Hugh ...
yes, you are correct. This thread started however with a given circuit on which was commented (I came in late), without immediately going to something different. Some values were changed using the same topology and tubes, which lead only now to the finding that the 12AT7 is after all not such a preference here although a favourite with many (including myself). This showed up when looking at how the signal of 100Vpp required for the KT88s would fit into the Va-Ia graph family for some triodes. This showed that, for an LTP, where the cathodes already loose some 100V off 460V, a low Va.sat was at a premium. That lead to my recent table and the sad fact that 12AT7 might not make the grade here.
Continuing:
It is perhaps high time (I cannot post regularly) that, having been 'wise', I better suggest my personal alternatives for the relevant circuit - mostly values. I again refer to the neat diagram from Taj (version 06, 7 February 2010).
As said above, I am lead to adopt the ECC88 for the LTP, although the E182CC also features strongly (only the Miller capacitance 😱!). Both these tubes can 'bottom' at exceptionally low anode voltages; I believe the E182CC might be more expensive. I have built up the circuit (sans KT88s) and tried to fit the 100Vpp signal into the anode swing as best I could. The following values do in fact, render an ECC88 capable of delivering a signal of 250Vpp on anode at visible setting in of distortion (overload). Thus, I would suggest the following component changes:
R14 = 470K
adding R? of 1,5Meg from EF86 pin 1 to common
R15 = 330 ohm
R21 = R22 = 27K (5W)
Also, the 'heater lifting' chain can draw less current:
R3 = 270K
In my construction the following d.c. voltages resulted (all measured with a 10M impedance DVM):
EF86 pin 1: 73V
EF86 pin 3: 0,82V
EF86 pin 6: 81V
ECC88 pins 1 and 6: 270V
ECC88 pins 3 and 8: 87V
Junction R3, R4: 42V
EF86 B+: 288V
HT+ : 460V
With these values I find the total gain 1140, thus suggesting a R17 = 39K for some 20dB NFB. As said before, C13 would depend on the OPT. It's value is best determined experimentally using a square wave; I would guess in the vicinity of 82pF. Likewise, C11-R12, if necessary.
Simulating the input capacitance of the KY88s, I found a h.f. -3dB frequency of only 32 kHz (expected higher!). This could be 'synthetically' increased by cancelling some of the ECC88 Miller effect. As anode 6 is of opposite polarity than anode 1, feedback from there would cancel that from anode 1 (Miller feedback). I found a resistor of some 1,2 megohm in series with 2,7pF from pin 6 to pin 2 (ECC88) to increase -3dB h.f. to 50kHz. This again will depend on the inter-capacitances of the final lay-out. Cut and try with the resistor value will yield best option, using a square wave signal. (Where these days low value ceramic capacitors are mostly <100V, this connection can be made from the cold side of C15 to pin 1.)
The LTP tail current was 14mA in my set-up. The best way to adjust the 10M45S will be by setting the voltages at ECC88 pins 1 and 6 to 270V. (10M45S will dissipate about 1,1W, so using a heatsink will be essential.)
Folks, that my contribution. As said before, I would suggest this as a chopping block. Other alternatives could come close. Those members with spectrum analyser facilities can be the final judge.
It is perhaps high time (I cannot post regularly) that, having been 'wise', I better suggest my personal alternatives for the relevant circuit - mostly values. I again refer to the neat diagram from Taj (version 06, 7 February 2010).
As said above, I am lead to adopt the ECC88 for the LTP, although the E182CC also features strongly (only the Miller capacitance 😱!). Both these tubes can 'bottom' at exceptionally low anode voltages; I believe the E182CC might be more expensive. I have built up the circuit (sans KT88s) and tried to fit the 100Vpp signal into the anode swing as best I could. The following values do in fact, render an ECC88 capable of delivering a signal of 250Vpp on anode at visible setting in of distortion (overload). Thus, I would suggest the following component changes:
R14 = 470K
adding R? of 1,5Meg from EF86 pin 1 to common
R15 = 330 ohm
R21 = R22 = 27K (5W)
Also, the 'heater lifting' chain can draw less current:
R3 = 270K
In my construction the following d.c. voltages resulted (all measured with a 10M impedance DVM):
EF86 pin 1: 73V
EF86 pin 3: 0,82V
EF86 pin 6: 81V
ECC88 pins 1 and 6: 270V
ECC88 pins 3 and 8: 87V
Junction R3, R4: 42V
EF86 B+: 288V
HT+ : 460V
With these values I find the total gain 1140, thus suggesting a R17 = 39K for some 20dB NFB. As said before, C13 would depend on the OPT. It's value is best determined experimentally using a square wave; I would guess in the vicinity of 82pF. Likewise, C11-R12, if necessary.
Simulating the input capacitance of the KY88s, I found a h.f. -3dB frequency of only 32 kHz (expected higher!). This could be 'synthetically' increased by cancelling some of the ECC88 Miller effect. As anode 6 is of opposite polarity than anode 1, feedback from there would cancel that from anode 1 (Miller feedback). I found a resistor of some 1,2 megohm in series with 2,7pF from pin 6 to pin 2 (ECC88) to increase -3dB h.f. to 50kHz. This again will depend on the inter-capacitances of the final lay-out. Cut and try with the resistor value will yield best option, using a square wave signal. (Where these days low value ceramic capacitors are mostly <100V, this connection can be made from the cold side of C15 to pin 1.)
The LTP tail current was 14mA in my set-up. The best way to adjust the 10M45S will be by setting the voltages at ECC88 pins 1 and 6 to 270V. (10M45S will dissipate about 1,1W, so using a heatsink will be essential.)
Folks, that my contribution. As said before, I would suggest this as a chopping block. Other alternatives could come close. Those members with spectrum analyser facilities can be the final judge.
Johan, did you get the email I sent yesterday with additional questions.
Thank's for all the work on this. You've gone above and beyond! I'm going to impliment all these schematic changes tonight, and bread board this in anticipation of my transformer shipment. I'll then scope it with a SQ wave for the final componet values for feedback. If all seems well i'll start the build.
Thank's for all the work on this. You've gone above and beyond! I'm going to impliment all these schematic changes tonight, and bread board this in anticipation of my transformer shipment. I'll then scope it with a SQ wave for the final componet values for feedback. If all seems well i'll start the build.
Last edited:
TubeMack,
That I did, Sir.
Pardon not replying immediately; I will do so soonest. The question was if there are any triodes for input stage use where preferred; yes - but I will reply shortly.
That I did, Sir.
Pardon not replying immediately; I will do so soonest. The question was if there are any triodes for input stage use where preferred; yes - but I will reply shortly.
Further,
Not sure whether the Steve Bench site supplied all the answers to the load line queries. His principles are in order, I choose the operating point rather differently though. If this is still unclear one could sum up in a single post for the uninitiated.
I would like to repeat that I have nothing against the 12AT7 per se as does Eli, but for this circuit it seems to be rather stretched to reach the desired KT88 grid amplitude comfortably. This was tested on an oscilloscope; one can easily describe how to judge this simply from inspecting the Ia-Va family of graphs. (But guys,
how come the available specs from different manufacturers so often differ enough to want one reach for antacid ... - as well as sometimes the measured values between 'new' examples. Wants one to go to semiconductors ...😱 OOPS!)
Not sure whether the Steve Bench site supplied all the answers to the load line queries. His principles are in order, I choose the operating point rather differently though. If this is still unclear one could sum up in a single post for the uninitiated.
I would like to repeat that I have nothing against the 12AT7 per se as does Eli, but for this circuit it seems to be rather stretched to reach the desired KT88 grid amplitude comfortably. This was tested on an oscilloscope; one can easily describe how to judge this simply from inspecting the Ia-Va family of graphs. (But guys,
how come the available specs from different manufacturers so often differ enough to want one reach for antacid ... - as well as sometimes the measured values between 'new' examples. Wants one to go to semiconductors ...😱 OOPS!)
Mmmmm.
And then there are yet others who believe that a valve input stage driving semiconductors is the thing.
I give up.
And then there are yet others who believe that a valve input stage driving semiconductors is the thing.
I give up.
Rene" Jaeger, designer of the Berkley Audio DAC and more recently chief analog designer at Mackie Designs, was showing off that very thing at our last amp builders club meeting.
A SE 300 B driving a 1 to .5/.5 interstage, driving a modified Gary Pimm Tabor PP power stage into a PP 6.6 kZ OPT for 20 watts of power. Had the 300B "beauty of music" character and so little distortion of any kind that even driving Maggies it sounded effortless, sweet and extremely detailed, with very deep and clear bass.
Surprised everyone there, absolutely no indication it was either a tube or solid state amp.
Bud
A SE 300 B driving a 1 to .5/.5 interstage, driving a modified Gary Pimm Tabor PP power stage into a PP 6.6 kZ OPT for 20 watts of power. Had the 300B "beauty of music" character and so little distortion of any kind that even driving Maggies it sounded effortless, sweet and extremely detailed, with very deep and clear bass.
Surprised everyone there, absolutely no indication it was either a tube or solid state amp.
Bud
(But guys,
Actually semiconductors seem to have the opposite effect of antacids. The little TO92's aren't too bad, but the TO3's hurt for hours. I recommend calcium carbonate as a less irritable alternative.
I'm not confident about the instructions, Johan, but I transliterated them to the schematic as best I could. (v0.7 attached.)
I should add that I don't understand pin 9 on the ECC88. Is the heater 12v?
..Todd
OOPS!
My bad in several ways. ECC88 has a 6V heater, pins 4 and 5 as per the EF86. The internal triodes each have 6.3V heaters, connected in parallel internally. That pin 9 then becomes an internal screen (shield, not G2!) - it should go to earth. Then C17 should rather go to the other side of the ECC88 grid stopper; i.e. pin6, EF86, instead of pin 2, ECC88. But mounted close 'over' the ECC88 socket. I also offered that as a later 'remedy'. As said, the value of R23 might differ. One should first get the circuit working well without them, then with the aid of an oscilloscope install them as a 'touch-to-make-contact' experiment (watch out for hand capacity) to make the rise edges of a square wave rectangular. That should also be measured on the grid side of the KT88s, not the voice coil output. This measure is not there to compensate for total amplifier shortcomings, just the ECC88 response, otherwise instability could result.
But more about that later if required. As said this is still a suggestion for others to comment on; the main thing being that one should use lower R21, R22 values and higher anode current to give the LTP some breathing space/headroom.
Also, I have kept to a mainly tube design as per original. There are alternatives by introducing semiconductors in certain places as has been suggested. In principle no problem with that; I personally prefer a tube amp as just that for the signal part, but that simply out of nostalgia, no scientific reasons for that.
My bad in several ways. ECC88 has a 6V heater, pins 4 and 5 as per the EF86. The internal triodes each have 6.3V heaters, connected in parallel internally. That pin 9 then becomes an internal screen (shield, not G2!) - it should go to earth. Then C17 should rather go to the other side of the ECC88 grid stopper; i.e. pin6, EF86, instead of pin 2, ECC88. But mounted close 'over' the ECC88 socket. I also offered that as a later 'remedy'. As said, the value of R23 might differ. One should first get the circuit working well without them, then with the aid of an oscilloscope install them as a 'touch-to-make-contact' experiment (watch out for hand capacity) to make the rise edges of a square wave rectangular. That should also be measured on the grid side of the KT88s, not the voice coil output. This measure is not there to compensate for total amplifier shortcomings, just the ECC88 response, otherwise instability could result.
But more about that later if required. As said this is still a suggestion for others to comment on; the main thing being that one should use lower R21, R22 values and higher anode current to give the LTP some breathing space/headroom.
Also, I have kept to a mainly tube design as per original. There are alternatives by introducing semiconductors in certain places as has been suggested. In principle no problem with that; I personally prefer a tube amp as just that for the signal part, but that simply out of nostalgia, no scientific reasons for that.
I see the source of my confusion with the heaters. The pin-out at tdsl.duncanamps.com shows them wired in series. The datasheets show otherwise. Internal shield to ground... Got it.
I think that also best reflects TubeMack's original intentions.
..Todd
Last edited:
Taj, I thought you had decided to stop documenting the project so guess what I spent 5 hours last night doing.
Thats cool, I really needed the practice! I guess I'll post it anyway since I worked so long on it. I don't have the time to learn Visio right now, so I had to use "Paint" with a symbol library I pasted together from other schematics.An externally hosted image should be here but it was not working when we last tested it.
I have temporarily removed the EF86 in anticipation of a better triode substitute from new findings by Johan. I have also tenatively picked a name for the amp. It was from my favorite classical piece that I happened to be listening to while doing the schematic. Adagio for strings!🙂
Cheers!
Last edited:
Tubemack,
Very Nice. If you want 14mA across the CCS in the ltp, I would suggest that R11 be set to 100R. To set 14mA should require about 180R.
I really do like the name, excellent choice.
Very Nice. If you want 14mA across the CCS in the ltp, I would suggest that R11 be set to 100R. To set 14mA should require about 180R.
I really do like the name, excellent choice.
Tubemack, If you are interested in continuing to draw the schematics with visio, I would be happy to send the base file to use as a template. It isn't very hard to learn
Taj, I thought you had decided to stop documenting the project so guess what I spent 5 hours last night doing.
An externally hosted image should be here but it was not working when we last tested it.
I have temporarily removed the EF86 in anticipation of a better triode substitute from new findings by Johan. I have also tenatively picked a name for the amp. It was from my favorite classical piece that I happened to be listening to while doing the schematic. Adagio for strings!🙂
Cheers!
R1 is too tall. 100 KOhms is about right. A large value in that position interacts adversely with the triode's CMiller to cause HF rolloff.
The unlabeled cathode bias resistor must either be bypassed or replaced by LEDs. A low impedance path to the cathode for injection of loop NFB is essential.
You have 271 V. on the LTP anodes. A 150 V. anode to cathode differential requires a little less than a -4 V. grid to cathode differential. That (as it must) places the voltage amplifier anode at "117" V. It seems an IR emitting diode biased 6GK5, with IB = 3.5 mA., will do. 10M45S load the 6GK5's anode, to ensure good open loop linearity.
Thanks!
R1 will be changed.
Led will be substituted.
Waiting on the last of Johan's findings for the Voltage amp, but the 6GK5 is still an option. If so, i'll impliment those changes for it. Hopefully soon!
Almost Done, but it appears I may be the only builder now? Not sure......
Last edited:
Almost Done, but it appears I may be alone in the build now?[/QUOTE]
I might be interested in building one also...I've been watching and waiting to see how it ends up.
I might be interested in building one also...I've been watching and waiting to see how it ends up.
Taj, I thought you had decided to stop documenting the project so guess what I spent 5 hours last night doing.
You've done well. I can't believe that's MS paint. What torture. 😱 Visio would be much easier (faster) if you have it.
Actually I didn't give up documenting. I gave up WAITING and participating in the technical discussion and decision making because it was moving in every other direction (mostly in a spiral) except forward. Johan has it moving in the right direction -- and kudos to Johan for that.
[And I also had some personal life circumstances that made participating difficult. That's resolved.]
It's a worthwhile project.
..Todd
Last edited:
Thats cool. Yeah it was kinda tough using paint, but still kinda therapuetic. Next project (12B4 Pre), I will attempt to document it myself. Look at the LED I had to do from scratch. The "Photon" had to be hand drawn. I think i'm going blind!
An externally hosted image should be here but it was not working when we last tested it.
Last edited: