Novanta,
The input and phase splitter will likely only draw a few mA each. The EL34s will likely be biased between 40 to 60mA each. So a stereo amp could draw as much as 250mA from the power supply. But you will want more than 250vA on your transformer B+ winding. Usually 20% to 40% more is recommended. Even with bias set at 50mA per output tube a 250vA rating on the power transformer is a bit too low. If you are building two mono amps, 250vA is more than enough.
Regards, S.
The input and phase splitter will likely only draw a few mA each. The EL34s will likely be biased between 40 to 60mA each. So a stereo amp could draw as much as 250mA from the power supply. But you will want more than 250vA on your transformer B+ winding. Usually 20% to 40% more is recommended. Even with bias set at 50mA per output tube a 250vA rating on the power transformer is a bit too low. If you are building two mono amps, 250vA is more than enough.
Regards, S.
GFgen,
This thread has hundreds of postings.
. . . Please let me know the Post number of the complete and accurate schematic, that you want to modify to use a CCS.
If some or all of the voltages, resistances, and currents are given for the phase splitter that has a long tailed resistor that needs to be replaced with a CCS, that helps me to selecting a solid state CCS chip.
In some cases, a negative voltage supply can be eliminated, in other cases the negative voltage supply can be set to less voltage.
In some cases, a negative voltage supply may have to be added, in order to keep the CCS chip from ever being at less than its Burden Voltage.
Selection criteria include the following CCS chip's characteristics:
Minimum operating voltage (Burden Voltage; maximum negative signal voltage peaks must not let the CCS have less than its Burden Voltage)
Minimum operating current.
Maximum operating current.
Maximum operating voltage.
Maximum dissipation.
Impedance of the CCS
With a pair of triodes, that do not have any grid current leakage, then with the cathodes connected together, whatever the signal current is in one triode, the other triode has the exact same signal current, only its phase is changed, not its amplitude.
The exception, is the signal current in the long tailed resistor or the CCS.
The higher the impedance of the CCS, the less current is stolen from the two cathodes, so the current match is near perfect.
The very high CCS impedance is the basis of matched plate currents.
Matched currents x matched plate load resistors = matched plate voltages (of opposite phase).
Who needs a dirty, mechanically jumpy, and intermittent wiper of a potentiometer?
Is 0.1 percent matched resistors, and 0.1% matched plate signal voltages good enough for most people?
There can be better resistor matching, but we are not designing a new Lamborghini engine that can go to 15,000 rpm, Right?
$0.03
This thread has hundreds of postings.
. . . Please let me know the Post number of the complete and accurate schematic, that you want to modify to use a CCS.
If some or all of the voltages, resistances, and currents are given for the phase splitter that has a long tailed resistor that needs to be replaced with a CCS, that helps me to selecting a solid state CCS chip.
In some cases, a negative voltage supply can be eliminated, in other cases the negative voltage supply can be set to less voltage.
In some cases, a negative voltage supply may have to be added, in order to keep the CCS chip from ever being at less than its Burden Voltage.
Selection criteria include the following CCS chip's characteristics:
Minimum operating voltage (Burden Voltage; maximum negative signal voltage peaks must not let the CCS have less than its Burden Voltage)
Minimum operating current.
Maximum operating current.
Maximum operating voltage.
Maximum dissipation.
Impedance of the CCS
With a pair of triodes, that do not have any grid current leakage, then with the cathodes connected together, whatever the signal current is in one triode, the other triode has the exact same signal current, only its phase is changed, not its amplitude.
The exception, is the signal current in the long tailed resistor or the CCS.
The higher the impedance of the CCS, the less current is stolen from the two cathodes, so the current match is near perfect.
The very high CCS impedance is the basis of matched plate currents.
Matched currents x matched plate load resistors = matched plate voltages (of opposite phase).
Who needs a dirty, mechanically jumpy, and intermittent wiper of a potentiometer?
Is 0.1 percent matched resistors, and 0.1% matched plate signal voltages good enough for most people?
There can be better resistor matching, but we are not designing a new Lamborghini engine that can go to 15,000 rpm, Right?
$0.03
Hi 6A3sUMMER,
The Schematic for the amp is in post #217.
I would like to eliminate the potentiometers used in this circuit. I think it is possible to use matched EL34 tubes; individual self bias 'resistors and individual bypass capacitors' in each EL34 cathode, only I don't know how to calculate for those values of resistors and capacitors. perhaps a ccs could be used here too?
A ccs in the phase splitter would be great, I think 0.1% would be more than good enough for this amp.
Output stage uses a matched quad of EL34's with bias set to 50mA. phase splitter has around 1.2 mA.
I would be most grateful for any help you can give with this! Thanks!
The Schematic for the amp is in post #217.
I would like to eliminate the potentiometers used in this circuit. I think it is possible to use matched EL34 tubes; individual self bias 'resistors and individual bypass capacitors' in each EL34 cathode, only I don't know how to calculate for those values of resistors and capacitors. perhaps a ccs could be used here too?
A ccs in the phase splitter would be great, I think 0.1% would be more than good enough for this amp.
Output stage uses a matched quad of EL34's with bias set to 50mA. phase splitter has around 1.2 mA.
I would be most grateful for any help you can give with this! Thanks!
I would definitely replace R13 with a CCS; plenty of headroom for it to work with at 87V. Immediately, that allows you to remove P1 and hand match R12/R14. Big obvious improvement if you are not opposed to semis.
If you have matched EL34's, I would not have a big problem removing P3/R17/R18/C12/C13. Not sure I would remove P2, though. It is handy to have as tubes age, and does not negatively affect the sound. I might move C11 to the wiper of P2, but 470uF is a bit large; kind of depends on the ripple at -55V.
C4 and R8 get adjusted based on the complete amp; I would not guarantee those values. A good starting point is R4=10*R8, then adjust C4 for nice square wave.
If you have matched EL34's, I would not have a big problem removing P3/R17/R18/C12/C13. Not sure I would remove P2, though. It is handy to have as tubes age, and does not negatively affect the sound. I might move C11 to the wiper of P2, but 470uF is a bit large; kind of depends on the ripple at -55V.
C4 and R8 get adjusted based on the complete amp; I would not guarantee those values. A good starting point is R4=10*R8, then adjust C4 for nice square wave.
Thanks for your reply, do you have a link to a ccs I could use in the phase splitter?
I will remove those components you suggest, and adjust R8 and C4 as per your instructions. Thanks!
I will remove those components you suggest, and adjust R8 and C4 as per your instructions. Thanks!
The phase splitter RL are 150k + 25k = 175k (P1 centered). The triodes will not have the same plate voltage unless each plate RL is 175k.
420V - 320V = 100V. 100V / 175k Ohms = 0.57mA.
And using 150k instead. of 175k, will reduce the gain by about 14%.
Set CCS to 1.14mA.
I had no idea of this circuit until I saw the schematic, but the phase splitter triodes are starved for current (even for an ECC83), in order to get high resistance RL, so there will be more gain.
That is ironic, because an LTP phase splitter's maximum gain is always Less than 1/2 of mu (u).
I am not sure what CCS can take 87V + peak signal swing if V1 plate.
And, it needs to operate at 1.14 mA.
Your estimate of the current was very close.
Solid State Experts . . .
. . . Does anybody know of a CCS that will work at the high voltage and low current for that spot?
Conversion to individual self bias for each tube:
You will get less output power, you are reducing the plate to cathode voltage by the amount of the self bias voltage
To have each EL34 tube at 50mA, then Individual self bias resistors need to be calculated according to Bias Volts / 0.050A.
Those individual self bias resistors will replace R24 and R25. They will need to have individual bypass capacitors across them.
I will estimate the voltage loss from 440V B+ to the EL34 plate to be 5V, and to the EL34 screen to be 2 Volts.
We need to find a graph of EL34 plate curves that has about 400 or 450V on the plate and on the screen, in order to estimate the bias voltage that will give us 50mA plate current.
From one set of curves, I found that 400V on plate and 400V on screen, to bias for 50mA plate current we need about -36V bias.
36V / 0.050A = 720 Ohms.
The EL34 cathode impedance is very approximately 90 Ohms.
200uF will give us about -1 dB at 20Hz, you might choose to use more capacitance than that. I have used 330uF and 1000uF to bypass capacitors.
With 36V self bias, the plate to cathode voltage of approximately 415V will be reduced 36V, so about 379V.
Because the screen to cathode voltage is also reduced by 36V, the bias required will be slightly less than 36V.
Start by using 680 Ohm self bias resistors instead of 750 Ohm resistors, you will probably be very close to 50mA plate current.
The EL34 tubes are not going to burn out with 680 Ohm resistors, you have lots of time to do measurements, and verify the current.
The plate and screen current are the total current in the 680 Ohm resistors.
Measuring the output transformer DCR from plate(s) to center tap, allows you to accurately calculate the current of each plate. Usually those DCRs are similar, but not equal.
Voltage (center tap to plate) / DCR = plate current.
Removal of parts:
Connect the non-grid ends of R16 and R19 directly to Ground.
Then disconnect all the other parts of the fixed adjustable bias circuit.
If you are starting from scratch, you have lots less parts to wire in.
Example, for the top EL34, you only keep C9, R20, R16, R26, and the 720 ohm resistor that replaces R24.
Repeat that set of parts for the bottom EL34.
Less parts in the grid circuit, no variable bias parts, and no -57V bias supply parts.
Using individual self bias; the same tube manufacturer, same vintage, matched EL34 tubes will eliminate the need to balance the output tube plate currents.
Other parts:
C4 "Lag" capacitor, and R8 are the open loop high frequency dominant pole, to aid the stability for the specific output transformer that was originally used.
C5, "lead" capacitor, and R7 are the global negative feedback parts that finish the job of making the amplifier stable for the specific output transformer that was originally used.
C5 (and perhaps C4, hopefully not) may need to be adjusted for the output transformers that you use.
Have fun building/modifying, testing, and listening!
$0.03
420V - 320V = 100V. 100V / 175k Ohms = 0.57mA.
And using 150k instead. of 175k, will reduce the gain by about 14%.
Set CCS to 1.14mA.
I had no idea of this circuit until I saw the schematic, but the phase splitter triodes are starved for current (even for an ECC83), in order to get high resistance RL, so there will be more gain.
That is ironic, because an LTP phase splitter's maximum gain is always Less than 1/2 of mu (u).
I am not sure what CCS can take 87V + peak signal swing if V1 plate.
And, it needs to operate at 1.14 mA.
Your estimate of the current was very close.
Solid State Experts . . .
. . . Does anybody know of a CCS that will work at the high voltage and low current for that spot?
Conversion to individual self bias for each tube:
You will get less output power, you are reducing the plate to cathode voltage by the amount of the self bias voltage
To have each EL34 tube at 50mA, then Individual self bias resistors need to be calculated according to Bias Volts / 0.050A.
Those individual self bias resistors will replace R24 and R25. They will need to have individual bypass capacitors across them.
I will estimate the voltage loss from 440V B+ to the EL34 plate to be 5V, and to the EL34 screen to be 2 Volts.
We need to find a graph of EL34 plate curves that has about 400 or 450V on the plate and on the screen, in order to estimate the bias voltage that will give us 50mA plate current.
From one set of curves, I found that 400V on plate and 400V on screen, to bias for 50mA plate current we need about -36V bias.
36V / 0.050A = 720 Ohms.
The EL34 cathode impedance is very approximately 90 Ohms.
200uF will give us about -1 dB at 20Hz, you might choose to use more capacitance than that. I have used 330uF and 1000uF to bypass capacitors.
With 36V self bias, the plate to cathode voltage of approximately 415V will be reduced 36V, so about 379V.
Because the screen to cathode voltage is also reduced by 36V, the bias required will be slightly less than 36V.
Start by using 680 Ohm self bias resistors instead of 750 Ohm resistors, you will probably be very close to 50mA plate current.
The EL34 tubes are not going to burn out with 680 Ohm resistors, you have lots of time to do measurements, and verify the current.
The plate and screen current are the total current in the 680 Ohm resistors.
Measuring the output transformer DCR from plate(s) to center tap, allows you to accurately calculate the current of each plate. Usually those DCRs are similar, but not equal.
Voltage (center tap to plate) / DCR = plate current.
Removal of parts:
Connect the non-grid ends of R16 and R19 directly to Ground.
Then disconnect all the other parts of the fixed adjustable bias circuit.
If you are starting from scratch, you have lots less parts to wire in.
Example, for the top EL34, you only keep C9, R20, R16, R26, and the 720 ohm resistor that replaces R24.
Repeat that set of parts for the bottom EL34.
Less parts in the grid circuit, no variable bias parts, and no -57V bias supply parts.
Using individual self bias; the same tube manufacturer, same vintage, matched EL34 tubes will eliminate the need to balance the output tube plate currents.
Other parts:
C4 "Lag" capacitor, and R8 are the open loop high frequency dominant pole, to aid the stability for the specific output transformer that was originally used.
C5, "lead" capacitor, and R7 are the global negative feedback parts that finish the job of making the amplifier stable for the specific output transformer that was originally used.
C5 (and perhaps C4, hopefully not) may need to be adjusted for the output transformers that you use.
Have fun building/modifying, testing, and listening!
$0.03
Last edited:
I am a fan of the typical cascode DN2540/10M90S design that has been beat to death on this site. If your currents are really small, something bipolar may be a more optimal choice. Plenty of those designs to be found here as well.
Thanks for that very comprehensive reply. lots to think about here.
I will start by removing the parts in the fixed bias circuit, whilst i try to find a ccs that will work for the phase splitter. though to be honest I'm not really sure what i need, i am pretty much a novice when it comes to this stuff. but i am eager to learn! but you guys have been a great help so far.
There are CCS that use 2 bipolar transistors, an LED, and a few resistors. They can take the voltage, and should be able to run at that low of a current.
Bottlehead.com had current sources that were used as plate loads, but I think they can be adjusted to work at about 1mA for cathode circuits.
Bottlehead fans, please check this out For GFgen.
Perhaps the original Bottlehead CCS circuit that is obsolete, that can be copied and built.
Bottlehead.com had current sources that were used as plate loads, but I think they can be adjusted to work at about 1mA for cathode circuits.
Bottlehead fans, please check this out For GFgen.
Perhaps the original Bottlehead CCS circuit that is obsolete, that can be copied and built.
I found this circuit for which boards are available. Do you think this circuit would work for my situation?
IMO you don't need a CCS. The existing 82k tail resistor is already large; large enough that the phase inverter balance will be excellent even without the trimpot, unless you happen to plug in a really unbalanced ECC83 which is very rare.
Hi Merlinb,
so are you saying to simply remove P1 and just keep the 2 plate resistors? doesn't R14 need to be of a lower value than R12 for the phase splitter to function properly?
so are you saying to simply remove P1 and just keep the 2 plate resistors? doesn't R14 need to be of a lower value than R12 for the phase splitter to function properly?
Yes, it wasn't needed in the original 5-20 and it isn't needed now.
Only when the tail resistor is rather small, which isn't the case here. Mullard knew what they were doing.
That is very interesting! I shall try the amp without the trim pot it will be interesting to see if that makes a difference.
Yes because there's 87V volts across the 82k then the AC variation is small and the LTP balance good. No need for a CCS in this situation. You tend to need the CCS when there is only a small voltage across the cathode resistor such as when the grids and biased to ground. In your case the grid bias is provided by the first stage.
If you really need a CCS then avoid the JFET or MOSFET ones as they have a big variation of device current depending on the sample used - hence they need a trim pot.
If you really need a CCS then avoid the JFET or MOSFET ones as they have a big variation of device current depending on the sample used - hence they need a trim pot.
If I remove the trim pot and leave the cathode resistor as is. then remove the fixed bias and use auto bias resistors and bypass caps in the output stage. then effectively i have reverted the design back tp the original Mullard circuit.
Therefore I will have undone all the upgrades to the circuit that Claus Byrith did. Were these upgrades not beneficial? was the original Mullard design superior? and should i rebuild the amp as per the Mullard circuit?
Therefore I will have undone all the upgrades to the circuit that Claus Byrith did. Were these upgrades not beneficial? was the original Mullard design superior? and should i rebuild the amp as per the Mullard circuit?
