Help! Newbuild Bassman with Asymmetrical clipping

[priesty]

Hello there everyone,

I've scratch built myself a 5F6A Bassman using 6550 output tubes, I'm having real problems with assymetrical clipping at high volume levels. I previously noticed that my supposedly "Matched" pair of output valves were drawing about 20ma difference in idle current, so installed a dual variable bias control circuit to perfectly match the idling current at 50ma per tube. I also increased the amount of negative feedback by replacing the 27K resistor with a 12K.

Putting the amp on the scope, clipping only appears from the output of the phase inverter forward, this remained the same when I changed the PI tube from a 12AX7 to a 12AU7. The clipping manifests itself on the scope as the "bottom" peak of the sine wave clipped with the top intact. Changing to the dual Bias control helped this a small amount, but not as much as I'd hoped. Distortion at full power can only be described as "farty"!

Any thoughts would be much appreciated!

[loudthud]

Does the clipping look the same if you swap the 6550s? Have you double checked the screen resistors?

[Jazz P Bass]

Power rail limit

I agree with LD. Swap the tubes.
The top & bottom "should" reach the power rail at the same time, when overpowering the tubes.
Look at the waveform before clipping occurs.(ie: clean) Is the width of the top of the curve the same as the bottom?
If not, that is a clue.
As to the distortion sound, the fact that the waveform is not symetrical, this may be what you are hearing.
Personally I have never understood the attraction of the sound of a waveform that reaches the limits of the power rail.
Can you say square wave?

[priesty]

Does the clipping look the same if you swap the 6550s? Have you double checked the screen resistors?

Loudthud, thanks for that, just wondering what I should be checking with the screen resistors? On both Grid stopper and screen resistors I have 1K 5 Watt units.

[tubeswell]

Putting the amp on the scope, clipping only appears from the output of the phase inverter forward, this remained the same when I changed the PI tube from a 12AX7 to a 12AU7. The clipping manifests itself on the scope as the "bottom" peak of the sine wave clipped with the top intact.

So you don't think this is grid current limiting happening in the PI? (That can cause the plate output to appear clipped on one side, and can cause farty distortion at higher vol). If it is happening in both output tubes, then its too weird to be originating in the output tubes. Cause might be too mass bass in the signal too early on in the circuit, so could try smaller coupling cap values, or 10k-100k grid stoppers on the pre-amp tubes, or 1k5 - 4k7 grid stoppers on the 6550s maybe. 2CW

[priesty]

So you don't think this is grid current limiting happening in the PI? 2CW

Gents, Thanks for your suggestions and your quick responces! In answer to your suggestions:

1 - I swapped power tubes, and clipping still occurred, maybe a tiny amount less clipping.

2 - Screen Resistors are 996ohm and 999ohm, pretty close match (but close enough??)

3 - Please see attached photos of waveform at OT secondary, just before distortion and just as distortion set in. Signal generator was my Jazz Bass Plating an open "A" (apologies for the quality!)

4 - Tubeswell -hows things in Windy Wellington???! It could well be grid current, Morgan Jones mentions it in his books, but don't fully understand the problem. Would I just need to add the grid stoppers to the PI tube as the rest of the preamp tubes have no clipping? As mentioned before, 6550's have 1K 5W grid stoppers (0.02% matched).

Cheers!

[Jazz P Bass]

Clipping

Yeah that second photo is clipping the rail.
One question. How loud is the amp before this clipping.
Where are the volume & tone controls set.
You get a lot of voltage gain on the preamp when you dime the tone controls.
Can you measure the AC wave at the speaker with a good DVM? It would be nice to know the wattage that this is happening at.
My best guess is that if you turn it up a little bit more, you will see both the top & the bottom clipping.
Do you have the 82K , 100K plate resistors on the PI?
That different resistor value was supposed to "help" with the symmetry issue.
(that does not say it solves it)
Then again you may be at the "clean" limit of the amp.
I have a Super Reverb on the bench that just starts to square wave at 4 on the volume. 100mv input. 30 watt output.
That is the way that Leo built them.

[priesty]

Yeah that second photo is clipping the rail.

Can you explain what you mean by "Clipping the Rail? It's not a term I'm familair with.

One question. How loud is the amp before this clipping.
Where are the volume & tone controls set.
You get a lot of voltage gain on the preamp when you dime the tone controls.
Can you measure the AC wave at the speaker with a good DVM? It would be nice to know the wattage that this is happening at.

To my ears, sounds like about 30 Watts the amp is clipping at. The 'scope traces you see are 5V/div so clipping is manifest at OT secondary at 10 Volt. is we take this as RMS of 7.1 volts into the 8 ohm load - 6.25Watts!!! All tone controls are set to halfway.

My best guess is that if you turn it up a little bit more, you will see both the top & the bottom clipping.

With Volume and tone controls maxed out, only the bottom part of the signal trace clips. The distorted trace I posted previously was of input 1 of "normal" channel with the volume flat-out.

Do you have the 82K , 100K plate resistors on the PI?

Yes, these are fitted as per 5F6A - I know because an earlier revision of my schematic had these both 100k, and I picked it up before "manufacture",

I certainly feel that I should have more clean power available with a pair of 6550's. I am looking at the phenomenon of blocking distortion, might this be the issue?

[Jazz P Bass]

Rail clipping

The power tube essentially modulates the power supply.
You cannot get any more than that.
When the input signal to the power tube is large enough, the waveform on the output "hits the rail" & flattens.
Quote: "The 'scope traces you see are 5V/div"
If this statement is true, I am seeing almost a 20 volt waveform.
That's 38 watts into your 8 ohm load.
If that was with the volume control dimed, what you see is what you get.
Turn it down.

[tubeswell]

Tubeswell -hows things in Windy Wellington???!

Bloody windy today (but it was fine last w/e so I didn't get much amp building done)

It could well be grid current, Morgan Jones mentions it in his books, but don't fully understand the problem. Would I just need to add the grid stoppers to the PI tube as the rest of the preamp tubes have no clipping?

There are two types of clipping:-

(i) Grid current limiting, which happens when the grid is driven more +ve than the cathode (on the +ve part of the signal swing) causing the grid bias voltage to drop and the output signal to appear clipped off. This is not plate clipping. because the plate is still amplifying what is appearing at the signal grid.

(ii) Cut-off clipping, which happens when the input voltage swing coming from the signal on the grid is so negative, that the flow of current in the tube is cut off.

In normal clean-biased stages the grid signal can swing happily up and down and the plate will amplify an unclipped signal. But if the input signal swing is too large, then you will ultimately get grid-current limiting happening on one side of the swing, and cut-off happening on the other side of the swing, resulting in clipping on both sides.

However if the stage is biased with a low bias voltage, you are more likely to get grid-current limiting clipping on one side, and if the stage is biased with a high bias voltage, you are more likely to get cut-off clipping on the other side.

Hi-gain amps use alternately biased stages to get that real creamy sustain.

However, if your bag is more cleaner signal, then the solution is to adjust the bias of the stage (by changing the cathode resistor) to the point where the signal clips evenly, and then attenuate the input signal so that there is enough headroom to ensure that clipping doesn't happen.

Now if the frequency response of each stage is set up to pass maximum bandwidth, you are more likely to get something called 'blocking distortion', which is a farty/blatty type sound heard especially when the amp is turned up and big ('power') chord, or bass notes are struck. Blocking distortion happens when (bigger value) coupling caps get so much charge in them that they can't release the charge fast enough (through the grid leak resistor), and the grid bias voltage of the following stage goes positive enough to cause undesirable levels of grid current. The solution for blocking distortion is a combination of signal attenuation between stages, and reduction of coupling cap values, and use of grid stoppers to prevent undesirable levels of grid current getting through the following stage and causing excessive grid current limiting. Values ranging up to 100k are quite workable for grid stoppers without affecting the tone too noticeably. Taking a bit of bass out of the entire signal (i.e.; by using smaller - e.g.; 2.2uF - cathode bypass cap at V1) can also help, as this has a cascading effect along the rest of the signal path.

As mentioned before, 6550's have 1K 5W grid stoppers (0.02% matched)

You could try 1k5-4k7, or you could experiment with reducing the value of the coupling caps from the PI. But from what you describe, it sounds to my humble hobbyist point of view like the problem is before/going into the PI. Others might have a different view. Try experimenting and see.

[priesty]

Now if the frequency response of each stage is set up to pass maximum bandwidth, you are more likely to get something called 'blocking distortion', which is a farty/blatty type sound heard especially when the amp is turned up and big ('power') chord, or bass notes are struck.

.

Tubeswell, thanks for your detailed description, it is much appreciated! I've done a bit of digging on this Grid Current "Blocking Distortion" and think this might be what is happening at the inlet to the Phase inverter. I intend to change the 1Meg grid leaks to a lower value and add grid blocking resistors to the phase inverter stage, I've got tons of gain, so the small losses associated will be worth it for the clean signal that will be fed forward to the Power tubes. Fingers crossed!!!

[MWJB]

Yes, you should have more than 30W with 2x6550. What is your power tube plate voltage (& current). What are the PT ratings? I'm not surprised that power is a little low, because you dropped the value of the NFB resistor. With an 8ohm load I'd be looking more towards 56K-100K (bigger, not smaller)? You are currently feeding back over a quarter of your output voltage (3.4 to one NFB ratio)!

Measure the AC input to both power tubes at 3.85mVAC input & 3x12AX7 (not knowing what tube you have in V1, 100mVAC on the input could be turning the amp inside out with a 12AX7), all controls fully CW, what do you get?

2 options - larger PI tail resistor, if you have 6.8K try 10K, if you have 10K try 22K. And/Or, tweak plate resistor values (reduce value on the overdriven side), especially if the amp starts to get too loose with a 22K tail.

Out of interest, have you tried a 12AT7 in V3, what tube do you have in V1?

"I intend to change the 1Meg grid leaks to a lower value and add grid blocking resistors to the phase inverter stage" I'd hold off on that until the balance issue is sorted and you're getting reasonable W RMS from the amp. Preamp grid stoppers should not be necessary with a good layout. 60-70W RMS would be more typical with a plate voltage of 470vdc+.

Any pics of the layout?

[priesty]

With an 8ohm load I'd be looking more towards 56K-100K (bigger, not smaller)? You are currently feeding back over a quarter of your output voltage (3.4 to one NFB ratio)!

Yes, I will change NFB , I was scratching around to improve the distorsion without really knowing what I was doing, more NFB definatly wasn't the cure here!! Even toying with the idea of reinstating the 27K resistor and taking the feed from the 4ohm OT tapping - not sure how much gain margin I have though.

V1 is the old 12AY7, didn't want too much gain for a bass amp!!

I will try altering the tail resistor first - it is currently 10K (see my attached schematic - apologies for the PriestyAMP Branding - a joke with my friends!!)

I tried a 12AT7 in V3 - it was a bit harsher than the 12AU7 and actually sounded brasher than the JJ ECC83 I had in V3 as per design.

Power section: Plate voltage is 411V, max current (RMS from fluke) was around 130mA. Plate voltage sags to about 380V with the dual GZ34 rectifier design. Power transformer is rated at 243VA, it's a lump of metal! Hammond 372JX if you want to look up particulars.

Please don't laugh at my layout photo! I know its not pretty, but no paracitic oscilation or RF problems (Inputs and V1 Grids are sheilded).

[loudthud]

Loudthud, thanks for that, just wondering what I should be checking with the screen resistors? On both Grid stopper and screen resistors I have 1K 5 Watt units.

The screen resistors just need to be within 10% or so. If one of them gets damaged in a way that makes it suddenly go to 10K, that would cause severe clipping on one side of the output.

Looking at your transformer set, the output transformer is too high an impedance for 6550s with that low B+. Try running with the next higher load impedance tap.

[MWJB]

OK your low B+ is going to limit power. You could try SS rectification and back of plate current to just stave off X-over? Your amp probably wont make much more power than if it was fitted with 6L6s, indeed there are undoubtedly 6L6 amps that will make more outright power.

"Please don't laugh at my layout photo! I know its not pretty, but no paracitic oscilation or RF problems (Inputs and V1 Grids are sheilded)." I'm biting my lip so hard, I'm drawing blood...but I haven't laughed yet...honest!:-) It's really not a good idea to have the power tubes between the board & the tone controls, much of your wiring is much longer than it would otherwise be. This & the layout can cause unwanted distortion, even if you don't have any PO.