Express Output Waveform Shots
Moderators: pompeiisneaks, Colossal
Re: Express Output Waveform Shots
Ah, the all important 3rd stage.....
You do not have the required permissions to view the files attached to this post.
Re: Express Output Waveform Shots
2nd stage. Up to the boring end of the amp now. Apart from some power rail sag, this is all standard gain stage stuff.
You do not have the required permissions to view the files attached to this post.
Re: Express Output Waveform Shots
Last of all the 1st stage. I was expecting a bit more reduction in 1st stage gain from the sag in the power rail. There is a tiny bit there. Also small shifts in relative phase between 1st stage signal and the speaker output I'm triggering to.
You do not have the required permissions to view the files attached to this post.
Re: Express Output Waveform Shots
There's a lot to digest there. And man I need to go to sleep!
Obviously I've put in some effort here, and quite a few hours. Hope it's of interest to everybody. It's been educational for me just capturing all this, let alone what I'll get out of it looking through it all again. And finally we've got a good set of Wreck Express waveforms for this forum.
Have a look through. I'm happy to answer any questions. When people are ready and if you're interested we can start discussing what we all see. Theories raised earlier in this thread all seemed to have a source of validity from people who know a lot of stuff. This full set of waveforms should have everything we need to prevent the discussion hitting dead ends.
May I also suggest rather than jumping all over the place maybe we could collectively start at the back of the amp and work our backwards. It would be good to collect observations and opinions section by section. Obviously the power grids, PI and 3rd stage are where the most interest will be. Don't know if that'll work or whether there's enough interest. Everyone might want to just take it away and draw their own conclusions, dunno. Anyway, enjoy!
Obviously I've put in some effort here, and quite a few hours. Hope it's of interest to everybody. It's been educational for me just capturing all this, let alone what I'll get out of it looking through it all again. And finally we've got a good set of Wreck Express waveforms for this forum.
Have a look through. I'm happy to answer any questions. When people are ready and if you're interested we can start discussing what we all see. Theories raised earlier in this thread all seemed to have a source of validity from people who know a lot of stuff. This full set of waveforms should have everything we need to prevent the discussion hitting dead ends.
May I also suggest rather than jumping all over the place maybe we could collectively start at the back of the amp and work our backwards. It would be good to collect observations and opinions section by section. Obviously the power grids, PI and 3rd stage are where the most interest will be. Don't know if that'll work or whether there's enough interest. Everyone might want to just take it away and draw their own conclusions, dunno. Anyway, enjoy!
Re: Express Output Waveform Shots
Man that's a lot of information to digest! In your recent remark, you said "quite a few surprises", could you please summarize what're differences that you noted or was that meant to be a quiz? 
I re-read the thread and there seemed to be different opinions on the PI's operation vis-a-vis the other amps, some had the opinion that the PI pretty much worked as it should, while you thought otherwise. Then there're the issues with screen grid and PI supply, which may or may not play a significant role in the TWE sound. Hope I got most of it...
I re-read the thread and there seemed to be different opinions on the PI's operation vis-a-vis the other amps, some had the opinion that the PI pretty much worked as it should, while you thought otherwise. Then there're the issues with screen grid and PI supply, which may or may not play a significant role in the TWE sound. Hope I got most of it...
Re: Express Output Waveform Shots
Thanks for the interest jazbo8! Quite a few surprises was a bit of a throw away line. I was coming into this having done a lot of waveform analysis on just the preamp & PI (built, not simulated, but not connected to a power section) and of the whole amp of a Wreck preamp mated to a smaller 6P1P power section. I was trying to get the level of PI signal into the 6P1P grids right using split plate resistors on both sides of the PI. I noticed some interesting things about how it worked but couldn't be sure of how to mate it properly without seeing the behaviour of a full size Express. I've pretty much built my Express because of a keen desire to understand the amp better, see what all the fuss is about, and having a kick arse amp I'll get to play out once in a blue moon.
As you say there was some disagreement at the start of the thread but none of us had the info to take the discussion any further. At this point I've even shelved what I said until I go through it all in more detail. My surprises came down to seeing something different in the output waveform that I wasn't expecting in the contribution of screen compression to smoothing the transition from power grid clipping dominant distortion to PI clipping dominant distortion that I wasn't getting with the smaller beam tetrode based amp. Second main one was that the behaviour of the 3rd stage and PI was exactly the same as I'd witnessed in the smaller amp.
OK, now you've got me started. I've read and re-read every thread I can find here which discusses the two hottest topics about the Express amp: clean to mean and why the 1K screen supply chain resistor is rated to 25W. There are a *lot* of differing opinions and some great discussion that never gets to really go anywhere. For the 1K resistor I've never seen heat dissipation calcs posted here, but they tell you everything. I think the most quotable line about clean to mean is how the amp clips progressively through power stage, then PI then preamp. There are so many classic amps that do the same from Marshall 18 Watters & Plexis to Fender Deluxe Reverb. Hell, I think even a Blues Junior does the same with the right settings. But none of them have the extremes of clean to mean like this amp. So what is really different? Secondly if you've every played around with waveforms and clipping thresholds, it's a very easy experiment to see/hear that a square wave sounds louder through a speaker than a sine wave of the same peak magnitude. It's double the power (which is not twice as loud but is 'a bit more' perceived volume) and has a heap of harmonics added, both of which make it sound louder. So what else is happening, and how does the Express reduce it's output headroom when overdriven compared to clean enough to make the two perceived volumes quite close if not the same? The answers are in these waveforms. And yes, the sag and response time constant of the PI supply node is a crucial part in the equation.
I also need to look over these for a while. I don't want to lauch into explainations prematurely, and others might see stuff I'm not seeing. Am happy to proceed step by step with some analysis but it would be a lot more beneficial if interested people out in TAG land joined in.
I had a good play through the amp again today (through my reactive dummy load with line out, and re-amped to home friendly volumes). Even attenuated it's just a pleasure to play and is like nothing else.
As you say there was some disagreement at the start of the thread but none of us had the info to take the discussion any further. At this point I've even shelved what I said until I go through it all in more detail. My surprises came down to seeing something different in the output waveform that I wasn't expecting in the contribution of screen compression to smoothing the transition from power grid clipping dominant distortion to PI clipping dominant distortion that I wasn't getting with the smaller beam tetrode based amp. Second main one was that the behaviour of the 3rd stage and PI was exactly the same as I'd witnessed in the smaller amp.
OK, now you've got me started. I've read and re-read every thread I can find here which discusses the two hottest topics about the Express amp: clean to mean and why the 1K screen supply chain resistor is rated to 25W. There are a *lot* of differing opinions and some great discussion that never gets to really go anywhere. For the 1K resistor I've never seen heat dissipation calcs posted here, but they tell you everything. I think the most quotable line about clean to mean is how the amp clips progressively through power stage, then PI then preamp. There are so many classic amps that do the same from Marshall 18 Watters & Plexis to Fender Deluxe Reverb. Hell, I think even a Blues Junior does the same with the right settings. But none of them have the extremes of clean to mean like this amp. So what is really different? Secondly if you've every played around with waveforms and clipping thresholds, it's a very easy experiment to see/hear that a square wave sounds louder through a speaker than a sine wave of the same peak magnitude. It's double the power (which is not twice as loud but is 'a bit more' perceived volume) and has a heap of harmonics added, both of which make it sound louder. So what else is happening, and how does the Express reduce it's output headroom when overdriven compared to clean enough to make the two perceived volumes quite close if not the same? The answers are in these waveforms. And yes, the sag and response time constant of the PI supply node is a crucial part in the equation.
I also need to look over these for a while. I don't want to lauch into explainations prematurely, and others might see stuff I'm not seeing. Am happy to proceed step by step with some analysis but it would be a lot more beneficial if interested people out in TAG land joined in.
I had a good play through the amp again today (through my reactive dummy load with line out, and re-amped to home friendly volumes). Even attenuated it's just a pleasure to play and is like nothing else.
TWE & gNFB
This has been covered before, but perhaps a refresher does not hurt... As mentioned earlier in the thread, the NFB (or rather no feedback) under overdriven condition, which is quite common in the TWE has to be one of the key ingrediants of the C2M behavior. So would it be possible to get a set of waveforms without gNFB?
Also some relavent discussion can be found here.
Also some relavent discussion can be found here.
Re: Express Output Waveform Shots
If anyone has a SPL meter they should measure the actual loudness of their amp when rolled back for clean vs. distorted. It would not surprise me if the distorted sound is about 3 dB louder than the clean sound.
Re: Express Output Waveform Shots
Dave, it wasn't a terribly accurate test but I've put a microphone in front of my amp and used the VU meters on my mixer as a level indicator. Distorted sound is very close to clean, not 3dB different.
We need tests, not conjecture.
We need tests, not conjecture.
Re: Express Output Waveform Shots
Yes, I think that was my suggestion.katopan wrote:We need tests, not conjecture.
Re: Express Output Waveform Shots
No worries. It would be great for someone else to verify the same with their Express too. I just used my mixer VU meter. Any VU monitoring and a microphone would do the same.
Re: Express Output Waveform Shots
Jazbo8, I saw your link after you posted and didn't have time to look at it. I've now seen it was only a short one anyway.
What would you like to see specifically without the NFB? Keep in mind that it took me over 10 hours to generate the set of waveforms I've attached. I'm not likely to take on doing another whole set without NFB. I agree NFB is an important part of the response of this amp and helps with making the clean to mean transition more dramatic. But I think the effect of the feeback both before and after clipping is fairly well understood.
I'm narrowing down in trying to understand how something is working here. In the output waveform the lower half of the wave (the wider side) has its trailing corner cut off in a fairly smooth fashion. Comparing the power grid waves, you can see that V4 grid clips and then backs out of grid clipping, where V5 is grid clipped the whole time. So the input to V4 backs off because of clipping and a reduction in headroom further back. This can also be seen in the PI plates. Pin 1 plate has the typical ramping up top indicating that the coupling cap to the power grids is charging up (normal behaviour for fixed bias push-pull) But the pin 6 plate ramps up for the first half and then flattens out. Comparing the DC PI plate waves to the PI power rail, you can see that the peak of pin 6 is higher than on pin 1 and that it runs out of headroom, hitting the PI power rail. That is the mixed mode distortion I've mentioned before. The speaker output has one side which is standard power grid clipping distortion, and the other side that transitions from power grid clipping to PI distortion half way through its half-cycle. The transition from power grid clipped to PI clipped is smoothed by the power valve backing out of clipping back through screen compression.
What would you like to see specifically without the NFB? Keep in mind that it took me over 10 hours to generate the set of waveforms I've attached. I'm not likely to take on doing another whole set without NFB. I agree NFB is an important part of the response of this amp and helps with making the clean to mean transition more dramatic. But I think the effect of the feeback both before and after clipping is fairly well understood.
I'm narrowing down in trying to understand how something is working here. In the output waveform the lower half of the wave (the wider side) has its trailing corner cut off in a fairly smooth fashion. Comparing the power grid waves, you can see that V4 grid clips and then backs out of grid clipping, where V5 is grid clipped the whole time. So the input to V4 backs off because of clipping and a reduction in headroom further back. This can also be seen in the PI plates. Pin 1 plate has the typical ramping up top indicating that the coupling cap to the power grids is charging up (normal behaviour for fixed bias push-pull) But the pin 6 plate ramps up for the first half and then flattens out. Comparing the DC PI plate waves to the PI power rail, you can see that the peak of pin 6 is higher than on pin 1 and that it runs out of headroom, hitting the PI power rail. That is the mixed mode distortion I've mentioned before. The speaker output has one side which is standard power grid clipping distortion, and the other side that transitions from power grid clipping to PI distortion half way through its half-cycle. The transition from power grid clipped to PI clipped is smoothed by the power valve backing out of clipping back through screen compression.
Re: Express Output Waveform Shots
Doing another whole set without NFB would be way too much work as you said, but perhaps you can post some shots where you think that NFB/no NFB play an important role... may be just before and after the NFB is broken with the overdrive.katopan wrote:Jazbo8, I saw your link after you posted and didn't have time to look at it. I've now seen it was only a short one anyway.
What would you like to see specifically without the NFB?
Thank you for the quick review of the clipping behavior of various stages, it is helpful as a guide for the tea-leaf reading:) I'll take a closer look...
LTP Tail
Just had another thought, nothing new I'm sure as so much had already been covered... But during your experiments, did you play with the value of the tail resistor? Was the 10k used purposely to introduce imbalance in the PI, and does making it even smaller make the C2M effect even more prominent or does the whole thing just collapse into a big mess? 
Re: Express Output Waveform Shots
It took a while but the penny finally dropped about what is shown in the PI cathodes waveform (standard LTP PI behaviour). OK, so I've not just parked the idea of the 3rd stage to PI coupling cap charging up, I've given up on it. I now understand that as soon as the PI input reaches grid clipping, it pulls up the cathode with signal instead of drawing grid current. That's shown by the positive peaks on the cathode. The extra current drawn through the cathode also flows in the anode and through the plate resistor. So that signal also appears on the lower bit of the plate waveform with its soft peak.
The other way around has the PI input swinging negative, and the cathode gets held up by the non-inverting side grid. This has the clipped feedback signal which has a flat top at this point, so the negative part of that side pin 1 plate signal is also flat, as different to the pin 6 side.
Found this thread that is relevant: http://music-electronics-forum.com/t11414/ Post #18 on page 1 nails it. Then you can think through what happens with signal going the other way as well.
I think I've narrowed down three things I really want to understand more. They are the duty cycle shift, effect of negative feedback, and the PI headroom difference on each side that gets created as it goes into clipping. I think the first and last play a part in generating the features we can see on the speaker output waveform. The NFB allows for the extreme of clean to mean, which I think we all understand but I'm happy to do some testing on.
So I'll try and get together some waveforms around the bit of interest with the NFB disconnected, and also think up what I can try to explore the other two items.
The other way around has the PI input swinging negative, and the cathode gets held up by the non-inverting side grid. This has the clipped feedback signal which has a flat top at this point, so the negative part of that side pin 1 plate signal is also flat, as different to the pin 6 side.
Found this thread that is relevant: http://music-electronics-forum.com/t11414/ Post #18 on page 1 nails it. Then you can think through what happens with signal going the other way as well.
I think I've narrowed down three things I really want to understand more. They are the duty cycle shift, effect of negative feedback, and the PI headroom difference on each side that gets created as it goes into clipping. I think the first and last play a part in generating the features we can see on the speaker output waveform. The NFB allows for the extreme of clean to mean, which I think we all understand but I'm happy to do some testing on.
So I'll try and get together some waveforms around the bit of interest with the NFB disconnected, and also think up what I can try to explore the other two items.