Hi Chris,
I agree with the proposed method even though a RC network across each diode is the most common proven and tried method, however it's not the diode's speed that's the problem_ after all you're just handling 50/60Hz which is sloooow_but the way the diode recovers from sudden switch off.
Unless you purposely bandwidth limit the reg_ easy enough with a classic tubed series reg_ you may, most likely will_ spot traces of HF hash on the scope with almost any untuned solid state series reg IME.
Fair enough, in practice you're not likely to pull the full amount of current from the rectifier with a tubetester but my point was that
from what I read in the manuals this 20' warm-up period is just a break-in cycle.
After that, 30 secs should do or that's what I keep hearing from the guys using these anyway.
Cheers,
I agree with the proposed method even though a RC network across each diode is the most common proven and tried method, however it's not the diode's speed that's the problem_ after all you're just handling 50/60Hz which is sloooow_but the way the diode recovers from sudden switch off.
Unless you purposely bandwidth limit the reg_ easy enough with a classic tubed series reg_ you may, most likely will_ spot traces of HF hash on the scope with almost any untuned solid state series reg IME.
Fair enough, in practice you're not likely to pull the full amount of current from the rectifier with a tubetester but my point was that
from what I read in the manuals this 20' warm-up period is just a break-in cycle.
After that, 30 secs should do or that's what I keep hearing from the guys using these anyway.
Cheers,
FYI, series resistance will only kill regulation. In a series circuit, such as the winding, rectifier and cap, additional resistance simply adds to the ON resistance, winding resistance (major) and cap ESR (and ESL being a dynamic situation). However, a capacitor across the diode, or better yet, a snubber of some sort will *kill* it. 0.01uF should be enough; too much and you'll waste VARs, heating the transformer.
Tim
Tim
Hi Tim,
By using a resistance lower than the normal tube rectifier resistance you will have better performance to begin with. The idea is to reduce the surge current and lower the peak charging current. Now, even if you use an equivalent resistance to the tube, your voltage is still higher and regulation is the same. This is without the heater current so the transformer will run cooler. The series ESR of the capacitor will be more pronounced with higher peak currents, but in itself does not get any worse with higher series resistance in the voltage source. When looking at values on the order of 47 - 100 uF this is normally not an issue. Higher capacitance will be very hard on a tube as you most probably will exceed the hot switching current rating.
You are, of course correct when you say that small caps in parallel with the diode kill the switching noise.
-Chris
By using a resistance lower than the normal tube rectifier resistance you will have better performance to begin with. The idea is to reduce the surge current and lower the peak charging current. Now, even if you use an equivalent resistance to the tube, your voltage is still higher and regulation is the same. This is without the heater current so the transformer will run cooler. The series ESR of the capacitor will be more pronounced with higher peak currents, but in itself does not get any worse with higher series resistance in the voltage source. When looking at values on the order of 47 - 100 uF this is normally not an issue. Higher capacitance will be very hard on a tube as you most probably will exceed the hot switching current rating.
You are, of course correct when you say that small caps in parallel with the diode kill the switching noise.
-Chris
anatech said:
This never was, and never is an issue (those SS dudes with their hideous cap input filters are testament); if it ever were, you'd simply use a smaller C1 and more LC filtering after (i.e., pi filter).
Well, with a tube rectifier, peak currents have to be limited. But those are tweaked by capacitance, not resistance, so it doesn't fit the context here.
Er... if you add resistance to beef up silicon to the same effective resistance a tube diode presents, the regulation and voltage will be the same as with the tube.
Well, not exactly because you have a linear rather than 3/2 power law characteristic, but close enough, certainly within range of a certain load current.
Tim
I'm not seeing it.
So, did anyone actually say what the difference between SS and tube rectification is?
I mean if you bandwidth limit the reg., no HF hash, then that would lead you to logically conclude that, the SS reg and the Tube reg now have the same characteristics..... somehow I don't think it's as simple as this.
Perhaps it has something to do with the inherent qualities of SS and tubes reg. I.e. Exponential vs. 3/2 power.
If this is true, then why?
David ... who really wishes someone one use a high powered oscilliscope to make some data known.
So, did anyone actually say what the difference between SS and tube rectification is?
I mean if you bandwidth limit the reg., no HF hash, then that would lead you to logically conclude that, the SS reg and the Tube reg now have the same characteristics..... somehow I don't think it's as simple as this.
Perhaps it has something to do with the inherent qualities of SS and tubes reg. I.e. Exponential vs. 3/2 power.
If this is true, then why?
David ... who really wishes someone one use a high powered oscilliscope to make some data known.
Tubes have inherent resistance (somewhat variable due to 3/2 power) whereas diodes, due to their exponential "turn-on" characteristic, drop no more than 1.5V peak. The majority of resistance in a tube rectifier is in the tube itself, whereas for SS, the diode resistance is so diminishingly small that the resistance of the winding and to a lesser extent, the capacitor, is greater. For example, you might read 10Vrms drop on a winding of say 200 ohms (100mA DC output) and 300VAC output for both tube and SS rectifiers. However, the output voltage for the tube might be 320V for a 5Y3, whereas the SS might read 400V. The 10Vrms drop corresponds to the RMS voltage generated across the winding due to charging current (V = IR). Since you can assume a 1V drop for the diode when turned on, the peak of the sine wave input is clipped to 401V, instead of the real 424.26V peak. (If anyone does measure or simulate these figures, remember I'm just SWAG'ing ballpark figures here.)
Since the SS has a full-load voltage much closer to actual peak voltage, regulation is *very* good. 20V good in this case. Whereas the tube drops 100V from peak, which if left unloaded, it will attain. (I.e., any load causes a DC output voltage less than sqrt(2) * VAC.)
Tim
Since the SS has a full-load voltage much closer to actual peak voltage, regulation is *very* good. 20V good in this case. Whereas the tube drops 100V from peak, which if left unloaded, it will attain. (I.e., any load causes a DC output voltage less than sqrt(2) * VAC.)
Tim
still missing the point
Soo, in this particular aspect, the 3/2 vs. the exponential resistences are what define the rectifier behavior then.
The majority of the posts are hinting at or stating outright that tubes have this quality about them that, while hard to explain in words, is very noticable. While I'm not disputing this, I just want to know what causes this proclaimed effect.
If you did answer my questioin, I'm sorry I'm being dense. The only thing I could come up with, from your explanation, is that the higher resistence and corresponding voltage drop creates a type of DC feedback which would cut down on the hash, ripples, or whatnot.
Would a tube rectifier be Voltage, Current, or Power constant?
The thing I'm struggling with is that I find it hard to believe that it is impossible to get the same effect from SS regulation.
David
P.S. Hell I'd go through the trouble just for that sexy purple glow
Soo, in this particular aspect, the 3/2 vs. the exponential resistences are what define the rectifier behavior then.
The majority of the posts are hinting at or stating outright that tubes have this quality about them that, while hard to explain in words, is very noticable. While I'm not disputing this, I just want to know what causes this proclaimed effect.
If you did answer my questioin, I'm sorry I'm being dense. The only thing I could come up with, from your explanation, is that the higher resistence and corresponding voltage drop creates a type of DC feedback which would cut down on the hash, ripples, or whatnot.
Would a tube rectifier be Voltage, Current, or Power constant?
The thing I'm struggling with is that I find it hard to believe that it is impossible to get the same effect from SS regulation.
David
P.S. Hell I'd go through the trouble just for that sexy purple glow
I think the main difference is, (as have been mentioned earlier) that a tube rectifier doesn't have any reverse conduction as ordinary SS diodes have and therefore there will be less RF hash produced. My experience is that it is possible to remove the RF hash by using snubbers over each SS diode, but
that these snubbers are necessary, (it also depends on the bandwidth of your amp, with a wide band OTL amp the problem is easier to notice)
Another difference is of course the higher resistance in tube rectifiers which limits the current peaks which is also helpful in order to eliminate noise, using a rersistor in series with each SS diode limits the current pulses.
Regarding measurements on RF hash from SS diodes, I found this http://www.hagtech.com/pdf/snubber.pdf , the problem with making measurements is that you need a current probe which is not what most DIYers have at home, however to see that there is some RF hash is easy with an ordinary oscilloscope.
Regards Hans
Hi all,
Philips PM3070 good enough? Had a Tek with current probe too. After much looking & tinkering I find that the biggest fault with SS rectifiers are the "pips" on the DC waveform and distorted AC waveform (more noise). When designing a low noise supply, you must look at the application and all circuits together. A low current supply with many 1,000 uF will allways sound bad. To reduce noise, try not to make noise to begin with.
That means that you increase the resistance in the leads of the diodes to cut the peak current and di / dt. You place snubbers across each diode to damp transients. You use a low noise regulator to isolate the raw DC from the circuit. Three terminal regulators don't "cut it". And use realistically sized capacitors for the application. To much capacitance is a bad thing, just as too little capacitance is ineffective.
High current supplies are a different animal.
At this point, improvement in all types of high voltage supplies have been found with good solid state regulators. Gas discharge tubes are too noisy (neons as well). That means filtered zeners. Transistors can be quieter than tubes, so why not? After a good regulator, tube or SS rectifiers can not be heard. Not unless you have other problems which need to be solved as well.
-Chris
Philips PM3070 good enough? Had a Tek with current probe too. After much looking & tinkering I find that the biggest fault with SS rectifiers are the "pips" on the DC waveform and distorted AC waveform (more noise). When designing a low noise supply, you must look at the application and all circuits together. A low current supply with many 1,000 uF will allways sound bad. To reduce noise, try not to make noise to begin with.
That means that you increase the resistance in the leads of the diodes to cut the peak current and di / dt. You place snubbers across each diode to damp transients. You use a low noise regulator to isolate the raw DC from the circuit. Three terminal regulators don't "cut it". And use realistically sized capacitors for the application. To much capacitance is a bad thing, just as too little capacitance is ineffective.
High current supplies are a different animal.
At this point, improvement in all types of high voltage supplies have been found with good solid state regulators. Gas discharge tubes are too noisy (neons as well). That means filtered zeners. Transistors can be quieter than tubes, so why not? After a good regulator, tube or SS rectifiers can not be heard. Not unless you have other problems which need to be solved as well.
-Chris
Hi Tim,
There are many tube rectifier circuits with series resistance to limit the peak current.
Anyhow, all this is within the context here. I see the biggest problem in mis-application of capacitance. How many tweaks do you see that are mainly an increase in supply capacitance? This puts the peak currents well above recommended levels. In properly designed supplies I don't see much difference in sound quality unless you "crank it". There will be more voltage sag with the tube units over SS with resistance and snubbers. There will be a difference in sound that may be audible. Unpleasant? Up to you.
If you have time, remove the large dropping resistor in a push pull amp. Something around 30 - 50W is good. Replace it with a SS regulator in a classic zener on emitter, error voltage to base, collector to pass transistor setup. Take the base drive for the pass transistor from an LED constant current source and filter the rest in good practice. Listen to the noise floor drop and the sound remain clear at high power levels. The voltage amp stages are now getting clean DC unaffected by the main supply. If you had a tube rectifier, try now setting up a SS compensated rectifier (use a standby switch). I'm not saying there won't be any difference, but most of the nasties should be gone (if not all).
-Chris
There are many tube rectifier circuits with series resistance to limit the peak current.
Anyhow, all this is within the context here. I see the biggest problem in mis-application of capacitance. How many tweaks do you see that are mainly an increase in supply capacitance? This puts the peak currents well above recommended levels. In properly designed supplies I don't see much difference in sound quality unless you "crank it". There will be more voltage sag with the tube units over SS with resistance and snubbers. There will be a difference in sound that may be audible. Unpleasant? Up to you.
If you have time, remove the large dropping resistor in a push pull amp. Something around 30 - 50W is good. Replace it with a SS regulator in a classic zener on emitter, error voltage to base, collector to pass transistor setup. Take the base drive for the pass transistor from an LED constant current source and filter the rest in good practice. Listen to the noise floor drop and the sound remain clear at high power levels. The voltage amp stages are now getting clean DC unaffected by the main supply. If you had a tube rectifier, try now setting up a SS compensated rectifier (use a standby switch). I'm not saying there won't be any difference, but most of the nasties should be gone (if not all).
-Chris
Hi,
I don't think so.
In fact I'd prefer a glow discharge over a zener, or string of zeners, for any application requiring highish voltages.
The noise from a glowdischarge tube is far more broadband_less peaky_ and much less obtrusive than that from a zener.
Moreover these tubes don't suffer the tempco problems most zeners do either.
As for the SS diodes: yes, putting the correct snubber network across them does help reducing switching transients and RF noise considerably. A series resistor before and after the diodes may help kill some more misbehaviour but most will agree that the best sounding rectifiers are damping diodes and the now trendy mercury vapour types.
The damper diodes offer much lower voltage drop due to their low internal resistance and don't nearly pollute the circuit with RF spuriae as much as their SS counterparts.
As you put it yourself, no need to filter out noise if you don't introduce it in the first place.
For sensitive circuitry such as phonostages and mic preamps you don't need any current to speak of, so why not opt for what does the better job in the first place instead of applying bandaid upon bandaid?
You could use Schottky diodes for the LT circuits such as bias and heaters, tubes for anything HT.
A decent PS choke does a very good job at isolating the AC part of the PS, add a well design tube regulator where stability is required and you're pretty close to simulating a battery PS without its practical disadvantages.
Using an isolation xformer is certainly a very good idea, the endresult will surely depend on the quality of the xformer(s) themselves...Quite another topic that's being overlooked. Regretably so.
Designing high performance tube circuits while sticking to engineering theories based on published specsheet figures will rarely ever yield an outstandingly good sounding design.
Engineering for audio can only be learned through years of trial and error and a will to listen to the results, not just by looking at an o-scope IMHO.
Cheers,
I don't think so.
In fact I'd prefer a glow discharge over a zener, or string of zeners, for any application requiring highish voltages.
The noise from a glowdischarge tube is far more broadband_less peaky_ and much less obtrusive than that from a zener.
Moreover these tubes don't suffer the tempco problems most zeners do either.
As for the SS diodes: yes, putting the correct snubber network across them does help reducing switching transients and RF noise considerably. A series resistor before and after the diodes may help kill some more misbehaviour but most will agree that the best sounding rectifiers are damping diodes and the now trendy mercury vapour types.
The damper diodes offer much lower voltage drop due to their low internal resistance and don't nearly pollute the circuit with RF spuriae as much as their SS counterparts.
As you put it yourself, no need to filter out noise if you don't introduce it in the first place.
For sensitive circuitry such as phonostages and mic preamps you don't need any current to speak of, so why not opt for what does the better job in the first place instead of applying bandaid upon bandaid?
You could use Schottky diodes for the LT circuits such as bias and heaters, tubes for anything HT.
A decent PS choke does a very good job at isolating the AC part of the PS, add a well design tube regulator where stability is required and you're pretty close to simulating a battery PS without its practical disadvantages.
Using an isolation xformer is certainly a very good idea, the endresult will surely depend on the quality of the xformer(s) themselves...Quite another topic that's being overlooked. Regretably so.
Designing high performance tube circuits while sticking to engineering theories based on published specsheet figures will rarely ever yield an outstandingly good sounding design.
Engineering for audio can only be learned through years of trial and error and a will to listen to the results, not just by looking at an o-scope IMHO.
Cheers,
Hi,
Do you mean the resistors in series with the plate circuit?
If so, these are only there to limit inrush current and won't affect current capability of the PS proper.
Perhaps so for class AB1 or class B designs but that's only a sign of bad PS design and often done on purpose in guitar amps...
With a class A amp such as the currently fashionable SE miniwatt amps this issue is rather moot.
Provided the PS is well designed, peak current demand should be met by the reservoir capacitance, not being drawn straight from the mains socket so to speak.
Cheers,
Do you mean the resistors in series with the plate circuit?
If so, these are only there to limit inrush current and won't affect current capability of the PS proper.
Perhaps so for class AB1 or class B designs but that's only a sign of bad PS design and often done on purpose in guitar amps...
With a class A amp such as the currently fashionable SE miniwatt amps this issue is rather moot.
Provided the PS is well designed, peak current demand should be met by the reservoir capacitance, not being drawn straight from the mains socket so to speak.
Cheers,
Hi Frank,
I listen to the supplies too. I use a blocking cap to the input of whatever amp is a bench amp at the time. I'm looking for obvious noise. Very revealing.
I generally agree with you, but get there my way. No bandages in my stuff. A great deal depends on circuit layout and ground paths as well. Any topology depends on the execution of the design.
The one thing I like about VR tubes is the gentle drift, rather than the positive tempco. The thermistors go a 10V drift at 275VDC down to 2V. This I can live with. I've tried 100V zeners and 33V zener stacks. The temperature drift is the same. The zener is bypassed, as is the positive voltage sense resistor (RC).
I design both ways but I've had better success with SS regulation, both specs and sound. The way I see it, I'm attempting to apply the best devices to a particular job, whatever the devices may be. Keeping an open mind has paid off well for me.
-Chris
I listen to the supplies too. I use a blocking cap to the input of whatever amp is a bench amp at the time. I'm looking for obvious noise. Very revealing.
I generally agree with you, but get there my way. No bandages in my stuff. A great deal depends on circuit layout and ground paths as well. Any topology depends on the execution of the design.
The one thing I like about VR tubes is the gentle drift, rather than the positive tempco. The thermistors go a 10V drift at 275VDC down to 2V. This I can live with. I've tried 100V zeners and 33V zener stacks. The temperature drift is the same. The zener is bypassed, as is the positive voltage sense resistor (RC).
I design both ways but I've had better success with SS regulation, both specs and sound. The way I see it, I'm attempting to apply the best devices to a particular job, whatever the devices may be. Keeping an open mind has paid off well for me.
-Chris
Hi Frank,
Just got your last post. I'll count myself out of the single ended discussion. Class AB and class A (not classic) are my areas of interest. The exception is a voltage gain stage with tubes.
Voltage sag in a power AB or B design is a fact of life to some degree. The current variation forces this to occur unless you install a power regulator as well. I'd rather have the extra voltage and isolate the voltage amp stages to minimise the effects of this.
I shutter to think how much power I'd require dissipated to run PSB Stratus Gold's, never mind more power hungry speakers that exist, in a class A tube amp design. A little 35W/ch tube amp does okay here. Class AB.
Some tube rectifiers I have seen use 500 ohm per plate. That matters. I guess I'm just pointing out some bad designs patched to hang together. (This particular unit is a preamp, uses two cetrons per channel that occassionally arc - designer on drugs I think) Anyway, I was referring to the hot switching current limiting, as you pointed out.
Just got your last post. I'll count myself out of the single ended discussion. Class AB and class A (not classic) are my areas of interest. The exception is a voltage gain stage with tubes.
Voltage sag in a power AB or B design is a fact of life to some degree. The current variation forces this to occur unless you install a power regulator as well. I'd rather have the extra voltage and isolate the voltage amp stages to minimise the effects of this.
I shutter to think how much power I'd require dissipated to run PSB Stratus Gold's, never mind more power hungry speakers that exist, in a class A tube amp design. A little 35W/ch tube amp does okay here. Class AB.
Some tube rectifiers I have seen use 500 ohm per plate. That matters. I guess I'm just pointing out some bad designs patched to hang together. (This particular unit is a preamp, uses two cetrons per channel that occassionally arc - designer on drugs I think) Anyway, I was referring to the hot switching current limiting, as you pointed out.
Hi Chris,
Sure, that's one of the paradoxes of audio: linear speakers (not that they really exist) versus high effieciency at the expense of BW extension and perhaps some other oddities.
However, if you can live with a 35W class AB amp_I assume it's a PP design_ you could certainly do away with the crossover distortion these AB amps invariably run into and enjoy the effortless power of a true class A amp.
The difference in mere musicality between the two can be amazing as I'm sure you're aware of.
Naturally, there are good and not so good designs of either conviction but all else kept equal, the class A amps, more often than not, are the ones that are the more respectful of that ultimate illusion called music IME.
Yep...470R is a commonly used value.
Hmmm...What type of tube are those Cetrons and what's the current draw of the cct?
Cheers,
Sure, that's one of the paradoxes of audio: linear speakers (not that they really exist) versus high effieciency at the expense of BW extension and perhaps some other oddities.
However, if you can live with a 35W class AB amp_I assume it's a PP design_ you could certainly do away with the crossover distortion these AB amps invariably run into and enjoy the effortless power of a true class A amp.
The difference in mere musicality between the two can be amazing as I'm sure you're aware of.
Naturally, there are good and not so good designs of either conviction but all else kept equal, the class A amps, more often than not, are the ones that are the more respectful of that ultimate illusion called music IME.
Yep...470R is a commonly used value.
Hmmm...What type of tube are those Cetrons and what's the current draw of the cct?
Cheers,
Although a little bit late...!
follow this link where my friend Aristidis describes his amp (using stand by - soft start)
http://aca.gr/paper31_build.htm
follow this link where my friend Aristidis describes his amp (using stand by - soft start)
http://aca.gr/paper31_build.htm
Hi Frank,
A Cetron is a make or brand of tube. The rectifiers in this case are 5U4GB's on steriods. 5U4's explode in this silly preamp.
The preamp in question is a Michael Elliot / Counterpoint thing. It has two outboard heavy power supplies (one per channel). These ultimately run 3 6DJ8's per channel. He uses 5881's as shunt regulators, 'cause these are faster.xeye: ) Every heater is shunt regulated as well. This thing gives off more heat than you can imagine. I forget the model, but the description will be enough.
Another one of Counterpoints monsters use the 500R plate resistors. 20W jobs that mount on the chassis with grease. When this preamp goes, so do those resistors sometimes. They run very warm normally.
These are my examples of very poor supply design.
Class A designs can be good. I like low feedback AB tube or transistor designs. They don't sound "squashed" either, very musical.
Thanks Frank -Chris
A Cetron is a make or brand of tube. The rectifiers in this case are 5U4GB's on steriods. 5U4's explode in this silly preamp.
The preamp in question is a Michael Elliot / Counterpoint thing. It has two outboard heavy power supplies (one per channel). These ultimately run 3 6DJ8's per channel. He uses 5881's as shunt regulators, 'cause these are faster.
xeye: ) Every heater is shunt regulated as well. This thing gives off more heat than you can imagine. I forget the model, but the description will be enough.Another one of Counterpoints monsters use the 500R plate resistors. 20W jobs that mount on the chassis with grease. When this preamp goes, so do those resistors sometimes. They run very warm normally.
These are my examples of very poor supply design.
Class A designs can be good. I like low feedback AB tube or transistor designs. They don't sound "squashed" either, very musical.
Thanks Frank -Chris
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.