Hi, I have been reading conflicting bits and bobs about grid chokes...I have a pair of 640h but at the moment because of space I cannot try it for myself. I have also been reading about how "the best capacitor is no capacitor" and have had the idea of applying the same thoery of grid chokes to replace the output caps on a srpp stage.
As I am just a layman I make the assumption that DC will pass thru a choke BUT AC will not. So if I removed the 10uf output caps and put a choke in parrallel to the hot pin on the output socket surely, as electricity takes the easiest path to ground, the unwanted DC will go to eath and the wanted AC pass thru to the amplifier? I would of created a sort of passive DC filter and would not have to remortgage to get a decent output cap?
As I am just a layman I make the assumption that DC will pass thru a choke BUT AC will not. So if I removed the 10uf output caps and put a choke in parrallel to the hot pin on the output socket surely, as electricity takes the easiest path to ground, the unwanted DC will go to eath and the wanted AC pass thru to the amplifier? I would of created a sort of passive DC filter and would not have to remortgage to get a decent output cap?
The only way to remove caps is to DC couple the circuit, such as by using a bipolar power supply, so the load is returned to the middle voltage, instead of being returned via capacitor to +V or -V.
Fortunately, anyone who tells you "capacitors are bad" is full of it. We are all free to use this wonderful electronic component anywhere it is needed.
Besides, you'll most likely get much worse results using chokes, especially a lot of them. Chokes are awful components. Tons of resistance, capacitance and really hard-to-model effects like distributed capacitance, loss and saturation.
In fact, even the best chokes are worse as inductors, than the worst electrolytics are as capacitors. I would rather use an electrolytic capacitor than a choke!
Tim
Fortunately, anyone who tells you "capacitors are bad" is full of it. We are all free to use this wonderful electronic component anywhere it is needed.
Besides, you'll most likely get much worse results using chokes, especially a lot of them. Chokes are awful components. Tons of resistance, capacitance and really hard-to-model effects like distributed capacitance, loss and saturation.
In fact, even the best chokes are worse as inductors, than the worst electrolytics are as capacitors. I would rather use an electrolytic capacitor than a choke!
Tim
so would you not even recommend to use chokes in series in the B+ to the anode? And if we for a minute ignore the "sound quality": would my reasoning work , that is ,remove the dc? Plus if the choke is in parrallel to the sound signal, surely as the AC does not pass thru it : the chokes quality/capacitance/saturation et al would not affect the AC . Please do not get the wrong end of the stick.I am not asking these questions as argument, I am just trying to learn about audio electronics.I guess if it truely worked every one would be doing it.
chokesrule,
You have got the right idea, but your implementation is all wrong. The capacitor has a very specific function in the circuit you describe. It blocks DC voltage from the driver getting on the following tube's grid. There is no way to get rid of it without making the driver tube's plate supply the same voltage as the grid. The grid choke serves a totally different function. It replaces the grid resistor of the output tube. So you can see two different components two separate functions.
That said, my own personal experience (unlike the poster above I will not make any absolute statements) is that triode output tubes like the 2A3 and 300B pass a lot more grid current than we normally expect and so it is a good thing to have a very low resistance to ground so that the grids do not start to collect electrons and affect the bias. So, that is the function of the grid choke to provide a low resistance to ground for the grid while providing a high impedance to the AC signal voltage. If you have a SET amp I would urge you to try the grid chokes. It will not provide the advantages of direct coupling, but it will provide some advantages over the grid resistor. Eventually, you really need to try direct coupling for the best control of the grid.
You have got the right idea, but your implementation is all wrong. The capacitor has a very specific function in the circuit you describe. It blocks DC voltage from the driver getting on the following tube's grid. There is no way to get rid of it without making the driver tube's plate supply the same voltage as the grid. The grid choke serves a totally different function. It replaces the grid resistor of the output tube. So you can see two different components two separate functions.
That said, my own personal experience (unlike the poster above I will not make any absolute statements) is that triode output tubes like the 2A3 and 300B pass a lot more grid current than we normally expect and so it is a good thing to have a very low resistance to ground so that the grids do not start to collect electrons and affect the bias. So, that is the function of the grid choke to provide a low resistance to ground for the grid while providing a high impedance to the AC signal voltage. If you have a SET amp I would urge you to try the grid chokes. It will not provide the advantages of direct coupling, but it will provide some advantages over the grid resistor. Eventually, you really need to try direct coupling for the best control of the grid.
Last edited:
In a general audibility sense there is no difference between excellent capacitors and excellent transformers. A choke is a transformer, if you use a power smoothing choke it will sound significantly worse than even an average capacitor. If you use a distributed gap, or amorphous core with gap choke, with low DC resistance, thus rather large, it will be equal in sonic quality to the best capacitors available. This is a matter of loses in the working materials.
The same is true in interstage transformers VS capacitors, except that an interstage capacitor will provide a "bloom" for loud sounds, they will seem to move towards you within the illusion of a reproduced sound field. The interstage transformer will not provide this slight dislocation. Otherwise, no audible differences. Expense wise, the capacitor wins, hands down.
The same is true in interstage transformers VS capacitors, except that an interstage capacitor will provide a "bloom" for loud sounds, they will seem to move towards you within the illusion of a reproduced sound field. The interstage transformer will not provide this slight dislocation. Otherwise, no audible differences. Expense wise, the capacitor wins, hands down.
I have had excellent results using followers to drive output tubes. Apply bias to the grid (or gate) of the follower and DC couple to the output tube. Sounds awesome and can get more power out due to positive grid drive potential. Getting the coupling cap out of there is a very good idea as it eliminates the possibility of blocking distortion (bias shift due grid current). I was surprised to see music waveforms from my own push-pull amp with severe clipping that did not sound nearly as bad as I would have expected. Instantaneous recovery from overload is a very neat feature to have in an amplifier.
I typically use like 50k or less load resistance for the follower so I am in very good control of tube bias even on tubes with less than perfect vacuum.
I have never used a grid choke. How high is reactance at 20Hz in a typical circuit?
I typically use like 50k or less load resistance for the follower so I am in very good control of tube bias even on tubes with less than perfect vacuum.
I have never used a grid choke. How high is reactance at 20Hz in a typical circuit?
You are just a layman, so surely it would be better to ask questions rather than make assertions?
The problem is that it's easy in electronics to run foul of the law of unintended consequences. Very often the unintended consequence is a big flash, a bang and the release of some magic smoke, and this is almost certainly what will happen if you take your own advice in this case. Your reasoning is not wrong, it's just that the DC is supposed to be there, that's why the cap is there, to allow the DC to be there without passing into the next stage or anywhere else. A choke has a low DC resistance, so if you connect one as you are suggesting then although the current through it may be small at first (due to its inductance) it may rise to a value sufficient to vapourise the part of the circuit it is flowing in, which will certainly not continue to work as intended with the choke in place, even if it survives.
Chokes are common (as are output transformers) in valve circuits where it's hard to do without them. They're common in RF. As Sch3mat1c says, however, they're a PITA, so we try to avoid using them if at all possible.
Capacitors are common in all kinds of circuits, they're a PITA (not as much as chokes, however) so we try to avoid using them if at all possible, but it's hard to do without them.
Don't get caught up in 'modification fever'. You need instruments and the wit to use them otherwise you run the risk of making things worse even though you have convinced yourself the sound is better.
It's worth remembering that even after 3 years university education the very large majority of EE graduates are incapable of building even a single-stage amplifier. Perhaps one in a hundred will become so if they are vitally interested
w
If the signal quality matters, then no, I wouldn't.
Fortunately, we are only interested in destroying *any* signal on B+, so the quality doesn't matter very much.
Consider the effects of parallel capacitance on the choke: this makes a capacitive voltage divider against the following filter cap. Even a simple choke won't have more than a few nanofarads parasitic capacitance, so the effect versus the ~100uF filter cap is still negligible.
This is still true when we consider the loss resistances (which are in parallel with the inductor, and in series with the capacitor (ESR)). In that case, the cap's ESR might be a few ohms, which still pales in comparison to the >100 ohms of the inductor or its loss component. (Even a very lossy inductor, with a Q of only 5, is still a fair filter, though a Q truely this abysmal would make one question why use an inductor at all. Real inductors are usually Q > 10, so it's not too bad.)
Not completely, and definitely not in any useful way. Consider the circuit you've built: you have a series choke (or resistor or whatever) from B+, then it's shorted to ground through this grid inductor. That leaves no room for the proceeding tube to work at (its plate voltage is around 0V!), so you've just broken your amp.
It's not completely true, because practical inductors* have nonzero resistance, and grid chokes have far more than most. That 640H choke of yours might be several kilohms! It starts looking like a pretty good resistor! The deceptive part now becomes: maybe your circuit still works, but with way more distortion and less gain than would justify using so many chokes, and now you've half shorted out the power supply and the whole thing is cooking itself to pieces!
*Practical, not "real". There are real inductors with zero resistance -- using superconductors. But they aren't practical, at least not for a tube amp!
Ah, but AC does pass through it! The signal gets shorted to ground. So actually, it's *stealing* your signal, which reduces gain, power output and increases distortion.
The worst sort of load is a resistor, which steals fully half of your signal. This is precisely why a class A, resistor-coupled amplifier can only deliver a power output of 25% the total static dissipation: half the signal disappears in the resistor! Replace the resistor with an inductor and you can suddenly get 50% efficiency. (Typical class AB amplifiers are in the 40-60% range.)
As loads go, you have three choices: resistors, inductors or CCSs.
Resistors are good because they are cheap, and very, very well characterised. A resistor is a resistor is a resistor, from DC to MHz and beyond. There isn't a single inductor that has the same dynamic range!
Small resistors are essentially ideal up to 1GHz. Carbon composition resistors are known to be noisy around small signals, and nonlinear around high voltages, so don't use them in phono stages or around high voltages (typically >300V drop). Most small resistor types look like very small inductors up in the GHz range; carbon comp are an exception, and are slightly capacitive. Needless to say, this has no effect whatsoever on audio frequencies.
Power resistors are most often wirewound, which obviously adds a bit of inductance. This is important as low as 100kHz for inductive wirewounds. Conclusion: buy noninductive resistors for load resistors when testing amplifiers. Noninductive resistors are good into the MHz range, which is more than enough for tube amps.
Disadvantages are as mentioned: they consume power! This isn't a big deal for small-signal stages, where you aren't interested in power, only voltage. There are only a few unusual circumstances where resistors cannot be used, like high voltage swings for a given supply voltage (a 6AS7 needs ~300Vpp grid drive, which suggests >400V supply -- an inductor would allow you to use only a 200V supply, which might be the same supply on the 6AS7 -- much easier to design).
Inductors are theoretically better than resistors and CCSs, because they consume "zero" power. A theoretical inductor has zero DCR, so it wastes no DC, and has no AC losses, so it wastes no signal. Unfortunately, real inductors aren't as nice, as mentioned.
CCSs drop DC voltage, so they consume power; however, they perform better than inductors over all frequencies (a CCS works at DC, an inductor does not!). CCSs have better AC performance than real inductors because they do not posess all those icky characteristics of resistance, capacitance and distortion. Needless to say, they also perform better than resistors, except in one matter, which is power consumption. A stage biased at a given voltage and current will dissipate the same power whether using CCS or resistor, but it will perform better with the CCS, because it doesn't have a resistor sucking away half the output power.
No problem, these are good questions for a beginner. Keep them coming.
Tim
Oops, I missed that you gave the inductance value of your chokes in your original post. It looks like that gives about 80k reactance at 20Hz. Not too terribly bad. I still like putting the follower in there a lot better, though. Then you can bias the follower with a large resistor (1M or more) which presents a very light AC load to the previous stage. 1M is a lot easier to drive than 80k.
I personally prefer to use CCS loads over choke loads for voltage amplification stages simply because they work so well from DC to well above audio frequencies. I think that increased power dissipation from a voltage amplification stage is a small price to pay for such good performance. Others disagree. Others apparently don't mind increased distortion and decreased gain at low frequencies for practical chokes.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.