I don't dispute that some degree of low-order distortion cancelling can be achieved. The issue is whether doing this results in more higher-order distortion, which is generally reckoned to sound more unpleasant. I agree that reducing the dominant distortion enables the application of NFB to improve things, but SE people often avoid NFB and PP allows more NFB because it starts with less distortion.
Of course it's about cascaded circuits which is what I described in the previous post. It's not a fixed amount but increasing amount with signal level. At low level is much less and at some point won't work if the amount to be cancelled is too much. The amount one can cancel depends on so many things that giving numbers is pointless.
I think it's rarely used because it's not easy. In the case of commercial stuff, even in those cases where people have all the knowledge, the experience and the tools, it's not economically convenient mainly because it requires time (working hours have become quite expensive for employers this side of the world, I think). It is specific to the actual amplifier when properly applied to a fine degree.
I took up my notes about the amplifier I presented above.
Some test results:
1. Fully optimized output stage alone with 50 % UL-connection:
- THD = 1.8 % @ 3 W
- THD = 2.5 % @ 4.5 W
2. The same output stage with 6N9S voltage amplifier, optimized for minimum total THD:
- Output THD = 0.80 % @ 3 W, when THD of 6N9S = 2.05 %
- Output THD = 1.08 % @ 4.5 W, when THD of 6N9S = 2.50 %
So, the total THD was decreased more than 50 %.
Essential component to adjust in order to get greatest distortion cancelling is the grid resistor of the output tube too.
Some test results:
1. Fully optimized output stage alone with 50 % UL-connection:
- THD = 1.8 % @ 3 W
- THD = 2.5 % @ 4.5 W
2. The same output stage with 6N9S voltage amplifier, optimized for minimum total THD:
- Output THD = 0.80 % @ 3 W, when THD of 6N9S = 2.05 %
- Output THD = 1.08 % @ 4.5 W, when THD of 6N9S = 2.50 %
So, the total THD was decreased more than 50 %.
Essential component to adjust in order to get greatest distortion cancelling is the grid resistor of the output tube too.
That is also true. PP requires less NFB but can handle more. SE requires more but can handle less.MrCurwen said:
I was planning to use no more than 12 db for this amp, because if I use more, I don't have the necessary amplification (with 6SN7's) to deliver the 44V RMS required by the kt90 grids at full volume. Assuming I plod through and end up with a pretty good design, is 12dB FB adequate? If I go higher, I need a different tube, or another stage.
Thanks,
Mike
Thanks,
Mike
Here is the LTspice schematic. I'm still messing around with operating points. The left CCS is set at 8ma, the right one at 16ma (8ma per tube.) Vi = .35Vrms, which is what I figure Vgk will be after feedback. Vo=45Vrms, giving A=128.
I have breadboarded this circuit and get a little more gain than with LTspice.
I don't have the quad of kt90's included because I haven't figured out how to model an output transformer correctly.
Mike
I have breadboarded this circuit and get a little more gain than with LTspice.
I don't have the quad of kt90's included because I haven't figured out how to model an output transformer correctly.
Mike
Attachments
Can you send the .asc of the circuit and the tube models you have used.
Then we could simulate too.
Then we could simulate too.
The .asc is attached. The 6SN7 model is from pctube, the BJT models are in the standard LTspice library.
Thanks for your all your comments, guys.
Mike
Thanks for your all your comments, guys.
Mike
I made some modifications to your circuit. I replaced the CCS's with constant current sources because I did not have all the component models you used. But that does not make any essential difference.
I increased R8 to 470k, because there is no point to use CCS to get linear operation conditions and then ruin it with to heavy AC-load. Also 1 M could be used.
Now the THD of the voltage amplifier is some 0.29 % but the THD at phase inverter out put is only 0.16 %, so some distortion cancelling takes place.
The total gain is 43.4 dB and the gain of 1st 6SN7 is 26.3 dB.
Not a bad circuit at all, but I would use 6SL7 as the 1st tube to get some more gain. You can now fine tune it further.
I increased R8 to 470k, because there is no point to use CCS to get linear operation conditions and then ruin it with to heavy AC-load. Also 1 M could be used.
Now the THD of the voltage amplifier is some 0.29 % but the THD at phase inverter out put is only 0.16 %, so some distortion cancelling takes place.
The total gain is 43.4 dB and the gain of 1st 6SN7 is 26.3 dB.
Not a bad circuit at all, but I would use 6SL7 as the 1st tube to get some more gain. You can now fine tune it further.
Adequate for what? Reducing distortion? Reducing output impedance? Reducing hum? In each case, you need to know what the non-feedback value is and what value you would like to achieve.mskl99 said:
It depends. For the Vixen design, 12dbv of NFB works just fine distributed between gNFB and lNFB. For the Le Renard, 6dbv is more like it. That design has variable gNFB only, and the full 13dbv of gNFB is definitely tending to a "solid statey" sound. Backing off helps bring Techno and Metal to life. 12dbv would be a good, general purpose, level of gNFB if you weren't adding variability.
As for how KT90s are, no idea. It depends on the harmonic distortion profile, what your preferences are, how the finals interact with the front end. It's basically a case of try, listen, evaluate, and adjust.
Attached is a PP-OPT model with 40 % UL taps. The primary inductance is 40 H.
You can adjust the turns ratio/ impedance by setting proper secondary inductance.
This model does not include coil resistances, but those can be added easily.
Given similar design competence, similar components or financial outlay, an open-loop PP amp will have significantly less distortion than a similar open-loop SE amp. The PP therefore needs less feedback. However, the significant distortion of SE means that it might not satisfy the criterion: make it good, then add feedback. If the open-loop distortion is high then feedback will create much more re-entrant distortion - this may be part of the reason why SE fans are often also 'anti feedback' fans.
I suppose if you had a really good OPT and rather clever circuits you could just have lots of loop gain and slap on lots of NFB to linearise the SE output, but why bother when PP is so much simpler and better?
I suppose if you had a really good OPT and rather clever circuits you could just have lots of loop gain and slap on lots of NFB to linearise the SE output, but why bother when PP is so much simpler and better?
😡
Since the whole ethic of most single ended design is ZERO negative feedback I believe that most people go that route to achieve the sound that they require.
The theory tha SE needs more feedback does not relate to real life.
I feel that this post is falling into the design it on computer
but build nothing and NEVER just listen with your ears camp.
The blind theory that lowest distortion at any cost will sound better usually comes from people who build nothing!
From the owner of 2A3 SE amp!!😱 (no feedback)
Since the whole ethic of most single ended design is ZERO negative feedback I believe that most people go that route to achieve the sound that they require.
The theory tha SE needs more feedback does not relate to real life.
I feel that this post is falling into the design it on computer
but build nothing and NEVER just listen with your ears camp.
The blind theory that lowest distortion at any cost will sound better usually comes from people who build nothing!
From the owner of 2A3 SE amp!!😱 (no feedback)
Zero negative feedback would strictly mean no triodes are allowed, but I assume you mean no extra added feedback in the circuit. If the sound that you require includes significant percent low order distortion and you are happy to use speakers with high mechanical damping (or prefer a strong bass resonance) then fine.ColinA said:
The problem is where the "just" appears in that sentence. 'Just listen with your ears' and 'listen with just your ears' can be very different things, yet people often confuse them. By the way, where did computers come in? I never mentioned them.
Who asked for lowest distortion at any cost? Not me. Other things being equal, lower distortion will sound better for those people who prefer low distortion (which includes me). Those who prefer more distortion, which may be a significant fraction of music lovers, will obviously choose or design equipment to deliver what they prefer. The problem only comes when such people try imagine or claim that their preferred distortion is somehow less distorting than less distortion.
I understand the part about SE needing more NFB than PP but not the bit about PP being able to handle more NFB than SE.
Cheers
Ian