Perhaps it's made in the middle of the week. 🙂
Monday: recovery from alcohol poisoning.
Friday: preparation for weekend drinking.
I bought (in the seventies) the new car (russian Lada). I took it to the service center because something rattled in the front door. There was a half empty bottle of russian vodka. 😉
Last edited:
EML300B publishes actual measured curves, so you can see for yourself.
source page
The sound from these is superb; even at $800, they are better value than lower quality, or used DHTs - IMHO.
7W with 5k load is unreal.
With 1k25 ALMOST possible the 6.5-7W (depends of OPT loss), but with enormous distortion.
Sadly EML -if I know correctly- only available via Jacmusic ... and his reputation highly variable. 😡
I had no trouble, though Jac does not issue much in the way of after-sales communication. Many buyers here on diyAudio have not been happy with that, for sure.
But EMLs are also available at tubesusa.com or in EU: esoteric-hifi.com (Warszawa), and at a number of other distis:
EML distributors:
But EMLs are also available at tubesusa.com or in EU: esoteric-hifi.com (Warszawa), and at a number of other distis:
EML distributors:
Distortion? Noise? I’d take a datasheet with a pinch of salt until it’s in a design and measured on an analyser 🙂
Suppose you have a single ended amplifier that is working perfectly, and satisfies you, but you want 2 x the power output.
One method:
Provide the same B+ voltage at 2 x the current.
Replace the original output transformer with one that has 2 x the laminations, 1/2 of the primary impedance, 1/2 the primary DCR, 1/2 the secondary DCR, the same frequency response, and 2 x the DC current capability.
Make sure the drivertube circuit can drive 2 x the capacitive current of a single output tube (or modify it so it can).
Parallel the tubes, but use separate biasing, set each of them at the current of the single tube you started with (total = 2 x the original current).
Use 2 tubes that are very well matched.
Follow the other rules, especially things like grid stoppers for parallel tubes.
Do not cheat on any of the above. You should get the same sound, distortion, etc. at 2 x the output power.
Build, Test, Listen.
Enjoy!
One method:
Provide the same B+ voltage at 2 x the current.
Replace the original output transformer with one that has 2 x the laminations, 1/2 of the primary impedance, 1/2 the primary DCR, 1/2 the secondary DCR, the same frequency response, and 2 x the DC current capability.
Make sure the drivertube circuit can drive 2 x the capacitive current of a single output tube (or modify it so it can).
Parallel the tubes, but use separate biasing, set each of them at the current of the single tube you started with (total = 2 x the original current).
Use 2 tubes that are very well matched.
Follow the other rules, especially things like grid stoppers for parallel tubes.
Do not cheat on any of the above. You should get the same sound, distortion, etc. at 2 x the output power.
Build, Test, Listen.
Enjoy!
Last edited:
@klauss00,
if you really want some power from 2A3 PSE without going into grid current (class A2) and without spending a fortune in tubes I suggest a tried and tested (by myself) solution: 6C4C PSE (equivalent to 6B4G but better specs) with Lundahl LL1627/90 mA and 2.3K plate load. I had this amp for a couple of years and then sold it. It was very good. I still have one more pair of LL1627. Maybe I build another one...
The 6C4C is built like a tank, in fact max specified anode voltage is 360V vs 300V of all other 2A3/6A3/6B4G types. You can run it at 18W plate dissipation without problems and it will last at least 6000-7000 hours, possibly 10000....
So, the 6C4C PSE with 2.3K plate load will give you 10W if run the tubes at 360V/50mA each. The total DC current is 100mA. The Lundahl is specified for 90 mA but it will work just fine. There will be a bit more DC induction and this means that the power rating at 30Hz goes from 13W down to 10W (which is the power of this amp). Below 30Hz distortion will increase because the core will get closer and closer to saturation but most SE amps do not have this performance at low frequency.
You will need VERY CLEAN 145V peak-to-peak to drive the output tubes.
The distortion of this PSE amp at ANY power level will always be much less than a 2A3 SE. When the typical 2A3 amp is about to clip at 3W output, the 6C4C PSE will have about 1%. I also measured 0.3% THD at 1W.
It's not difficult to get matched 6C4C's and they can still be found at around $40-$45 each. If you are lucky, even less....
if you really want some power from 2A3 PSE without going into grid current (class A2) and without spending a fortune in tubes I suggest a tried and tested (by myself) solution: 6C4C PSE (equivalent to 6B4G but better specs) with Lundahl LL1627/90 mA and 2.3K plate load. I had this amp for a couple of years and then sold it. It was very good. I still have one more pair of LL1627. Maybe I build another one...
The 6C4C is built like a tank, in fact max specified anode voltage is 360V vs 300V of all other 2A3/6A3/6B4G types. You can run it at 18W plate dissipation without problems and it will last at least 6000-7000 hours, possibly 10000....
So, the 6C4C PSE with 2.3K plate load will give you 10W if run the tubes at 360V/50mA each. The total DC current is 100mA. The Lundahl is specified for 90 mA but it will work just fine. There will be a bit more DC induction and this means that the power rating at 30Hz goes from 13W down to 10W (which is the power of this amp). Below 30Hz distortion will increase because the core will get closer and closer to saturation but most SE amps do not have this performance at low frequency.
You will need VERY CLEAN 145V peak-to-peak to drive the output tubes.
The distortion of this PSE amp at ANY power level will always be much less than a 2A3 SE. When the typical 2A3 amp is about to clip at 3W output, the 6C4C PSE will have about 1%. I also measured 0.3% THD at 1W.
It's not difficult to get matched 6C4C's and they can still be found at around $40-$45 each. If you are lucky, even less....
klauss00,
I said:
"Make sure the drivertube circuit can drive 2 x the capacitive current of a single output tube (or modify it so it can)."
Then you said:
"can you be a little more specific?"
Two 2A3 tubes in parallel has 2 x the Miller Effect capacitance, and 2 x the grid to filament capacitance as a single 2A3 tube has.
Driver tubes have to be able to drive the 2A3 Miller Effect capacitance and 2A3 grid capacitance, or there will be problems with the driver performance.
The slew rate will be limited, high frequencies will be distorted, and the high frequency response will be rolled off.
Parallel tubes require 2 x the current to drive the doubled capacitance, versus a single 2A3.
I hope that explains it.
I said:
"Make sure the drivertube circuit can drive 2 x the capacitive current of a single output tube (or modify it so it can)."
Then you said:
"can you be a little more specific?"
Two 2A3 tubes in parallel has 2 x the Miller Effect capacitance, and 2 x the grid to filament capacitance as a single 2A3 tube has.
Driver tubes have to be able to drive the 2A3 Miller Effect capacitance and 2A3 grid capacitance, or there will be problems with the driver performance.
The slew rate will be limited, high frequencies will be distorted, and the high frequency response will be rolled off.
Parallel tubes require 2 x the current to drive the doubled capacitance, versus a single 2A3.
I hope that explains it.
A 2A3 PSE like the one I suggest will have about 55 pF Miller capacitance + stray. A single 300B in "equivalent" amplifier will typically have 65-70 pF +stray. By "equivalent" I mean: same output power similar operational conditions. For example 350V/80mA (-74V bias) with 2.2K for 9.6W output at 5% THD.
So, the 2A3 PSE is slightly easier to drive for both voltage swing and Miller capacitance. Never worse in any case.....
To expand on 6A3sUMMER's description some more, the driving valve must supply some current to charge and discharge a capacitance at its anode. Some is stray capacitances (surprisingly, even the tiny capacitances in the air around parts and valve sockets and such matter somewhat) and some is a multiplied capacitance effect from the gain of the output valve on its internal capacitance (called Miller effect). These are all tiny capacitances, pF, meaning 10 exp-12 Farad, tiny but significant here.
Folk lately talk about "slew rate" as if it applies to audio amplifiers, but that's really an oversimplification. Slew rate is a measurement of an amplifier in severe overload, not in voltage but in input frequency. It's tested by driving the input with a vertical edge, like a single sample of a square wave, and measuring the rise time of the output. The amplifier is not even remotely in its linear operating range, so this is not very directly related to linear operation.
Slew rate calculations seen recently about required driver current are not so much false as they are misleading. They should be considered as a worst case, amplifier in overload, man the lifeboats, condition. But! that number is linearly related to the elliptical loadline that the driving valve sees in practice, so can sometimes be assumed to be a shortcut. We (hopefully) have become used to drawing reactive loadlines for output valves, or at least trying to be conservative enough to accommodate some real-world loading. We should adopt the same rigor for driving valves, but with the advantage that loading is definable and predictable. Many classical designs are not conservative, and need re-examination.
All good fortune,
Chris
Folk lately talk about "slew rate" as if it applies to audio amplifiers, but that's really an oversimplification. Slew rate is a measurement of an amplifier in severe overload, not in voltage but in input frequency. It's tested by driving the input with a vertical edge, like a single sample of a square wave, and measuring the rise time of the output. The amplifier is not even remotely in its linear operating range, so this is not very directly related to linear operation.
Slew rate calculations seen recently about required driver current are not so much false as they are misleading. They should be considered as a worst case, amplifier in overload, man the lifeboats, condition. But! that number is linearly related to the elliptical loadline that the driving valve sees in practice, so can sometimes be assumed to be a shortcut. We (hopefully) have become used to drawing reactive loadlines for output valves, or at least trying to be conservative enough to accommodate some real-world loading. We should adopt the same rigor for driving valves, but with the advantage that loading is definable and predictable. Many classical designs are not conservative, and need re-examination.
All good fortune,
Chris
Chris, and 45,
All Good points!
Some amplifier measurement "Tools":
Slew rate is just one.
Rise Time, Fall Time; low frequency bandwidth, damping factor, and distortion; high frequency bandwidth, damping factor, and distortion; mid frequency performance (same measurements as low and high frequency as above), square wave response, and many more.
In regard to overloading the amplifier (Hi Fi Amp, not Guitar Amp), there are multiple differences not only in distortion, but in recovery time.
Single ended or push pull; With and without global negative feedback; local negative feedback; RC coupling, DC coupling, Interstage coupling; Fixed bias, self bias, battery bias, etc.; single triode, SRPP, etc.; and the rest of the circuit topology.
There is an old term, called "sticking", not a bad term to remember.
And, some of the 2 most important tests:
Listening
Looks (Just an added item; level of importance belongs to the beholder).
And . . .
Does a single 300B look better than parallel 2A3 tubes?
Do 2.5V DC filaments sound better than 5V DC filaments?
Drivers that drive +/-45V to +/-50V (2A3 250V 60mA to 2A3 300V 50mA); versus driving a 300B +/- 61V to +/- 85V.
In that case, capacitance is not the major issue anyway.
Successful designs for all those voltages are multiple.
All Good points!
Some amplifier measurement "Tools":
Slew rate is just one.
Rise Time, Fall Time; low frequency bandwidth, damping factor, and distortion; high frequency bandwidth, damping factor, and distortion; mid frequency performance (same measurements as low and high frequency as above), square wave response, and many more.
In regard to overloading the amplifier (Hi Fi Amp, not Guitar Amp), there are multiple differences not only in distortion, but in recovery time.
Single ended or push pull; With and without global negative feedback; local negative feedback; RC coupling, DC coupling, Interstage coupling; Fixed bias, self bias, battery bias, etc.; single triode, SRPP, etc.; and the rest of the circuit topology.
There is an old term, called "sticking", not a bad term to remember.
And, some of the 2 most important tests:
Listening
Looks (Just an added item; level of importance belongs to the beholder).
And . . .
Does a single 300B look better than parallel 2A3 tubes?
Do 2.5V DC filaments sound better than 5V DC filaments?
Drivers that drive +/-45V to +/-50V (2A3 250V 60mA to 2A3 300V 50mA); versus driving a 300B +/- 61V to +/- 85V.
In that case, capacitance is not the major issue anyway.
Successful designs for all those voltages are multiple.
Last edited:
2A3 grid to plate 16.5 pF
2A3 rp = 800 Ohms
2A3 u = 4.2
Primary 2500 Ohms; (2500/(2500 + 800)) x 4.2 = gain of 3.8
16.5 pF x 3.8 = Miller Effect capacitance of 62.7 pF
(one 2A3 62.7pF; But 125.4pF for parallel 2A3 tubes).
The driver has to drive 125pF.
That is a lot more than a single 300B Miller Effect capacitance.
2A3 rp = 800 Ohms
2A3 u = 4.2
Primary 2500 Ohms; (2500/(2500 + 800)) x 4.2 = gain of 3.8
16.5 pF x 3.8 = Miller Effect capacitance of 62.7 pF
(one 2A3 62.7pF; But 125.4pF for parallel 2A3 tubes).
The driver has to drive 125pF.
That is a lot more than a single 300B Miller Effect capacitance.