Strange PI voltage problem

[HBamps]

I am wiring up a new project this evening and have run across a very weird problem. The circuit is base on a blackface styled deal with GS - tonestack – GS – LTPI – PA (nothing complicated). On fire up I was checking my voltages and was looking at the plate voltages for the phase inverter (long tail pair). Rail voltage is about 314V, one side the plate was about 200V and the other was 312V.

I tried 3 different tubes, checked conductivity in the wiring, tried to re-tension socket, measured voltages all around it and I am STUMPED! To me, the 312V reading means it’s not connected to anything (open) as there is no current going though the plate resistor to drop the voltage.

Might any of you have any idea what is happening? FWIW, it’s the grounded side of a long tail pair (for lack of a better way of saying it). Could it be a bad grid coupling cap? (cathode and grid voltages seem normal)

Thanks for looking

[daz]

I would suspect a plate resistor that changes value under load. It may measure the correct resistance when you measure it, but with voltage supplied it may fail.

[HBamps]

I would suspect a plate resistor that changes value under load. It may measure the correct resistance when you measure it, but with voltage supplied it may fail.

So if a resistor fails, does it short? Then it would act like a wire. Wonder how the tube may operate in that manner.

Another fact is that the amp is very noisey (with or without guitar plauged in) and distorts with the vol on 2, past half and it makes a very nasty sound.

[Chuck H]

the amp is very noisey (with or without guitar plauged in)

Probably a grounding issue

and distorts with the vol on 2,

With no attenuation at all even a "normal channel" BF type amp has much more gain than an actual "normal channel" BF. With the normal channel BF amps there is a series resistor, the load of the trem circuit and the load of the other channels plate resistor that attenuate the signal prior to the PI. Also, if your using a 12ax7 as the PI tube that will have more gain than the Fender circuit with a 12at7. Also, often when people build this type of amp they use PI resistor values that are more Marshall than Fender. This will also add gain. Also, with this kind of build most will use a .022 or larger coupling cap into the PI instead of the 500p or 1000p cap that Fender used and that adds gain (in the very low end = mush). I like to use a split plate load of 22k/78k in series for the last preamp stage plate load with the 78k on the plate and the coupling cap connected to the junction of the two resistors.

past half and it makes a very nasty sound.

Probably a parasitic oscillation due to lead dress and component location.

The voltage imbalance on your PI is most likely a wiring error such that the triode reading high isn't conducting. Even though you've checked it, check it again... And again, until you find the error. Just because everything is wired up and the component values check out doesn't mean it's wired correctly. It may also be a bad tube. Try a known good tube in that socket.

Chuck

[Enzo]

One side of your tube is not conducting. You tried different tubes, so lets assume the tube is OK.

Either the circuit is open or the tube is cold.

Look closely, are BOTH heaters running inside that tube? For example what if the wiring for heaters went to pins 9 and 5, but the jumper to pin 4 from 5 was open? One side of the tube would not be lit.

And the circuit? You have 300v at the plate instead of 200, fine, what is at the cathode? Zero? Typically the two cathodes are wired together, is that the case, and is that wiring intact?

Your triode is not running, we don;t need complex scenarios, this is a straight DC problem.

And remember that in a tube socket, you solder wires to the little solder tabs, the tube pins meet little female soket pins. Each solder tab is supposed to be part of a female pin grip, but a pin can break in two inside, so the solder tab is not making continuity with the pin.

[HBamps]

With no attenuation at all even a "normal channel" BF type amp has much more gain than an actual "normal channel" BF. With the normal channel BF amps there is a series resistor, the load of the trem circuit and the load of the other channels plate resistor that attenuate the signal prior to the PI. Also, if your using a 12ax7 as the PI tube that will have more gain than the Fender circuit with a 12at7. Also, often when people build this type of amp they use PI resistor values that are more Marshall than Fender. This will also add gain. Also, with this kind of build most will use a .022 or larger coupling cap into the PI instead of the 500p or 1000p cap that Fender used and that adds gain (in the very low end = mush). I like to use a split plate load of 22k/78k in series for the last preamp stage plate load with the 78k on the plate and the coupling cap connected to the junction of the two resistors.

Chuck

Thanks Chuck. For the 22k/78k split: I am not sure what that does. together it's 100k for the full voltage drop but the plate only see's the 78k, therefore less gain?

Check out this link (http://ax84.com/media/ax84_m143.jpg). it's almost exactly what I built but mine has cathode biased PA and the coupling cap to the PI is .005. The clips of this amp don't seem overly gainy. I pretty sure I have done something wrong or there is a broken wire or something.

[Chuck H]

Thanks Chuck. For the 22k/78k split: I am not sure what that does. together it's 100k for the full voltage drop but the plate only see's the 78k, therefore less gain?

Almost. The coupling cap connected to the junction means there is 78k of series resistance and a 22k resistor to a near 0 impedance ground (impedance not resistance). Think of it as if that plate load were a 100k pot with one outer leg at the B+ and the other outer leg is on the plate. If you connect the coupling cap to the center tab the pot would act as a gain control. This circuit is like that but using fixed resistors. It acts as a voltage divider to the AC signal feeding the PI. So you get attenuation with fewer parts and no clumsy extra loads.

I might change the coupling cap to the PI to 2200p. That will at least roll off anything below 70hz or so that you really can't use anyway freeing up watts for more productive frequencies.

Also worth noting... Being cathode biased means that you can effectively subtract the cathode voltage from the plate voltage when you determine the plates working voltage. If you have 400V on the plates and 30V on the cathode your 6V6's will play and feel like they only have 370V on them. Since power tubes run at high voltages sound tighter and brighter than power tubes run at lower voltages you should expect fewer dynamics. The cathode bias also produces sag that will soften dynamics, especially on pick attack. So that's also going to effect the sound. Especially because your also tube rectified and that adds sag also. Something you could try would be to simply plug in a solid state rectifier. That will raise voltage and reduce sag somewhat countering the effect of the cathode bias. If your 6V6 plate voltage is already over 400V you don't want to do this though. It's almost as easy to build a bias supply for the 6V6's and run them fixed in that case. Another easy thing you can try is paralleling the cathode resistor with a pair of 5W zener diodes in series. Each one equal to half the standing cathode voltage. This would also be fixed bias, but you would still only have the plate voltage minus the cathode voltage as your working voltage. But you will get less sag making the amp more dynamic.

Chuck

[HBamps]

Almost. The coupling cap connected to the junction means there is 78k of series resistance and a 22k resistor to a near 0 impedance ground (impedance not resistance). Think of it as if that plate load were a 100k pot with one outer leg at the B+ and the other outer leg is on the plate. If you connect the coupling cap to the center tab the pot would act as a gain control. This circuit is like that but using fixed resistors. It acts as a voltage divider to the AC signal feeding the PI. So you get attenuation with fewer parts and no clumsy extra loads.

Cool, that seems pretty simple. I never thought of a voltage divider that way. I am always thinking of it on the AC side with one side grounded. This idea seems like a very useful one. Before I read your post, I installed a 22n cap – 220k/220k divider to gound – to the 5n cap – to the PI. (Seems like a lot of extra stuff that I could get a lot easier your way). Thanks!

I might change the coupling cap to the PI to 2200p. That will at least roll off anything below 70hz or so that you really can't use anyway freeing up watts for more productive frequencies.

I’ll try this.

Also worth noting... Being cathode biased means that you can effectively subtract the cathode voltage from the plate voltage when you determine the plates working voltage. If you have 400V on the plates and 30V on the cathode your 6V6's will play and feel like they only have 370V on them. Since power tubes run at high voltages sound tighter and brighter than power tubes run at lower voltages you should expect fewer dynamics. The cathode bias also produces sag that will soften dynamics, especially on pick attack. So that's also going to effect the sound. Especially because your also tube rectified and that adds sag also. Something you could try would be to simply plug in a solid state rectifier. That will raise voltage and reduce sag somewhat countering the effect of the cathode bias. If your 6V6 plate voltage is already over 400V you don't want to do this though. It's almost as easy to build a bias supply for the 6V6's and run them fixed in that case. Another easy thing you can try is paralleling the cathode resistor with a pair of 5W zener diodes in series. Each one equal to half the standing cathode voltage. This would also be fixed bias, but you would still only have the plate voltage minus the cathode voltage as your working voltage. But you will get less sag making the amp more dynamic.

Chuck

Thanks for all this, but I really like cathode biased amps. The idea about the zeners is new to me though.

[Chuck H]

Here's a neat trick then...

You can measure your cathode voltage as the amp starts to break up, let's say it's 14 volts. Bridge a 5W/15V zener (cathode toward the cathode pin) across the cathode bias resistor. Now you have an amp that gives you that cool cathode bias compression when played clean but tightens up into fixed bias as it starts to overdrive. This works brilliantly to maintain note articulation and keep the bass from going weak and mushy. It also reduces crossover distortion because you have a much lower bias shift. So you can have your cake and eat it too. 5W zeners can be bought in DO type case also so they're easy to manage. No chassis mounting like some high watt zeners. A 5W should be good in this app for 25W+ or so but if you want to be extra, over rated sure you can series two 5W/7.5V zeners making a 10W/15V (good for 50W) and still avoid chassis mounting. For a couple of bucks it's a fun experiment to see if you like it.

Chuck

[HBamps]

I tried the split load 78k/22k with a 3n coupling cap and it works REALLY well! I think this way of shaving gain is a MUCH better method than the coupling cap - voltage divider to ground - coupling cap - PI (which seems to change the tone and kind of load things down). The split load it essentially sounds the same (open and springy) but with lower gain, just what I was looking for. Thanks so much for this tip. Is this the same principal that's going on in the 2nd gain stage of the Brown Deluxe (6G3)?

Now I finally have the amp stable enough to fine tune the eq and such.

Almost. The coupling cap connected to the junction means there is 78k of series resistance and a 22k resistor to a near 0 impedance ground (impedance not resistance). Think of it as if that plate load were a 100k pot with one outer leg at the B+ and the other outer leg is on the plate. If you connect the coupling cap to the center tab the pot would act as a gain control. This circuit is like that but using fixed resistors. It acts as a voltage divider to the AC signal feeding the PI. So you get attenuation with fewer parts and no clumsy extra loads.

I might change the coupling cap to the PI to 2200p. That will at least roll off anything below 70hz or so that you really can't use anyway freeing up watts for more productive frequencies.

Also worth noting... Being cathode biased means that you can effectively subtract the cathode voltage from the plate voltage when you determine the plates working voltage. If you have 400V on the plates and 30V on the cathode your 6V6's will play and feel like they only have 370V on them. Since power tubes run at high voltages sound tighter and brighter than power tubes run at lower voltages you should expect fewer dynamics. The cathode bias also produces sag that will soften dynamics, especially on pick attack. So that's also going to effect the sound. Especially because your also tube rectified and that adds sag also. Something you could try would be to simply plug in a solid state rectifier. That will raise voltage and reduce sag somewhat countering the effect of the cathode bias. If your 6V6 plate voltage is already over 400V you don't want to do this though. It's almost as easy to build a bias supply for the 6V6's and run them fixed in that case. Another easy thing you can try is paralleling the cathode resistor with a pair of 5W zener diodes in series. Each one equal to half the standing cathode voltage. This would also be fixed bias, but you would still only have the plate voltage minus the cathode voltage as your working voltage. But you will get less sag making the amp more dynamic.

Chuck