I just went thru the Edcor power transformers again, and the one we've got still seems like the best choice. If we want a convenient bias tap, the choices get very limited quickly. The virtual CT can be made with a pair of 100R's tied together per SY's suggestion, and the CT can be grounded or lifted to 60V....or we can do a separate 12V transformer, which would also allow a simple human powered startup/warmup delay if desired.
One other thought....for a $20 one-time fee, Edcor will create a new P/N and make a custom, so perhaps we could entertain losing the 5V tap and adding a CT to the 6.3V winding. Either way, the lead time for Edcor transformers is usually 3-4 weeks.
One other thought....for a $20 one-time fee, Edcor will create a new P/N and make a custom, so perhaps we could entertain losing the 5V tap and adding a CT to the 6.3V winding. Either way, the lead time for Edcor transformers is usually 3-4 weeks.
Last edited:
Hammond 166N12=12.6V CT @ 4A
Hammond 166Q12=12.6V CT @ 6A
http://www.radiodaze.com/hammond05.htm
Less 60hz coupling, and allows for human powered (or relay) delay warm up if desired.
Last edited:
I think SY is right. A separate transformer simplifies the parts purchasing for future hobbyists that want to build this. And it solves a few other minor problems.
We could always use the power xformer's low voltage windings to run a large matrix of red LEDs as homage to SY's contribution.
..Todd
We could always use the power xformer's low voltage windings to run a large matrix of red LEDs as homage to SY's contribution.
..Todd
Sy, wonder about the omit of input the cap, that's very good idea indeed, however I wonder how much offset can such a DC coupled input stage tolerate, I guess it's not really the same deal as with SS as we have an OPT preventing the speakers from seeing any DC. Further on the LTP wouldn't be very much affected as it is limited by the CCS, and oh, the KT88 do already have a decoupling cap on g1, so besides a bit offset bias working point I guess it's not any catastrophic..?
To take it one step further I'm thinking whether the input tube could be connected in such way it would have a better 'DC auto-biasing' which could null out the input DC offset if so just slightly that would be a nice feature without sacrificing too much on other qualities, we have 3 grids to play with.
Cheers Michael
To take it one step further I'm thinking whether the input tube could be connected in such way it would have a better 'DC auto-biasing' which could null out the input DC offset if so just slightly that would be a nice feature without sacrificing too much on other qualities, we have 3 grids to play with.
Cheers Michael
Last edited:
... then the voltages given do not work out.
Fine, I realise this is only preliminary, but somewhere things must begin to get together. With the latest schematic (I hope; one must move fast here to keep up - different time zones and all that!), a family of EL86 graphs (staying with EF86 for now) would yield the following voltages, starting with VB+ = 460V as given: VB+/2 = 330V; Va = 170V; Vg2 = 195V; Vc = 4,2V. These voltages are unnecessarily high, limiting the headroom of the phase inverter. If I might contribute to this exercise with the following suggestions:
Firstly, in a low output voltage amplifier stage, getting Vg2 to lower than Va would cut down on the partition noise (the extra noise generated by pentodes having a 2nd grid in the electron stream). My suggestion would be for the following values instead: R3 = 68K (a more general value, although not an important difference from 75K); R13 = 100K; R14 = 560K; R15 = 470 ohm. R16 can be eliminated (see later).
With these components the relevant voltages then become, with VB+ = 460V and still an EF86: VB+/2 = 300V; Va = 110V; Vg2 = 84V; Vc = 1,1V. With cathode and g2 both bypassed the stage gain is about 240, decreasing to about 130 with unbypassed R15. It is then unnecessary to use R16, applying NFB over R15 (R17 going to the EF86 cathode). If maximum stage gain is desired, R16 could go down to 10 ohm or such with appropriate values for R15 and R17 to keep the tube voltages as suggested. (The above figures have been checked by a quick hook-up.)
The open-loop gain for this arrangement used with an 12AT7 is quite high, being about -3dB around 50k - 70kHz depending on other constants, allowing easy proportioning of R12/C11 for stability.
I hope these suggestions will be considered useful.
Correction: Last paragraph should of course read: "The open-loop bandwidth" ... apology!
Last edited:
Given "normal" values of offset (<10mV), I can't see this as a problem at all, other than potential thumps or pops with power on/off or connecting/disconnecting. Reality is that nearly all sources have caps in their output, making it a moot point. The input cap is useful if there's a volume control- wipers and DC do not get along.
Thank you Johan! This is the kind of input/feedback I've been hoping for. I'm only a writer, artist, librarian at this point and need to be told EVERYTHING, if it wants to end up on the schematic. Without specifics like that, the schematic cannot change. I'll update the schematic as best I can later on.
I'm not sure I'm able to draw the biased heater transformer circuit without more guidance (a napkin drawing).
I'm not going to tackle the 6GK5 schematic until this version gets approval from everyone (and hopefully someone has built and tuned it for minimal smoke).
Thanks again.
..Todd
I'm not sure I'm able to draw the biased heater transformer circuit without more guidance (a napkin drawing).
I'm not going to tackle the 6GK5 schematic until this version gets approval from everyone (and hopefully someone has built and tuned it for minimal smoke).
Thanks again.
..Todd
Last edited:
Taj: napkin drawing attached for basic heater biasing voltage divider. The values in my earlier post were based on tapping off of the 240V B+.....If that changes, new R values will be needed to achieve a +60V or so heater reference. 60V isn't critical, but a rough target.
This works with any filament voltage obviously........at this point I don't know if we are doing 6.3V or 12.6V. I did not include values here in case we need to adjust B+ per Johan's post.
Last edited:
How about reducing even more resistors..
Nope, that increases common mode noise considerably.
Oh my, I have made a blunder... I was thinking DC from the filament XT, what a brain f**t!
EDIT: I guess I had DC on my mind all the time as I have decided to go DC for two reasons, one is no mains hum at all and the other is I am going to use a regulator with a small cermet trimmer so I can run the first burn in hours on nominal filament voltage to get rid of residual gasses inside the tube, once that is done I would trim down the DC Voltage by 5% to prolong tube life. But I guess as a best compromise for the group an AC supply will do also.
Are you considering some kind of soft start? Using a separate filament transformer makes it a bit easier as it can be allowed to heat up immediately the tube filaments while the main transformer could have a resistor in series on the primary which could be shorted with a relay after a certain time so B+ is raising slowly.
Cheers Michael
EDIT: I guess I had DC on my mind all the time as I have decided to go DC for two reasons, one is no mains hum at all and the other is I am going to use a regulator with a small cermet trimmer so I can run the first burn in hours on nominal filament voltage to get rid of residual gasses inside the tube, once that is done I would trim down the DC Voltage by 5% to prolong tube life. But I guess as a best compromise for the group an AC supply will do also.
Are you considering some kind of soft start? Using a separate filament transformer makes it a bit easier as it can be allowed to heat up immediately the tube filaments while the main transformer could have a resistor in series on the primary which could be shorted with a relay after a certain time so B+ is raising slowly.
Cheers Michael
Last edited:
I must say i'm just thrilled with the direction of this thread. Lots of group knowledge on amp design being passed down here, and a finished product designed and tweaked by so many experts. This should be quite the amp when done. Could it even be a CJ, or ARC killer? Eli once stated that a well done Mullard LPT certianly could be.
Last edited: