martin manning wrote:For those interested in the effect of the PI changes, I ran some simulations and found that D's version has about 2 dB less gain at 10 Hz and 1dB less gain from 200Hz on up so the BW is actually reduced a bit. The size of the cap in the LNFB controls that additional LF roll-off, and if I make it 100n it's -1 dB across the board. The trimmer of course allows balancing the outputs, which is a very useful feature.
Martin
Thanks for running those numbers.. I assume you were comparing the stock 5E3 Cathodyne to HAD's (lower gain)version..
I recently modded a BF Princeton reverb with this PI and noticed an increase in headroom..
I imagine with a 110k plate (more swing) and the better balanced PI and the local feedback loop that HAD's version would have more headroom too..(which also seems to be a theme with this amp, ie,better filtering and split cathodes)..So if one were to measure output power (the old fashioned way) at the onset of clipping you would be able to get more power from HAD's version..(this is what I think KWS is hearing with his..Mo headroom..IMO)So not to confuse gain with power output..Of coarse we really wouldn't know for sure until we scoped and measured..But this is what my ears tell me!
Tony
" The psychics on my bench is the same as Dumble'"
martin manning wrote:For those interested in the effect of the PI changes, I ran some simulations and found that D's version has about 2 dB less gain at 10 Hz and 1dB less gain from 200Hz on up so the BW is actually reduced a bit. The size of the cap in the LNFB controls that additional LF roll-off, and if I make it 100n it's -1 dB across the board. The trimmer of course allows balancing the outputs, which is a very useful feature.
Martin
Thanks for running those numbers.. I assume you were comparing the stock 5E3 Cathodyne to HAD's (lower gain)version..
I recently modded a BF Princeton reverb with this PI and noticed an increase in headroom..
I imagine with a 110k plate (more swing) and the better balanced PI and the local feedback loop that HAD's version would have more headroom too..(which also seems to be a theme with this amp, ie,better filtering and split cathodes)..So if one were to measure output power (the old fashioned way) at the onset of clipping you would be able to get more power from HAD's version..(this is what I think KWS is hearing with his..Mo headroom..IMO)So not to confuse gain with power output..Of coarse we really wouldn't know for sure until we scoped and measured..But this is what my ears tell me!
Tony
Good call Tony! Kinda semi version of a Precision PS!
martin manning wrote:The trimmer of course allows balancing the outputs, which is a very useful feature.
...I imagine with a 110k plate (more swing) and the better balanced PI and the local feedback loop that HAD's version would have more headroom too..
Tony
How best to adjust the PI trim with a cathode-biased output? In the Ampeg SVT, this same PI balance is adjusted by observing balance on the output tube cathode reference resistors to ground. To balance the output, including the effects of the output tubes, and not just the PI tube, the optimum setup might be to split the cathode resistors. Or just dial to taste - listening at high volume with everything turned up.
The sim shows that the PI in this circuit clips sooner than the standard 5E3 by a wide margin- at 31V (peak) input vs. 41V at 400Hz with a 250v supply. The output swing is still quite adequate at 25V (vs. 38V), though, and I suspect that the observed increase in head room is due to the slightly lower gain as well as not driving the power tubes so hard. Output power at the onset of clipping should not be affected if that is the case.
To adjust the PI trimmer I think I'd start by measuring AC volts across the PI outputs with signal applied. Adjust to zero, then tweak by ear.
Edit: Increased supply voltage to 250V.
Last edited by martin manning on Thu Jun 19, 2014 8:22 am, edited 1 time in total.
The sim shows that the PI in this circuit clips sooner than the standard 5E3 by a wide margin
So by raising the plate voltage from 250v to 309 (22k to 12K) also raising both the plate rails (swing) from 56K to 110k and the cathode from 65k to 103K (reduction in gain)....Balanced PI (lower THD) and LNFB (decreasing harmonic distortion)..Results in a PI with less headroom,than the stock 5E..Not doubting your findings just wondering why these changes would result in a PI with less headroom? What am I missing?
Tony
Last edited by talbany on Thu Jun 19, 2014 5:36 pm, edited 1 time in total.
" The psychics on my bench is the same as Dumble'"
I ran them at constant supply voltage. If I raise the D-type to 310 then I can get to about 38V peak input at clip. Remember the AC load doesn't go up as much as the Ra and Rk would suggest because you still have the power tube grid leaks in parallel.
martin manning wrote:I ran them at constant supply voltage. If I raise the D-type to 310 then I can get to about 38V peak input at clip. Remember the AC load doesn't go up as much as the Ra and Rk would suggest because you still have the power tube grid leaks in parallel.
OK Thanks!!..(Cathodynes already have a good amount of headroom!)...Yeah but the grid leaks should be the same with the stock 5E as well correct..We know we are not driving the grids as hard with the D-version..Not trying to be a smart ass here just trying to figure what is it about the D-version that causes it to have considerably less headroom than the stock 5E where one could think all these changes HAD did (for me) points to a PI with (slightly) increased headroom..Any ideas there?
Tony
" The psychics on my bench is the same as Dumble'"
Hello, just to throw in my .02 cents. You guys are WAY over my head with the math side of this ( I have been told by a good friend of Dumble that he is VERY into the math side also) but I always thought that a cathodyne phase invertor was rather inefficient(sounds like I maybe wrong). I always attributed this as why I like the way Blackface Princetones and Tweed Deluxes break up. Dumble has kind of filtered the heck out of this thing and also raised the voltages up with the GZ34 rectifier. Both those things in my experience make Tweed Deluxes not want to give it up as easily. Maybe the trimmer is there not to balance the PI but inbalance it for sonic reasons.
What am I missing?
