Hi revintage (Lars),
I think the anode follower is the way forward for me as you noted gain is 'tunable' with R2. so it looks the perfect solution.
So as described in post 6 & #11 I built an preamp based on a trioded D3a using r1=100k and r2=150k from a circuit I found on the web (Can't remember the source of it now but found it on my PC HD from some months ago).
My measurements on oscilloscope though looked strange in that the output from the buffer was not inverted but it did show gain of ~1.6. (Note I have buffer connected directly to 100k log pot then into power amp)
I was wondering if the reason for the output not being inverted was because of the way I connected the D3A pentode as a triode hence the earlier post #15.
I have only limited knowledge of tube circuits so I'm not able to design any circuits from scratch that is why I went ahead with this D3a triode anode follower from a 'known' design.
In conclusion I guess the best way forward is
1. 'resolve' how to connect d3a pentode in triode mode. (g2, g3 to anode or g2 to anode, g3 to cathode)
2. clip lead different values for r2 until I reach satisfactory gain in my system with different sources.
3. Finish build
Thanks for your help
Cheers
Colin
I think the anode follower is the way forward for me as you noted gain is 'tunable' with R2. so it looks the perfect solution.
So as described in post 6 & #11 I built an preamp based on a trioded D3a using r1=100k and r2=150k from a circuit I found on the web (Can't remember the source of it now but found it on my PC HD from some months ago).
My measurements on oscilloscope though looked strange in that the output from the buffer was not inverted but it did show gain of ~1.6. (Note I have buffer connected directly to 100k log pot then into power amp)
I was wondering if the reason for the output not being inverted was because of the way I connected the D3A pentode as a triode hence the earlier post #15.
I have only limited knowledge of tube circuits so I'm not able to design any circuits from scratch that is why I went ahead with this D3a triode anode follower from a 'known' design.
In conclusion I guess the best way forward is
1. 'resolve' how to connect d3a pentode in triode mode. (g2, g3 to anode or g2 to anode, g3 to cathode)
2. clip lead different values for r2 until I reach satisfactory gain in my system with different sources.
3. Finish build
Thanks for your help
Cheers
Colin
Generally, the total noise from the tube can be specified as an equivalent noise resistance, where it's given specifically as the resistance at room temperature that would generate the same amount of noise as the tube at its normal temperature. So you need to compare the noise from the control to the ENR of the tube. For an ECC88/6DJ8 sort of tube at normal operating currents, it will be a couple hundred ohms. So a 10k control at midpoint will totally overwhelm the noise from the tube.
Hi Colin,
Begin by connecting screen to anode and g3 to cathode. You can later try strapping g3 the other way.
But also give your D3a a chance as a pentode.
The stage will absolutely be phase-reversed.
Probably you want gain below 1 for your CD.
I would also recommend trying diodes instead of cathode resistor/capacitor.
Contrary to what the others say, do not use high ohmic series resistors. This also means, do not go over 50k volume pot if you want to have it all. It is not about noise....
Begin by connecting screen to anode and g3 to cathode. You can later try strapping g3 the other way.
But also give your D3a a chance as a pentode.
The stage will absolutely be phase-reversed.
Probably you want gain below 1 for your CD.
I would also recommend trying diodes instead of cathode resistor/capacitor.
Contrary to what the others say, do not use high ohmic series resistors. This also means, do not go over 50k volume pot if you want to have it all. It is not about noise....
revintage said:
If it's not about noise perhaps you can suggest to me another mechanism whereby an attenuator decreases dynamics?
In fact the noise contribution of an attenuator in this position is insignificant anyway as the signal is huge in comparison. I only introduced it as it is THE ONLY WAY in which a resistive divider can impact on dynamic range.
A resistive divider divides signals LARGE and SMALL by the same amount. This is what we understand by a LINEAR RELATIONSHIP such as expressed in the equation V=IR.
The whole thread is based on a false premise. The sound is not FLAT or LIFELESS it is simply NOT AS LOUD.
What is a volume control if not a resistive divider? Volume controls are not infrequently in the megohm range.
revintage said:
Yeah. Pigs will fly.
SY - you cannot make an active attenuator with a lower noise contribution than a passive attenuator.
Recall the laws of thermodynamics:-
1. You can't win
2. You can't break even
3. You can't get out of the game
I realise that some of you are struggling with these ideas, but sometimes you try my patience.
w
Where did I say that you could? I was merely pointing out the obvious- why sweat tube noise when it's lost compared with the Johnson noise of the attenuator.
I'm fairly comfortable with thermodynamics, having spent some years teaching it at a university. The more accurate statement of the Three Laws is:
1. You can't win, you can only break even at best.
2. You can only break even at absolute zero.
3. You can't get to absolute zero.
wakibaki said:
What about decreasing the previous stage's source current availability, thus the damping factor and slew rate available for driving the input impedance and Miller of the next stage (possibly interconnect capacitance too, if between components), by adding extra series resistance part with an L pad? Valid thinking?
In loudspeaker HF horn drivers or tweeters an auto former tap is always subjectively more dynamic than a resistor L pad of equivalent attenuation in my experience. For instance, BBC used auto former taps in crossovers of their original monitors, much dearer to make than simply attenuating with resistors, and they had serious listening panels to compare and arrive at that. Also all LS3/5A collectors know that the best crossover versions are the ones with the auto former taps. Each time I have to add or increase a gate stopper or a grid stopper to prevent some possible oscillation in use with unknown equipment or interconnects, the dynamics subjectively lessen and the tone gets muddier. There is no voltage attenuation taking place there, just the source impedance rises.
Why all those findings? Any thoughts?
Also, some times pigs fly, but not very often.
If you guys are really serious, and you appear to be, can you enlighten me as to why all the resistive attenuation that happens inside complete circuits is OK, but apparently once you move it outside the magical preamp or power amp box, it kills dynamics? I'm not entirely sarcastic, I'd like to hear a reason why if you have one.
To my experience, higher impedance is not OK internally too. The higher the impedance of a Riaa filter, or the higher the stoppers, the more it dies subjectively. Also each time some load resistor or RC coupling is substituted by a load choke or an interstage transformer, it always livens up. I don't know too. I inquire for why the subjective findings.
sorry to jump in your threat. i am also has question in this voltage divider circuit.
is it because the mismatched impedance that cause the lost of dynamic?
here is the page about voltage divider that also discuss about impedance matching...
http://sound.westhost.com/vda.htm
i would like to try this into my too much gain systems as well
i would like to try out the 24k and 1 k resistors deviders that gives 28db attenuation, Zin=25K and Zout=960ohm
while my pre Zout is 800 ohm, hence i think this keep the Zout to the amplifier almost the same while still keeping the 10X golden rule of impedance matching. think this will take care of the impedance mismatched? can some confirm this?
erwin
is it because the mismatched impedance that cause the lost of dynamic?
here is the page about voltage divider that also discuss about impedance matching...
http://sound.westhost.com/vda.htm
i would like to try this into my too much gain systems as well
i would like to try out the 24k and 1 k resistors deviders that gives 28db attenuation, Zin=25K and Zout=960ohm
while my pre Zout is 800 ohm, hence i think this keep the Zout to the amplifier almost the same while still keeping the 10X golden rule of impedance matching. think this will take care of the impedance mismatched? can some confirm this?
erwin
From that linked article: ''Conclusion...It is used to reduce voltages, not current or power (although both are affected, that is a side effect, and not the real purpose).
Make sure that the resistance values you use are "sensible", and do not impose excessive loading or introduce excessive output impedances''
Obviously current and power are affected as a side effect.
Make sure that the resistance values you use are "sensible", and do not impose excessive loading or introduce excessive output impedances''
Obviously current and power are affected as a side effect.
SY said:
Oh. Well I wasn't. Sweating, that is. In fact I was quite handily winning this argument under my own steam, without the benefit of an intervention which appeared to be an objection to my position, and was poorly enough expressed for me to mistake it as such.
I'm simply pointing out that, however small the contribution of an active device, all other things being equal, using one in these circumstances actually results in (trivially) poorer dynamic range than a simple divider.
Why anybody would imagine that it was preferable to add a valve stage is incomprehensible to me if only in economic terms.
As to the question of 'impedance matching' as discussed in the ESP article, this is not true impedance matching as it is understood in RF circuits but rather measures taken to ensure that a resistive divider does not excessively load a source circuit and in turn is not excessively loaded by a sink.
Although this can have an impact in audio circuits the implications are more that the signal levels will not be precisely those expected from calculation rather than distortion or compression. In this case it is probably sufficient merely to take a pot in the tens or hundreds of kilohms (sensible) range, connect it up and tweak the position until a satisfactory level is achieved.
When attenuation is required in a power circuit, such as an antenna feeding a receiver, a Pi or T attenuator can be designed which presents the same impedance (usually 50 ohms) to both the source and sink. This is an example of true impedance matching.
salas, there is something to be said for the points that you raise but none of them were being addressed by the suggestions being made previously and I refer you to leadbelly's contribution which reminds us all that we gaily change the impedances seen by various stages in a system by changing volume or tone control settings without concern for muddying or loss of dynamic range in the sound.
erwin, if you need 28dB, then these values will be fine.
w
I guess I wasn't looking at who was winning, I just thought that your statement "Any active stage you put in there will increase the noise floor more than a passive attenuator because it's hotter. Remember Boltzmann?" could easily be misunderstood to mean that the Johnson noise contribution from the tube was somehow calculated at a higher temperature (that's what "hotter" means to me) than that of the attenuator and gave a higher noise voltage. That is not the case- ENR is specified as input-referred and at room temperature. The tube's contribution is negligible compared to the attenuator for any reasonable tube, so if it is desired to (for example) drive cables without having the frequency response vary with volume setting, a tube (or other) buffer brings benefits without any significant deficit.
Of course, the notion that, say, a 6dB resistive divider (unless engineered by a bonehead) will "kill dynamics" is also nonsense.
wakibaki said:
Overlooking something in one lower current domain where it manifests more discerningly or even below some subjective threshold, if kind enough, does not justify sidestepping it in another higher current domain like loudspeakers, where BBC's preference for autoformer taps instead of Lpads is a strong testament of some possible subjective loss.
We may merrily drive a PP valve output stage with a scrawny phase splitter many times in classic circuits for instance, but that does not mean that it will not flesh out and kick if we buffer it with a Mosfet or a gutsy cathode follower. This is how ''it sounds'' and it may be debated, some will prefer the classic circuitry tone, but just try to use a single ended valve amp with 5R output on a 4R speaker. There it will not even make it half decently. Its just a not so good practice, magnifying stage by stage IMO. And yes, the output Z forms an Lpad with the speaker's Z, modifying its frequency response strongly according to its varying impedance, on top of forming minus damping factor to strongly upset the bass alignment .
SY said:
Given that I am simply trying to assist someone in achieving a rational and economic solution to a problem with the minimum of effort on my part, it would have been more useful had you posted your last statement first and gone on subsequently to elaborate on the situation regarding thermal noise.
It sure would have saved me a load of typing.
Next time try to focus on the big picture.
w
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.