Design question about 6JC5 beam pentode....
I see that in the Vertical Deflection Amplifier section of its data sheet, the maximum grid leak value is 2.2M ohms when used with cathode bias.
I also see in other similar pentodes' data sheets that happen to have class A amplifier sections, that for class A amp usage, the max grid leak value is about half of what it is in the horizontal or vertical deflection section. Can that be used as a rule of thumb? In other words...
For these beam pentodes, is the max grid leak resistor value for class A audio amp usage about half of the value stated in the vertical deflection usage section?
I see that in the Vertical Deflection Amplifier section of its data sheet, the maximum grid leak value is 2.2M ohms when used with cathode bias.
I also see in other similar pentodes' data sheets that happen to have class A amplifier sections, that for class A amp usage, the max grid leak value is about half of what it is in the horizontal or vertical deflection section. Can that be used as a rule of thumb? In other words...
For these beam pentodes, is the max grid leak resistor value for class A audio amp usage about half of the value stated in the vertical deflection usage section?
Last edited:
Thanks, and understood. I'll keep the value of the grid leak resistors to no higher than 2/3 of the stated maximum.
The 6JC5 max value is stated to be 2.2M with cathode bias, so I'd go no more than 1M, or maybe knock that down to 680k.
In this RCA SP-10 circuit, the previous circuit is the current-starved 6AU6 driver with the 330k plate load resistor. The original circuit specifies 330k for the 6V6 grid leak resistor, making the source : load ratio close to 1 : 1. Hence the desire to increase the value of the output tube's grid leak resistor, to avoid loading down the driver stage too much.
The 6JC5 max value is stated to be 2.2M with cathode bias, so I'd go no more than 1M, or maybe knock that down to 680k.
In this RCA SP-10 circuit, the previous circuit is the current-starved 6AU6 driver with the 330k plate load resistor. The original circuit specifies 330k for the 6V6 grid leak resistor, making the source : load ratio close to 1 : 1. Hence the desire to increase the value of the output tube's grid leak resistor, to avoid loading down the driver stage too much.
Ultimately, I gotten the best out of grid chokes for my PP amps. It does require something more than a .4mA 6AU6 to drive...LOL They're also single-pair output designs. They're not without challenges; get enough L to present a decent load, and avoid the capacitance that degrades their high frequency performance. Present output tubes excluded, there are lots of entertaining outputs that have no where near that high a grid circuit rating.
Douglas
Douglas
I'm trying to decide on the output tube to use. The choice is narrowing down to between 6JC5 and EL86 (actually 6P43P-E).
6JC5 has up to 19W plate dissipation, but I'd use it down around 15W. It also has a very high max value of grid leak R (2.2M !), which should help the 6AU6 drive the outputs. I was going to go with 1M at the most.
EL86/6P43P looks like it can make 20W output from a 5k p-p primary OPT, which I happen to have available, so no purchases necessary. I also have a 500VCT 250mA power transformer available, which should get me 300VDC after rectification. Smoothing will drop that a few volts, but that's OK because EL86 should have only about 250V plate-cathode, which is attainable from a raw 300V B+.
EL86 max grid leak R with cathode bias is 1M, so 680k should work, I think.
Also, I have a hand-me-down aluminum chassis that was punched for PP EL84s, so I can plop the EL86s in with no modification. I will have to punch extra holes for the 6AU6 (or 6CB6) drivers.
Today I'm leaning towards the EL86/6P43P solution, for the convenience, and because it's looking ridiculously good in simulation (super-low THD, H2 dominant, and >15W of power per channel).
6JC5 has up to 19W plate dissipation, but I'd use it down around 15W. It also has a very high max value of grid leak R (2.2M !), which should help the 6AU6 drive the outputs. I was going to go with 1M at the most.
EL86/6P43P looks like it can make 20W output from a 5k p-p primary OPT, which I happen to have available, so no purchases necessary. I also have a 500VCT 250mA power transformer available, which should get me 300VDC after rectification. Smoothing will drop that a few volts, but that's OK because EL86 should have only about 250V plate-cathode, which is attainable from a raw 300V B+.
EL86 max grid leak R with cathode bias is 1M, so 680k should work, I think.
Also, I have a hand-me-down aluminum chassis that was punched for PP EL84s, so I can plop the EL86s in with no modification. I will have to punch extra holes for the 6AU6 (or 6CB6) drivers.
Today I'm leaning towards the EL86/6P43P solution, for the convenience, and because it's looking ridiculously good in simulation (super-low THD, H2 dominant, and >15W of power per channel).
Generalization:
When it comes to the resistance of Rg that might allow thermal runaway . . .
Then with all other things equal, fixed/fixed adjustable bias is more likely to allow thermal runaway.
Self bias is less likely to allow thermal runaway.
When it comes to the resistance of Rg that might allow thermal runaway . . .
Then with all other things equal, fixed/fixed adjustable bias is more likely to allow thermal runaway.
Self bias is less likely to allow thermal runaway.
I'm hoping 6P43P works out well.
I have a pair of the old Dynaco Z565 OPTs, which have a primary that's about 8k plate-to-plate, but only has 8 and 16 ohm secondary taps. Unfortunately, the speakers I use are nominally 6 ohm impedance, but their impedance curve dips down to 4 ohms at several frequencies. So I'm thinking of using the Z565 transformers with the almost 4 ohm speakers connected to the OPT 8 ohm taps. I figure that will change the impedance seen by the output stage to more like 4k ohms (plate-to-plate), which should work for a pair of EL86/6P43P in pentode. I'm hoping this will work out.
EDIT TO ADD: I think the plan to use the 8 ohm tap on a 4 ohm speaker will only work if the primary inductance is large enough to allow reasonable bass frequency output. I'm gambling that the Z565's 100H primary inductance is adequate for this task. If not, I guess I'll have to switch to a different OPT with a 4 ohm secondary tap.
I have a pair of the old Dynaco Z565 OPTs, which have a primary that's about 8k plate-to-plate, but only has 8 and 16 ohm secondary taps. Unfortunately, the speakers I use are nominally 6 ohm impedance, but their impedance curve dips down to 4 ohms at several frequencies. So I'm thinking of using the Z565 transformers with the almost 4 ohm speakers connected to the OPT 8 ohm taps. I figure that will change the impedance seen by the output stage to more like 4k ohms (plate-to-plate), which should work for a pair of EL86/6P43P in pentode. I'm hoping this will work out.
EDIT TO ADD: I think the plan to use the 8 ohm tap on a 4 ohm speaker will only work if the primary inductance is large enough to allow reasonable bass frequency output. I'm gambling that the Z565's 100H primary inductance is adequate for this task. If not, I guess I'll have to switch to a different OPT with a 4 ohm secondary tap.
Last edited:
A simple test that everyone ought to do . . .
Take a DMM and accurately measure the DCR of your loudspeakers.
That gives you an idea of the impedance your amplifier probably will see at some frequency ranges.
For a ported enclosure, those frequencies may typically be 3 frequency bands/ranges.
Below first woofer peak, between woofer peaks, and above the upper woofer peak.
For a closed enclosure. those frequencies may typically be 2 frequency bands/ranges.
Below woofer the woofer peak, and above the woofer peak.
Take a DMM and accurately measure the DCR of your loudspeakers.
That gives you an idea of the impedance your amplifier probably will see at some frequency ranges.
For a ported enclosure, those frequencies may typically be 3 frequency bands/ranges.
Below first woofer peak, between woofer peaks, and above the upper woofer peak.
For a closed enclosure. those frequencies may typically be 2 frequency bands/ranges.
Below woofer the woofer peak, and above the woofer peak.
Are you asking for the DCR of the total speaker system from + to - terminals? Or the DCR of the individual drivers (woofer and tweeter)?
DCR between the + and - speaker system terminals:
JBL Studio 530 = 4.8 ohms
Sonus Faber Concerto = 6.2 ohms
Snell Acoustics E-III = 3.9 ohms
Snell Acoustics C = 7.1 ohms
DCR between the + and - speaker system terminals:
JBL Studio 530 = 4.8 ohms
Sonus Faber Concerto = 6.2 ohms
Snell Acoustics E-III = 3.9 ohms
Snell Acoustics C = 7.1 ohms
Last edited:
rongon,
Thanks for doing that!
As you can see, there might be some low impedances at some frequencies, or range of frequencies.
Even if the minimum impedance is 10% higher than the DCR you measured, none of your speakers would have a minimum impedance that is as large as 8 Ohms.
Even if the number was 30% larger than the DCR, two of them would be less than 8 Ohms minimum impedance.
A lot of speaker systems connect an inductor between the input + terminal and the woofer voice coil.
There may be many other components in the woofer crossover, but the usual DC path is + terminal, inductor, woofer voice coil, - terminal.
This is what causes the DCR that I am talking about.
Thanks for doing that!
As you can see, there might be some low impedances at some frequencies, or range of frequencies.
Even if the minimum impedance is 10% higher than the DCR you measured, none of your speakers would have a minimum impedance that is as large as 8 Ohms.
Even if the number was 30% larger than the DCR, two of them would be less than 8 Ohms minimum impedance.
A lot of speaker systems connect an inductor between the input + terminal and the woofer voice coil.
There may be many other components in the woofer crossover, but the usual DC path is + terminal, inductor, woofer voice coil, - terminal.
This is what causes the DCR that I am talking about.
Impedance is to AC signal, not simply DC resistance. In moving coil speakers, impedance usually varies a lot with frequency.
You can see it in measurements of speaker impedance with frequency. For instance, for the JBL Studio 530 (from Audio Science Review):
The impedance dips down to <5 ohms between around 140Hz to 250Hz, and is well below 8 ohms at 45Hz,180Hz and 13kHz. Contemporary speakers are designed to be driven by amplifiers with essentially zero output impedance, so they almost always have these kinds of very bumpy impedance curves. To compensate, I figure it's best to design tube amps to work into a 4 ohm load, even for speakers with nominal 6 ohm or 8 ohm impedance ratings.
My understanding is that the speaker system DCR will result from a combination of the DC resistance of the any inductors connected in series and the voice coil DC resistances of the speaker drivers. From what I've seen, most woofers rated as 8 ohms impedance will have a voice coil DCR of about 6 to 7 ohms. Then the system DCR can be influenced by any resistances connected in parallel, such as from a Zobel network, low-pass filter, notch filter. Here's the crossover for the JBL Studio 530:
There are no DC resistances going straight from signal to ground, and I don't know how to figure the DCR for the crossover circuit itself. But you can see L1 (2.3mH) connecting from the input + terminal to the + terminal on the midwoofer (LF). L1 is an iron core inductor, so its DCR is probably only about half an ohm.
And what does all this have to do with an amplifier design? Well... I have a pair of very nice Dynaco Z565 OPTs, which do not have a 4 ohm secondary tap (they have 8 ohm and 16 ohm). Can I get these to work with 4 ohm speakers?
My understanding is that the transformer's stepdown ratio remains constant, so halving the load impedance on the secondary halves the load impedance of the primary. I.e., my 8k:8 ohm transformer becomes a 4k:4 ohm transformer when I connect a 4 ohm load to its 8 ohm secondary tap. The primary inductance doesn't change. So perhaps the 100H primary inductance of the Z565 will save the day?
You can see it in measurements of speaker impedance with frequency. For instance, for the JBL Studio 530 (from Audio Science Review):
The impedance dips down to <5 ohms between around 140Hz to 250Hz, and is well below 8 ohms at 45Hz,180Hz and 13kHz. Contemporary speakers are designed to be driven by amplifiers with essentially zero output impedance, so they almost always have these kinds of very bumpy impedance curves. To compensate, I figure it's best to design tube amps to work into a 4 ohm load, even for speakers with nominal 6 ohm or 8 ohm impedance ratings.
My understanding is that the speaker system DCR will result from a combination of the DC resistance of the any inductors connected in series and the voice coil DC resistances of the speaker drivers. From what I've seen, most woofers rated as 8 ohms impedance will have a voice coil DCR of about 6 to 7 ohms. Then the system DCR can be influenced by any resistances connected in parallel, such as from a Zobel network, low-pass filter, notch filter. Here's the crossover for the JBL Studio 530:
There are no DC resistances going straight from signal to ground, and I don't know how to figure the DCR for the crossover circuit itself. But you can see L1 (2.3mH) connecting from the input + terminal to the + terminal on the midwoofer (LF). L1 is an iron core inductor, so its DCR is probably only about half an ohm.
And what does all this have to do with an amplifier design? Well... I have a pair of very nice Dynaco Z565 OPTs, which do not have a 4 ohm secondary tap (they have 8 ohm and 16 ohm). Can I get these to work with 4 ohm speakers?
My understanding is that the transformer's stepdown ratio remains constant, so halving the load impedance on the secondary halves the load impedance of the primary. I.e., my 8k:8 ohm transformer becomes a 4k:4 ohm transformer when I connect a 4 ohm load to its 8 ohm secondary tap. The primary inductance doesn't change. So perhaps the 100H primary inductance of the Z565 will save the day?
Call the primary L constant for now. It is far from a single number, but for this discussion, we should treat it that way. Take that 100 Hy you mention as its L value. In || with an 8kOhm primary load, there is a low frequency behaviour. With a 4kOhm primary load in || with this same 100 Hy, which one will deliver better low frequency performance?
Douglas
Douglas
I would expect the higher 8k impedance across the primary inductance will allow more efficient energy transfer at low frequencies. It will also provide more damping to the output tubes, resulting in lower THD. It will limit output power as the load line on the output tube is made flatter, increasing gain but decreasing the increase in plate current as signal swings the tube down towards 0V bias.
4k ohms in parallel with the primary inductance would shunt more signal (energy), so there would be less output at low frequencies. and the output at low frequencies would be more distorted.
However, the plate resistance of EL86 with feedback around it will be very low. That means more efficient energy transfer from the EL86 output tubes to the speaker load at mid- and high-frequencies, so more power out at 1kHz (for instance).
So it's a very mixed bag. But I would expect bass response to suffer, and THD to increase by reducing the effective primary impedance from 8k down to 4k. However, power output possible at mid-frequencies would increase.
Am I close? Am I wrong?
4k ohms in parallel with the primary inductance would shunt more signal (energy), so there would be less output at low frequencies. and the output at low frequencies would be more distorted.
However, the plate resistance of EL86 with feedback around it will be very low. That means more efficient energy transfer from the EL86 output tubes to the speaker load at mid- and high-frequencies, so more power out at 1kHz (for instance).
So it's a very mixed bag. But I would expect bass response to suffer, and THD to increase by reducing the effective primary impedance from 8k down to 4k. However, power output possible at mid-frequencies would increase.
Am I close? Am I wrong?
??
Bass response will be better with a 4 ohm load across the 8 ohm secondary tap?
Power output will go down?
THD will go down?
Remember, the DCR of the OPT windings do not change.
The turns ratio does not change, therefore the impedance ratio does not change.
The primary inductance doesn't change (much).
The load line will be steeper with the lower primary impedance. Usually that means lower gain and more distortion. But that's when thinking about using a resistive load on the plate, not a transformer.
Well, I'm really missing something. I'll look at the RDH4 tonight. Would there be an explanation of this in Valley & Wallman?
Bass response will be better with a 4 ohm load across the 8 ohm secondary tap?
Power output will go down?
THD will go down?
Remember, the DCR of the OPT windings do not change.
The turns ratio does not change, therefore the impedance ratio does not change.
The primary inductance doesn't change (much).
The load line will be steeper with the lower primary impedance. Usually that means lower gain and more distortion. But that's when thinking about using a resistive load on the plate, not a transformer.
Well, I'm really missing something. I'll look at the RDH4 tonight. Would there be an explanation of this in Valley & Wallman?
Primary ("magnetizing") inductance appears in parallel across the output valves' anodes, as a parasitic load which must (also) be driven. At large signal and low frequencies it contributes to distortion in the output valves. At small signals the inductance and a parallel sum of output valve "plate resistance" and reflected load impedance cause an RL high pass pole.
FWIW, at one time Dynaco said that these had 1000H primary inductance, but without specifying test conditions, so ambiguous. Still..
All good fortune,
Chris
ps: by parallel sum I just mean reciprocal of the sum of reciprocals, in resistor fashion
FWIW, at one time Dynaco said that these had 1000H primary inductance, but without specifying test conditions, so ambiguous. Still..
All good fortune,
Chris
ps: by parallel sum I just mean reciprocal of the sum of reciprocals, in resistor fashion
Last edited:
4 ohms on the 8 ohm tap creates a lower primary Z and yes, a steeper load line as the current is higher. But that means higher power and higher distortion. It's easy to remember that a lower resistance, lower impedance (Z), creates more current. Going the other way, puting 8 ohms on a 4 ohm tap reduces power but also has lower distortion. Doubles the primary Z, lowers the load line angle as the current is lower. So the general rule is don't put a lower ohm speaker than the tap's rating. There is a within reason exception. 6 on 8 wouldn't be a problem in most cases. Power pushers can break anything. And then other cases are when you want to match a tranny to a tube it's primary Z isn't suited for. Not the best way to solve that issue, IMO.
Last edited:
I measured the primary at 500 Henries, using 10V, 20 Hz, but that was the limit of the bridge I used. (a DynaClone from Triode did 700, but with slightly higher leakage inductance) Dropping impedance by half gives another octave in the bass, potentially losing an octave at the high end. 21 mH leakage inductance is 2.6K at 20 KHz, effectively in series with the 4K primary. About 4800 Ohms, +33 degrees. The other downside is resistive loss, which will double as a percentage. 88% efficiency becomes 76%.
As far as distortion, it depends on the transfer function of the tube. Pentodes typically have an impedance "sweet spot" with increasing distortion higher and lower. Push-pull EL86 have a null in the distortion at about 3K and 25W. See GE data sheet.
As far as distortion, it depends on the transfer function of the tube. Pentodes typically have an impedance "sweet spot" with increasing distortion higher and lower. Push-pull EL86 have a null in the distortion at about 3K and 25W. See GE data sheet.