Asymmetrical Clipping

[B. Bop]

Is asymmetrical clipping a normal characteristic of single ended amps? I have a SE amp that starts flattening the positive peaks of the output waveform at less than half of full output power (full output power being defined as the maximum peak to peak amplitude). Is this normal?

 

[phsyconoodler]

What do your ears tell you about the sound?Your scope is a tool,yours ears are the final judge.

 

[B. Bop]

The amp sounds great. It's just a little lite on clean tones. The amp is one that I built myself so I am not sure if the asymmetrical clipping is the result of some design error or if it's normal for this type of output section.

 

[B. Bop]

I am using a KT88 with 590V on the plate, biased at 90% of 40 watts.

 

[phsyconoodler]

Try some different cathode resistors til you get what you want to hear.Does it have negative feedback?If so,tweak the resistor til you get the cleans you want.

 

[antik]

It is very common to have asymmetrical clipping in SE amp. If you are going for maximum power with maximum unclipped symmetrical signal, adjusting bias might not be enough. Changing the load line (output impedance) might be required. Plus changing screen voltage might also be needed to change the curves. (Hopefully the transformer is not saturating and causing the clipping)

 

[mark norwine]

many single-ended amps clip asymmetrically because [as others have eluded], the B+ is too high. Because of how high it is, the current needs to be lowered to keep from cooking the output tube. The result of the lowered current is the asymmetry you see.

A blackface champ is a perfect example. B+ is way too high.

 

[John Phillips]

It sound like it's biased way too hot.

Just because it's a single-ended and Class A does NOT mean that you need to bias it up to 100% of its dissipation rating, or anywhere near it. The dissipation rating is simply a measure of how much power the tube will take before it starts to suffer damage, it tells you nothing whatever about the correct operating point - contrary to almost everything you will hear or see written about 'how to bias an amp'.

Correct bias is determined by the operating point and NOT the tube dissipation, but its difficult to get this across in the face of almost universal wrong information about it...

Biasing to 70%, 90% or whatever only works if the amp has been designed to use the maximum possible power for the tube, and even then are simply rough upper limits for tube reliability depending on the configuration.

If the amp is designed to put out far less than the maximum capability of the tube, you don't want to bias it anywhere near as hot as it will go, it does nothing other than increase the load on the power supply and the DC current in the OT, and you will also cause it to clip at the forward end first. You could build a 1W KT88 amp if you wanted, and you would NOT idle the tube at 40+ watts... would you? Then the signal would be like a tiny ripple on top of a flood of totally wasted power, and if the transformers were designed for a 1W amp, you'd certainly blow them.


Later BF and all SF Champs as Mark mentioned are a good example of a wrongly-biased amp too - but it's not because they're biased too cold to cope with the high B+, it's the exact opposite. The B+ is high, but actually within the usable range, and the reason they clip too soon at the forward end is because they are biased too hot as well. If you raise the cathode resistor value (usually to 820 ohms or 1K - and remember to change the cap to take the increased voltage), you can center the waveform and get both more clean power and better tube life.

 

[B. Bop]

Put the amp back on the bench and replaced the 825 ohm +100 ohm pot cathode resistor with a 1K ohm resistor + 100 ohm pot. Tested with 1Khz signal. Original output was 6 volts clean output for about 4.4 watts clean with the bias at 70 ma and 532 V plate to cathode. After increasing the cathode resistor to 1K the clean output dropped to 4.5V for 2.5 watts clean with the bias at 63 ma and Vp=595 and Vp-k=533. Plate dissipation with 825 ohm cathode resistor = 37.2 watts. Plate dissipation with 1K cathode resistor = 30 watts. The amp is putting out about 25W at maximum output waveform amplitude when biased with 70 ma. and Vp = 590V. My observation was that reducing the bias reduced the output across the board. Mabe this is all the clean output I can get from this amp.

 

[John Phillips]

Put the amp back on the bench and replaced the 825 ohm +100 ohm pot cathode resistor with a 1K ohm resistor + 100 ohm pot. Tested with 1Khz signal. Original output was 6 volts clean output for about 4.4 watts clean with the bias at 70 ma and 532 V plate to cathode. After increasing the cathode resistor to 1K the clean output dropped to 4.5V for 2.5 watts clean with the bias at 63 ma and Vp=595 and Vp-k=533. Plate dissipation with 825 ohm cathode resistor = 37.2 watts. Plate dissipation with 1K cathode resistor = 30 watts. The amp is putting out about 25W at maximum output waveform amplitude when biased with 70 ma. and Vp = 590V. My observation was that reducing the bias reduced the output across the board. Mabe this is all the clean output I can get from this amp.

Something is very wrong with either the bias or the voltage swing available. You will still get a high maximum power output with the bias way off, but only at the expense of having one side of the waveform clipped almost totally... but that's not the same as the maximum clean output.

You would normally expect the static dissipation to be just over twice the clean audio power output if the bias is correct.

Try raising the cathode resistor value far enough to center the waveform between cut-off and clipping (which is the correct operating point for Class A, FWIW) and see what you get. You will need a much larger value of cathode bypass cap too probably, since otherwise the audio signal will have to pass more through the resistor.

 

[B. Bop]

Increasd the cathode resistor until the bias current was down to arouind 30 ma. Caused the plate voltage to go to just over 600V. Also increased the bypass cap to 100 uf. Same type of output waveform occured only at a much lower amplitude. The output waveform with the original 70 ma. bias at 590 Vp yielded a signal that went 6V positive and 26V negative. Got roughly the same thing, but at a lower amplitude evel. Also checked Vg1 = 0V.

 

[John Phillips]

Increasd the cathode resistor until the bias current was down to arouind 30 ma. Caused the plate voltage to go to just over 600V. Also increased the bypass cap to 100 uf. Same type of output waveform occured only at a much lower amplitude. The output waveform with the original 70 ma. bias at 590 Vp yielded a signal that went 6V positive and 26V negative. Got roughly the same thing, but at a lower amplitude evel. Also checked Vg1 = 0V.

Hmmm....

If the positive voltage swing available is so much less than the negative, this means that the operating point is still far too high... but if reducing the tube current to under half doesn't cure it... or even make much difference... what OT primary impedance do you have?

 

[B. Bop]

Primary impedance is 2.3K. Tried 5K, 4K, 3K and connected a Hammond 1628SE, 5K (11 Lb. transformer) for 2.5K to see if more iron was the answer. Get the highest output with the 2.3K, Mercury Magnetics transformer that is in the amp now. Just now tried lowering the screen voltage to see what would happen. Reduced the output and seemed to have a small effect on centering the waveform better. But the max clean output stayed the same. 4.5 watts clean, 25 watts max.

 

[AdmiralB]

590Va is WAY too high to be able to run true class A. The GEC sheets don't illustrate SE or class A operation, but the 6550A specs don't go over 400Va.

Champs and Vibro Champs aren't truly class A either, in the strict definition that both sides clip simultaneously. Voltage is too high.

 

[AdmiralB]

Are you looking at the output waveform on the primary or secondary side? If you're looking at the secondary side, how do you know it's the positive waveform that's clipped?

 

[B. Bop]

590Va is WAY too high to be able to run true class A. The GEC sheets don't illustrate SE or class A operation, but the 6550A specs don't go over 400Va.

Champs and Vibro Champs aren't truly class A either, in the strict definition that both sides clip simultaneously. Voltage is too high.

I am using a Winged C KT88 made by Svetlana. I down loaded the KT88 data sheet from Svetlana and it shows a maximum "DC PLate Voltage" of 800V and a "Plate Dissipation" of 42 Watts. I was hoping to make a fairly high powered SE Class A amp so I am trying to get maximum output from the tube. However, the performance curves listed in the data sheet dont match my results. The curves in the sheet are for a Vscreen of 300V and a Vscreen of 140V. I am running a screen voltage of 562V with a plate voltage of 586V. This is pretty much at the 8 watt limit of the screen. I found that the higher the screen voltage the more output I got.
The data sheet has a table of operating conditions for a 19 watt single tube class A amp with 400V on the plate and 87 ma bias current. The Grid No. 1 DC (control) voltage listed is -16.5 volts which would lead me to believe that the specs are for a fixed bias configuration rather than cathode biased. I think that cathode biased amps have 0V on g1. Is it possible that the cathode biasing is causing the asymmetrical clipping?

 

[B. Bop]

Are you looking at the output waveform on the primary or secondary side? If you're looking at the secondary side, how do you know it's the positive waveform that's clipped?

I have a 8.2 Ohm, 10 watt resistor wired to a 1/4" plug, plugged into the output speaker jack with a true RMS meter and an Oscope attached to the resistor. The positive leads of the meter and the scope are connected to the positive output lead (the side comming from the output transformer secondary lead that is not grounded). When displayed on the scope the positive half of the waveform is the one that gets clipped early.

Question: Does the output transformer invert the signal. In other words is the tube clipping at cutoff or saturation if the positive half of the wave on the secondary of the transformer is clipped.

 

[AdmiralB]

Your Va is way too high. In order to keep the plate dissipation below limits, you're running on the 'cold' side of the halfway point.

The OT inverts the signal. You're clipping the negative side, and the only way you'll get symmetrical clipping is to get the Va down to around 400V or lower.

The voltage limits for true class A are really quite low, around 275V or so for EL34s and 6L6. You can run SE amps hotter but they clip asymmetrically - which violates the definition of class A.

 

[B. Bop]

Your Va is way too high. In order to keep the plate dissipation below limits, you're running on the 'cold' side of the halfway point.

The OT inverts the signal. You're clipping the negative side, and the only way you'll get symmetrical clipping is to get the Va down to around 400V or lower.

The voltage limits for true class A are really quite low, around 275V or so for EL34s and 6L6. You can run SE amps hotter but they clip asymmetrically - which violates the definition of class A.

I thought that Class A was defined as the tube conducting for all 360 degrees of a sinewave? If the tube is clipping on the negative half of the wave, then it must be going into cutoff. So what you are saying is that the bias current is too low allowing the tube to cut off? I will try to variac the amp down to a 400 V plate voltage and see if this fixes the clipping.

 

[AdmiralB]

That's correct, but the other side of the coin (as OTM mentioned earlier) is that for textbook class A, the idle point must be exactly halfway between cutoff and saturation - so any clipping occurs symmetrically.

Yours is idling below halfway, so that the tube cuts off at the negative input swing but stays clean on the positive. You can't bias it hotter to get to the halfway point because you'll exceed the limits of the tube....so the Va has to come down.

Or, you can live with the asymmetrical operation. Champ owners have been doing it for forever.