Regulated Supplies in Tube Amps
I am new to this forum. My first few posts and responses were in a thread about a 5692 line stage design. Part of the conversation in this thread involved the need for regulated power supplies. I've searched the forum as best I could to see if I'm duplicating something that is already here. I don't think that I am, but if so let me know before I make a real mess. :)
There appear to be differing opinions about the value of regulated supplies and I admit that I havenít had an opportunity to do any kind of A/B comparison to judge for myself.
However, it is possible to get some idea of what this is all about and I thought, if the members of this forum will indulge me, that I might give it a try. Perhaps this will be helpful to forum members who donít really understand whatís happening here. And I certainly hope (and expect) that the top notch circuit guys will be free with their comments and criticisms. I freely admit to being merely a physicist who can use Maxwellís equations to solve problems, but really doesnít have a lot of circuit design experience.
To study the problem of PS impedance, Iíve simulated a handful of circuits to be explained in later posts. Iíll be putting up a series of posts instead of one long one. Iím using simulations because itís too hard for me to set up the differential equations using Kirchoffs laws, do the Laplace transforms, and study the transfer functions. :scratch: Besides, the simulators do this for you. I use PSpice 9.2 Demo and Beige Bag BSpice. Both are very good simulators. Iím using PSpice here because I have more tube models for it at the moment.
There will be some limitations on the modeling because I donít know how to handle certain aspects of power supplies yet. To get things started, here is the basic circuit with an empty, three-terminal regulator in it:
In succeeding posts Iíll be putting different regulators into the design to see how they perform.
The load is simply a grounded cathode stage using a 6dj8. When this stage is fed with a 1V peak signal it will swing 5mA pp. In the simulations, Iím placing an AC source at the input of this amplifier and doing small signal AC analysis on the amplifier/PS combination.
One major limitation of this little study is that the line, transformer, and rectifier portions of the PS are replaced by a perfect voltage source. Much information is left out. Nevertheless, we can still get some idea of how the PS impedance affects the audio circuit.
The AC analysis will reveal (within the limitations of the simulation) the incremental behavior of the PS. To understand the affect of the PS on the circuit we have to look at both the real (resistive) and the imaginary (reactive) behavior of the system. For the first exploration, Iíll be looking at the resistive behavior of the PS. In particular, the diagram shows a delta-V and delta-I at the output of the PS. Taking delta-V/delta-I for the real parts of V and I will give us this resistive behavior. The point marked PHASE is for later.
If at any time any of you think Iím wasting your time, please donít hesitate to let me know and Iíll stop. And if Iím wrong, donít let me make a fool out of myself for too long please. Or maybe even help me out. :)
The first ďregulatorĒ is the addition of a simple RC stage. Here it is:
The analysis will look at the response of the total circuit from 10Hz to 1MHz. The Y-Axis is delta-V/delta-I or, basically, the resistance of the supply to AC signal demands.
The results of the analysis are:
What this says is that at 10Hz the power supply looks like ~50ohms to the current demand of the amplifier. At 100Hz it looks like 5ohms. And so forth. To see this better at the higher frequencies, this log plot will be clearer.
No surprise here. This is a straight line because the ďregulatorĒ is a simple RC network. At 10KHz the resistance is about 20milli-ohms. Etc. etc.
The basic feature of the situation is, however, that even at 10Hz the PS shows only 50ohm AC resistance, which means that at 5mA pp, the voltage will vary only by .25Vpp. However, this may be important to the audio quality of the amplifier. I donít know the answer to this.
Just a couple general questions, which will certainly get some flames aimed again at my direction; I've made no secret of my opinion that most of the power supply stuff that we sweat is overblown. You've got a line amp there which has to swing, oh, maybe 2V to clip any power amp it's attached to (in a line stage, you don't need to swing 5 ma with a 10K resistor; at those levels, the nonlinearities of the tube will swamp out any reasonable power supply's effect anyway). It's going to draw a constant current. It's got a 10K plate resistor. It will be followed by a buffer. It's not going to drive any significant reactance, just a few pF at most. It's used in a high-level stage, not a phono amp.
1. Why would you sweat whether the source Z is 10 milliohms or an ohm (or a dozen ohms)? If that's important, better temperature stabilize that 10K resistor. And use only those magic tubes that have constant rp with current and don't change characteristics with age.
2. What advantage accrues from connecting a regulator directly to the plate resistor, rather than preregulating to stabilize the B+ and then using a simple RC or LC to get the noise out? This is a class A circuit...
3. Why is it that even with simple regulators, or with the RC/LC in Q2, probing the supply lines with signal running through the circuit at intended levels shows pretty much nothing other than DC?
In my own preamp, when I run signal through the circuit, I see nothing on the B+ or B- rails with a scope, and hear nothing with headphones coupled to the supply rails through a capacitor. It's reflected in the output, where the S/N is better than -85 dB (limit of my measurement capability), despite me using nothing fancy in the way of regulators. Admittedly, I'm not running high-level 10 Hz signals through it, just music.
SY, I am not taking a position on this at the moment. I'm just trying to see show data that makes sense to me to see what other's opinions of it might be.
This is a class A circuit, but many, if not all, of the line stage amps are class A and many folks put regulators on them.
I'm going to continue with one more analysis and then wait to see if I should keep going.
Regulators won't necessarily hurt, it's just not where you'll get the most bang for the buck in this application. IM not-so HO.
Yes, lots of people do it. But lots of people also pay $100 for a jar full of rocks which promise to improve your sound. Popularity of an approach (especially when considered with respect to a particular subculture) doesn't say anything as to whether or not it's useful.
This regulator is a fairly common design for a series pass regulator using a triode/pentode pair (although the pentode is in triode mode).
The regulator is set to deliver 200VDC using a 40V reference zener diode. Note the capacitor in the lower right hand corner. The value of this capacitor will have an impact on the behavior of the regulator.
For the first simulation, Iím setting C=0. This shows the behavior of the ďrawĒ regulator.
Notice that the resistance starts at about 6ohms at 10Hz, drops to 3ohms to well above the audio spectrum, and then begins to rise again. So, whatís happening above 1MHz? This:
There is a resonance in the PS at about 40MHz. No big deal, weíll put C=470u and try again. Like this:
Notice that the resistance is about 4ohms at 20Hz, drops to 3.5 and then stays very constant until above 100KHz. This regulator will give a very even response to the demands of the circuit over the entire audio range and beyond.
Typically, however, we have C=largevalue. Letís try C=330u. Here it is:
This is an interesting curve and looks similar to the one for the simple RC ďregulatorĒ. In fact, it is. Here is the log plot:
Except, the series regulator shows improvement at low frequencies. The resistance at 10Hz is now about 5.5ohms almost 10 times less than the simple RC design. Otherwise, above about 100Hz, they look the same.
Does this matter to the circuit?
Iím going to stop here to let everyone else respond. And then, I guess, you all will tell me what to do. :)
From a technical standpoint, for class A amplifiers, regulation makes no difference. Just as long as the circuit gets the power it needs, it will operate and conditions will not change. This will be true in a line stage which handles small signals, but an output stage (or any stage driven near its maximum capability) will experience a slight increase in current at maximum output, thus requiring regulation. Of course this need is pretty small, and can easily be supplied by a capacitor-input type rectifier, and minimal resistance thereafter (say, choke filtering, no resistors).
For all other classes of operation, current varies widely with signal level (the higher the class, the worse the change), and better regulation is required. For most applications, a cap-input or choke-input circuit will suffice, with advantage being had with silicon diodes (tubes suck at rectification... face it guys :o ). However, a good hi-fi amp will require regulation of at least screen voltage, especially if a large voltage drop is needed, say 200V on the screens, whereas the voltage supply may be 400 or 500V. (Nevermind that if it's all triode..)
Regulating the entire circuit is a waste of time and space, and implies a very badly designed power transformer, rectifier or filter, because they should have sufficient regulation to suit.
That's my technical word on it. We now return you to your regularly scheduled programming of metascience and psychoacoustics... :p
Interesting stuff. Just out of interest, have you evaluated any SS regulators for tube amp power supplies?
I am considering using regulated screens on a 6V6 amp, but don't have room for another tube...
Also, are there any good links to regulator schematics (of all kinds) out there?
My stash of models is empty of high voltage SS regulators. I do have low voltage types that would be used for heater supplies.
However, it would be fairly easy to model BJT series pass regulators of increasing complexities. After I get through with the tube regulators (one more to come), I will give this a try.
I personally am not up to speed on resources for regulator schematics, although there must be a million of them out there. I'm sure the folks in the forum can help here.
SY, Tim, I do realize that I'm pushing the circuit beyond reasonable operating conditions. I just needed something that would exercise the PS a little. Besides, it's a circuit that everyone understands and, if I keep going with this, it will lend itself nicely to the imaginary part of the discussion. ;)
There hasnít been too much response. I guess that this thread is a yawn. :o Nevertheless, Iíll continue a little longer . . . .
We can also use a shunt regulator. Here is a tube shunt regulator using another 6dj8. The idle point is set by the 6.2K resistor and the AC regulation is performed by the tube.
Capacitors are not typically placed across shunt regulators, so for the first simulation Iíll set C=0. Using the same delta-V/delta-I analysis as before, we get this:
This graph shows that, basically, the shunt regulator acts a series resistance of about 100ohms for all of the audio spectrum and beyond. This is twice as much as the series regulator at low frequencies and much more than that at higher frequencies. At 5mA pp, the voltage variation will be about 0.5V pp. Still small compared to the 200VDC at the rail.
What happens if we bypass the regulator with a large cap? As before, C=330u? We get this.
This result is almost identical to the basic RC regulator because the effects of the large capacitor dominate the behavior of the system. Here is the log plot for better clarity.
The regulator adds nothing to the basic RC stage. However, the higher resistance of this regulator when there is not capacitor indicates that the regulator doesnít have enough gain. This regulator could be improved, but this is good enough for this post.
Next will be simple series pass BJT regulator.
|All times are GMT. The time now is 11:28 AM.|
vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2016 DragonByte Technologies Ltd.
Copyright ©1999-2016 diyAudio