How to Bias a Marshall JCM800
Biasing any Marshall has many pitfalls, and it will be a daunting task to try and demonstrate the ‘proper’ way to set up your Marshall (or any other amplifier equipped with EL34’s). In a nutshell, here are a few things to watch out for.
- Biasing strictly by the ‘crossover method’ will often leave your EL34’s dangerously close to exceeding maximum idle plate dissipation. There should always be a very soft notch present, even at ‘clean’ levels. I have yet to fully understand why this is so, but perhaps the fact that the EL34 is a true Pentode has something to do with this phenomenon.
- Many early Marshall amplifiers left the factory without any Screen Grid resistors. This isn’t a big deal, providing you have an ample stash of premium NOS Mullard EL34’s to keep your baby fed.
- After you believe you have the idle Plate current set ‘properly’, you should monitor the Screen Grid current, to make sure the output tubes you purchased aren’t ‘flaky’, and to see exactly how ‘matched’ they really are. It isn’t so much excessive Plate current that burns your tubes; it is excessive Screen Grid current as well.
- Many ‘gurus’ believe any ‘mod’ to your amplifier that makes it louder, or brighter, is a wonderful mod. Likewise, many will bias the output stage ‘hot’, resulting in an amplifier that drives easier. Since you own a Marshall in the first place to be loud and have distortion, many players love these ideas as well. The very early version of the book ‘How To Service Your Own Tube Amp’ (Tom Mitchell) stated to set your idle plate current at 60mA. Of course this is very high, and should demonstrate a point. Modern EL34’s aren’t as robust, and should be treated with due caution. I will still list the Weber VST and the Tube Amp Bias Calculator suggestions, but be forewarned that I will not be attempting to reach any of the bogey values listed.
- As much as the Plate current, and specifically excessive amounts, can be detrimental to the health of your tubes, the output transformer often gets overlooked. Many older output transformer spec sheets listed maximum amounts of current that could safely be carried through the primary winding. Today, it is often only an output wattage listed (where math is needed to determine the primary current handling capabilities), and even this is usually an arbitrary number conservatively guesstimated. I always ‘bench run’ an amplifier for a short time, and determine if either of the transformers are running hot.
With that out of the way, let’s put a 1982 JCM800 50-watt head up on the bench, and get to work.
There are many bias ‘calculators’ available ‘on line’ for you to investigate. For the curious, or for those that wish to compare the parameters selected for the different calculators, you may visit the following alternatives.
- Tube Amp Bias Calculator. I haven’t checked out the other pages to this web site, but so far it looks interesting.
- Dreamtone has a lot of interesting pages to their web site, but this link is strictly for the bias calculator.
- Duncan’s Amp Pages has a plethora of information available; a tube data base, tone stack ‘calculators’, and this ‘anode load calculator’.
If anyone knows of any other similar web sources, a ‘heads up’ my way would be most appreciated. Now, for those that insist on measuring plate current (or what you believe to be plate current), please consider the following warning.
WARNING! For those of you who insist on measuring plate current, be advised of the following. Measuring by way of the ‘transformer shunt method’, the indicated plate current (or what is purported to be the plate current) will be lower than if you insert an ammeter in series with the plate connection, or use a 1-ohm resistor in the cathode of the output tubes. In one experimental setup, the transformer-shunt method indicated 33mA of idle plate current. Inserting an ammeter in series with the plate lead yielded a reading of 36mA. Measuring the voltage drop across a 1-ohm resistor in the cathode gave a reading of 31.5mVDC, translating to a current of 31.5mA through that particular tube. Carefully measuring the resistance value of the purported 1-ohm resistor, I ‘discovered’ the value was slightly higher than 1-ohm. Hopefully, you see the pitfalls with this method which is highly touted as being the only ‘accurate’ method.
The first set-up I chose was a complete set of new ‘Groove Tubes’. The phase inverter tube chosen showed a close matching between halves, and the power tubes (EL34S, and a performance rating of ’10’) were tested to see that their transconductance matched within 5%. The power tubes appear to be from Svetlana. Many ‘gurus’ will not even own a tube tester, much less an oscilloscope, so this exercise is going far and above the ‘normal’ practice. Besides, with ‘Groove Tubes’, they should already be matched, right? Using a B&K E310B signal generator, and an 8-ohm 450-watt dummy load, the bias is set high (-44.2VDC), and the amplifier is hooked back up to our Leader LBO-508A oscilloscope. I do advise that a 20MHz bandwidth is about my minimum ‘standard’. The trace usually appears to be sharper, clearer, and helps bias up the amplifier ‘better’. Check out the photo below.
What you may notice is that the waveform isn’t too terrible ‘as is’. The measured idle plate current was 20mA with a bias voltage of -44VDC. Now, adjusting the controls to achieve a clean sine wave is a chore in itself, and setting the bias only complicates matters. Here, I reduced the bias carefully, until the waveform below was seen.
The above waveform was obtained by lowering the bias voltage to -41.4VDC. The measured plate current was 31mA for one tube, and 33mA for the other tube. So much for premium ‘matched’ tubes. The measured plate voltage was 456VDC. A visit to the Weber Bias Calculator and the Tube Amp Bias Calculator both suggested a bias measurement of 38.3mA, so I believe I had hit this sucker as close as I’d care to. Rather than the ‘guru approved’ 70% target, I will just have to learn to live with these tubes biased at only 61% of maximum plate dissipation. I mentioned a few tips on what you can check to confirm you did not purchase ‘flaky’ tubes, and to be sure they are biased ‘properly’, and here is the list. It is guaranteed to be neither complete nor authoritative. This is just what I do with certain EL34 amplifiers.
- Compare Plate voltages. A lower voltage indicates the one tube is drawing more current than the other(s), and should not be used in that particular set. A one-volt difference isn’t worth worrying about (and is actually common), but a discrepancy of more than several volts is worth worrying about.
- Compare Screen Grid voltages. A lower voltage indicates the one tube is drawing more current than the other(s), and should not be used in that particular set. A one-volt difference isn’t worth worrying about (and is actually common), but a discrepancy of more than several volts is worth worrying about. If you have the capabilities, measuring Screen Grid current at idle and full output is a helpful indicator of whether or not you have serious troubles. This is especially true for a Marshall that likes to ‘eat tubes’, or blow fuses. Be aware that you could have parasitic oscillations, and that is a whole other story.
- Compare bias voltages, and on both ‘sides’ of the feed resistors. A one-volt difference isn’t worth worrying about (and is actually common), but a discrepancy of more than several volts is worth worrying about.
- Play the amplifier, and watch the tubes closely. Once you learn that particular tube and amplifier, you can almost eliminate this step. The following is a quick, abbreviated chart for the ‘typical operation’ and ‘design-center maximum’ ratings of a pair of EL34/6CA7 tubes, courtesy of the RCA Receiving Tube Manual (RC30). I only ‘bother’ with these measurements servicing a Marshall amplifier, or an EL34 equipped amplifier that likes to ‘eat’ output tubes.
|Zero-Signal Plate Current||120mA|
|Maximum-Signal Plate Current||143mA|
|Zero-Signal Grid #2 Current||20mA|
|Maximum-Signal Grid #2 Current||44mA|
There are only two things to worry about in biasing a guitar amp. The first… is not to set the bias current so high as to exceed plate dissipation at any portion of the tubes operation… The second is to not set the bias current so low as to sound bad…. There is no single ‘correct’ bias point.
Randall Aiken, aikenamps.com
I mentioned earlier that you do need to be careful in setting the amplifier controls when attempting to bias any Marshall. Below is a photograph of the same JCM800, but now the controls are set just a little too high for our purposes.
What you should note is that the pesky cross-over notch has actually returned. The square wave also shows ’tilt’, which in this instance is an indication of a sharper leading edge. A square wave is made up of a fundamental plus an infinite number of odd-order harmonics. This makes for an excellent ‘detective’ when you want to know about an amplifier’s frequency response. However, keep in mind that an amplifier without negative feedback cannot deliver a true square wave at any frequency, and the time constant of the feedback circuit will affect the results1. The waveform above could have low-frequency attenuation, or it could have a feedback circuit with a long time constant. However, this was using a clean sine wave input, so the technician should know better, and set his amplifier controls in an effort to avoid the waveform seen above. Below is the same signal generator, plugged into the same JM800, and now set for as clean a square wave as I could muster.
What you should ‘see’ is the lack of a cross-over notch. The frequency response of the amplifier isn’t very ‘flat’, or the output transformer is possibly at fault. Therefore, it isn’t prudent to discuss what is happening in this circuit, since I did not take the time to investigate specifically what was causing the poor square wave response. Besides, we are simply interested in ‘learning’ how using an oscilloscope biases any tube amplifier so cold that the average guru can see his breath standing next to the amplifier. For the curious; the above waveform would indicate poor low-frequency response, and there would be no ‘phase shift’ at this particular test frequency. However, as I have mentioned before, the output transformer could be at fault just as easily as the circuit design.
Now I would like to take another ‘step up’ in my oscilloscope choice, and see what I can learn about my JCM800. Next is an Elenco S-1345. This is a 40MHz oscilloscope, and from the photograph below, you should note that even the crossover notch appears sharper and clearer. I will keep the same set of Groove Tubes, and look at the waveform with the bias voltage returned to -44.2VDC.
Carefully jockeying the controls and the bias voltage, I achieve the following waveform.
The measured plate voltage is 456VDC, and the measured plate currents are 33.8mA and 35mA with -37.8VDC worth of bias. We have pretty well ‘nailed it’, so I feel pretty good about this oscilloscope. I decide to test play the amplifier before I think I am ready to try a different set of EL34’s.
The obvious problem is when has (the crossover notch) ‘just disappeared’? Most folks do just a bit more than ‘just disappeared’ and get their (output tubes) too hot causing shortened tube life and overheating. Not very accurate or repeatable.
R.G. Keen, geofex.com
Substituting ‘everything’, I select an 80MHz Sencore SC3080 oscilloscope. Next, the Groove Tubes are swapped out for a set of JJ EL34’s, I return the bias voltage to -44.2VDC. Just to keep things really interesting, I also switch to a 4-ohm dummy load. Hooking up the 1KHz test signal, I achieve the following waveform.
The Sencore SC3080 is a really nice oscilloscope with a few extra tricks up its sleeve. You can select a digital readout to display the frequency, peak-to-peak voltage, or the DC voltage measurement. Since the peak-to-peak voltage is of more interest to me, I chose that option. By using math (ugh!), we can say 37 volts peak-to-peak is approximately 12.95 volts RMS. Taking the next step, 12.95 X 12.95 / 4 = 41.76. This is to say this Marshall is achieving 41.76 watts, even biased pretty ‘cold’. This is all perhaps ‘useless information’, but I’d thought I’d throw it in there, just in case anyone cares. Carefully adjusting the bias voltage, the following waveform is seen.
‘Properly’ biased Marshall cannot achieve ‘extra’ output power, but the tone will definitely improve.
The measured plate voltage is 455VDC, and the plate current is 35mA and 36mA, or about 64% of maximum idle plate dissipation. Playing the amplifier at clean settings (and who does that with a Marshall?), the bottom-end seems as full and as tight as any Marshall I have heard, so I can safely assume this procedure went smoothly. Measuring the Screen Grid voltages and Control Grid voltages did not hint at future melt-downs, so for now we can close the book on this JCM800. I will do some ‘tweaking’ to the circuit, and those modifications can be seen at The Ultimate JCM800?
What, if anything, can be concluded from all of this? I have a few points to make, and you may come away from this ‘Lesson’ with a few of your own observations. This is all perfectly acceptable, and hopefully this will lead you toward doing your own experimenting.
- Most EL34 amplifiers biased strictly by the ‘crossover notch’ method can dangerously exceed the ‘70%’ bogey value of maximum idle plate dissipation, especially using extremely ‘high-end’ oscilloscopes. Accept this fact, and your life will be a lot less stressful. There should always be a very soft notch present, even at ‘clean’ levels.
- You should be aware of exactly when the crossover notch disappears, and back off the bias voltage slightly.
- The control settings on the amplifier made a tremendous difference in how accurate the bias adjustment can be. It cannot be over-emphasized; we must set the controls for a symmetrical waveform, and maximum unclipped output.
- The accuracy is also affected by the meter you use, and the method you use to measure idle plate current. If you are using 1-ohm resistors in the cathode, the 1-ohm resistors have to be matched, and be exactly 1-ohm. I opened up the plate connection, to avoid the screen current and insure accuracy. Measuring voltages with a fancy Fluke DMM gave ‘different’ results when compared to voltages measured with a ‘garden variety’ DMM purchased from my local Radio Shack. Therefore, your measurements will absolutely vary from mine.
- Although the bias voltage is reduced (less negative) until the notch ‘almost’ goes away, it returns ‘softly’ when the waveform is seen to start clipping from increasing the Volume control. Also, you can ‘tweak’ the bias voltage, by decreasing it slightly after the notch has just disappeared; beyond a point the waveform will not increase in size, and the top half will begin to flatten out. This is a good time to stop, and increase the bias voltage slightly. Remember to make sure your output tubes plates are not glowing a nice cherry-red, or that you have exceeded maximum plate dissipation. This is how you will learn exactly what you are doing, and the possible consequences.
- One critic argues that the negative feedback in the amplifier affects the bias adjustment when biasing with an oscilloscope, and removing the feedback loop negates the bias adjustment. I suppose if you cannot dazzle me with technically brilliant arguments, the next best thing is to baffle me with bullshit. I haven’t met a Marshall head yet having a feedback loop that liked to go AWOL when I least expected it, negating my bias adjustment.
- Oscilloscopes with a lower bandwidth and/or accelerating voltages yielded a trace which appeared ‘fuzzy’, making it difficult to distinguish exactly when the crossover just disappeared. As a theory, in many cases the crossover may not have completely disappeared, resulting in an amplifier biased on the cold side. Truth or fiction? The oscilloscopes with a good, clear trace showed the crossover notch disappearing more clearly, and resulted in an amplifier biased much ‘better’. If you learn to understand the waveform presented to you from your old Heathkit oscilloscope, you can ‘compensate’ for the fuzzy waveform, and make sure the crossover notch has disappeared. This is a criticism to the oscilloscope ‘method’, and I cannot argue that criticism. However, until you own a ‘good’ oscilloscope, you will simply have to learn to make compromises and compensations. Using your ears to fine tune the bias adjustment is another way to make sure you have it ‘right’. In the end, you will almost always have an idle plate current that falls somewhere between 10mA and 40mA, so why criticize how I got my highly accurate number? Also, I have yet to figure out where this ‘magic number’ of 70% of maximum plate dissipation comes from.
- The ‘dummy load’ you use will only affect your work if you are ‘sweeping’ the audio range, and plotting frequency response. A ‘better’ mousetrap involves a ‘load’ designed to emulate the speakers frequency response and varying impedances. Do you really need this kind of accuracy? Not really, but if your are the curious sort, investigate this circuit by CLICKING HERE. An excellent ‘article’ on this very topic can be found by CLICKING HERE. It is a very brief missive, but still makes some very good points. Recommended reading, to be sure.
A general rule of thumb is that class AB amplifiers are usually operated at no more than 70% of the maximum plate dissipation of the tube…….. This doesn’t mean you should automatically bias all tubes to 70% of maximum dissipation…….. they can be biased at any lower current if desired, and many people prefer a point of around 50% to 60% of the maximum plate dissipation, which contributes to longer tube life.
Randall Aiken, aikenamps.com
The whole point of this exercise was to show you that using an oscilloscope to bias your amplifier is not as bad as many gurus will have you believe. However, there are pitfalls that have to be addressed. Back at Articles That Didn’t Quite Make The Cut, you were given a hint that any oscilloscope can lead you astray. Make sure the oscilloscope you use is in good operating condition, and use the appropriate probes. The same carries over to your signal generator. Misunderstanding or misusing these techniques can also result in a very disappointing experience.
1) Audio Design Handbook, H.A. Hartley.