Hi Bigun, I am completely wrong on the line level explanation. I suppose posting replies at 3:17am isn't a good idea 
So for commercial audio sources:
Vrms at 0db has a Vpk of 1.414V
This suggests a Vrms of 1.414/sqrt2 =1V
There. I feel better.
In any case, my opinion on the whole question regarding the AC signal and bias is unchanged. Its running in class A and the bias is fixed. AC signal will not cause a grid leak.
So for commercial audio sources:
Vrms at 0db has a Vpk of 1.414V
This suggests a Vrms of 1.414/sqrt2 =1V
There. I feel better.
In any case, my opinion on the whole question regarding the AC signal and bias is unchanged. Its running in class A and the bias is fixed. AC signal will not cause a grid leak.
Last edited:
Maybe we need a real engineer to weigh in on this though.
I don't care about RMS. All that matters here is the max. amplitude of the signal coming from the source. Modern audio sources will often go quite a bit higher than the "consumer level". My DAC will do +/- 3 V easily, and other sources may even give more output. It is important that the DC bias point accommodates enough headroom to allow the full signal swing without the grid swinging close to where grid current starts. Note I am not referring to grid leakage (that's something else).
Let's see what the "real engineers" out there have to say about grid current:
Rod Elliott: "...As an example, the cathode may run at +1V DC - see below for the design decisions that have to be made. The input voltage to the grid must never exceed 1V at the positive peaks, or the control grid will become positive with respect to the cathode and it will act like a diode (which it is). This means that the absolute maximum input voltage is 707mV RMS (sinewave), but preferably no more than 500mV. If your input voltage is greater than that, the valve will distort because grid current will be drawn during positive peaks. For high input levels, it is essential that the signal is attenuated before it reaches the grid to ensure that there is no chance of grid current. This is one of the main reasons that vintage valve preamps have a much higher sensitivity (lower input levels) than a modern opamp based equivalent."
Morgan Jones in his book "Valve Amplifiers", 3rd edition, p. 161: "Although grid current occurs at Vgk = 0 V in an ideal valve, practical valves enter grid current a little earlier due to the thermocouple effect of junctions between dissimilar heated metals within the valve, and the average energy of the electrons in the electron cloud above the cathode surface. Typically, grid current begins at Vgk ≈ -1 V, and this is known as the contact potential."
He then goes on to demonstrate the effect by looking at the waveform you get from driving a tube into grid current. If the source can't provide the current to drive the grid, the result will be a clipped signal.
I know that there are many designs out there that bias the input tube grid at approximately -1 V or so (see also Morgan Jones' comment above). This may be fine with signal sources that don't have a lot of voltage swing and are able to provide the current if the tube grid is driven to the range where grid current starts. But not all signal sources are like this, so I'd like to design the input stage such that this situation is avoided as far as possible.
Yes, I know these references very well. This is why I have designed to bias at cathode +2V or so. Some people use cheap red LED's which have a forward bias of 1.8V and those amplifiers perform exceptionally well.
I use a stepped attenuator on the input as well, and hardly ever get up to 0db gain. When I do, the signal is really clean and distortion is still very low. I know because I measured it. Even at +1V cathode bias there is no significant effect on measured distortion (as Morgan Jones notes in his book).
I don't see the point in +3V cathode bias unless you are using an additional pre-amplifier that doesn't act just as a buffer and 'boosts' the signal. I don't use pre-amplifier that boosts the signal, but they were popular in the 1990's. If you anticipate someone doing this then I would not use this sort of direct coupled design as you previously posted.
In this case, I would definitely use a higher cathode bias and allow a higher plate voltage on the input valve. Then I would cap couple to a source follower (with CC sink, something like Allen Wright did) to direct couple to the grid of the DHT. You can get fancy here if you wish too, especially if you have another winding to provide a negative rail... For most pentodes like El34, KT66, KT88 there is no advantage to doing this (imho) and I would just cap couple to their grids.
Ian
Last edited:
With the grid biased -2 V below the cathode, it takes only 1 V of signal swing to get the grid up to -1 V. That's where grid current tends to kick in according to Morgan Jones. With my signal sources, I need more than just 1 V of headroom. This means I can't use the pentodes (triode strapped) mentioned above as the driver tube if I want to keep the driver voltage at 130 V or below in order to avoid a B+ voltage of 500 V or more. Designing is the art of making the right compromises... I guess it's a bit of a personal choice in the end.
Last edited:
Hi Matthias
For my My 2a3 the Plate idles around at 420V. The cathode is at 175V.
On my input valve, the plate is at 130V (held by gyrator). The cathode is at about +2.1V. I am currently using C3g's but will probably go back to E280F or EC8010. I have a bunch of stuff to breadboard including a crazy phono cascode design I came up with last year while on a long business flight...
Pulling the 2a3 during operation does not destroy any part of the circuit. Pulling the input valve/tube does not destroy the 2a3. Sticking in a 'dead filament' 2a3 has no negative impact either.
THD at 'full volume' using commercial line level measures in at around 1% when using 'better quality' made in USA or Japan 2a3's. The cheapest made in china 2a3's in my stash of parts can easily bring that value up a few percent though. Admittedly, there are more expensive ones from china that are probably much better. I am not biased against made in china.
My phono stage puts out pro audio level, but the ceedee player is standard commercial audio level. You are welcome to come over for a visit some time!
Best regards
Ian
For my My 2a3 the Plate idles around at 420V. The cathode is at 175V.
On my input valve, the plate is at 130V (held by gyrator). The cathode is at about +2.1V. I am currently using C3g's but will probably go back to E280F or EC8010. I have a bunch of stuff to breadboard including a crazy phono cascode design I came up with last year while on a long business flight...
Pulling the 2a3 during operation does not destroy any part of the circuit. Pulling the input valve/tube does not destroy the 2a3. Sticking in a 'dead filament' 2a3 has no negative impact either.
THD at 'full volume' using commercial line level measures in at around 1% when using 'better quality' made in USA or Japan 2a3's. The cheapest made in china 2a3's in my stash of parts can easily bring that value up a few percent though. Admittedly, there are more expensive ones from china that are probably much better. I am not biased against made in china.
My phono stage puts out pro audio level, but the ceedee player is standard commercial audio level. You are welcome to come over for a visit some time!
Best regards
Ian
Last edited:
Your DAC has Vpk of 3V at 0db? This suggests Vpp of 6V and 0db Vrms of 2.1v... It is indeed greater than standard line level.
Is it a commercial product? Maybe I am just a dinosaur... In any case, if I used your DAC, I would need to keep the volume lower, as to not bother my swiss neighbors any more than I currently do.
The attenuated input signal from your DAC would not cause increased distortion with my amplifier. Maybe it would sound very nice though.
Back in the late 80's and 90's when certain valve/tube pre-amps could put out at Vpp of 20V, you never saw anyone dial it up past 10 o'clock.
Ian
Last edited:
Your DAC has Vpk of 3V at 0db? This suggests Vpp of 6V and 0db Vrms of 2.1v... It is indeed greater than standard line level.
Is it a commercial product? Maybe I am just a dinosaur... In any case, if I used your DAC, I would need to keep the volume lower, as to not bother my swiss neighbors any more than I currently do.
It's a DDDAC. That's just a passive I/V converter stage with no active buffer or amplifier at the output. Just a 2:1 step-up transformer. I just checked the signal level again, and it's even a bit higher than the 3 Vpk I mentioned.
I am not sure what you mean by "commercial" products.
Audiophile high end, Box store products, professional sound recording, or sound reinforcement products.
I either use a remote stepped volume control in front of my power amps, or have a potentiometer in the power amp input circuit.
I have tested at least 5 CD players or more (except for my two AA battery Walkman CD player). They all have 2.1Vrms out, 2.97V peak, when they have the 0dB 1kHz tone selected from the Denon Technical Audio test CD (16bit full scale).
Most of my driver and output stages do not need anywhere near that full output of those CD players to clip the amplifiers.
I try and design the input stage to have at least -0.75V grid to cathode at amp clipping (I shoot for -1V).
Tests that one of my associates has run multiple times from medium impedance sources on different input tubes suddenly has more distortion when the grid is about -0.75V to -0.5V grid to cathode.
Try it yourself on your amp (measure the distortion versus signal level for just the input stage, not the later stage(s).
You mileage will improve (to closer to 0V grid to cathode) when you use a 50 Ohm generator, but who has "commercial" music sources that have true 50 Ohm output impedance?
Audiophile high end, Box store products, professional sound recording, or sound reinforcement products.
I either use a remote stepped volume control in front of my power amps, or have a potentiometer in the power amp input circuit.
I have tested at least 5 CD players or more (except for my two AA battery Walkman CD player). They all have 2.1Vrms out, 2.97V peak, when they have the 0dB 1kHz tone selected from the Denon Technical Audio test CD (16bit full scale).
Most of my driver and output stages do not need anywhere near that full output of those CD players to clip the amplifiers.
I try and design the input stage to have at least -0.75V grid to cathode at amp clipping (I shoot for -1V).
Tests that one of my associates has run multiple times from medium impedance sources on different input tubes suddenly has more distortion when the grid is about -0.75V to -0.5V grid to cathode.
Try it yourself on your amp (measure the distortion versus signal level for just the input stage, not the later stage(s).
You mileage will improve (to closer to 0V grid to cathode) when you use a 50 Ohm generator, but who has "commercial" music sources that have true 50 Ohm output impedance?
Last edited:
wow. If I hooked up your DDAC to my 2a3 amplifier and 94db speakers, the volume control would probably not get far beyond the -24db mark since it would be simply too loud. This would happen before any real distortion could be measured, much less observed. The lowest setting I have on my stepped attenuator is -62db. That would leave about 11 steps for input control.
So no more distortion would be measured or observed, the only problem would be the fewer steps left to control the level. That would indeed be sad...
One thing that is very nice with my amplifiers is that I can easily handle the terribly low level iphone/ipad sources that my daughters like to share with me. Maybe I am not a dinosaur after all?
Ian
So no more distortion would be measured or observed, the only problem would be the fewer steps left to control the level. That would indeed be sad...
One thing that is very nice with my amplifiers is that I can easily handle the terribly low level iphone/ipad sources that my daughters like to share with me. Maybe I am not a dinosaur after all?
Ian
Last edited:
Well, one more question... what's the role of R7 in the schematic I attached in post 66? The schematic was drawn for a load of 3 mA, but we'll end up with a much higher current in the driver of this 45 amp. Can I make this resistor smaller? A lot smaller? Remove it?
R7 is "R-MU". Its actual equivalent impedance is increased due to the bootstrapping effect of the lower jFET. Like a cathode/source follower. The bootstrapping depends on the actual transconductance of the FET (gfs) which it also depends on the current (Ia). The higher the bootstrapping the more linear the stage is. However, you pay the price of more volts dropped across the resistor and higher HT required.
In practice I found 470R to be a good value when the driver has a current of 20mA or more. For low currents anything between 470R to 1K5 works (and sounds) well. Also the quality of this resistor is important.
In practice I found 470R to be a good value when the driver has a current of 20mA or more. For low currents anything between 470R to 1K5 works (and sounds) well. Also the quality of this resistor is important.
On your website you wrote "R7 is the mu-follower resistor which typically is set to 1/gm for optimal value." Why is this?
Assume I was going to use this with a 6H30 at 30 mA, and instead of the 2SK170 I'd use a DN2540 with a specified gm of around 0.3 mho (yes, that's variable depending on the voltage and current, but let's use this number as a rough guide). This would translate to R7 = 1/gm = 3 Ohm or so. Really!? That would be cool because it would hardly drop any voltage between the FET and the plate of the driver tube.
What do you mean by this?
Hi Matthias
You need to have some boostrapping effect or the output impedance will increase, won't it? With a 3 ohm resistor, it will be simply the same as taking the signal off the plate.
I personally just worked on this experimentally. I started with around 220 ohm, then increased it. I've been using 1k ohm for around 14mA. Sounds like a weak excuse, determining a component just experimentally...
Also, the dn2540 will work, but it won't sound as good as some other devices. Its one of those 'sand' things perhaps?
Ian
You need to have some boostrapping effect or the output impedance will increase, won't it? With a 3 ohm resistor, it will be simply the same as taking the signal off the plate.
I personally just worked on this experimentally. I started with around 220 ohm, then increased it. I've been using 1k ohm for around 14mA. Sounds like a weak excuse, determining a component just experimentally...
Also, the dn2540 will work, but it won't sound as good as some other devices. Its one of those 'sand' things perhaps?
Ian
Exactly! I wonder where the "1/gm rule" rule comes from.
Nothing wrong with determining things empirically. But I'd like to understand how things work before I start designing the active load for the driver on paper.
These are specified to 60 mA and 25 mA max. current (if I am reading the data sheets correctly). I am looking at 30 mA idle current for the driver tube, so the BF862 won't work, and the J310 spec. may be a bit too close. Some more headroom would be nice.
Output impedance from mu-follower is 1/gm. You can play with the maths and derive it yourself. Also would be great if anyone can jump with the analysis and explain the 1/gm ratio needed for best power transfer, not voltage.
I posted recently some measurements for gfs at lower current. You may find them useful. I’d go for anywhere between 470R and 1K5.
I found the quality of this resistor to contribute to the sound, as well as the bootstrapping cap C1.
Cheers
Ale