New guy here, I am rebuilding an old Capehart amp that was originally a push-pull 6F5/6L6 mono. I would like to convert it to a SE stereo using the original componets as much as possible. Can anyone post a schematic for a SE 6F5 (with grid caps) and 6L6 outputs...new to tubes but not electronics, I have read the tube manuals and have a fair idea of what's going on (kinda-sorta) but any imput would be welcomed. The rectifier is a 5Z3, I have refurbished the mains xformer and the choke, just need output xformers, schematic and some guidance....
thanks
jt
thanks
jt
Attachments
Last edited:
A look at the 6F5 data sheet indicates that it is electrically similar to the triodes in the 12AX7. It's a very good voltage amplifier, but can't fight its way out of a wet paper bag.
Other than its UX4 base, the 5Z3 is the same as a 5U4G.
Those phenolic wafer sockets are crap and should be replaced.
You have 19 W. plate dissipation 6L6Gs. A look at the data sheet should provide some ideas about operating conditions.
The Edcor CXSE25-8-3.2K could be suitable. Definitely take advantage of the ultralinear tap. Put a 330 Ω current limiting resistor in the line between g2 and the UL tap.
I'm thinking constant current loading for the voltage amplifier and DC coupled ZVN0545A MOSFET buffering too. Remember that the 6F5 is a wimp.
Other than its UX4 base, the 5Z3 is the same as a 5U4G.
Those phenolic wafer sockets are crap and should be replaced.
You have 19 W. plate dissipation 6L6Gs. A look at the data sheet should provide some ideas about operating conditions.
The Edcor CXSE25-8-3.2K could be suitable. Definitely take advantage of the ultralinear tap. Put a 330 Ω current limiting resistor in the line between g2 and the UL tap.
I'm thinking constant current loading for the voltage amplifier and DC coupled ZVN0545A MOSFET buffering too. Remember that the 6F5 is a wimp.
Thanks for the reply and the links. I was thinking the sockets left something to be desired but as they were from the "good ole days" I figured they were OK...I'll replace them with ceramic.
As for the 6F5's, I just scaned the spec sheet's, not really knowing all the in's and out's they looked OK to me.
Perhaps I need to reasses this project and look to a more modern, well documented circuit for my first tube build. I would like to keep this project all tube and not have to use SS parts to compensate for a poor preforming desgin...maybe I don't want it as original as I thought.
Looks like I have much to learn...
As for the 6F5's, I just scaned the spec sheet's, not really knowing all the in's and out's they looked OK to me.
Perhaps I need to reasses this project and look to a more modern, well documented circuit for my first tube build. I would like to keep this project all tube and not have to use SS parts to compensate for a poor preforming desgin...maybe I don't want it as original as I thought.
Looks like I have much to learn...
Eli knows his stuff. I'd follow his instructions as they will lead at worst to a reasonable performance.
The only issues you have left to deal with at this early stage are operating points and the ability of your power transformer to reach them.
Sure, there is now a bunch of math to do, but oddly enough thats a part of the fun.
Ask away, I'm sure plenty are watching and just waiting their turn to jump in with advice...
The only issues you have left to deal with at this early stage are operating points and the ability of your power transformer to reach them.
Sure, there is now a bunch of math to do, but oddly enough thats a part of the fun.
Ask away, I'm sure plenty are watching and just waiting their turn to jump in with advice...
If you are willing to put the 6F5s away for a future project, I think we can come up with a 100% SS free design.
Ultralinear mode "finals" require some loop NFB, in order to obtain a satisfactory damping factor. IMO, you would be hard pressed to find a better package of high gain, high gm, and low RP than the 6GK5, whose data sheet is here. The 6GK5 has "stones" and needs no buffering help. To use the 6GK5, cut rhombic shaped pieces of sheet Aluminum of about the same size as the Octal phenolic wafers. "Break" the corners, for safety's sake. Mount 7 pin mini sockets in the pieces and mount the pieces where the 6F5 sockets now are. The technique is known as hole shrinking.
Circuit topology will be quite straight forward. Both the driver and the "final" will be self biased. Simple RC interstage coupling will be used. GNFB will be taken from the speaker tap and applied to the driver's cathode.
Do you have an o'scope, which can be used to optimize NFB phase compensation? If not, a brute force method of phase compensation is available, which delivers pretty good results. So, if push comes to shove, the only instrument absolutely necessary to finish the project is a decent multimeter.
Eli - The 6F5's can wait for another application, I really did like the looks of the grid cap wires though...
As far as "hole shrinking" it's not an issue, I'm much better at metal work than tube amp design.
I pretty much understand the Ultralinear and GNFB approach and really don't have any issues with it, but I keep hearing some folks talking about the virtues of pure SET sound. Can I have a switchable setting for this, or am I misinformed...the end goal is to make some really good noise.
I really appreciate your time and sharing of your knowledge. I guess to get started I will measure the xformer voltages and test the 5Z3...I'll post the results.
thanks
jt
As far as "hole shrinking" it's not an issue, I'm much better at metal work than tube amp design.
- This is a good thing. I can do good fab/assembly work, but I'll need a good schematic to work from.
I pretty much understand the Ultralinear and GNFB approach and really don't have any issues with it, but I keep hearing some folks talking about the virtues of pure SET sound. Can I have a switchable setting for this, or am I misinformed...the end goal is to make some really good noise.
- Yes, I have a vintage Dumont single channel that works well and access to modern equipment if necessary.
I really appreciate your time and sharing of your knowledge. I guess to get started I will measure the xformer voltages and test the 5Z3...I'll post the results.
thanks
jt
JT,
IMO, 2 things about that circuitry stand out. Notice the use of a speaker field coil as a PSU filter element. High flux AlNiCo permanent magnets are a post WW2 development. Pre-war, electromagnets had to be used to obtain high flux. Capehart used a differential, AKA LTP, phase splitter. LTPs are quite in vogue, currently.
You will retain the CLC filter topology Capehart used. A 20 μF./500 WVDC Sprague "atom" in the 1st filter position will give you better performance than OEM, without over stressing either the 5Z3 or the power trafo. As the filter choke shields the rectifier and power trafo from the remainder of the PSU filter, it is safe to pile up the energy storage in the 2nd cap. position.
IMO, 2 things about that circuitry stand out. Notice the use of a speaker field coil as a PSU filter element. High flux AlNiCo permanent magnets are a post WW2 development. Pre-war, electromagnets had to be used to obtain high flux. Capehart used a differential, AKA LTP, phase splitter. LTPs are quite in vogue, currently.
You will retain the CLC filter topology Capehart used. A 20 μF./500 WVDC Sprague "atom" in the 1st filter position will give you better performance than OEM, without over stressing either the 5Z3 or the power trafo. As the filter choke shields the rectifier and power trafo from the remainder of the PSU filter, it is safe to pile up the energy storage in the 2nd cap. position.
This is what I have so far, since this is no longer a restoration, are the 6L6 outputs still a good choice? Feel free to fill in the blanks
PARTS LIST
C1 - 20 μF./500 WVDC Sprague "atom" Cap.
C2 - ?
.
.
.
DRIVERS - 6GK5 (2)
.
.
OPT - Edcor CXSE25-8-3.2 (2)
thanks
jt
An externally hosted image should be here but it was not working when we last tested it.
PARTS LIST
C1 - 20 μF./500 WVDC Sprague "atom" Cap.
C2 - ?
.
.
.
DRIVERS - 6GK5 (2)
.
.
OPT - Edcor CXSE25-8-3.2 (2)
thanks
jt
This is what I have so far, since this is no longer a restoration, are the 6L6 outputs still a good choice? Feel free to fill in the blanks
An externally hosted image should be here but it was not working when we last tested it.
PARTS LIST
C1 - 20 μF./500 WVDC Sprague "atom" Cap.
C2 - ?
.
.
.
DRIVERS - 6GK5 (2)
.
.
OPT - Edcor CXSE25-8-3.2 (2)
thanks
jt
Dude,
I meant it, when I said pile up the energy storage after the choke. Back in the days of yore, the high value 'lytics we take for granted did not exist. Wire a pair of Panasonic EEU-ED2E221 (220 μF./250 WVDC) 'lytics in series and add parallel voltage equalizing resistors. You can use 2X series wired Panasonic EEU-ED2E470, with voltage equalizing resistors, in the 1st position. Those ED series parts a 105o C./10000 hour rated and really good stuff.
I've uploaded the beginnings of a signal schematic (1 channel). For now, not all parts values are known. The dots near the O/P trafo indicate phasing.
Attachments
Eli - I had read somewhere (can't find it now) in an old RCA manual that the 5Z3 should be limited to 36uF in the CLC filter...guess I read wrong. Thanks for the schematic, this is exactly what I need I do much better with visuals. I found the ED series caps, when you say the 1st position you mean after the choke? A PSU diagram would help me greatly.
Sorry to seem like such a baby about all this but as I said in the beginning I'm no designer, but I can build and work with HV. I'm hoping to use this project to understand what's really going on at each component and connection and believe it or not I have made progress from just a short time ago.
I'm in no rush here and want to do this correctly...please take your time, I do appreciate your efforts
jt
Sorry to seem like such a baby about all this but as I said in the beginning I'm no designer, but I can build and work with HV. I'm hoping to use this project to understand what's really going on at each component and connection and believe it or not I have made progress from just a short time ago.
I'm in no rush here and want to do this correctly...please take your time, I do appreciate your efforts
jt
I've uploaded a PSU schematic. For better channel separation, the small signal portion of the supply is pseudo dual mono.
Caps. in series follow a reciprocal law, like resistors in parallel. The net for 2X 47 μF. parts in series is 23.5 μF., which does not exceed the 5Z3's published limit.
Change the signal schematic to make the interstage coupling cap. value 220 nF./400 WVDC and the 6L6 grid to ground resistor 220 KOhms. This change is made, after a review of the 6L6 data sheet.
Caps. in series follow a reciprocal law, like resistors in parallel. The net for 2X 47 μF. parts in series is 23.5 μF., which does not exceed the 5Z3's published limit.
Change the signal schematic to make the interstage coupling cap. value 220 nF./400 WVDC and the 6L6 grid to ground resistor 220 KOhms. This change is made, after a review of the 6L6 data sheet.
Attachments
...OK, This is what I have so far:
Feel free to fill in the blanks...
So...can we address the signal input components and inserting a volume control.
Do I need to start assembly in order to determine the remaining component values?
Thanks again for your time...If anyone else has input it is welcome...seems the more I learn the more I need to learn
jt
Hi jt,
R6 function is to set the amount of feedback. This must be chosen by ear, it's down to personal preference and required gain.
'Phase compensation' applies only to C3 and must be chosen experimentally after choosing R6 (connecting a dummy load + oscilloscope at the output and a 10kHz square wave generator at the input, adjust for minimal ringing without rounding the step of the square wave too much)
R8 defines the output tube bias and idle dissipation, this will also have some effect on the sonics. You should try different values (without exceeding the dissipation of the tube).
C4 must be chosen so its reactance is negligible compared to R8 at the lowest frequency you want to reproduce (usually 20Hz).
Kenneth
R6 function is to set the amount of feedback. This must be chosen by ear, it's down to personal preference and required gain.
'Phase compensation' applies only to C3 and must be chosen experimentally after choosing R6 (connecting a dummy load + oscilloscope at the output and a 10kHz square wave generator at the input, adjust for minimal ringing without rounding the step of the square wave too much)
R8 defines the output tube bias and idle dissipation, this will also have some effect on the sonics. You should try different values (without exceeding the dissipation of the tube).
C4 must be chosen so its reactance is negligible compared to R8 at the lowest frequency you want to reproduce (usually 20Hz).
Kenneth
The volume control should be 10 KOhms. Commercial CDPs are quite capable of driving that load. The 150 KOhm 6GK5 grid to ground resistance comports well with the 1:10 driving to driven impedance ratio rule.
Reduce R4 to 10 Ω. 6GK5 IB will be 8 mA. 100 Ω for R4 will upset 6GK5 bias.
R6 and R4 form the gain set pair. The fraction of the net O/P voltage fed back = R4/(R4+R6). The setup is a textbook voltage divider.
"Belt and suspenders" indicates increasing R9 to 390 Ω. A Carbon film part is good here.
Noise and/or accuracy suggest 1% tolerance metal film parts for R1, R4, R5, and R6. Vishay/Dale RN65 is my preference.
CC means Carbon composition. That construction is both non-metallic and non-inductive, which is "ideal" in the grid stopper role.
While the value of R8 remains unknown, its construction will be hefty wire wound. Substantial power will be dissipated here.
Metal oxide resistors, of appropriate voltage rating, are good in the PSU.
An interesting possibility is to replace R3 with a 10M45S constant current source (CCS). That tweak is good for linearity and makes sizing R23 and R24 very easy.
If a NOS Panasonic ECQ-P(U) can be sourced for C2, do so. Otherwise, a 716P series Orange Drop is good.
Reduce R4 to 10 Ω. 6GK5 IB will be 8 mA. 100 Ω for R4 will upset 6GK5 bias.
R6 and R4 form the gain set pair. The fraction of the net O/P voltage fed back = R4/(R4+R6). The setup is a textbook voltage divider.
"Belt and suspenders" indicates increasing R9 to 390 Ω. A Carbon film part is good here.
Noise and/or accuracy suggest 1% tolerance metal film parts for R1, R4, R5, and R6. Vishay/Dale RN65 is my preference.
CC means Carbon composition. That construction is both non-metallic and non-inductive, which is "ideal" in the grid stopper role.
While the value of R8 remains unknown, its construction will be hefty wire wound. Substantial power will be dissipated here.
Metal oxide resistors, of appropriate voltage rating, are good in the PSU.
An interesting possibility is to replace R3 with a 10M45S constant current source (CCS). That tweak is good for linearity and makes sizing R23 and R24 very easy.
If a NOS Panasonic ECQ-P(U) can be sourced for C2, do so. Otherwise, a 716P series Orange Drop is good.
Eli, I've heard you make a similar statement at least once in the past. Regrettably, I fear I am aware of neither the enlightened nor the brute method. Is there a pointer towards a little education?
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.