• WARNING: Tube/Valve amplifiers use potentially LETHAL HIGH VOLTAGES.
    Building, troubleshooting and testing of these amplifiers should only be
    performed by someone who is thoroughly familiar with
    the safety precautions around high voltages.

My Wave Isn't Square.

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
The top trace is the input signal at about 3.8v P-P, 1K hz. The bottom trace is 10v P-P after the coupling cap.

-6 dB NFB has reduced the over shoot and most of the ringing. The issue now is the top and bottom of the wave are not flat. They both increase with time. I understand this represents a emphasized low hz response.

I presume the differance in the angle of the top and bottom is due to nonlinearity.

The output trace (not shown) is an inverted image with the highs over emphasized, decrease with time, and with a bit more angle. The angle increases as volume increases.

The amp is very quiet and clear but it's a bit bright.

How would I correct this type of frequency response?
 

Attachments

  • Sch.jpg
    Sch.jpg
    25.1 KB · Views: 1,550
  • SW 6SN7.JPG
    SW 6SN7.JPG
    107.9 KB · Views: 1,522
You are looking at high pass filter behavior, which is typical when looking at square waves. The lower you go in frequency, the more pronounced this will become. The higher in frequency, the less pronounced.

Schematic would help to identify the capacitor that is forming this HPF, but I would say you're fine if you are happy with the bass response of the amp.
 
Administrator
Joined 2004
Paid Member
The top trace is the input signal at about 3.8v P-P, 1K hz. The bottom trace is 10v P-P after the coupling cap.

-6 dB NFB has reduced the over shoot and most of the ringing. The issue now is the top and bottom of the wave are not flat. They both increase with time. I understand this represents a emphasized low hz response.

I presume the differance in the angle of the top and bottom is due to nonlinearity.

The output trace (not shown) is an inverted image with the highs over emphasized, decrease with time, and with a bit more angle. The angle increases as volume increases.

The amp is very quiet and clear but it's a bit bright.

How would I correct this type of frequency response?

You haven't touched on the OPT you are using which is certainly responsible for most of what you are seeing. FWIW that is a fairly good looking SQ wave. No OPT is going to have sufficient HF response to pass a perfect square wave except at relatively low frequencies, (a couple of kHz or so) and that presupposes that the primary inductance is sufficient not to load the tube at the frequency of interest.
 
Administrator
Joined 2004
Paid Member
Why do you want to 'correct' the response? The negative feedback is doing, presumably, what you want it to do. It corrects the output (precisely, the NFB sampling point) by modifying the input. You are simply seeing the modification. You would see far 'worse' waveforms inside other amps.

Indeed it is, and it would have been more interesting/useful to see the square wave at the output. I also agree having seen far worse waveforms in other amps as well.
 
Administrator
Joined 2004
Paid Member
Instead of measuring the amplifier square wave performance you should do a simple frequency response measurement to determine whether there is anything beyond subjectivity in your comment about the amplifier sounding bright.

I suspect what you will find is the amplifier's power response is down several dB or more at 20Hz (or even higher) resulting in the subjectively bright sound.

Your source could be a bright/hard sounding CD player/dac and this amp is just telling it as it is.

Your speakers could have HF drivers that are inherently bright, analytical or....

It is also possible that you just don't like the sound of a pentode amp even with some global feedback applied. I know I don't in many cases.. :D

You really need to give us a lot more information.. :D
 
Last edited:
How do the square waves look at the output?

I've seen the same thing myself. If you connect the probe to a Hi-Z point (as you did here) the apparent loss of high frequencies could be an artifact of cable and o'scope deflection amp capacitance.

I got some hideous-looking square waves, and what looked like an Fh of barely 20KHz by o'scoping a Hi-Z point. Connected to the more appropriate Lo-Z point (the output of the cathode follower grid drivers) the square waves were nice and flat, and the Fh measured to be some 117KHz.
 
Thanks for your input

The first photo is input (1K hz, 4V P-P) and output (10V P-P) with -6db NFB. The second no NFB. I still have a little over shoot to work on.

The OPT's are 10w 5K-8 Edcor's (GXSE10-8-5K).

It's good to know I'm in the ball park. It's also good to know there is room for improvement. I will play around a bit more with the NFB. Mullard references NFB in the -20db range for pentodes in their Tube Circuits for Audio Amplifiers book. The sweep is a good idea.

I'm consistant with the listening test. Same source, speakers, tracks, room, volume. You might right, It may just be the new pentodes. I'm used to my old Mullards.
 

Attachments

  • Output-With NFB.JPG
    Output-With NFB.JPG
    111.5 KB · Views: 1,335
  • Output-No NFB.JPG
    Output-No NFB.JPG
    90.4 KB · Views: 1,280
Administrator
Joined 2004
Paid Member
The result shows heavy non-linearity, not only frequency roll-off. This is from ECC82, 'Aikido-like' circuit.


I see evidence of a relatively low slew rate which implies bandwidth limitations, and asymmetry in the rise and fall times which relate to a variance in the ability to sink and source current in a typical resistively loaded amplifier stage. Some of this could be the result of your probe/cable capacitances. I don't really see a "heavy non-linearity" as presented in your waveform, but I guess my standards are lower than yours.. (I've seen much worse actually) :D
 
Troncone's scope may need some adjustment- when you look at rise and fall times, you can see some negative time...:D It's a simple screwdriver adjustment on most scopes; with a flat trace, adjust for the line being perfectly horizontal.

The -3dB point for 0u047 and 470k is a little high (7 Hz) but not enough to cause that much bass loss. I'd suspect insufficient primary inductance for use with a pentode.
 
Administrator
Joined 2004
Paid Member
If you change your coupling capacitor (first post picture) from 0.047uF to 0.47uF it will look more like a square wave.


Closed loop this isn't going to be the case, and that LF corner is at 7.2Hz so no change with a 1kHz square wave. Really not hard to understand that the transformer is the culprit here.

Edit: I see SY beat me to the punch.. :D
 
I don't really see a "heavy non-linearity" as presented in your waveform,

In my waveform you can only see exponential rising and falling edges, both same in shape and time. This is perfectly linear behaviour, same as for R-C passive circuit. The scope has even calculated rise time and fall time of the step response for you. In the waveform presented by the member, there is an overshoot on rising edge and no overshoot on falling edge. This is a prove of dynamic non-linearity of the measured circuit. Thank you for asking, feel free to put further professional question.
 
Administrator
Joined 2004
Paid Member
Troncone's scope may need some adjustment- when you look at rise and fall times, you can see some negative time...:D It's a simple screwdriver adjustment on most scopes; with a flat trace, adjust for the line being perfectly horizontal.

<snip>

Looking closely I now notice the same thing, some temporal laws are being violated.. :p I see it particularly on the rise time side of things, something seems to be going on with the scope timebase and I'm not completely certain that the probe compensation will take care of it. Does this scope have an astigmatism control? (Not sure that is even relevant)
 
Last edited:
Administrator
Joined 2004
Paid Member
In my waveform you can only see exponential rising and falling edges, both same in shape and time. This is perfectly linear behaviour, same as for R-C passive circuit. The scope has even calculated rise time and fall time of the step response for you. In the waveform presented by the member, there is an overshoot on rising edge and no overshoot on falling edge. This is a prove of dynamic non-linearity of the measured circuit. Thank you for asking, feel free to put further professional question.

I don't see the overshoot on the rising edge perhaps a quirk of my video display system. (or my eyes?) :D
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.