Hoffman Amplifiers Tube Amplifier Forum
Amp Stuff => Tube Amp Building - Tweaks - Repairs => Topic started by: mpayne on April 19, 2010, 07:35:18 pm
-
Sorry to keep bugging you guys but I'm at a loss again. When I place my multimeter across either of the 1 ohm resistors and power on my amp, the current slowly starts to increase as the tubes heat up and just keeps going to ~100 ma. I then turn the amp off for fear of damaging something. I've set the pot R36 such that the resistance from R40 to ground is 22k and both bias still sky rocket when I turn the amp on. I've probed the amp (with power off and caps discharged) to check for bad connections, something grounded, wrong resistor values, cold solder joints and I've found none. Any suggestions on what I should check next?
Thanks for the help, I really appreciate it
Mike
-
Before trying to measure tube current you need to check your bias voltage. Just pull the output tubes and lay them aside until the bias voltage is good. Connect one meter lead to ground and connect the other to V4 pin 5. Adjust the bias pot for maximum negative voltage. Now move the meter lead to V5 pin 5. You should have the same negative voltage that you measured on V4 pin 5.
How much voltage did you measure?
-
Since the amp does not generate bias voltage until after the standby switch is on I would remove the power tubes and then power up the amp (both power and standby) and measure the voltage at pin 5 of each power tube socket and see if you have a negative bias voltage to each power tube socket and note what that voltage is. If one or both of the sockets does not have bias voltage at pin 5 then trace back through the bias circuit and see where the voltage stops. Could be an open circuit/bad solder joint in the bias circuit. That would be a good place to start - confirm that you have bias voltage at the power tube grids.
I would really consider moving that bias voltage tap (just after the standby switch and before the rectifier diodes) to the other side of the standby switch so that bias voltage will develop to the power tubes before the main B+ is applied. As it is wired now you have no bias voltage to the tubes until you throw the standby switch. It may not be a problem but I'd prefer full bias voltage up and running before I applied B+ to the tubes.
-
I would really consider moving that bias voltage tap (just after the standby switch and before the rectifier diodes) to the other side of the standby switch so that bias voltage will develop to the power tubes before the main B+ is applied.
Good catch. I strongly agree.
-
The voltage at pin 5 for V4 and V5 started out at -6.8v. I left the 3 12AX7 in, but pulled the power tubes. As the time went by the voltage started increasing and after about 5 minutes it was at -5.8 for both. Does that sound right?
Thanks,
Mike
-
The voltage at pin 5 for V4 and V5 started out at -6.8v. I left the 3 12AX7 in, but pulled the power tubes. As the time went by the voltage started increasing and after about 5 minutes it was at -5.8 for both. Does that sound right?
Sounds a little on the low (too positive) side to me. We know it's low since your power tubes are pulling 100ma each. While you are measuring your bias voltage with the power tubes removed adjust the bias voltage pot back and forth. Is it possible you have it set to minimum instead of maximum? What is the bias voltage range at pin 5 or each tube when you adjust the pot from one end of its sweep to the other? If it swings when you adjust the pot and you do have it at maximum and are getting -6 or -5 VDC then you must change some of the resistor values in your bias circuit to get more negative voltage to the grids.
-
I was going to say have you tried changing R40 to a 68K or R84 to a 180K? Thats where your problem is. I have built countless amps and i can not remember but maybe one time where the bias resistor values on the schematic had me up and running right in the correct bias range. Change either one of them one at a time and tell us your reading
-
Quote:"Sounds a little on the low (too positive) side to me. "
No kidding!You need to have a range of -50v to about -30v to even start with a pair of big bottle tubes.
Get the bias supply in that range with no tubes installed and then you are ready to put the tubes in and check.
-
Doug has a Library of information on this site, here is one of the pages on biasing http://www.el34world.com/charts/fenderservice5.htm its geared towards fender amps but it applies to most any amp. Its a lot of info but you really should have some understanding of what your trying to do as your doing it. Had you stared here it would have been obvious when you measured that -6.8v not to go on and adjust the bias range resistor, (sounds like you may even have the wrong value resistor in R40 or R84) You need to adjust those resistors to get the negative voltage on pin 5 to around -50v before putting the power tubes in. Read thru it and try understand it. Doug recomends reading the entire Library before building an amp which is probably a good idea but I dont know that anyone has ever done it. Take your time & double check the Resistor values around the bias board. .. Joe
-
I know there's an accurate formula somewhere for this rule of thumb, but just take my word for it...
With tubes like EL34 or 6L6, always look at your screen voltage (G2, which is pin 4). Divide that number by 10, then stick a minus sign in front of it. This is a good starting point for how much bias voltage you need, and you want to be able to adjust it to have some range above and below.
For example, say you have 450v on both plate and screen. 450/10 = 45, so -45v is what you want on pin 5. It would be nice to be able to adjust between -35v to -55v.
The rule of thumb is loosely based on the triode amplification of a pentode, sometimes seen as mu g1-g2 (mu between the control grid to screen grid). This is around 8-10 for a 6L6, and is used to estimate a bias voltage as a starting point. This is not an optimum bias voltage, because it doesn't account for what class you're running in, and doesn't have the fudge factor that probably goes along with the "right way" of figuring this. But it does tell you immediately that -6.8v is waaaay wrong and needs to be investigated.
-
Changing R84 to 180k made no difference. It stayed at -6.7. Changing R40 to 68k made it -7.8. Also, pin 4 is 513 VDC, which seems a little high to me. I've checked the bias pot (R36) and it's set to 25k (i.e. I measured the ohms between R40 and ground and it's 25k). I've probed that bias circuit several times. I can't find any shorts, all the values are correct, the diode is working. What could I've done wrong to make this happen.
Thanks again,
Mike
-
Change R84 to 100k Change R40 to 68K. What is your reading?
-
Changing R84 to 180k made no difference. It stayed at -6.7.
What does it read when you rotate the pot fully in the oppsite direction? ( with out the power tubes in)
Can you take a good pic and post it?
I think you can try removing the one leg of the pot that goes to ground as in this layout http://www.el34world.com/Hoffman/images/Plexi50Old.gif that way its a variable resistor instead of a voltage divider. .. Joe
-
Now, I don't claim to be much of an expert but here is where I would start..Pull you Power tubes, turn the amp on and then measure the voltage at the connection between R28 and R27. Rotate you pot and see if you are getting the changes you would expect..If so then check C10 and C11 and make sure that they are not leaking. Take a hard look at R28 and R27 and make sure they are the correct value.. Just some thoughts from a rookie.
-
"I was going to say have you tried changing R40 to a 68K or R84 to a 180K? Thats where your problem is."
If he raises the value of R84 isn't that going to deliver less voltage to the bias network? Wouldn't you lower R84 and then raise R40 (if necessary) to get the basic voltage range where you need it?
-
Now, I don't claim to be much of an expert but here is where I would start..Pull you Power tubes, turn the amp on and then measure the voltage at the connection between R28 and R27. Rotate you pot and see if you are getting the changes you would expect..If so then check C10 and C11 and make sure that they are not leaking. Take a hard look at R28 and R27 and make sure they are the correct value.. Just some thoughts from a rookie.
Also C10 and C11 need to be positive side to ground since the bias network is negative voltage.
-
I've been skimming through a bunch of schematics looking at fixed bias networks and that R84 dropping resistor at 120K seems to be huge in comparison to everything else I am seeing. Most of what I am seeing are anywhere from 1K to 5K. I'm betting that resistor is just way too big for this circuit and is dropping all of your voltage before it ever gets to the diode.
-
Changing R84 to 100k and leaving R40 at 68k caused the voltage to go to -8.3 I've attached a picture of that part of the board. I don't think C10 and C11 are polar, but let me know if I'm wrong.
Thanks again for your time,
Mike
-
C22 & 35 are in backwards, the positive end should go to ground in a Bias circuit. . Joe
-
mpayne - Yes, you're right about c10 and c11. I got off track and confused those coupling caps designations with the bias ckt caps c22 and c35. I apologize for the confusion. I also thought you were building an amp from scratch and not working on an existing amp. Did it ever work at one time?
I think I'm just muddying the water and will drop back a bit and let the others help you sort this out.
-
Bingo! Thanks soooo much. Yes the caps were backwards. Turned them around and the voltage jump to ~-80v (or I guess dropped to -80v). You guys are great! I"m going to read up on tube biasing before I continue and make sure I don't mess something else up.
Thanks again,
Mike
-
Fantastic! Good job guys. Now put them bias resistors back as they are on your schematic and give us both the low and the high readings from pin #5 when turning your trimmer pot
A pic is truly worth a thousand words!
-
Okay, I've put back the original resistors and when the trimmer is at max ohms it's -62v at pin 5 and at min ohms it's -46v at pin 5. Is that about right?
Thanks again for all the help,
Mike
-
I've put back the original resistors and when the trimmer is at max ohms it's -62v at pin 5 and at min ohms it's -46v at pin 5. Is that about right?
Yes. Set the bias pot for max negative bias voltage on pin 5. Then put your power tubes back in. Now you should be able to safely set the cathode current to the value you want.
-
It Works!!!!!!!!!!!! Turned it on, bias was about 9mv across the 1ohm resistor. I adjusted it to ~30mv and it sounds just fine. Thanks again for all of your time. I really appreciate it.
- Mike
-
Feels good to find and fix a problem with a little help doesnt it?
-
Yes :grin:
Thank you sooo much for your time. I really really appreciate it.
- Mike
-
> I know there's an accurate formula...
No.
In principle you can plot on tube data curves. In fact that's too much work, fudging for high G2 voltage. And the low-current area of the data is most variable tube-to-tube and decade-to-decade.
> EL34 or 6L6, always look at your screen voltage (G2, which is pin 4). Divide that number by 10, then stick a minus sign in front of it.
That is correct for "cut-off".
Assuming that Mu stays "10" at low current.
And assuming we want cut-off.
For smoke-testing a fix-bias amp, we DO want cut-off to start out.
But Mu falls-off at low current. (Mu is zero at zero current.) So Vg2/10 won't actually give cut-off. mpayne had more, -62V bias, and got 9mA. Which is pretty cold for a big bottle. Not cut-off, but a good start-up value.
For fat Class A we want a lot more current. So we need to be able to set a lesser grid voltage. About half-way from cut-off to max. Max happens at zero grid voltage. We never want to bias hotter than half of max.
> say you have 450v on both plate and screen. 450/10 = 45, so -45v is what you want on pin 5.
Then I would want more than 45V available. And I would never want to be able to turn to less than half of that, 22.5V. And in hi-power gitar amps, not even to that "half".
> It would be nice to be able to adjust between -35v to -55v.
Yeah, usually a good zone for your assumed G2=450V condition.
Since we did not know what else was wrong with mpayne's amp, the -62V was a good extreme. It turned out that everything was fine except, as predicted, -62V was mighty cold (9mA). The amp is safe, and plays, but a bit rough at low level. Turning toward -50V or so brings current up to a sweeter level.
FWIW: mpayne's report of 9mA at ~~460V on G2 and -62V on G1 does not agree with the 1963 Tung-Sol 6550 data. It "should be" 30mA. But I don't trust the old data down at the bottom, and 6550s are not made in Bloomfield NJ any more, they are made by very different people on very different machines. We do demand that a "6550" can pass 500++mA when driven hard; we don't care so much about teeny currents.
-
I swear there's a formula given for "optimum bias" in the front matter of the RCA Tube Manual. It's something like 0.56*mu*Eb (assuming a triode). So something like 0.56 * mu(g1-g2) * Screen voltage should be workable. EDIT: Nope, that forumla is wrong, as you'd have to divide by mu, not multiply. I do notice that that 0.56 fudge factor moves the bias away from cutoff and towards a working bias point.
Of course, this makes no allowance for class A, lean Class AB, etc. On the face of it, it appears a working point for class A operation (assuming that's allowed by the plate and screen voltages).
-
> there's a formula given
Yeah, like "a 14 foot rafter for a 10 foot span". That works great on a 45 degree roof, runs a little long on a low-pitch roof, and is way wrong for a steep-pitch roof.
It gives "Class A", pretty near.
Figure Vg2/Mu as cutoff voltage CV.
Then 1 * CV is cutoff, 0 * CV is full-on.
For Class A you want to idle half-way between.
If the tube were perfectly linear, you would bias to 0.5 * CV.
Tubes turn-on better than they turn-off. So for half-current we have to lean towards cutoff.
0.6 * CV is a good approximation for most tubes.
> assuming that's allowed by the plate and screen voltages
And load impedance.
-
And you were right, I boogered the formula.
What I was thinking of was 0.56Eb/Rl, which gives an "optimum idle current" that is slightly higher than halfway to cutoff. You would still have to refer to the data sheet to see what bias voltage this translates to.
Sorry!
-
RCA does suggest either 0.5 or 0.6 (depending on decade) cutoff as trial bias for a power stage.
"0.56Eb/Rl" looks like plate current for a voltage amplifier pentode? For maximum output? In a simplified situation you bias the plate to about half the supply voltage for equal swing both ways. If you have a load your up-swing is a bit less; since tubes won't turn-on dead-short your down-swing is a bit less. Usually the up-swing is the problem, so you aim a bit lower than half the supply voltage. "0.56Eb/Rl" sets the plate at 0.44 times B+. Someone thought this was a good general guide, and it got into the literature.
Of course when you throw 20%-step parts (100K, 120K, 150K, 180K) and +/-20% tubes together, you will really get anywhere from 0.4 to 0.5 times B+. All of which work fine too. Dunno why they didn't give the number as "0.55".... it is a LOT easier to find on the slide-rule (and a bit quicker on a calculator).
-
Yeah, the formula was to arrive at an optimum idle current, not a bias voltage. And it was given as a general guide, obviously to give a working point for a class A amp. So I remembered wrong the whole way round! Oops... :rolleyes: