Right And Wrong Ways To Bias?

Discussion in 'The Workbench' started by bigyinuk, Aug 8, 2016.

  1. bigyinuk

    bigyinuk Active Member

    Joined:
    Jun 5, 2016
    Messages:
    214
    Likes Received:
    152
    I've always biased my amps using the grid voltage method - This was how a Marshall tech told me they do the amps. i.e set the grid voltage to between -37V and -42V.

    All well and good I thought until today when I was trying to bias my recently purchased 2558 and the bias voltage would only adjust to -35V, it wouldn't get to -37V.

    So, I reverted to the "proper way" and popped an ammeter into the cathode circuit and checked the current.

    With the pot fully clockwise, and the grid voltage at -35V the cathode current was 27mA, way too low for an EL34.

    I tweaked the pot until the cathode current settled at 37.6 mA.

    Out of curiosity I then checked the grid voltage and that measured at -34V.

    Kinda disagrees with what I'd expect to see.

    I'm wondering if the voltage reading was low on my meter because it might have a lowish internal resistance? The grid current would be quite low so I guess that internal resistance (impedance) could effect the voltage reading?

    Any words of wisdom?

    Regards

    Jon
     
    Last edited: Aug 8, 2016
  2. neikeel

    neikeel Well-Known Member

    Joined:
    Feb 12, 2014
    Messages:
    4,855
    Likes Received:
    3,252
    That is how I thought Marshall biased their amps (-ve voltage on the grids method).
    So you have a clip-on ammeter - an unusual bit of kit unless you are a pro, or did you break the circuit and insert a meter into the gap (quite brave!).
     
  3. bigyinuk

    bigyinuk Active Member

    Joined:
    Jun 5, 2016
    Messages:
    214
    Likes Received:
    152
    It's easy to get at everything on the 2558, so I unsoldered the cathode connection, put some wire in the circuit and connected up my ammeter :)

    I used the formula: Valve power rating / cathode voltage x 700 to give me the cathode current in mA.
     
  4. danman

    danman Active Member

    Joined:
    Apr 27, 2014
    Messages:
    462
    Likes Received:
    155
    Location:
    Richmond, Va
    I would speculate that back in the day when tubes were very consistent, it was normal to set by grid voltage alone. With the wide variance in today's tubes, it is best to check the actual current draw.
     
    bigyinuk likes this.
  5. bigyinuk

    bigyinuk Active Member

    Joined:
    Jun 5, 2016
    Messages:
    214
    Likes Received:
    152
    Unless my multimeter is affecting the grid voltage reading, that certainly looks to be the case. I'll have to check all my other amps as I have re-biased most of them using that method. May also have to make up or invest in a cable to make connecting up easier.
     
  6. HAMPAMP TUBE AMP SERVICES

    HAMPAMP TUBE AMP SERVICES Member

    Joined:
    Apr 25, 2009
    Messages:
    168
    Likes Received:
    10
    Location:
    Black Mountain N.C.
    When it comes to bias, nothing is written in stone. The voltages specified my Marshall (or any other company for that matter) are meant to be a general guideline. Brand new parts in brand new amps vary in value but must meet a tolerance and guidelines are usually set for a certain amp with a certain tube from a certain manufacturer. After years of use, these tolerances can drift considerably a lot of times rendering the factory specs useless except as a good starting point and then only if you use the recommended tubes, ie...the brand that the manufacturer shipped the amp with..... The reason why the current was low could have been due to this tolerance drift but more than likely the fact that the factory settings using the -voltage setting are not set for the much higher 70% dissipation that most seem to prefer (which came to us from the Hi-Fi guys) but rather to a much lower point to improve reliability and long life for the tubes.
     
  7. bigyinuk

    bigyinuk Active Member

    Joined:
    Jun 5, 2016
    Messages:
    214
    Likes Received:
    152
    @HAMPAMP TUBE AMP SERVICES

    That's a sensible explanation - thanks very much.

    I can see considerable sense in setting the quiescent current as low as possible these days as valves are not as well made as they used to be and likely won't last nearly so long.

    I will have to experiment with lowering the bias on my amps and seeing at what point I can hear (unwanted) distortion creeping in from the crossover point.

    Is there a "recommended" lower setting ?

    Regards
     
  8. sloan_amps

    sloan_amps Member

    Joined:
    Jun 24, 2014
    Messages:
    83
    Likes Received:
    69
    Many manufacturers today set bias using the grid voltage most likely because it's fast and easy. It's terribly inaccurate in my opinion because adjustment by even 1 volt can sometimes result in a variation in current draw by several mA. It may be why I see so many amps with a very cold bias from the factory, which manufacturers seem to prefer in order to save on warranty issues.

    I have a pair of bias probes that I use, one from Eurotubes and one from Tube Depot. My problem with the Tube Depot probe is that is has a 1 ohm resistor built in that it reads across, but I find that the resistor measures more like 1.2 which results in a typical variance of 3 to 4mA, however it does let me know I'm close plus I mainly use it to monitor plate voltage, which it's perfect for. The Eurotubes probe though is great because it works the same way as inserting your ammeter into the circuit by reading the actual current draw passing through it.
    These are $25 each last I checked and make bias adjustments quick and easy.

    Of course, there is the old oscilloscope method where you monitor the waveform and adjust to get rid of crossover distortion, but this method is obviously more complicated and time consuming so is rarely used today. It used to be my primary method but I save so much time with the bias probes and setting for 70% dissipation, and generally get very good results that way.
     
  9. RickyLee

    RickyLee Well-Known Member VIP Member

    Joined:
    May 22, 2010
    Messages:
    10,038
    Likes Received:
    3,466
    Location:
    Riverside CA U.S.A.
    I am surprised that your valves are not running a bit on the hotter side at that control grid voltage setting. Most of the various EL34's I have need to see -40VDC to -45VDC to bias up around 65% idle.

    My preferred bias measurement is taking the voltage drop across each side of the output transformer winding and dividing that into the resistance of said output transformer winding. But of course you usually have to pull the chassis to do this. I do regret typing this info in the forum sometimes as I think some newbie might read it and try reading resistance on the OT winding with the amp powered ON.

    :ugh:
     
  10. bigyinuk

    bigyinuk Active Member

    Joined:
    Jun 5, 2016
    Messages:
    214
    Likes Received:
    152
    @sloan_amps Marshall also recommend using the one ohm resistor method as it protects your output transformer should something untoward happen. That resistor does need to be a very close tolerance though. It is possible to get very high accuracy resistors of course, so might be worth changing the one in your probe?

    @RickyLee I played around with the bias on my 2558 and to get the grid voltage to -37V I had to change the resistor in series with the adjustment pot (very common apparently). Setting it to this voltage however resulted in a quiescent current draw of 27mA. (50% power) Rather on the low side for an EL34.

    I still suspect that my ammeter is confusing the issue as its internal resistance is lowish and could potentially be affecting the results when reading the low current grid voltage.

    I think the cathode current method is the best way forward anyway as after all that's what you're adjusting for.
     
    Last edited: Aug 13, 2016
  11. ampmadscientist

    ampmadscientist Well-Known Member

    Joined:
    Jun 26, 2014
    Messages:
    17,865
    Likes Received:
    12,058
    Location:
    Bio-Isolation Lock Down
    Marshall biases amps w/ a scope, resistor load and sine wave generator. There is no bias probe, etc...
    Then, there is a check for input sensitivity. It has to meet the specification.

    The bias is adjusted hot enough to eliminate crossover distortion, prior to power amp clipping.
    After the power amp clips, (just after) then some crossover distortion appears in the waveform.

    Sensitivity: It's really important to a guitar player...but the bias probe has no way of knowing.
    Gain is not the same as sensitivity. Although frequently confused....

    Groove Tubes: They go for a small amount of distortion, just before clipping. (this would be colder bias)
    But, the tubes would last longer using this method. (less failures under warrantee)

    Sensitivity Tests:
    5 mv sine wave at input would drive the power amp to full power (this is 2204 specification).
    10 mv at input would drive the power amp to full power, (this is Plexi specification).
    Hotter bias = more sensitivity....the hotter it is, the more sensitive it becomes.
    Preamp tube gain will affect the sensitivity also....allowing higher or lower sensitivity.

    Bias Formulas: (for bias probes)
    Very inaccurate, because it does not take in to account low, medium, high current tubes.
    Tubes, being all different - require different bias settings. This is why they are matched into SETS.
    Without a scope, you really have no accuracy in the adjustment of bias.
    Low current tubes - could be biased too hot.
    High current tubes - could be biased too cold.
    The formula = only in the rough ball park....too much room for error.

    ma at cathode: That's ALL of it. That IS the final number. There is no "adjustment" based on plate voltages.
    And that is the flaw in your "formula."
    (I think your "formula" is a load of crap)

    Where did I learn this from?
    From an engineer, at the Marshall factory.
    And I have done it his way...ever since.
    This is how the Marshall factory checks your bias setting. There is no "formula," there is no "bias probe."
    But what there is: a reality check....no guessing.

    So, get your test equipment,,,and get w/ the program.
     
    Last edited: Aug 13, 2016
  12. mickeydg5

    mickeydg5 Well-Known Member

    Joined:
    Sep 30, 2011
    Messages:
    25,211
    Likes Received:
    12,166
    Location:
    US of A
    I was waiting to see if anyone would address this.
    He has an ammeter in circuit. How is he measuring cathode voltage?

    The negative control grid voltage should be tracked and an average can be used to ball park things but is not a good bias monitor because all tubes are different.
     
  13. bigyinuk

    bigyinuk Active Member

    Joined:
    Jun 5, 2016
    Messages:
    214
    Likes Received:
    152
    Interesting posts :)

    That "formula" is something that comes up all over Google when you search for EL34 bias. Yes, its only a rough guestimate as the 70% is only a approximation, but at the time all I needed was a guide to set the bias as I don't currently have an oscilloscope to look for crossover distortion.

    @micketdg5 Yes, I had an ammeter in circuit when measuring the current but before that I measured the plate voltage.

    I've also repeated these tests measuring the voltage drop across a 1 ohm resistor - easier to measure plate voltage.

    I've been having a look at the 6100, which has a 489V plate voltage.

    Using Marshall's suggested method of biasing by setting the grid voltage on Pin 5 to between -42V & -38V I looked at the current.

    Results:

    At -38V grid I measured 30mA and at -41V grid I measured 25mA and I couldn't *hear* any discernible difference.

    In the absence of an oscilloscope to do it 100% properly, I guess I'll go with -41V on the grid and 25mA current as presumably my valves will last a little longer that way?

    Regards

    Jon
     
    Last edited: Aug 13, 2016
  14. ampmadscientist

    ampmadscientist Well-Known Member

    Joined:
    Jun 26, 2014
    Messages:
    17,865
    Likes Received:
    12,058
    Location:
    Bio-Isolation Lock Down
    Plate dissipation in watts = you forgot a few things.

    1. Your Output Transformer Winding is dissipating power - but you left that out of the formula.
    2. Your speaker load is dissipating power - but you left that out of the formula too.
    3. The wattage is not "just" being dissipated by the tubes. What about the rest of the circuit?
    4. Your "formula" assumes that the full power is dissipated by the tubes ----but clearly, that's not correct.

    Wattage = Volts X Amps into a load ----wrong.
    1. Volts X Amps into a load = Volt Amps (VA) This is "Apparent Power." Not True Power.

    2. You never considered the efficiency of the load. This is the only way to calculate "Watts," which is TRUE power.

    3. So you assumed that "plate dissipation" was in Watts ---but you never realized it was VA instead. You never came up with a true number for "Watts...." Because you have no idea what the efficiency of the load IS!

    4. You also assumed that "all tubes are the same..." "all tubes bias the same."
    But that is also far - removed from reality. The reality is that: tubes are all different.

    :monkey:And THAT is why your "formula" fails....... so miserably.:monkey:

    (whoops)
     
    Last edited: Aug 13, 2016
  15. bigyinuk

    bigyinuk Active Member

    Joined:
    Jun 5, 2016
    Messages:
    214
    Likes Received:
    152
    I remember all that from my City & Guilds days back in 1980.
     
  16. sloan_amps

    sloan_amps Member

    Joined:
    Jun 24, 2014
    Messages:
    83
    Likes Received:
    69
    My probe has a 1% tolerance resistor in it just like the ones included in amps with bias test points, and I've gone through a lot of the other 1% resistors I have and none of them are close enough for me, because even a 1% variation in the resistor can have a much greater swing in the bias reading. However, I could just measure what the actual resistance is and multiply that by whatever the meter reads and get much closer.

    The other problem with reading current draw across a resistor on the cathode is that you are actually measuring cathode current which would require you to subtract the screen current from that in order to truly get the plate current. The Eurotubes probe I use actually just allows me to insert my meter (or ammeter to be more correct) and read the plate current draw directly. This is actually the most accurate method I have found, and over the years I've tried them all.
     
  17. sloan_amps

    sloan_amps Member

    Joined:
    Jun 24, 2014
    Messages:
    83
    Likes Received:
    69
    My summation of this subject:
    Everyone has their preferred method of biasing an amp...some are more accurate than others of course, but each person needs to find the method that they are comfortable with performing safely and to the level of accuracy they are happy with. For instance, not everyone is comfortable using the shunt method, and not everyone can afford a signal generator and scope, but this does not mean they can't bias their own amp with some degree of accuracy.

    I can tell you from my experience, in the almost 30 years I've been doing this, I've tried every method and I prefer biasing by reading the actual plate current. For years, I used a sig gen and scope, but the amount of extra time and effort using that method didn't have any better results than the way I do it now. Besides, I've seen Marshalls, Peaveys, and Mesas, right off the shelf that could not be biased out of crossover distortion without modifying the amp, and I've seen amps many times ended up with extremely cold bias that way and sounded like crap. However, I'm pretty sure that rising labor costs and the push to control expenses are the main reason why these factories quit using the scope method except in the design/prototyping phase. All of them I've dealt with in the last few years now either bias according to the negative grid voltage or cathode current, which is generally what they'll put in their manuals, if they even share the info.

    It's also important to note that not all probes are created equal. Probes with resistors in them need to be measured to determine the actual value of the resistor in them and factor that into the current reading you get as well as take into consideration the screen current if you're reading across the cathode. In my opinion, the best way still is to get one that will directly measure the plate current.

    Of course, how accurate do you need to be? In my opinion, accurate enough to find a balance between tone and tube life expectancy. After all, we're talking about guitar amplification, not hi-fi (and both of those are a long way from the tube-powered communications & radar equipment I worked on in the military years ago!). I can tell you some amps sound their best when biased hot, yet you end up replacing tubes every 6 months, so that's where you have to decide what to compromise. That's essentially why someone came up with the idea of biasing a tube at 70% of its dissipation. I will sometimes vary that by 5% to 10% depending on the amp and customer's wishes.

    For a really great article on biasing, read Randall Aiken's article here:
    http://www.aikenamps.com/the-last-word-on-biasing
     
    neikeel likes this.
  18. RickyLee

    RickyLee Well-Known Member VIP Member

    Joined:
    May 22, 2010
    Messages:
    10,038
    Likes Received:
    3,466
    Location:
    Riverside CA U.S.A.
    I replaced the little 1/4W 1 Ohm bias resistors in my TSL100 with 2W metal flameproof. Bummer is they are not the precision type and have a 5% tolerance. But for me that is close enough on the bias. 5% of 33mA is only about 1.6mA. Not enough to break the bank for me.

    So yesterday I wanted to check those resistors. I let the TSL100 run for a bit then checked bias from those 1R resistors via the test points on the back of the amp. I had 66mA on each side of the output transformer (two EL34's each side). I then checked the bias by measuring the voltage drop across each side of the OT divided into the resistance of the windings. I had 61mA on each side of the OT after the calculations. Each EL34 had aprox 2mA of screen current at idle. That gave me 65mA on each side and only 1mA off from the method using the 2W 1 Ohm bias resistors.

    On a side note, when I had the stock 1/4W 1 Ohm bias resistors in the amp, each side was off by plus 3mA measuring via those little 1/4W resistors. And one side was actually almost at plus 4mA.
     
  19. bigyinuk

    bigyinuk Active Member

    Joined:
    Jun 5, 2016
    Messages:
    214
    Likes Received:
    152
    Thanks, You can go to extreme lengths but I think that para hits the nail on the head.
     
  20. mickeydg5

    mickeydg5 Well-Known Member

    Joined:
    Sep 30, 2011
    Messages:
    25,211
    Likes Received:
    12,166
    Location:
    US of A
    That is why your formula seemed confusing. I believe you meant "Valve power rating / plate voltage x 70% to give me the cathode current in mA" to be stated.
    Like 25W / 460V x 70% = 38mA for example. Note that if measuring at the cathode then the screen current and plate current should be witnessed.

    If your bias is on the low side you may not hear much or any difference. When you bias higher it may bring out a bit more presence and power particularly at lower volumes in the output from the power tubes. The 30mA and 25mA are far lower than 38mA plus. At 25mA the amplifier is on the verge of getting closer to crossover distortion although -41V is not bad and it may just be less efficient power tubes being the cause.

    I tested by substituting a half dozen sets or more a while back in an amplifier with the negative control grid set at -38VDC for each. The idle current ranged from the lowest set at 36mA to the highest at 52mA. I am sure with some of todays tubes the variance will be greater. It has been noted and discussed in threads on this board.
     
    Last edited: Aug 13, 2016

Share This Page