But still something bugs me. They say that the peak Ik should be 57mA when it hits class B. So at full power the cathode voltage would be 0.057x130ohm=7.4V. But the datasheet gives a driving signal of 10Vrms. That means the grid is gonna swing 3.6V positive at some point. That'll shoot the tube straight into saturation or something, I don't know. My tubes are currently working with a 270ohm resistance and I still have an idle current of 37mA, does that make it class AB or purely class B?
57mA is per tube - that mean the cathode voltage is 2 x 0.057 x 130 Ohm= 14.82V.
10V RMS = 10 * SQRT 2 = 14.14V
Ah, so that's the common cathode resistor example. No, man that's the same thing as having 2x270ohm separate resistors, I've tried it, it brings a little more distortion though, somewhat pleasant, probably second order, but that's not what I meant. I've seen circuits with separate resistors of 150ohm.
Secondary taps are worst thing to do. I am curious, if somebody will explain why.
Few companies seems still doing it. Lazy? Cost cutting too much?
Having interlaced primary and sec. windings is key for max performance. (high feedback between windings)
Now look again at sec.tap scheme; when you use only 0-4R then rest is unused, that part of primary will have no (or reduced) magnetic feedback to secondary - which in laymans terms means more "parasitic inductance" on primary.
What does more parasitic inductance to freq. response, i think people know already. ):
You can see on google, some freq. response of hammond, etc..; how each sec.tap changes amp response.
--using thinner and more pieces of wire in paralell, is better than single thicker wire. (allows more interleaving) See Vanderveen http://www.amplimo.nl/images/downloads/ds vdv/vdv3050se.pdf He found a way with multiple secondaries in paralell, to overcome this problem.
--using separate wires is better than taps, look at Sowther or Lundahl SINGLE ENDED OUTPUT TRANSFORMERS
Now look again at sec.tap scheme; when you use only 0-4R then rest is unused, that part of primary will have no (or reduced) magnetic feedback to secondary - which in laymans terms means more "parasitic inductance" on primary.
What does more parasitic inductance to freq. response, i think people know already. ):
You can see on google, some freq. response of hammond, etc..; how each sec.tap changes amp response.
--using thinner and more pieces of wire in paralell, is better than single thicker wire. (allows more interleaving) See Vanderveen http://www.amplimo.nl/images/downloads/ds vdv/vdv3050se.pdf He found a way with multiple secondaries in paralell, to overcome this problem.
--using separate wires is better than taps, look at Sowther or Lundahl SINGLE ENDED OUTPUT TRANSFORMERS
Last edited:
> They say that the peak Ik should be 57mA when it hits class B.
No, the _DC_ Ik per tube is 57mA, or 114mA total Ik.
You can "NOT" run cathode resistor bias for "class B" because of the large change of current from 8mA to 57mA. For class B you need a negative 15V bias supply.
> 0.057x130ohm=7.4V
_TWO_ times 0.057A, 114mA, in 130 Ohms. 14.8V bias.
> driving signal of 10Vrms
10Vrms is 14.14V peak. About a half-volt -negative- of the dreaded positive grid range.
> That'll shoot the tube straight into saturation or something
There's no big BANG when you drive a grid positive. A negative grid is a near infinite load, say 100Megs. Shunted by the 100K-500K grid resistor, still a light load. When the grid goes positive its impedance drops way down, around 1K.
If you drive with a big strong source, you "can" pump big current into the grid, and melt it. However other tubes with beefier grids are *made* to be driven positive grid for more output (and much more grid drive current).
In 99% of small audio amplifiers, the grid is driven with an R-C network. If you try to force grid positive, the high grid current loads-down the source, and ultimately charges-down that capacitor. The average DC grid voltage goes negative. Possibly to cut-off. No signal passes. "Grid blocking". It will recover at the R-C time rate. In Hi-Fi this is distressing. At least one Famous Musician's trademark is pushing his power bottles right to the brink of cut-off for an exaggerated tone.
> My tubes are currently working with a 270ohm resistance and I still have an idle current of 37mA, does that make it class AB or purely class B?
Two EL84? At 250V-300V supply? 8K-10K load?
That sounds WAY COLD.
Philips EL84 datasheet
Back in 1953 some junior engineer in a white shirt with pocket-protector *sweated* over a hot breadboard to get you these numbers. His goal was to give you GOOD working conditions, with minimum effort on your part, so you would buy a boat-load of EL84 and Philips would pay his salary.
Separate cathode resistors, without bypassing, give better DC balance but much lower gain, higher drive. While the EL84 may distort less, the driver must work harder and distort more.
No, the _DC_ Ik per tube is 57mA, or 114mA total Ik.
You can "NOT" run cathode resistor bias for "class B" because of the large change of current from 8mA to 57mA. For class B you need a negative 15V bias supply.
> 0.057x130ohm=7.4V
_TWO_ times 0.057A, 114mA, in 130 Ohms. 14.8V bias.
> driving signal of 10Vrms
10Vrms is 14.14V peak. About a half-volt -negative- of the dreaded positive grid range.
> That'll shoot the tube straight into saturation or something
There's no big BANG when you drive a grid positive. A negative grid is a near infinite load, say 100Megs. Shunted by the 100K-500K grid resistor, still a light load. When the grid goes positive its impedance drops way down, around 1K.
If you drive with a big strong source, you "can" pump big current into the grid, and melt it. However other tubes with beefier grids are *made* to be driven positive grid for more output (and much more grid drive current).
In 99% of small audio amplifiers, the grid is driven with an R-C network. If you try to force grid positive, the high grid current loads-down the source, and ultimately charges-down that capacitor. The average DC grid voltage goes negative. Possibly to cut-off. No signal passes. "Grid blocking". It will recover at the R-C time rate. In Hi-Fi this is distressing. At least one Famous Musician's trademark is pushing his power bottles right to the brink of cut-off for an exaggerated tone.
> My tubes are currently working with a 270ohm resistance and I still have an idle current of 37mA, does that make it class AB or purely class B?
Two EL84? At 250V-300V supply? 8K-10K load?
That sounds WAY COLD.
Philips EL84 datasheet
Back in 1953 some junior engineer in a white shirt with pocket-protector *sweated* over a hot breadboard to get you these numbers. His goal was to give you GOOD working conditions, with minimum effort on your part, so you would buy a boat-load of EL84 and Philips would pay his salary.
Separate cathode resistors, without bypassing, give better DC balance but much lower gain, higher drive. While the EL84 may distort less, the driver must work harder and distort more.
An externally hosted image should be here but it was not working when we last tested it.
okay this is the final circuit I managed to concoct, I took a completely different approach on almost everything, added a few local feedbacks here and there, how does it seem now. And yeah I eventually got that the EL84 is given 130ohms in the datasheet as a common resistor, anyway, my tubes are currently working at 10.2V cathode voltage with 270ohm resistor each, what do you mean cold biased, they're way into class AB1 region, having an idle power dissipation at around 11W?
Hi
Just some comments on this new design :
1 - the PP stage is in Ultra-linear mode, available power will be a little bit less than 17W
2 - feedback on the first stage (C2-R3) : why ? The EF80 already works under feedback with the global feedback loop coming back on its cathode. And the level of feedback will depend on the source impedance. Furthermore, R2 = 1 M will introduce a low pass filter with parasitics capacitances around
3 - global feedback: R29 = 220 K should be a mistake ? Too high value, it should be in the range of 2.2 K
IMO, a local feedback with the UL mode and a global feedback (secondary -> EF80 cathode) should be satisfactory.
Jacques
Just some comments on this new design :
1 - the PP stage is in Ultra-linear mode, available power will be a little bit less than 17W
2 - feedback on the first stage (C2-R3) : why ? The EF80 already works under feedback with the global feedback loop coming back on its cathode. And the level of feedback will depend on the source impedance. Furthermore, R2 = 1 M will introduce a low pass filter with parasitics capacitances around
3 - global feedback: R29 = 220 K should be a mistake ? Too high value, it should be in the range of 2.2 K
IMO, a local feedback with the UL mode and a global feedback (secondary -> EF80 cathode) should be satisfactory.
Jacques
The feedback on the first stage I found useful, to keep its gain low, I only need it to have a gain of 4-5. It's got little to do with distortion, since I chose a pretty linear bias point for the EF80. The whole capacitance deal, well I actually consider this an advantage. According to the datasheet, EF80 has an input capacitance of 7.5pF, so worst case scenario the -3dB point is at 21kHz (with 1Mohm), so it'll minimise the possibility of oscillations right from the first stage. The other local feedback is the UL itself, I think with 20% taps it should give around 15W. The global feedback is adjustable, it'll be easier to test and find an appropriate permanent value like this. I'll swap it later with a fixed resistor, once I determine the precise value I need. Thanks anyway, I didn't think of the input capacitance thing when I designed it. Could've turned out bad, I'll pay more attention to that the next time.
thanks for your answer
If you want to keep first stage gain low better to use a triode with low mu (12BH7, 12AU7, 6SN7,...) and to remove the local feedback (C2-R3). And linearity of the first stage should be better.
And furthermore, 2 feedbacks on the first stage (local + global) will make difficult to find the correct response stability because they are dependant on each other (they both act on the gain of the first stage).
Jacques
If you want to keep first stage gain low better to use a triode with low mu (12BH7, 12AU7, 6SN7,...) and to remove the local feedback (C2-R3). And linearity of the first stage should be better.
And furthermore, 2 feedbacks on the first stage (local + global) will make difficult to find the correct response stability because they are dependant on each other (they both act on the gain of the first stage).
Jacques
I'll think of something, thanks. I'll give this a go first, because up to now I've only used triodes as preamps, and I wanna give pentodes a go, see If I like how they sound, and then I'll decide what suits me best. Alternatively I can try a triode cascade preamp. It has pentode characteristics without the inherent hiss of a pentode.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.