A sawtooth wave whose polarity has been inverted will not look like a typical sawtooth unless it's played backwards. Either way it will have the same harmonic content as a non inverted sawtooth. It will also sound the same. Non repetitive signals like percussive music may or may not sound the same when inverted.
The music synthesis community is somewhat divided on whether absolute polarity, or even small phase differences make an audible difference when two dissimilar signals are summed. I have heard valid examples of both sides of the argument, so neither must ALWAYS be correct.
The statements I made about phase shift and polarity are correct when speaking about repetitive waveforms that are symmetrical about the X axis, sines, triangles and square waves with a 50% duty cycle. A sawtooth waveform is an example where a polarity reversal does not equal a phase shift. Once multiple signals are summed together one cannot always assume that an inverted signal can be the same as a phase shifted signal.
We however tend to test and simulate our audio amps with sine waves, and the analogies I used will help one understand why phase shifts can make an amplifier unstable.
And yet we can reverse 3-phase motors with a switch (standard in 3-ph world but not stocked at HomeDepot).
Ahhhhh, I lied:--- HD lists one but is out of stock tonight.
https://www.homedepot.com/p/ASI-3-P...Motors-Red-Handle-P0202518S-RH-EKIT/310673347
Take a 2 channel scope, and a 1 kHz sine wave generator. Apply the sine wave to an amplifier input and to channel 1 of the scope.
Connect channel 2 to the amplifier output. Turn on the scopes Automated Phase Measurement feature.
A. First, use a very wide bandwidth amplifier that has 3 triode stages cascaded (not cascode). It will have will have 3 x 180 degrees.
But the scope will show the input to output phase to be 180 degrees (3 x 180 = 540 degrees).
The input to output phase is 180 degrees.
B. Second, use a very wide bandwidth amplifier that has 2 triode stages cascaded (not cascode). It will have will have 2 x 180 degrees.
But the scope will show the input to output phase to be 0 degrees (2 x 180 = 360 degrees).
The input to output phase is 0 degrees.
Connect channel 2 to the amplifier output. Turn on the scopes Automated Phase Measurement feature.
A. First, use a very wide bandwidth amplifier that has 3 triode stages cascaded (not cascode). It will have will have 3 x 180 degrees.
But the scope will show the input to output phase to be 180 degrees (3 x 180 = 540 degrees).
The input to output phase is 180 degrees.
B. Second, use a very wide bandwidth amplifier that has 2 triode stages cascaded (not cascode). It will have will have 2 x 180 degrees.
But the scope will show the input to output phase to be 0 degrees (2 x 180 = 360 degrees).
The input to output phase is 0 degrees.
An interesting thread on polarity vs phase.
https://www.audiosciencereview.com/.../analytical-analysis-polarity-vs-phase.29331/
https://www.audiosciencereview.com/.../analytical-analysis-polarity-vs-phase.29331/
Gotta say, that's a serious bit of wanking. Nobody would argue that some abstract simplified model couldn't be made where anything equals any other thing.
But, putting down the bong for a minute those guys (and of course, they're guys) need to answer one question; in a real amplifier with one or more 180 deg phase shifts, how does feedback work?
All good fortune,
Chris
But, putting down the bong for a minute those guys (and of course, they're guys) need to answer one question; in a real amplifier with one or more 180 deg phase shifts, how does feedback work?
All good fortune,
Chris
Wow. I am impressed how many of the experts here can be so neglecting with the phase and polarity terms. Yes, they are simple concepts. No, they are totally not the same.
Just compare the meaning and relation of these terms considering sinusoidal and non-sinusoidal (or even aperiodic) signals. The polarity thing works the same for all types of signals. The 180° relation between polarity and phase goes poof with non-sinusoidal signals.
Why do so many here use "phase" when they mean "polarity"? Are the engineering parts of their brains hardwired to sinusoidal signals only?
Just compare the meaning and relation of these terms considering sinusoidal and non-sinusoidal (or even aperiodic) signals. The polarity thing works the same for all types of signals. The 180° relation between polarity and phase goes poof with non-sinusoidal signals.
Why do so many here use "phase" when they mean "polarity"? Are the engineering parts of their brains hardwired to sinusoidal signals only?
Perhaps the proper name for a really good low time delay "Phase Inverter" is "Polarity Inverter".
The historical name, phase inverter is grandfathered; but polarity inverter is more accurate.
Although the two polarities of the above circuit are opposite, the time of an impulse event, like the transition from zero volts to the "top" of the impulse; when compared to the inverted polarity output from zero volts to the "bottom" of the 'now-inverted' impulse; the voltage transitions happen at the same time (no delay).
That works for a sine wave, sawtooth, square wave, music, etc.
The historical name, phase inverter is grandfathered; but polarity inverter is more accurate.
Although the two polarities of the above circuit are opposite, the time of an impulse event, like the transition from zero volts to the "top" of the impulse; when compared to the inverted polarity output from zero volts to the "bottom" of the 'now-inverted' impulse; the voltage transitions happen at the same time (no delay).
That works for a sine wave, sawtooth, square wave, music, etc.
Last edited:
Just a note. A transformer loaded common cathode stage should not invert the phase, as it is depicted in the first post schematic.
50AE,
The schematic, Post # 1 shows one polarity at the output tube's grid g1.
The output tube plate has the opposote polarity of g1.
The Plate drives the top of the primary, with B+ at the bottom.
The 16 Ohm secondary tap is at the top, the common is at the bottom.
B+ and Common do not move, and the Plate and 16 Ohm are of the same polarity (*).
The polarity (phase markings) of the primary and secondary are not marked with a dot. Therefore, you should default to the most often used convention as in (*).
According to that, the schematic is correct.
The schematic, Post # 1 shows one polarity at the output tube's grid g1.
The output tube plate has the opposote polarity of g1.
The Plate drives the top of the primary, with B+ at the bottom.
The 16 Ohm secondary tap is at the top, the common is at the bottom.
B+ and Common do not move, and the Plate and 16 Ohm are of the same polarity (*).
The polarity (phase markings) of the primary and secondary are not marked with a dot. Therefore, you should default to the most often used convention as in (*).
According to that, the schematic is correct.
I've figured out the phase stuff, I know what to do.
The transformer I'm using cannot have it's leads swapped to correct the phase, however this is not a problem as I will send the GFNB back to stage 2 rather than 1 (something that was not shown in my original diagram as I was still thinking about it).
Anyway spent several days calculating all the resistor values and cap values and did a new schematic.
So I've now started the prototype. So far so good, get 525V with a 5U4GB on the B+ when unloaded (expecting 430V with the load).
I like to do things different, this amp will be different when finsished (probably too complicated).
One of four things will happen:
1. It will be amazing when I finish it
2. I'll never finish it (have been meaning to make this for over a year, spent £1000 on components and equipment so far)
3. It will work but be noisy and crap
4. I'll electocute myself
Depending on how things go in the next few weeks, if people are interested I will create a progress thread with more info. Fingers crossed my calculations and theory are all correct and this will work and not a waste of time.
The transformer I'm using cannot have it's leads swapped to correct the phase, however this is not a problem as I will send the GFNB back to stage 2 rather than 1 (something that was not shown in my original diagram as I was still thinking about it).
Anyway spent several days calculating all the resistor values and cap values and did a new schematic.
So I've now started the prototype. So far so good, get 525V with a 5U4GB on the B+ when unloaded (expecting 430V with the load).
I like to do things different, this amp will be different when finsished (probably too complicated).
One of four things will happen:
1. It will be amazing when I finish it
2. I'll never finish it (have been meaning to make this for over a year, spent £1000 on components and equipment so far)
3. It will work but be noisy and crap
4. I'll electocute myself
Depending on how things go in the next few weeks, if people are interested I will create a progress thread with more info. Fingers crossed my calculations and theory are all correct and this will work and not a waste of time.
mdpaudio,
1. Check the polarity of a battery. Use your scope probe.
2. Check the phase of a 120V-Neutral-120V power mains.
Use a two channel scope, and 2 probes.
Caution! Connect the scope ground clips to ground, do not connect the ground clips to neutral. Neutral is fractions of a volt, to several volts higher than ground, and the resistance of that voltage source is very low. A probe ground clip that is connected to Neutral may smoke the probe ground wire.
Do Not float your scope!
Connect one probe tip to the one 120 volt source and the other probe tip to the other 120 volt source.
The phase is 180 degrees. . . . OK.
Now, notice that each 120V source has Positive polarity and Negative polarity, as a function of time.
Phase
Polarity
Lesson 1
1. Check the polarity of a battery. Use your scope probe.
2. Check the phase of a 120V-Neutral-120V power mains.
Use a two channel scope, and 2 probes.
Caution! Connect the scope ground clips to ground, do not connect the ground clips to neutral. Neutral is fractions of a volt, to several volts higher than ground, and the resistance of that voltage source is very low. A probe ground clip that is connected to Neutral may smoke the probe ground wire.
Do Not float your scope!
Connect one probe tip to the one 120 volt source and the other probe tip to the other 120 volt source.
The phase is 180 degrees. . . . OK.
Now, notice that each 120V source has Positive polarity and Negative polarity, as a function of time.
Phase
Polarity
Lesson 1
Last edited:
Update:
I built the prototype, left channel only and all the phase stuff is as expected.
Due to the phase being flipped I have to apply NFB to the 2nd stage.
I got 6W RMS before any NFB, I then added a 5.6K resistor from the 16ohm output to the R6 resistor after splitting it in 2.
It seems adding NFB to the 2nd stage allows me to get 9W RMS before distortion.
No feedback: 0.22VPP input = 19.6VPP out
Feedback: 0.3VPP input = 24.4VPP out
Here's my ratnests build, I have created multiple star grounds as it's just a test prototype.
Sounds great, however I think the EF86 is making a hissing sound, you have to put your ears to the speakers.
Amp currently running in UL mode. KT88 bias voltage of 41V and plate is 406V. Bias resistor is 560ohm.
Dang these things get bloody hot, first time building something like this.
I built the prototype, left channel only and all the phase stuff is as expected.
Due to the phase being flipped I have to apply NFB to the 2nd stage.
I got 6W RMS before any NFB, I then added a 5.6K resistor from the 16ohm output to the R6 resistor after splitting it in 2.
It seems adding NFB to the 2nd stage allows me to get 9W RMS before distortion.
No feedback: 0.22VPP input = 19.6VPP out
Feedback: 0.3VPP input = 24.4VPP out
Here's my ratnests build, I have created multiple star grounds as it's just a test prototype.
Sounds great, however I think the EF86 is making a hissing sound, you have to put your ears to the speakers.
Amp currently running in UL mode. KT88 bias voltage of 41V and plate is 406V. Bias resistor is 560ohm.
Dang these things get bloody hot, first time building something like this.
Not a rats nest it’s more of an alligator farm! Putting the feedback to the second stage works but it’s technically not global feedback anymore! It’s multistage feedback As global is output to input. Feedback can be useful but too much is a bad thing too. Listen to various levels of it before deciding what you like.
Haha yes. It allows me to try different components easily. I've removed some of the clips and used soldered connections since.
Also second channel is now done.
I'm surprised how well it works like this.
I will play about with the feedback resistor, I manged it get it to oscillate, might need a mica capacitor across the resistor.
Also second channel is now done.
I'm surprised how well it works like this.
I will play about with the feedback resistor, I manged it get it to oscillate, might need a mica capacitor across the resistor.
This is not so complicated.
1. You have oposite phase at the input ! This is first issue,
2. Every amplifying stage doing 180deg phase shift - make the opposite phase that one at the input.
3. In this example You have 3 amlifinig stages. So the signal eill be inverted with 0deg at the input.
4. Last amplifing stage has transformer load.
5. Just reverse output polarity @ transformer secondary. COM pin is going to speaker +. And SPEKER pin going to - speaker terminal, AND optionaly to ground.
.Thats all to have in this 3 stage ampifing unit to have phase in = phase out.
cheers
.
Try to correct phases from the input to output and it will be more clear...
cheers
Just one thing:
What is the value of negative grid bias for ouput tube?
What is the measured +V value @ R9/C7 point to katode of KT88?
I am asking that because I think that You can rearrange the circuit it can be as 2 stages?
Example:
persume that the +V@K of KT88 is +40V
that alowing smaler value of input signal to the output tube to be less than 2x40Vp-p
IF the input signal is say 6Vp-p
Then You will need input stage amplification of 80/6=13.3 times?
That is little bit smaller than say ECC82 as input tube.
You have many others too to choose.
Than You can have phase in = phase out and with that connection of the speaker terminals as You sketc first?
Cheers.
I dont know precisly, but
i think in this version You will have small sensitivity, maybe less than 1Vp-p for max power.
And even few hunderds of mV IF You employ ECC83 in the SRPP stage before output tube.
One is obvious - it is huge mistake to state ecc82/ecc83 in that design as "option"
ECC82 has mju factor of 18cca and ECC83 has mju of 100... 🙁
.
What is the value of negative grid bias for ouput tube?
What is the measured +V value @ R9/C7 point to katode of KT88?
I am asking that because I think that You can rearrange the circuit it can be as 2 stages?
Example:
persume that the +V@K of KT88 is +40V
that alowing smaler value of input signal to the output tube to be less than 2x40Vp-p
IF the input signal is say 6Vp-p
Then You will need input stage amplification of 80/6=13.3 times?
That is little bit smaller than say ECC82 as input tube.
You have many others too to choose.
Than You can have phase in = phase out and with that connection of the speaker terminals as You sketc first?
Cheers.
I dont know precisly, but
i think in this version You will have small sensitivity, maybe less than 1Vp-p for max power.
And even few hunderds of mV IF You employ ECC83 in the SRPP stage before output tube.
One is obvious - it is huge mistake to state ecc82/ecc83 in that design as "option"
ECC82 has mju factor of 18cca and ECC83 has mju of 100... 🙁
.
@Zoran
KT88 bias is 39V, I'm trying 470 - 560 ohms, the higher the loadline the lower the distortion.
I purposely decided to go with this design. The EF86 is configured as a triode so the gain is reduced as well as the distortion.
I'm not too worried about not having GFNB.
I'm trying to drive this to peak with a low level input, this design achives this. 0.3VPP gets me the full 9W per channel output.
I need to try some different non Russian EF86's, the ones I have are definately microphonic.
If only I had an audio analyser to verify what my ears can hear.
KT88 bias is 39V, I'm trying 470 - 560 ohms, the higher the loadline the lower the distortion.
I purposely decided to go with this design. The EF86 is configured as a triode so the gain is reduced as well as the distortion.
I'm not too worried about not having GFNB.
I'm trying to drive this to peak with a low level input, this design achives this. 0.3VPP gets me the full 9W per channel output.
I need to try some different non Russian EF86's, the ones I have are definately microphonic.
If only I had an audio analyser to verify what my ears can hear.
You probably have one, a sound card with line input can be used as an audio analyser using freely available software like REW.