CAUTION CAUTION CAUTION -------------------------------------------------------------------------------- Both the filament transformer and the transformer under test will have at least AC line voltage on them, an may well have much higher voltage, several hundred volts on one or more windings. You are therefore in danger of being KILLED if you are not both knowledgeable and careful about how you do these tests. DO NOT TRY THIS IF YOU DO NOT HAVE THE KNOW-HOW AND EXPERIENCE TO WORK SAFELY WITH THESE VOLTAGES. IF YOU HAVE ANY QUESTION IN YOUR MIND WHETHER YOU CAN DO THIS WORK SAFELY, YOU CAN'T. Seek experienced help if you have any question in your own mind. The tests run like this. Identify which wires are which by color code, circuit connection, or by using an ohmmeter to find which connects to which. Label the wires. From the same ohmmeter test, write down the resistances you measured on the windings. Generally, windings with resistances over a few ohms are high voltage windings, either a power transformer primary or high voltage output, or an output transformer primary. Note that it is common for primary windings on power transformers to have from two to six wires, with the wires over two being taps to adjust for various line voltages from 110-117-120-125-208-220-240. Secondary windings on power transformers and primaries on output transformers will have either two or three leads, and secondaries on output transformers will have to to four leads. Also note if any winding is shorted to the transformer core. Sometimes an internal shield will be deliberately connected to the core, but if a multi-lead winding is connected to the core, this is usually an internal short, and a dead transformer. Once you have identified the windings, hook up one and only one winding to either 1/2 of the 6.3VCT or to the variac. Try to select a low voltage winding, one that has low resistance from the ohmmeter test. Make sure that no other leads are connected (or shorted together, or touching your screwdriver on your bench or... well, you get the idea). A turn of plastic tape on each wire end you're not using at the moment is a good idea. Set your voltmeter on this winding, and the current meter to measure the current through it, and bring the circuit up. The voltmeter should measure 3 volts AC, the light bulb (if used) should NOT be lit brightly, and nothing should be humming or smoking ;-). There should be little current going through the winding. If the voltage is lower than 3 volts, or you are pulling amps of current, then there is a load on the transformer, internally since you have disconnected all the leads, meaning that there is an internal short. You should try to select a winding for this test that is normally a low voltage winding, either a filament winding in a power transformer, or a secondary in an output transformer. If all is well, measure the voltage that now appears on the other windings. The voltages will be equal to the ratios of the voltages that will appear on these windings in normal operations. B. Where can I get a good replacement output transformer for my vintage DoppelBanger amp? Dixie Sound Works, Gunthersville, Alabama has a great reputation for (re)winding quality vintage re-makes. The company that made the amp may have service parts. The quality is variable from company to company and time to time, though. There are a number of companies that have entered the transformer market in the last year, so expect that there will be new places to get quality rewinds and replacement transformers C. I want to make my own power and output transformers. How do I do this?/ Where can I find information about this? Designing and hand winding transformers is not terribly difficult, but it does require information and skills that are relatively hard to find. You are unlikely to save a whole lot of money unless used or broken parts are cheaply available to you. You may want to do this if you feel that you were selected by some deity to take this on as a life work. First, take a transformer apart. A burned out tube-type power transformer will do. Do this carefully and slowly, imagining how you would have put it together in the first place to get it the way it was. This is an excellent introduction to the manual skills and materials needed to successfully produce one on your own. Learn about how transformers are designed from one or more of the following, in this order: "Transformers for Electronic Circuits", Grossner (check your library) "Radiotron Designer's Handbook, fourth edition "Audio Transformer Design Manual", Wolpert, $36, privately published, available from: Robert G.Wolpert 5200 Irvine Blvd. #107 Irvine CA 92720 "The Williamson Amplifier" D.T.N Williamson, reprint available from Old Colony Sound Labs Handbook of Transformer Design and Applications by William Flanagan (second ed.) "rewinding transformers with CAD" by Hugh Wells W6WTU Ham Radio Dec '86 p.83 "Fast Optimization of Transformer Design" EDN Nov '62 by Davis, J. H. These sources will help. They are NOT a complete cookbook. Note that it is very possible to make a transformer that will operate relatively well, but may break down unexpectedly and KILL you if it is not constructed with safety in mind. D. Should I replace my stock transformer with a new/old/vintage/purple one for better clean/grunge/grit/etc. sound? Unless you REALLY know what you're doing and have heard the transformer you'll be swapping in and like it, no. There are a huge number of variables in the "sound" of a transformer, and you should exhaust other means first. You might not get that magic sound after all that work unless your ears - and amp tech - are really good. Are potted / impregnated transformers better? Is potting / impregnation necessary? Necessary for function? No. Necessary for long term reliability? Yes. If transformers are not potted or impregnated in some way, they will eventually have problems with slight repeated movements of the wire wearing the insulation and interleaving material, and with moisture infiltration. In some climates, moisture infiltration will let molds and mildew get started inside the transformer if it's not used almost daily. "Potting" is not the same as "impregnating"."Impregnation" means getting some kind of insulating gook soaked all through out the windings and spaces inside, while potting means putting the whole assembly in a can and filling the can (and transformer, as in impregnation) with goop. To be reliable for years of use, a transformer must at least be impregnated with varnish or epoxy. Almost all commercial transformers of any size are treated this way. A do-it-yourself way to do this is to dunk the whole transformer in a bucket of varnish and pull a vacuum on it to expand the air out of the windings. A few vacuum cycles will get varnish well infiltrated into the windings. Then bake it at low temperatures (under 140 to 200F) for a long time, until the varnish is truly hardened. -------------------------------------------------------------------------------- What is the easiest way to get tube sound at a good price? Obtain an old piece of tube gear, perhaps intended for another purpose, like mono hifi, at no or low cost. Modify this to duplicate to a certain extent the circuit of an existing amplifier. Tinker to your heart's content. There is a document on exactly this at http://www.wwu.edu/~n9343176/docs/old2new.html The document goes into excellent detail on the in's and out's of building from old tube gear and the possible and useful variations of which stages with how much gain go where in the amp. Build a tube preamp from scratch, and use this to drive another larger amplifier which does not necessarily have to be tube based. I have designed things like this, so have others. Good tube sound, and inexpensive. Really convincing tube distortion, especially if you add some lowpass filtering to simulate the high frequency cutoff of guitar speakers. This is what the Hughes and Kettner Blues Master and Cream Machine tube preamps did (they've been discontinued). These were entire tube amplifiers with maybe 2 or 3 watts output, a simulated load, and a line level output in addition to the speaker output. They did a VERY respectable job. -------------------------------------------------------------------------------- How can I modify my tube amp to ... ? (also see recommended mods, below) Get lower hum? Replace the defective humming tube Replace or improve the power supply filter capacitors Fix the defective internal ground wiring, as on a reverb tank, or previous "improvements" and modifications Run the preamp filaments on regulated DC, not AC, starting with the input tubes Rewire the grounding so the amp is star grounded, and does not use the chassis as a ground bus Move the signal wires around, nearer/farther from the chassis or 60 Hz AC carrying wires Use coax cable in the signal path, at least in the early sections where noise counts the most. Tie one end of the shield to ground and terminate the other end with some shrink tube so it cannot touch anything. This way the coax shield acts as an antenna and conducts the RF to ground (as well as Faraday shielding hum out). If you tie both ends to ground you set up some capacitance (and the possibility of ground loops) you're better off without. The shield should be tied to the star ground point individually, and bypassed to the chassis locally with a good RF cap of about 0.001 to 0.01. George notes "You may already use this in your own amps but I thought we might share it with the rest of the tinkerers - it's especially useful for people that are trying to add extra gain stages. I even use it between the input jacks and the first stage since in most Fender amps it has to traverse the width of the board. (Kaschner) Have higher gain/more distortion? Install an extra gain stage by Using an unused tube section if one exists Adding another tube to the chassis Using the reverb tubes as additional gain stages Using a power MOSFET as a cathode follower to drive tone control and volume controls for lower loss Using a power MOSFET to replace an existing cathode follower, freeing up that tube section for more gain Remove the feedback on the power amp stage; newer Fenders and other amps use feedback on the power amp to reduce distortion. Removing this increases gain and and distortion, and makes the distortion start at lower volumes. On Fenders, it's generally a white wire from the 'ext speaker' jack to a 2.2k resistor. Cut this wire, or lift it at one end. To be really slick, put in a toggle switch. (Edelman) Use the alternate channel for more gain, perhaps jumpering two channels together have a smoother, less buzzy distortion? Use a lowpass filter somewhere inside the amp in the signal path to cut higher harmonics; perhaps a capacitor to ground from the final preamp tube grid or plate -or- Use series grid resistors to cut the high frequencies in and after distortion stages Use a lowpass filter after the amplifier and before the speakers to cut out some of the higher overtones. -------------------------------------------------------------------------------- When should I bias my amp and how do I do this? A. What is "bias"? "Bias" in this context refers to the amount of voltage held on the grids of the output power tubes. This controls the amount of current the output tube(s) conduct exclusive of the signal current, or, looking at it another way, the amount of overlap where both tubes are conducting simultaneously. I will talk about the output tube current since the terms "underbiased" and "overbiased" are confusing with tube amps. A technician who works with only tube amps will usually refer to the voltage which sets the operating current in the tubes. In these amps, the bias is a negative voltage, so "overbiased" to such a technician would mean that the tubes are held in a condition of too little current, just backwards from the solid state terms most of us are familiar with. "Underbiased" would mean that the tubes have too little negative voltage on their grids and are conducting too much current simultaneously. The idle current in the output tube and the degree to which the output tubes overlap in conduction is what you're trying to adjust, not how many volts go on the grids; you just have to use the grid volts to change the current and conduction angle. The whole topic of bias is tied up with the "Operating Class" the power amp is designed for. There are only three classes useful to us in tube amps, Classes A, AB1, and AB2. Class A means that the output tubes are biased so that both tubes are always conducting. Even on maximum signal peaks, the tube driven most "off" will still be conducting some current. In both class AB's, the bias is set so that on a signal peak, one of the tubes can be driven completely off for some part of a signal cycle. In class AB1, no grid current flows into the grid of the tube, and in class AB2 some grid current is driven into the grid of the tubes. There is a class B, where both tubes never conduct current at the same time, only alternately. The point of all this is this: The Class of the amplifier is determined by how much bias current is present. If there is a lot of bias voltage, the grids are held 'way negative, then only the tube which is driven by the positive going half wave of the signal at any moment is conducting. This is class B. It sounds ugly because the point where the signal crosses over from positive to negative and begins to drive the other tube is not reproduced cleanly, and creates [surprise!] crossover distortion. You can look at the output signal with an oscilloscope and see crossover clearly as you make the bias voltage too negative for both tubes to conduct at the same time. As the bias voltage is made less negative and allows both tubes to conduct a little, the crossover notch diminishes swiftly, and you are in class AB2; a little less negative, and they both conduct more, and you have class AB1. If you go further, you get to the point where both tubes always conduct, making the amp work in class A, which has the least crossover distortion of any of these operating conditions. Too little simultaneous conduction in the output devices puts them in the most nonlinear region of their transfer characteristic, so crossover distortion is high; but as you increase the amount of simultaneous conduction, the power used and dissipated by the outputs goes up, perhaps to a disastrous degree. You are trading standby current and power dissipation in the output devices off against distortion. If both outputs are biased almost totally off at idle, crossover distortion is very bad. As the simultaneous conduction is increased, crossover goes down rapidly, until it gets smaller than the residual THD of the amp itself, and becomes much less audible. There is a fairly broad sweet spot where the crossover distortion is comparable to the THD and the idle current and idle power dissipation are reasonably low. This is the region you're looking for. Lots of bias, both tubes conduct all the time - and eat a lot of power, get hot, other Class A kinds of things. Little bias, both tubes overlap less, get less hot, put out more total power - and produce crossover distortion, which sounds especially unpleasant. Power tubes individually have slightly different DC gains, so the same bias voltage on two different tubes produces two different current levels. "Matched pairs" are two tubes selected to be close together. Groove Tubes grades tubes from 1 to 10 so that any two "3"'s for instance are close enough to sub for any other "3", so you don't need to rebias if you keep buying the same number from them. Note that you may not want matched pairs, depending you your taste. See section D. below. B. When should I bias my amp? You should re-bias the amp whenever you change power tubes or modify the power amp circuits. Each power tube needs a certain bias current to keep it operating at the point where the amount and type of distortion under normal conditions is well controlled. Individual tubes vary widely in the grid bias that sets the correct idle bias current. If you change tubes or tinker with the circuit, you need to make sure the tubes are set back into operation in a way that sounds good and does not cook the tubes. Amps typically provide only one adjustment point for bias, assuming that you will have bought matched sets of power tubes. It is possible to modify your amp to "match" unmatched tubes by setting the bias voltage and AC drive level of each tube individually. This may require some serious soldering, though. See section D. below for a discussion on matching, and the mods section for what you have to change. C. How do I bias my amp? -------------------------------------------------------------------------------- CAUTION CAUTION CAUTION -------------------------------------------------------------------------------- Keep in mind that tube amps use high voltages, and they can *kill* you if you don't know what you're doing. So, if in doubt, leave the job to a qualified technician. How do you correctly bias an amp? There a few different approaches but first hook up a speaker or a passive load to the output and remove any input signals; tube amps need to have a load or they can sometimes become unstable. Check and make sure the proper size fuse is installed. Output Transformer Shunt Method The most common and simplest procedure is to hook a current meter from the plate (anode) across half of the primary of the output transformer; this is called the "output transformer shunt method." The idea here is that milliammeters commonly have a very low series impedance so that when placed in parallel to half of the primary, almost all of the current flows through the ammeter. When you hook things up this way, your meter is floating at the voltage level of the plate, which is typically hundreds of volts -- be very careful! You could open the wire from each plate to the output transformer and hook in a meter in series with the plate temporarily, but that is a terrible amount of work for the small gain in accuracy. Adjust the bias pot so that the current reading is the appropriate value for the type of tube (see the table below). Let the amp warm up and note if the bias changes significantly. If so, select a compromise bias point. Keep in mind that if your circuit uses more than one tube per side, the bias current you're reading is multiplied by the number of tubes (e.g., if you're reading 60 milliamps and there are two power tubes per side, if the tubes are matched each of the two are getting nominally 30 milliamps). Check the other side of the circuit to confirm that the two sides are close (within 5 milliamps) to each other. If your ammeter has too high a series impedance, the shunt method won't work because the bias current gets significantly split between the meter and the transformer; the meter has no idea how much current is going through the transformer. You'll know it's not working because the current values you'll be reading will be much too low no matter how far you adjust the bias pot, the tubes will be glowing hot, and when you note that you'll reach quickly for the power switch! If you don't reach it quickly enough, you might blow a fuse. Don't despair: you can use another method called the "cathode resistor method." Cathode Resistor Method If the circuit already has a resistor in-line between the cathode and ground, use it. If the circuit has the cathode hooked up directly to ground, insert a low value resistor (say 1 Ohm/1 Watt) [even 10 ohms will work well, as the currents in a tube circuit will cause only a volt or so max across a 10 ohm resistor, not enough to change the circuit operation a lot.] in between the cathode and ground. This doesn't have to be a permanent change to the circuit; you can make a little adapter that fits between the tube and its socket that runs all the signals straight through except for the cathode lead -- that path gets the low value resistor in-line. If you make the adapter, you don't even have to drop the chassis from the amp to set the bias. Just pull a tube, install the adapter, and adjust. Hook up a voltmeter across the resistor and measure the voltage. For a 1 Ohm resistor, if you read 30 millivolts Ohm's Law says that you have 30 milliamps running through it. If you have some other value resistor, make the appropriate calculation. Easy! But since the current at the cathode is the sum of the bias current and some other leakage currents, you need to compensate the reading a bit, typically 5 to 10 milliamps. What's nice about the cathode resistor method is that you're not dealing with high voltages. The cathode sits very close to ground so the chance of a dangerous mistake is lessened. You're also reading each tube's bias current individually. Other Methods Some of the manufacturers say to set the bias voltage to some specified voltage, without any other measurements. Presumably some designer somewhere decided how much was good for you and wrote down "Set the bias to xx volts" as a good compromise for all the tubes s/he expected. This method ignores the variability of transconductance in output tubes, and only gives good results for matched sets that happen to be exactly like the "typical" ones the designer thought they'd get. Note that Gr@@ve Tubes tries to help by providing matched tubes with a bias number from 1 to 10. If you have GT's with a "4" bias number, and you replace with a GT "4" set, they will have selected only tubes that are properly biased at that level, and no rebiasing will be necessary. Of course, GT expects to be repaid a fair profit for this service to you... Another way to set bias is to use a test signal, typically a sine wave. Monitor the output waveform on an oscilloscope and adjust the bias for minimum crossover distortion. The obvious problem is when has it "just disappeared"? Most folks do just a bit more than "just disappeared" and get their outputs too hot causing shortened tube life and overheating. Not very accurate or repeatable. You can also use a special purpose instrument that nulls the input signal out of the output signal so that you can monitor just the distortion products. You then adjust the bias to get the distortion to a realistic minimum without making it dramatically less than the residual THD. This is the premium method, but requires a distortion analyzer - big bucks. These methods can be more accurate than the first two methods but they require expertise and tools that most folks don't have. If you are a circuit hacker, and live on solder fumes and cold coffee, you can modify the amp with solid state servo bias adjusters that twiddle the bias to each output tube on the fly on a continuous, real time basis to keep each tube -* exactly *- where it ought to be. Only recommended for real wiring fanatics... GENERAL BIAS GUIDELINES (from Tremolux@aol.com) Currents Per Tube - Class AB1 Operation (most musical instrument amps are designed to run in class AB1) 6L6 - 30 to 35 ma 6V6 - 22 to 27 ma EL-34/6CA7 - 35 to 40 ma., sometimes even higher! 6550 - 40 to 50 ma EL-84/6BQ5 - 22 to 27 ma Class A currents will be higher. Example is 50 ma for a 6L6. Don't try to run an amp designed for AB1 in pure class A, it will overheat and probably blow. To handle the higher idle currents, Class A amps usually run at lower plate voltages. D. Matched output tubes - do you need them? Do I always have to buy matched pairs of output tubes? The issue of "matching" output tubes, either by buying carefully matched pairs or by tweaking the bias levels and drive signals per output tube is not a settled one. It used to be common wisdom to simply buy matched tubes. A few people noticed, however, that they had a favorite pair of output tubes, which made their amp sound much better than others. The common assumption was that these tubes were better matched somehow. When these tubes get measured, though, it usually turns out that they are NOT matched, at least not matched for AC gain characteristics. The concept of matched output tubes comes to us musical amp types from the hifi community, where they are trying to get the lowest possible distortion. This was true from the start, when Fender was trying to build low distortion amps and copied hifi circuits. The concept has simply clung to us, largely through inertia. It is relatively well accepted even in the hifi circles now that even-order distortion is euphonic, sounds good to our ears. It is very likely that the even-order distortion produced when mismatched output tubes are used sounds better than perfectly matched tubes. If you have modified your amp so you can independently set the DC bias and the AC drive signal, you can tune almost any pair of tubes into AC and DC matching. You can also tune in a selective amount of AC drive mismatch to experiment with the selective mismatching sound. There are technical reasons for matching. Getting enough turns of wire on the primary of an output transformer to get the right primary inductance and still using as little iron and copper as possible to do the job properly is an engineering problem that almost always results in Class AB output transformers being smaller for proportional power outptu than a Class A output transformer would be. The (relatively) smaller transformer and wire size makes a class AB (most guitar amps) output transformer susceptible to burning out if one of the half-primaries carries too much current. If one side of the transformer carries significantly more current (like double) than it would otherwise in "normal" operation, it is possible it will overheat or open, effectively killing the transformer. Tubes that are so mismatched that to get the right total current for a pair means that one is carrying more than 50% over the nominal DC current for a matched pair is getting into the region where you ought to worry about output transformer damage. If you mismatch, try to get the DC current the same in both sides of the output transformer, and an imbalance in the AC gain of the tubes. The logical limit of this AC mismatching is to remove all the AC drive from one output tube, which is a technique used by at least one commercial amp maker. This effectively keeps the output transformer happy with respect to DC, and gives you a single ended output stage; this also costs you a large amount of your available output power, but, hey, we're after tone, right? Note that the commercial tube suppliers have good reason for wanting to sell us matched sets at a premium. I would expect their opinion to be that matched sets are absolutely crucial. As in all musical matters, let your own personal ears be your guide. If you have a set of tubes you know are not matched, or if you have modified your amp to be able to set the bias and drive levels on each output tube separately so you can either match or not match the tubes at will, you might want to try un-matching them and see how it sounds to you. Q: I have found that if the bias for push-pull with cathode bias design, is set with the bias at 50 mA for EL-34 tubes when idling(i.e. no sound input), over time the cathode bias resistor will blow. Why is this? A: This is true if the amp is cathode biased into Class AB; in this class, the average current in the outputs rises with signal. On Class A biased amps the current is already at max for that bias point, and should not drift up except if the tubes drift from thermal effects. Note that cathode bias of an amp into class AB IS possible. Cathode bias is not equal to Class A. Setting the bias point for a little less standing current (which is an unambiguous description for bias) is an OK solution if your cathode resistors are undersized as long as you can live with the increase in crossover distortion at lower sound levels. The "sweet spot" is wide, so that may be fine. If you were previously in Class A and dropped the current a little, this can move you slightly into AB for large signals. Q: Does amplifier stability have anything to do with the temperature of the output tubes? Can tubes go into thermal runaway? A: Yes. Emission in tubes increases with temperature, but not a whole lot, as the tube gets hotter. The predominant effect is that as the tube gets hotter, you cause outgassing from the metal, glass, and other materials in the tube. The gasses are attracted to the grid as the most negative point in the tube and stick to the negatively charged grid, causing a decrease in grid bias. If the tube is too gassy (which it can get to by being too hot) you can get into a condition where the grid leak current changes the bias in the direction of more current, which makes the tube hotter, which causes more current. The solution here is to lower the value of the grid leak resistors. This increases the available current to the grids and keeps the tube out of runaway. --------------------------------------------------------------------------------