm’s of 6 to 10 mA per volt.
So what are we measuring if we are not measuring bias voltage? We are measuring the current in the output tubes as a voltage drop across a resistor. We don’t really care what value of negative voltage is applied to the grid via the bias pot that you are adjusting. As long as it is enough to dial the tube in. One maker, whose initials are ARC, purposely limits the range on bias adjustment thus requiring output tubes to fall in a rather narrow range (less than 10 volts) of grid bias. Because we get all the bias and Gm data from our tube tester, Ram Tubes can select tubes in the required range for these amps. (We sell the same tubes at lower prices). When I test 6550’s only half of the tubes will fall within a 10 volt range. I designed the RM-9 to give the user a very wide bias adjustment range from –32 to –64 volts (32 volt range). This range also allows the user to use EL-34’s, KT-77’s, KT-88’s, 6550’s and any other tube that can be biased within that range of negative grid voltage.
The ST-70 had a 15.6 ohm 1% resistor in series with the cathodes of each output pair. That’s a rather odd value. Why not 15, that’s a standard value. Well David wanted to have exactly 100 mA through the two tubes and 15.6 ohms yields (by Ohm’s Law) 1.56 volts in that situation. He even bothered to get a precision resistor. When its all biased up you can take your meter and read the negative grid voltage that you adjusted on the bias pot and it will be somewhere between negative 30 to negative 40 volts with average tubes. Now if you said to the radioman. “I bias my tubes at –35 on the grid for 50 mA of plate current per tube.” He would say… gotcha.
End of Part 1.
Stay tuned for Part 2. Where we will look at the terms: autobias, fixed bias, cathode bias, self bias. These are terms I feel need clarification. It makes discussing things easier.
* The “WayBackMachine was a creation of the cartoon show “Bullwinkle”.
-
Good stuff, Roger! :thumb: (It did take me a while to remember what this is all about, thank you for the refresher!)
-
John,
Thanks for being the first reply. I fear I went too deep for most readers.
I would appreciate hearing from others. I'm testing the waters here.
-
Hey Roger, if the water didn't get deep sometimes, nobody would learn how to swim....
I appreciate your information. Since the last amp I picked up happened to be a Stereo 70, you piqued my interest.
Since I started messing with tube amps (again) I've had a lot of 'brain freeze' in learning, and the more exposure I get to functional facts (as opposed to technical manuals) the more sinks in, hopefully.
Thanks for the efforts!
Regards,
DeadFish
-
It could be complete waffle for as much as I know about it, but Roger's style kept me reading to the end.
Love the asides and anecdotes - keep it up and we can learn as we are entertained :thumb:
Jim
-
Excellent Roger, thank you! I'm with Jampot, I love to learn about all the background info, asides and anecdotes.
-
Thanks to all for reading and writing.
Gordy, what's the name of the cartoon character?
-
I got my first tubes 1 year ago, thanks for the very nicely put usefull information... :thumb:
-
Hi Roger,
He's Marvin The Martian, from Chuck Jones' Loonytoons... http://en.wikipedia.org/wiki/Marvin_the_Martian
(http://www.angelfire.com/pa/lkmarvin/Pictures/marvredblk.jpg)
I noticed yours as well, Plato from Raphael's The School, very nice!
-
I fear I went too deep for most readers.
I would appreciate hearing from others. I'm testing the waters here.
Deep but informative.
I apologize for simply reading your posts & not responding.
Your posts have that Trifecta Blend of the 3-E's
Entertaining, educating & enlightening. Please continue.
Your notion of a weekend Santa Barbara 'primer' sounds appealing.
However, since most of us are at many different levels of knowledge
I'm curious how you'd balance that out....remedial for some of us, while
in depth for others.
-
Yeah, that was fun. Nice work! My Cary biases in mA, my Manleys bias in mV across a 10ohm resistor, same result. I have lots of bias questions, but I will wait patiently for the sequel.
Thanks Roger!
Rich
-
Roger, this is great. I come to the table with not much more than V=IR, but I have an appetite to learn more. Please keep going.
Thanks,
Steve
-
Thanks to all for the encouragement. For people fooling with tubes it's great to know Ohm's law. I had a student that wanted to understand electronics without Ohm's law. I told her... Good Luck!
One of the things I encourage my students to do is think about the voltage, current, resistance.... wattage (if you care to go that far) of simple things, like turning on a lamp. If you close the switch on a 100 watt incandescent bulb, what happens? What is the current flowing through the switch, wire, circuit breaker... all the way back to Hoover Dam if you care to.
Here's something to consider. If you have an ohm meter, measure the resistance of a 100 watt lamp. What current will flow when you close the switch?
Roger
-
Well, I have a 30 Watt bulb on hand. Measuring this with my multimeter -- it takes a beat to settle down -- I get 37 Ohms. With around 120 Volts AC from the socket, V/R would be more than 3 Amps. I find that pretty surprising as I would have expected much less.
Well then, isn't the power in Watts given by IV? If so that would say the power dissipated is about 390 Watts -- which must be way off for a 30 Watt bulb. To get a 30 Watt result for 120 Volts gives 30/120 = 1/4 amp. That sounds more like it.
So I'm not sure what's going on with this. Maybe the bulb is more than just simple resistance? Or maybe the resistance varies with the applied voltage? What am I missing?
-
heat?
-
Maybe so. Taking your lead, that means the filament becomes more resistive as it heats up. And so for this 30 Watt bulb it would then start out cold at 37 Ohms and increase to something like 480 Ohms.
-
Flatmap got this one. The resistance does go up, about 10 times when the filament is hot. If you hold a cold bulb in your hands you can actually see the resistance go up a tad from the heat of your hands. So every time you turn on a 100 watt lamp there is close to a 10 amp inrush. The heaters in your tubes behave the same way but the ration is not quite 10 to 1 because they never get up to the temp of a lightbulb. Once a glowing body gets in the visible range it goes from dull red to red to orange to yellow then white. Studio photographers rate the color of light in degrees Kelvin. Fluorescent lamps are also rated on this temperature scale.
Here is good time to comment on the yellow-white flash you see near the base of many 12AX7's and other small tubes. This flash is perfectly normal and does not affect the life of the tube. What is happening is the small uncoated section of the filament gets hot faster because the assembler has to scrape away the white filament insulation in order to spot-weld the filament wire to the pin.
-
By same token, things get more conductive when cooled. I think that even a strawberry will become a electromagnet and levitate over a permanent magnet if chilled cold enough. Is this correct?
-
I've been messing about with tube schematics and design for the better part of the last two years, so I knew the message you were telling but I still found the historical context really interesting to read.
Ohm's law, Kirchoff's law (law of preservation of currents), some elementary algebra and problem solving get you most of the way in playing with tubes.
-
Thanks everyone for your participation. A few notes: There are materials whose resistance stays constant with temperature, nichrome wire is one. It's the wire you see laced back and forth that glows when it makes your toast. It's virtually the same resistance hot (glowing) and cold. Here's an interesting tip, from Bruce DePalma again. If you are ever in need of a high power roughly 8 ohm load. A 1200-1500 watt portable heater, toaster or any nichrome wire heating element is very close to 8 ohms cold and the same hot. If it has a fan and you have a big enough amp to get above 60 volts you might even get the fan to run. How much power would your amp be putting out at that point?
As far as law an order: I hardly ever (maybe never) use Kirchoff, but Thevenin is a very good thing to know. The trouble with Kirchoff's law is it gets very cumbersome and you have to solve a matrix, ugggg. Thevenin on the other hand is precisely what we need to understand the effects of loads on amps and preamps. It's one I use every day. It's quick, nothing more than adding, subtracting, multiplying, dividing. Throw me a bone and I'll get into that one.
Here's a little history: http://en.wikipedia.org/wiki/Thevenin. Here's a quote from that article.
"Appointed as a teaching inspector at the École Superieure in 1882, he became increasingly interested in the problems of measurement in electrical circuits. As a result of studying Kirchhoff's circuit laws and Ohm's law, he developed his famous theorem, Thévenin's theorem, which made it possible to calculate currents in more complex electrical circuits".
-
Well yes, I'd love to hear more about what Thévenin's theorem is about!
-
In brief, Thevenin realized that any source could be modeled as an ideal source with a resistor in series (for a voltage source) or in parallel (for a current source). For AC one can extend the theorem to include inductors and capacitors, but let's stick to single resistors first.
Consider a 12 volt car battery vs the little 12 volt battery that is in most car remote controls that we carry on our key chain. They are both 12 volts but you can't start the car with the little battery in the remote. Why is that, they are both 12 volts? Thevenin not only answers the question but gives a very simple circuit that will predict how each battery will behave under load. Both are modeled as a 12 volt ideal voltage source. That means a theoretical 12 volt source with no internal resistance that can supply an infinite amount of current. Now he adds the internal resistance in series with that voltage source. The starter battery might have an internal resistance of .03 ohm while the little remote control battery might be 24 ohms.
So now we have a big difference in the two batteries. If you short the car battery the current will be 400 amps (12/.03=400). If you short the transmitter battery the current will be 12/24 or 1/2 amp. Big difference. Since the starter of you car can draw 60 amps in sunny California (and way more in a Minnesota winter) we can actually predict the resulting votage right at the battery terminals. Also note that corroded terminals simply add resistance thus dropping the voltage further. When my 7 year old Mustang battery started to go I simply hooked up a meter to the battery while cranking the engine and saw the voltage fall from 12 to 7 though the headlights would only pull it down to 11 or so. The battery was still a 12 volt battery, but its internal resistance had gone up markedly making it unreliable. Note also that when the starter draws the battery voltage down you have less spark which further reduces your chances of getting going.
Once these resistances are known we can predict the battery terminal voltage under any load. Don't go looking for this resistor inside the battery, it's not a resistor but a resistance that is dependent on the surface area of the plates, condition of the electrolyte and any connection resistances within the battery. Thevenin says we can lump all those into one imaginary resistor that is in series with the ideal voltage source. Say we put a .1 amp load on each battery. The starter battery will fall .1 x .03 ohms or .003 volts. Hardly perceptable. But the little battery will fall .1 x 24 ohms or 2.4 volts. Big difference and easily calculated.
A little practical advise here. A very easy way to test your car battery and terminal connections is to have someone crank the starter while you have your meter probes poked well into the tops of soft lead terminals. Note the voltage while cranking. Then move the probes to outside of the clamps and measure that voltage. Whatever the difference is what you are loosing at the terminal to post interface. You can also measure separately from post to ring of the plus and then the minus. That will measure the voltage loss on that interface directly. You may find one is very good and one very bad. Leave the good one alone and fix the bad one. I never ever take a battery terminal off just because it might look bad on the outside. If it measures good then it's good and you can't make it better and will likely make it worse. Good clean terminals stay clean because they have a tight fit. Scraping, fileing and sanding destroys the fit getting you into a maintenance cycle that could be avoided.
Three other things about car batteries while we are on the subject. There is an old wives tale that you will cause a car battery to discharge if you put it on the concrete floor. I would love to hear any supportive argument for that notion. The other thing is that if you ever drop a wrench across the terminals and short them out the wrench will likely weld to the terminals and you should get out of there because the battery is likely to explode from the high current discharge. Never never use anything but distilled water. Check the levels twice a year and top them off. Many times I have seen a batter fail because one cell was only half full. No-maintenance batteries are no good and leaving a battery discharged for even a few days causes sufation and that's death to a battery.
I'll end with a question: If we want to consider how our wall outlet voltage is affected by our audio equipment loads how can we get a handle on that? One could add up all the resistances back to Hoover Dam while considering all the transformers, wire, generator winding resistances... on and on. A quite impossible task. Using Thevenin one can get the answer very simply with just a voltmeter and a light-bulb. Any suggestions?
Next time I will get into how Thevenin can be used to understand how loads affect our preamps, poweramps, signal sources, cables......and whatever else you might be curious about.
-
Roger, thanks for this introduction to Thevenin's theorem. I'm thinking about your question on how to deduce the thevenin resistance at the mains outlet.
Here's my attempt to answer: First measure V at the wall. Then connect a lamp to the outlet and measure the voltage drop across the bulb, call it Vbulb. I think the difference (V - Vbulb) would be the drop across the Thevenin resistance, if I'm tracking this correctly. So we'd need to also measure the current in the circult including the light bulb = Ibulb. Then the Thevenin resistance RT = (V - Vbulb) * Ibulb.
Is this the right idea?
-
Steve,
Yes, good work, that's right. In practice I would use the heaviest load I had such as a 1200 Watt electric heater. A 100 watt light bulb doesn't make much of a dent in the voltage and we have to consider that the line is fluctuating a little. Since it's a linear system the more current the better but don't blow the fuse.
-
So, just going with this a bit... Suppose I placed a resistive load across the speaker terminals of one channel of my amplifier and made the same set of measurements. Then this figure for Thevenin's R, say for a fixed sinusoidal output, would this give me what is referred to as the output impedance for the amp?
Steve
-
Yes Steve, that would give you what you are looking for. Just as we want to see good voltage regulation at out wall socket, we want to see good regulation at our amplifier output. I should mention here that wall socket regulation and amplifier output regulation are not dependent on one another.
Here's how I do it. Run the amp at one volt output, sine wave. Put the load of interest (8 or 4 ohms) across the amp and see how many dB the output falls. You can do the calculation to get the Thevenin resistance, but why bother, you already have the answer in dB. You can also make the test at various frequencies and see the variation with frequency. This doesn't work on high feedback solid state amps because the dB difference is virtually unmeasurable. There are other ways to get at it though.
Here are some ballpark numbers. If you want to go back to damping note that a 1 dB drop corresponds to a damping factor of about 10 and a 1/2 dB corresponds to damping of 20. If the output falls 6 dB when the load is connected you have a damping factor of 1. There the output impedance equals the load resistance.
We are told by many an author that "matching" the load to the plate resistance of a triode gives the highest power transfer. Any comments on that anyone?