dhaen said:
Agreed, but nonetheless quite feasible.
I think that there are often undiagnosed problems with the widespread use of LC (and LCLCLCLCL...) unregulated supplies. Don’t get me wrong, these kinds of supplies can sound glorious. They can also be very effective at removing line noise. I have a few LC projects around of my own design that sound great. But…
There can be unrecognized problems – in the low-end especially. The L and C form a parallel resonant circuit. At low frequencies, this will manifest as a peak in the supply’s output impedance (power supplies are supposed to present a ZERO output impedance at all frequencies by the very definition of a voltage source). If there are cascades of several LC sections in the supply, as is so popular among DIYers today, there will be a complex group of impedance peaks in the supply’s output. These peaks will modulate the supply’s output voltage in response to current demands by the audio circuit. The prevailing assumption is that these non-ideal effects occur way below the audio band where they are safely out of harm’s way. A well-design LC supply will ensure that. However, too often these effects are near enough to the bottom of the audio band that they will affect bass response, and possibly create intermodulation products way up higher in the audio band since there may be very high amplitude voltage and currents in the infra-bass range. Furthermore, the reactive LC supply impedance can interact with audio circuit reactances such as the impedance of cathode bypass cap as reflected through the plate, plate chokes and signal transformers, in complex ways that create deviations from the intended circuit behavior. Add capacitive coupling between multiple audio stages and some negative feedback, and it’s no wonder that bass sounds can vary so much. Sometimes even motorboating will result.
The widespread use of the otherwise-wonderful PSUD simulator can lull folks into thinking that if the simulation result looked good, the power supply will work fine. But PSUD, to my knowledge, doesn’t simulate the supply’s output impedance (as presented to the audio circuit) across the frequency band.
So this is one reason I usually use regulated supplies (and I’ve built both tube and SS versions that have worked fine). If designed properly, and it’s not all that hard to do so, the output impedance will be very low all the way down to “DC”. This aspect of performance removes the reactive interaction seen in raw LC supplies and provides a stable and clean basis for the design of the rest of the audio circuits. At higher audio frequencies, the finally shunt cap with its smaller bypass caps will usually define the supply impedance, no matter whether the supply is regulated or not, tube or solid state.
I suppose a sort of ideal situation would be a regulated supply with a choke input (LC) filter for the raw supply. This would keep current spikes from creating noise either by magnetic field radiation, or by AC line or ground conduction. In other words, the best of both worlds is possible.
Sorry for the rambling post…
Hi Brian,
I happen to agree with everything you've said. Your post handled a broad topic well.
Your point about stopping the high frequency content before it's created is something that I think may be lost on many designers. Monster caps are not required and work against a quiet power supply design.
-Chris
I happen to agree with everything you've said. Your post handled a broad topic well.
Your point about stopping the high frequency content before it's created is something that I think may be lost on many designers. Monster caps are not required and work against a quiet power supply design.
-Chris
astouffer said:
Hi,
Correct.
In the case of the regulator shown in the above link it does not matter.
One point though: while the regulator will do it's job outside of the intended original circuit it wasn't really designed to regulate in the usual sense in the actual circuit it was designed for in the first place.
As a matter of fact the original design had so much capacitance behind the regulator and a current draw so small that under no normal operating conditions the regulator had to work at all.
So what does it do?
Look at it as a means to mimmick a trickle charger for a battery supply. See it as a daft means of isolating the entire audio circuit from the mains....
Anyway you want but don't just go lifting parts of designs out of context unless you pretty well know what you're talking about, please.
Naturally, all electronic regulators have their drawbacks and designing a perfect one for audio applications is a near impossible task.
Fortunately you can put a decent amount of capacitance behind a series reg. so when you're design operates in class A current can be drawn straight from the reservoir caps instead of calling on the regulator to supply it at a constant voltage. Hence the comparison to a trickle charger above.
So much for that naysaying moderator whom I probably taught a thing or two ?
As for the reason of the ECC83 in the cascode? Well you'd want super high gain with no added noise from a penthode as an error amp. Low noise penthodes being too expensive to design with for production we were left with what you see.
Want to improve on that one?
Change the socket layout and drop in an E283CC for even lower noise or should you want razor sharp output stability you could find yourself some ECC807s. Both of which are plentiful but not in this world....Maybe valve nirvana, who knows.....
So, why am I writing this?
First of all, if you want decent sound do not rely on the regulator to provide it. The regulator can however do other things unregulated supplies can't.
Second, if you think you do not understand what regulators can and can't do for audio don't just go lifting one from a textbook and judge it in your circuit, what you'll probably be listening to is the change of Z of your PS across the audiospectrum. It will quite likely sound different from the unregulated supply but different does not necessarily equal better even if you like what you're hearing.
Better meaning closer to the original signal in my audio dictionary, i.e. not a dB louder nor a little bit sweeter than the truth, etc.
Thirdly, valve circuits being relatively high voltage devices their exact operating voltages aren't all that critical, within reason that is.
They do however like to see every change in operating voltage applied proportionally to their elements. So when you decide to regulate one rail you're well advised to look at what happens to the rest of the supplied voltages. Grid bias being one of them.
To give you a better perspective: it's absolutely necessary to regulate the B+ of a tube operating at 24VDC if you're entire design depends on a loadline that has to be maintained as precisely as possible. A tubed MC headamp is one example, microphone amps are another if you catch my drift.
Input stages of amplifiers working in class AB can benefit from regulated voltages to keep them isolated from the hogging of the output stage for example.
In short if you don't inderstand real world demand of your circuit nor the precise operation of it, forget regulators. You're quite likely better of without.
@ D'Haen: Thanks ole git.
Cheers,
I agree about fixed grid bias needing to be regulated. Screen regulation is common for pentode output amps but that's no good on its own if mains voltage drifts. You want to be sure that the quiescent current through the OP tubes stays constant under all conditions and you need to regulate the grid bias to achieve this.
Wavebourn, there is definately a reason to use a tube rectifier in a tube amp. In addition to the well known 'tube warmth' associated with tube amps, there is also a characteristic known as 'soft clipping' where tube amps create a compression effect, softening the blow on hard transients like drum smacks. My interest is using tube mic preamps for recording drums, so this compression effect is of more interest to me than the typical tube warmth sound. According to this article, the tube rectifier is partly responsible for the soft clipping / compression effect. So, your claim that solid state rectifiers can be 'better' is usually the opposite of what attracts people to tube amps in the first place, which are the 'flaws' in tube amplification. Quoted from the article:
"It also makes for a compression effect. Lowered voltage due to a transient causes all of the biasing throughout the amplifier to also lower, in turn lowering the overall amplification ability and introducing distortion effects of incorrect bias. So again the so-called squishing of tubes rears its ugly face. This is likely why many do not like solid state or regulated power supplies, because they reduce or eliminate the squishing effect, and reduce or eliminate the even order distortion produced by the change of bias due to dynamic transients."
http://members.tripod.com/~gabevee/mypwrsup.html
"It also makes for a compression effect. Lowered voltage due to a transient causes all of the biasing throughout the amplifier to also lower, in turn lowering the overall amplification ability and introducing distortion effects of incorrect bias. So again the so-called squishing of tubes rears its ugly face. This is likely why many do not like solid state or regulated power supplies, because they reduce or eliminate the squishing effect, and reduce or eliminate the even order distortion produced by the change of bias due to dynamic transients."
http://members.tripod.com/~gabevee/mypwrsup.html
Hi,
That effect is not caused by the use of a tube rectifier per se but by the unadequate power supply following it or the incorrect use of a tube rectifier vis a vis the expected current peaks.
There is no "warmth" to be associated with a correctly designed tube amplifier. There is however a preponderance of even order harmonic distortion versus odd order for most tube amplifiers.
Soft-clipping can be cute as an effect in studio use or for electric guitars but again, clipping is clipping so you're overloading the outputstage which is not what you want from an audio amp.
What this has to do with using a tube rectifier versus a diode as a rectifier is beyond me.
Cheers,
That effect is not caused by the use of a tube rectifier per se but by the unadequate power supply following it or the incorrect use of a tube rectifier vis a vis the expected current peaks.
There is no "warmth" to be associated with a correctly designed tube amplifier. There is however a preponderance of even order harmonic distortion versus odd order for most tube amplifiers.
Soft-clipping can be cute as an effect in studio use or for electric guitars but again, clipping is clipping so you're overloading the outputstage which is not what you want from an audio amp.
What this has to do with using a tube rectifier versus a diode as a rectifier is beyond me.
Cheers,
I don't, at least not as a general rule. And I don't think that's what Frank was saying.
Let's say you have a triode or UL amp. You regulate the bias but don't regulate the output stage B+ (regulated output stage supplies are the exception). The line voltage increases. The bias, being regulated, stays fixed, but the B+ to the output tubes' plate/screen increases. Then the tube's current will also increase.
Contrariwise, if the line voltage decreases and the bias stays fixed, the output stage idle current will decrease, resulting in higher distortion.
If the bias is left unregulated, it will increase and decrease along with the B+ supply, providing a bit of DC feedback.
Now if the output supply is regulated (or you have a pentode amp with a regulated screen supply), it indeed makes perfect sense to regulate the bias supply.
Isn't that what I said? It's what I meant, anyway: dont' regulate one without regulating the other.
Heck, if I have to rebut your argument then I may as well rebut Frank's at the same time. I wasn't going to bother but I think you're wrong, Frank.
Floating the heater supply in that regulator seems to me like too much of a gamble. At what potential is the heater supply going to sit if it's floated? It has to be at some potential, doesn't it? How can we be sure it will be just right for both halves of the 12XA7 and the EL86 so as not ot blow the heater-cathode insulation on one of 'em? Also, I have always understood that you need a high transconductance tube to make a decent cascode amplifier, which a 12AX7 is not. I wish Allen Wright was in this forum, because he can speak with far more authority on this than I can.
ray_moth said:
Isn't that what I said? It's what I meant, anyway: dont' regulate one without regulating the other.
Heck, if I have to rebut your argument then I may as well rebut Frank's at the same time. I wasn't going to bother but I think you're wrong, Frank.
Also, I have always understood that you need a high transconductance tube to make a decent cascode amplifier, which a 12AX7 is not. I wish Allen Wright was in this forum, because he can speak with far more authority on this than I can.
Hi,
Sorry but you think wrong on both counts in this case.
What's the heater to cathode potential?
Why would you need a decent cascode when all you need is a high amplification factor?
Cheers,
SY said:
It is the single reason for cascodes. The drawback is distortions: upper cathode sees relatively high dynamic resistance load compared to capacitance. They are good in UHF amplifiers where harmonics are out of the band, or distortions are exploited for goodness of a frequency conversion. Of course, they may be used in voltage regulators, but I think again about my main question: "Why do we need tube regulators?", that is the matter of a different topic.
SY said:
I don't see why that would matter when all the regulator has to do is amplify the voltage error it detects.
As I think I explained in a post above, the example is badly chosen as in the circuit it was designed for an error could occur at the input of the reg but not at output so it really doesn't apply. Which is why I put in a previous answer: "In this case...."
So, would I advise people to build this reg, for anything other than for what it was specifically designed for? NO, absolutely not.
If it's regulation you want look elsewhere, if you want it to do what I have described it does then yes, it does that very well.
Note that I also specifically mentioned the E283CC as an upgrade en lieu of the 12AX7A. The E2833CC should not be affected by Miller effect as much as an ordinary 12AX7A. Should you not care for Miller en vue of the goals you set for the regulator you could sharpen the error correction even further by using an ECC807.
Last of all, I stated these issues more than once on the forum and I am well aware of the fact that the circuit as an isolated regulator could be improved. But then it was never meant to play the role people usually associated with a typical regulator and, if I may add once again, you can do a lot more with a regulator than just regulate alone.
If you want a good sounding regulator for a preamp I would advise shunt regulation, not series regulation unless you implement the circuit the way I intended and have explained.
Cheers,
ray_moth said:Frank, please be patient!
Hi,
No problem, I'd rather be reading than discussing a regulator no one really needs anyway.....
Thanks for the welcome back, ray_moth.
Cheers,
Power supplies are, or should be, characterized by two measures: line regulation and load regulation. My strongest reason for using regulation for audio is for load regulation, NOT for line regulation. Line regulation comes for free. Generally I don’t care too much whether my B+ is 300 volts or 310 volts due to line voltage changes (although sometimes I do care in DC coupled designs…). But I always want the supply voltage to resist changing/modulating with varying signal current demands. That means thinking of the supply in terms of output IMPEDANCE (over frequency and with regard to linearity and phase). That seems to be the most meaningful parameter by which to characterize audio supplies, followed by noise performance (hum rejection, etc.). If I understand some of the arguments above, there is the suggestion that regulation is not necessary because line regulation is usually not critical. I couldn’t disagree on that basis alone. But if you think in terms of supply impedance, it’s pretty hard to design an unregulated supply that provides a very low impedance at very low frequencies. To maintain an arbitrary one ohm at 10Hz (never mind at one Hertz), the output cap in an unregulated supply would have to be 16,000uF. Got any 16,000uF at 450 volts caps? Not me. Think you need only ten ohms? OK, got any 1,600uF at 450 volts? Even if you could find such a bank of caps, and could keep it from resonating with a choke earlier in the filter path, you’d have to worry about ESR and ESL. If you bypass this giant cap with smaller caps, you’d risk additional parallel resonances between the big ESL and the smaller capacitances.
The low end is where active regulation can shine. Use good film caps to provide the current path at high frequencies.
Designing a perfect ANYTHING is nearly impossible! But I think it's probably easier to design a more nearly ideal regulator (whether tube-based or solid-state) than almost any other audio circuit.
The low end is where active regulation can shine. Use good film caps to provide the current path at high frequencies.
Designing a perfect ANYTHING is nearly impossible! But I think it's probably easier to design a more nearly ideal regulator (whether tube-based or solid-state) than almost any other audio circuit.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.