Mmmmmm.... no. Close, but no
A profound statement! Appreciated - but now ... how does that help us going forward? Care to kindly offer an explanation?
haha guys... the tubefare is heating up
A bit more on the matter on DC offset equalising around the KT88 pair, if we look at the latest rev 0.3 (http://www.diyaudio.com/forums/atta...-20-kt88-pp-blocks-kt88-ul-mullard-draft3.pdf) I noticed something peculiar around the Cathode resistors R34-37 formation, if tubes now behave as BJTs & FETs in the following way where an increased resistor value would decrease any offset in the tube, in that case the 2 resistors on the right side half which are connected c-c are a waste and in that case only the resistors on the left side connected via GND should be left to maximise the DC offset suppression.
EDIT: oh well, they need a common series resistor before connected to ground to have a DC offset suppressing function, a CCS would do a big difference!
Cheers Michael
A bit more on the matter on DC offset equalising around the KT88 pair, if we look at the latest rev 0.3 (http://www.diyaudio.com/forums/atta...-20-kt88-pp-blocks-kt88-ul-mullard-draft3.pdf) I noticed something peculiar around the Cathode resistors R34-37 formation, if tubes now behave as BJTs & FETs in the following way where an increased resistor value would decrease any offset in the tube, in that case the 2 resistors on the right side half which are connected c-c are a waste and in that case only the resistors on the left side connected via GND should be left to maximise the DC offset suppression.
EDIT: oh well, they need a common series resistor before connected to ground to have a DC offset suppressing function, a CCS would do a big difference!
Cheers Michael
Last edited:
Indeed - that is not what I meant. I simply tried to convey that a balancing pot (in the 12AT7 anodes, slider to +), while also cumbersome to some degree, enables one to adjust for equal signal amplitude at the KT88 anodes, also including somewhat of the OPT differences.
To put in perspective, the question was asked as to 'preference'. I have no flag to wave against a CCS here; it is one small matter compared to another. My personal choice would be for a LTP resistor and small balancing pot in the 12AT7 anodes, setting while watching on the KT88 anodes with a scope. Using a square wave signal intelligently has advantages.
Maybe a few other simplifications while discussion is still on: R23 is somewhat pious - it is in series with a CCS impedance orders higher, no real danger of supersonic instability there. The same for R18 and R25. On the other hand, R32, R33, R38 and R39 do have roles to play.
On the role of a low-cut filter in an RIAA circuit: Absolutely; I have posted on that elsewhere. A very useful 12dB/octave slope can be had below whatever cut-off frequency is chosen, without the addition of a single extra component in a feedback type RIAA equaliser; by simply proportioning what is already there according to feedback filter theory, instead of leaving the response continuing on below say 20Hz down to almost d.c. Why this is not done as a regular standard ....
A profound statement! Appreciated - but now ... how does that help us going forward? Care to kindly offer an explanation?
There's also the factor of load impedances- with a cathode CCS, get those matched and balance is guaranteed, at least at the O/P tube grids.
Johan Ok I get it, I seem to have misinterpreted/read to quick your post.
Resistors, on the matter of R23s role I am not really sure either, that's correct the output impedance of the CCS is in comparison very large so an additional 1kOhm resistor is adding virtually "zero" to it, a power dropper it isn't either as the value is too small, so my guess were on something else that the Miller capacitance Q1 could have some negative impact in conjunction with a tubes Cathode though that point is low impedive.
R18, 25 grid stoppers, I would think it is a good practice as it forms a low pass filter with the Miller cap for RF junk in the ether, tubes are very good in picking up RF junk, radio stations etc etc, (et al: if and when a grid stopper is used it should be placed as close as possible to the grid pin, radio waves are very short and slightest pin/wire length acts as an antenna!), but you think it's not necessary on the matter of 12AT7 in our drawing, what's your take on it?
The 5th RIAA time constant by JP?
(I gather that you are probably aware of the 4th "unknown" RIAA time constant at 50 kHz.)
A link btw: riaa backgrounder 1 - Introduction
Cheers Michael
Resistors, on the matter of R23s role I am not really sure either, that's correct the output impedance of the CCS is in comparison very large so an additional 1kOhm resistor is adding virtually "zero" to it, a power dropper it isn't either as the value is too small, so my guess were on something else that the Miller capacitance Q1 could have some negative impact in conjunction with a tubes Cathode though that point is low impedive.
R18, 25 grid stoppers, I would think it is a good practice as it forms a low pass filter with the Miller cap for RF junk in the ether, tubes are very good in picking up RF junk, radio stations etc etc, (et al: if and when a grid stopper is used it should be placed as close as possible to the grid pin, radio waves are very short and slightest pin/wire length acts as an antenna!), but you think it's not necessary on the matter of 12AT7 in our drawing, what's your take on it?
The 5th RIAA time constant by JP?
(I gather that you are probably aware of the 4th "unknown" RIAA time constant at 50 kHz.)
A link btw: riaa backgrounder 1 - Introduction
Cheers Michael
Last edited:
Definitely use grid stoppers on all high gm types, including the 12AT7/ECC81 and 6GK5. Remember, that the 2 types were engineered for service in VHF front ends.
IMO, only Carbon composition (CC) resistors should be used in grid stopper positions. Only CC construction is both non-metallic and non-inductive. As already stated, keep the amount of metal, which can act as an antenna, to the absolute minimum. Non-inductive construction avoids creating a "tank" circuit between resistor and all sorts of capacitances. Stoppers are supposed to suppress parasitic oscillation. So, do nothing to encourage the pests.
IMO, only Carbon composition (CC) resistors should be used in grid stopper positions. Only CC construction is both non-metallic and non-inductive. As already stated, keep the amount of metal, which can act as an antenna, to the absolute minimum. Non-inductive construction avoids creating a "tank" circuit between resistor and all sorts of capacitances. Stoppers are supposed to suppress parasitic oscillation. So, do nothing to encourage the pests.
Regarding R23,
I've attached a clip of the original schematic that includes a couple annotations in that section. They seem to imply (to my interpretation) that resistor is a current "fine tuner". I have no idea how one hand-matches a single resistor. I suppose you do it with the hand that isn't clapping.
..Todd
I've attached a clip of the original schematic that includes a couple annotations in that section. They seem to imply (to my interpretation) that resistor is a current "fine tuner". I have no idea how one hand-matches a single resistor. I suppose you do it with the hand that isn't clapping.
..Todd
Attachments
Last edited:
Hi,
SY stated a good deal of them.
But also there's the fact one triode is grid driven and the other cathode driven, which will naturally lead to imbalances. That is why many designers use unequal anode resistors for AC balance at the sacrifice of DC balance.
Using a common-mode feedback loop with the CCS in the tail can make the difference millivolts, even with unbalanced tube sections and with using equal anode resistors to maintain DC balance.
This method was developed by Fred Nachbuar (RIP, posted on audio places as fnachbuar or 807) for his RA-100 amplifier:
RA-100: LTP/CCS PA Driver Improvement
And I adapted it for a SS CCS a few years ago, the latest version is used in my DynaMutt driver board:
http://www.diyaudio.com/forums/blogs/geek/140-dynamutt-st-70-should-have-been.html
(scroll down to the driver)
And it can be used with a pure bipolar CCS too, as I did in a guitar amp design:
http://ckamps.com/docs/FuzzyPup_v0.gif
or for HiFi:
A 6V6 PP Project
Cheers!
A profound statement! Appreciated - but now ... how does that help us going forward? Care to kindly offer an explanation?
SY stated a good deal of them.
But also there's the fact one triode is grid driven and the other cathode driven, which will naturally lead to imbalances. That is why many designers use unequal anode resistors for AC balance at the sacrifice of DC balance.
Using a common-mode feedback loop with the CCS in the tail can make the difference millivolts, even with unbalanced tube sections and with using equal anode resistors to maintain DC balance.
This method was developed by Fred Nachbuar (RIP, posted on audio places as fnachbuar or 807) for his RA-100 amplifier:
RA-100: LTP/CCS PA Driver Improvement
And I adapted it for a SS CCS a few years ago, the latest version is used in my DynaMutt driver board:
http://www.diyaudio.com/forums/blogs/geek/140-dynamutt-st-70-should-have-been.html
(scroll down to the driver)
And it can be used with a pure bipolar CCS too, as I did in a guitar amp design:
http://ckamps.com/docs/FuzzyPup_v0.gif
or for HiFi:
A 6V6 PP Project
Cheers!
Last edited:
Regarding R23,
I've attached a clip of the original schematic that includes a couple annotations in that section. They seem to imply (to my interpretation) that resistor is a current "fine tuner". I have no idea how one hand-matches a single resistor. I suppose you do it with the hand that isn't clapping.
..Todd
hand matched to the one in the other channel..........
Geek: If you have a CCS in the tail and the loads are equal, the output is balanced. Period. The cathode versus grid thing is absolutely irrelevant. I once thought the same as you; EC8010 told me I was wrong, so I went and measured it for myself. Whoops. The unbalanced loads were traditionally used because people didn't have good CCS, they used very finite tail resistors.
Think of it this way: there's no grid current and the tail has essentially infinite AC impedance. That means the current through the loads has to be equal by Kirchoff's Law- there's no other place for it to go. So the balance is inherently determined by the degree to which the loads are matched and nothing else. In fact, you could even use two different tubes for each section and the balance would STILL be as perfect as the load matching.
Think of it this way: there's no grid current and the tail has essentially infinite AC impedance. That means the current through the loads has to be equal by Kirchoff's Law- there's no other place for it to go. So the balance is inherently determined by the degree to which the loads are matched and nothing else. In fact, you could even use two different tubes for each section and the balance would STILL be as perfect as the load matching.
I disagree.
Here's why:
Driver Board for the ST-70 Using Common Mode Feedback PI
My measurements say quite different.
Cheers!
Notice something wrong with the second scope photo? (actually, same problem on all of them, but it really shows up there) Get out that screwdriver!
I'll set up an extreme case this weekend (as I did when I tried to prove EC wrong and found out he was right) and post the scope photos. Kirchoff's Law really does work. In your circuit, critical matching includes not just the plate resistors, but also your feedback resistors
I'll set up an extreme case this weekend (as I did when I tried to prove EC wrong and found out he was right) and post the scope photos. Kirchoff's Law really does work. In your circuit, critical matching includes not just the plate resistors, but also your feedback resistors
I've actually used two scopes. Dunno where the original pics went, they were far better
Look forward to your results.
Part of the CMFB circuits value is to eliminate the need for such matching. The FB resistors, 5% is plenty good enough because it forms a small part of such a huge divider.
Cheers!
Geek's board has starting driving the amps..........
diytube.com :: View topic - Driver Board for the ST-70 Using Common Mode Feedback PI
diytube.com :: View topic - Driver Board for the ST-70 Using Common Mode Feedback PI