Home |
Search |
Today's Posts |
#1
|
|||
|
|||
"The Audibility of Power Supplies"
Hi guys,
Sometime ago I was looking at the Curcio web page concerning "The Audibility of Power Supplies". I was also in the process of restoring my Sherwood S8000 where I made a few posts on this group. My poor Sherwood sure had lots of power supply ripple, and regulation issues that I did not like at all. I decided to do something about the power supplies in my Sherwood. I was reading the White paper by Joe Curcio, and although I dare not disagree, I cannot understand the need to have a tight regulation loop with DC to 1 Mhz bandwidth. I must be missing something. I am more comfortable (maybe because it is easier) with a regulator that has DC to several kilohertz bandwidth, and use sufficient power supply decoupling capacitors that kick in to stiffen the buss before the regulator runs out of bandwidth. That way the regulator does not need to be ultra sophisticated, and the bus is clean and pure from DC to beyond the audio range. I also have a hard time reasoning how to make a tight wide bandwidth regulator that can only source current to a buss that is lightly decoupled with capacitors. Wouldn't this require a totem-pole design that can both source, and sink to maintain rs=0? Anyway, here is Joe's white paper that got me going on this: http://www.curcioaudio.com/Audibility_of_PwrSupply.htm My Sherwood has been evolving now for several months. The power supply is about done. A re-cap of all those black "Good-All" audio coupling capacitors is next. I was also thinking of beefing up the 7868 audio drive by replacing the 12AX7 driver (1/2 voltage amplifier, other 1/2 a hot cathode phase inverter) with a 6EZ8. This gives me three triodes (like 1.5 12AT7's) in a 9 pin bulb. That certainly opens up a few options. I have three regulated power supplies in my Sherwood now. The regulated screens for the 7868's are within 1/4 volt p-p of +375 DC when driving the tar out of the amplifier with all sorts of program material. It's the prolonged bass notes from rap music (I hate listening to that stuff, but good for testing your amp!) that causes the worst dip in voltage (1/4 volt), and that is because the voltage feeding the regulator is also dipping to sustain a full power 5 second 20 hertz bass note. Why do I need a active regulator with 1 Mhz bandwidth? Below is a link to my circuits, and a brief description. These circuits are my doing, and the product of lots of bread boarding, a few arc's, a couple wisps of smoke, and a shock or two. It sure was fun though! http://pages.prodigy.net/jcandela/Sh.../regulator.htm I welcome your comments. Jim Candela |
#2
|
|||
|
|||
Jim Candela wrote: Hi guys, Sometime ago I was looking at the Curcio web page concerning "The Audibility of Power Supplies". I was also in the process of restoring my Sherwood S8000 where I made a few posts on this group. My poor Sherwood sure had lots of power supply ripple, and regulation issues that I did not like at all. I decided to do something about the power supplies in my Sherwood. I was reading the White paper by Joe Curcio, and although I dare not disagree, I cannot understand the need to have a tight regulation loop with DC to 1 Mhz bandwidth. I must be missing something. I am more comfortable (maybe because it is easier) with a regulator that has DC to several kilohertz bandwidth, and use sufficient power supply decoupling capacitors that kick in to stiffen the buss before the regulator runs out of bandwidth. That way the regulator does not need to be ultra sophisticated, and the bus is clean and pure from DC to beyond the audio range. I also have a hard time reasoning how to make a tight wide bandwidth regulator that can only source current to a buss that is lightly decoupled with capacitors. Wouldn't this require a totem-pole design that can both source, and sink to maintain rs=0? Anyway, here is Joe's white paper that got me going on this: http://www.curcioaudio.com/Audibility_of_PwrSupply.htm My Sherwood has been evolving now for several months. The power supply is about done. A re-cap of all those black "Good-All" audio coupling capacitors is next. I was also thinking of beefing up the 7868 audio drive by replacing the 12AX7 driver (1/2 voltage amplifier, other 1/2 a hot cathode phase inverter) with a 6EZ8. This gives me three triodes (like 1.5 12AT7's) in a 9 pin bulb. That certainly opens up a few options. I have three regulated power supplies in my Sherwood now. The regulated screens for the 7868's are within 1/4 volt p-p of +375 DC when driving the tar out of the amplifier with all sorts of program material. It's the prolonged bass notes from rap music (I hate listening to that stuff, but good for testing your amp!) that causes the worst dip in voltage (1/4 volt), and that is because the voltage feeding the regulator is also dipping to sustain a full power 5 second 20 hertz bass note. Why do I need a active regulator with 1 Mhz bandwidth? Below is a link to my circuits, and a brief description. These circuits are my doing, and the product of lots of bread boarding, a few arc's, a couple wisps of smoke, and a shock or two. It sure was fun though! http://pages.prodigy.net/jcandela/Sh.../regulator.htm I welcome your comments. Jim Candela The audibility of PS depends on how much PS noise makes it into the audio output signal as part of the noise and distortion of the amp. If the PS causes noise and IMD, then you will hear that once its above a certain threshold. It is child's play to make decent linear PS with CLC filters for tube amps, where the output signal might have only 1% of its N&D measurement due to PS operation. Quad II has one of the worst PS of any commercial amplifiers ever made, with just a 16 uF cap to anchor the CT of the OPT, which is then charged by the rectifiers. Yet because of common mode rejection in the output stage, and class A operation, Quad II has very reasonable N&D measurements by standards of the era, and at low levels there is no audible PS hum. If we build an amp with 17mV of hum at the OPT CT instead of Quad's 17vrms, then it is good practice, and the amp will work into class AB with a low value RL with less N&D. When one tube is turned on, and the other off in the output stage, then the ripple voltage and signal are in series across the 1/2 primary of the OPT, and thus noise enters the audio path. Using a CLC filtered supply with C2 = 470 uF, the ripple is reduced to a tiny amount, and hence PS caused N&D. But also the IMD from frequencies within the amp caused by inadquate anchoring of the CT is also avoided; PS design isn't just about rectifiers and noise. Worst case for noise is an SET amp where RL is in series with a low Ra triode. With say a KT88 in triode, RL = 5k and Ra = 1k, so if there is 1 vrms of noise at the supply, 0.167mA of noise current flows, which is 0.83 volts across the OPT primary, and this might be reduced x 1/25 by the OPT to 0.033 vrms at the output. NFB of 12 dB would reduce this to 8 mV, but obviously is quite excessive and unacceptable. With noise in the same SET at say 10mV at the OPT, the noise at the output after FB would be 0.08 mV, which might be quite acceptable, and well below N&D from other regions. A 470 uF cap has 3.4 ohms impedance at 100 Hz. If some LF signal current in the3 SET was quite high, then the LFv signal vltage appearing at the 470 uF is going to be low, and the capacitor contribution to IMD caused by alteration to instantaneous supply voltage changes is going to be quite negligible, and less than that caused to HF due to changes in gain caused by the LF signal present. So any better regulation benefits are not going to add up to anything worth the trouble of a regulator. Regulators might have their place in some fancy shmancy gear where the mains supply is likely vary widely, of there is need to have DV stability, right down to 0Hz. I have one amp fitted with a solid state plate supply regulator, and although it has been reliable, I have always gone to plain CLC filters for power amps, knowing that whatever noise makes it to the output, in terms of measurable N&D, it is of such a trifling quantity that it ain't worth a worry. The amp with a plate supply regulator sounds no better. Relying on large value electros is to me a valid way to make low impedance power supplies. The 34 ohms of impedance that a 470 uF cap has at 10 Hz is very small compared to a typical 5k anode load which might be connected, so with 100vrms at 10 Hz at an anode at one end of the RL, the supply voltage change at 10 Hz is only 1.7 volts, and I leave you gentlemen to work out what IMD would be caused solely by this situation if a HF signal of 25vrms was also present, and the tube was a KT88 in triode. Or, put another way, what difference to the N&D would occur if the supply was from a regulated supply, whose Ro was say 3.4 ohms at 10 Hz instead of 34 ohms? I do not doubt your regulator circuits work as proposed. I find simpler emitter follower types using BU208 and a darlington connected smaller bjt to be more reliable than any mosfets. With preamps, I am quite happy with CLCRCRC filters. Patrick Turner. |
#3
|
|||
|
|||
Hi RATs!
What is noise? Yes, the AC 50/60 Hz is not music. Neither is the jibberish from your neighbors nuke box, so filtering is a good plan. Please understand that fabulous amplifiers are just products, with costs and benefits. After the filter, each amplifier stage uses the power on the signal. The signal is inverted at each stage, and the signal is not exactly the same in both channels. Well, OK, mono should be ... Anyway, whatever a stage does with the power is noise to all the other stages. And all the other stages in turn are noise sources to this one. I have done a couple of things which improve the sound I hear: Use stand alone filament supplies. The signal/noise from each stage does not need to be transmitted back into the B+ PS. Filament windings on the B+ power transformer are cheap, and they work OK, but OK is not what I am after My B+ has a single C1, first cap after the rectifier tube. I use this to set the desired B+ voltage. With 6F6G, my Now Playing, I am using 0.5 uF. C1 feeds four parallel L-C-L-C filters, one for each stage. The driver stages have different values than the power stages. Any stage is C-L-C-L-C-L-C-L-C from any other stage. Sounds OK Happy Ears! Al Alan J. Marcy Phoenix, AZ PWC/mystic/Earhead |
#4
|
|||
|
|||
Subject
Some say switching supplies suxz0r... I was just listening to my headphones amp off a flyback supply prototype I'm working on, sounds as good as otherwise... would help if it had a control loop besides me though Especially a dynamic one... darn power supply ripple is haunting me Regulated supplies DC to whatever frequency a cap can take over at is all that's possibly necessary, period, end of sentence, give him the award this show's done. DC to 1MHz in a regulator is a complete and utter waste of time and money, not to mention very difficult to do without bypasses to help it along anyway! Heh! Tim -- "That's for the courts to decide." - Homer Simpson Website @ http://webpages.charter.net/dawill/tmoranwms "Jim Candela" wrote in message ... |
#5
|
|||
|
|||
Thanks for the informative comments. Here is some more history leading to my decision to add regulators.When I was going through the Sherwood, I kept hearing a hard to describe distortion when listening to FM, and when cranking the audio. I made a attenuator to allow me to listen to my speakers at reasonable volume levels without getting that "turn down that darn thing" from my wife. It seems that when the output tubes transitioned from class A to Class Ab1, the current pulses started causing DC operating point shifts all over the place that were very low frequency. After a lot of head scratching, it became clear that the FM oscillator was shifting with small changes in supply voltage, and this was evident by the fact that the AFC voltage was moving around as a result. I had the AFC switch off. I also had instances where the frequency locked 38 kHz oscillator would on occasion lose lock, and a beat note would be heard. Early attempts concentrated on filtering the FM main DC feed which comes from the junction of the two 150 uf capacitors in the HV doubler circuit. Sherwood used a single R-C (something like 470 ohms, and 40 uf). At one point I tried a 10 Henry choke, and 180 uf. That cleaned it up the power line ripple quite a bit, but the low frequency swing from the output tubes remained. It was kind of like a whip being excited by the incoming voltage pulsing, and the effect of the choke seemed to do more harm than good. The solid state source follower completely eliminated the problem, and also provided 60 db of power line ripple rejection. I did put a 22 uf cap on the input of the regulator with a series 330 ohm resistor. The resistor was mainly used to drop some voltage to lower the thermal burden of the fet series pass, and the capacitor effectively filtered out a high frequency spike on the incoming ripple (I guess it's from the abrupt diode turn off in the voltage doubler). A few dollars worth of silicon seemed to be a very effective solution to this problem. The screen regulator was something I wanted to do anyway. The 7868/7591 type output tubes can sometimes go into a thermal runaway when running near maximum ratings, and then the line voltage jumps say from 120 to 125 volts..... It is difficult to keep the idle current constant (fixed bias) this way because it keeps changing. With a tightly stabilized screen voltage, and fixed bias source, the idle current drift is very small, even with my set of 20+ year old 7868's. The screen regulator also feeds the rest of the lower level audio chain with R-C networks with a time constant of 200ms, twice (1k, 200 uf and 10k, 20uf), and then nearly 1 second for the phono preamp section (47K, 20 uf). Looking at those power busses AC coupled on my scope at 50 mv / division, all I see is the line...no ripple, or wiggle when cranking the audio, or varying the line voltage from 110-125 vac. I kind of like that. This may be overkill, but isn't a wheel-barrow full of chokes, and capacitors also overkill? Thanks, Jim |
#6
|
|||
|
|||
Hello Patrick,
Relying on large value electros is to me a valid way to make low impedance power supplies. Sure it is, I agree. But, as an article from Thorsten Loesch suggests, there seems to be more than only low impedance to a good PSU than only low impedance. E.g. Mr. Loesch says, that the PSU is part of the signal path (at least for the SE amp circuits he is discussing in his article) and therefore additional considerations have to be made. I freely admit that his article is way ahead of the "problems" level (and circuits level) I am dealing with, so I cannot say anything on his considerations, except he made them. The article I am refering to is he http://www.fortunecity.com/rivendell...iertheory.html and the PSU related parts within it begin roughly 2/3 down from the start. Maybe this article is of interest to you and other RATs according to the subject of this thread. Tom -- Live is too short to be taken seriously. - Oscar Wilde |
#7
|
|||
|
|||
Hi RATs!
Overkill? Never! Overkill is a buzz word used by bean counters to justify cutting a penny from the PS build cost Happy Ears! Al Alan J. Marcy Phoenix, AZ PWC/mystic/Earhead |
#8
|
|||
|
|||
Jim Candela wrote: Thanks for the informative comments. Here is some more history leading to my decision to add regulators.When I was going through the Sherwood, I kept hearing a hard to describe distortion when listening to FM, and when cranking the audio. I made a attenuator to allow me to listen to my speakers at reasonable volume levels without getting that "turn down that darn thing" from my wife. It seems that when the output tubes transitioned from class A to Class Ab1, the current pulses started causing DC operating point shifts all over the place that were very low frequency. After a lot of head scratching, it became clear that the FM oscillator was shifting with small changes in supply voltage, and this was evident by the fact that the AFC voltage was moving around as a result. I had the AFC switch off. I also had instances where the frequency locked 38 kHz oscillator would on occasion lose lock, and a beat note would be heard. Does not this tell you that several stages of RCRCRC filtering are required to stop the B+ wandering affecting the RF stages? Sure, us a regulator, or replace the existing electros with ones that are 10 times the value. The B+ for the FM RF input stage is easily regulated with zeners and a large value cap. I don't have any trouble with my home designed tube FM set. Early attempts concentrated on filtering the FM main DC feed which comes from the junction of the two 150 uf capacitors in the HV doubler circuit. Sherwood used a single R-C (something like 470 ohms, and 40 uf). At one point I tried a 10 Henry choke, and 180 uf. That cleaned it up the power line ripple quite a bit, but the low frequency swing from the output tubes remained. Chokes are low impedance at very low F, so mains variations get changed into fluctuating DC levels which find their way down through the CLC filters with nothing to stop it. RCRC filters are better, because the R value remains constant even at DV, and RC will better smooth the LF jitter on the rails. Shunt regulating with zeners also will stop the B+ undulations to a large enough extent for most tube applications, where RC filters are used. It was kind of like a whip being excited by the incoming voltage pulsing, and the effect of the choke seemed to do more harm than good. It is important to arrange the power supply so that it doesn't form some kind of phase shift oscilator at some LF. The solid state source follower completely eliminated the problem, and also provided 60 db of power line ripple rejection. I did put a 22 uf cap on the input of the regulator with a series 330 ohm resistor. The resistor was mainly used to drop some voltage to lower the thermal burden of the fet series pass, and the capacitor effectively filtered out a high frequency spike on the incoming ripple (I guess it's from the abrupt diode turn off in the voltage doubler). A few dollars worth of silicon seemed to be a very effective solution to this problem. I have done similar things, and with similar results, but recently I have drifted away to simple RCRC filters, with large enough time constants, and coupling time time constants which prevent oscillation. The screen regulator was something I wanted to do anyway. The 7868/7591 type output tubes can sometimes go into a thermal runaway when running near maximum ratings, and then the line voltage jumps say from 120 to 125 volts..... It is difficult to keep the idle current constant (fixed bias) this way because it keeps changing. With a tightly stabilized screen voltage, and fixed bias source, the idle current drift is very small, even with my set of 20+ year old 7868's. Thats because anode current is mainly dependant on the G2 to K voltage. Its typical of all pentodes/tetrodes. The screen regulator also feeds the rest of the lower level audio chain with R-C networks with a time constant of 200ms, twice (1k, 200 uf and 10k, 20uf), and then nearly 1 second for the phono preamp section (47K, 20 uf). Looking at those power busses AC coupled on my scope at 50 mv / division, all I see is the line...no ripple, or wiggle when cranking the audio, or varying the line voltage from 110-125 vac. I kind of like that. I prefer to shunt regulate the screen supply, which is always at a lower value than the B+ for the anodes, and if a screen starts conducting too much current, its voltage simply falls because of the voltage dropped across the series R supply resistor. If series pass element regulation is used, the regulator tries to keep the screen voltage constant and the voltage x current rating of the screen can be exceeded, and poof goes a tube. This may be overkill, but isn't a wheel-barrow full of chokes, and capacitors also overkill? It depends how they are used. And at least a CLC supply in a power amp is required. But once initial CLC filtering is done, RCRC filters are fine. Elcaps are much higher value for the size than they used to be 35 years ago, so may as well take advantage of them. Zeners are cheap, and 5 watt 50v or 75v types can be seriesed for simple regulators to the input stages. I also never take the B+ for input stages via the screen supply, since if this voltage drops, it affects following supplies. So in a power amp where there is a screen supply, I keep that as one separate RC from the CT B+, and RCRC to the input stages, where I sometimes place a simple zener shunt regulator. The amp can be driven into gross overload, enduring a 15% B+ supply drop in the output stage, and there is no change to the input tubes DV conditions. I try to avoid having to actively regulate anything. Patrick Turner. Thanks, Jim |
#9
|
|||
|
|||
"TubeGarden" wrote in message
... Overkill is a buzz word used by bean counters to justify cutting a penny from the PS build cost I figured you'd say something like that. Any serious engineering is a total troll for you. ;-) Tim -- "That's for the courts to decide." - Homer Simpson Website @ http://webpages.charter.net/dawill/tmoranwms |
#10
|
|||
|
|||
Jim Candela wrote:
DC to 1 Mhz bandwidth. You can get that by using a small RF bypass cap, say 0.05uF, from the B+ line to ground. Regular electrolytic filter caps have a fair amount of inductance, and the bypass cap shunts that. Why you'd need that is another question. Maybe to keep transient current demands by the output stage from leaking into other parts of the amp. |
#11
|
|||
|
|||
"Robert Casey" Jim Candela wrote: DC to 1 Mhz bandwidth. You can get that by using a small RF bypass cap, say 0.05uF, from the B+ line to ground. ** At 1 MHz, a perfect 0.05 uF cap with no leads has an impedance of 3.2 ohms. Regular electrolytic filter caps have a fair amount of inductance, ** That stupid myth is not still going is it ??? Test one and see how wrong you are. and the bypass cap shunts that. ** At best it will merely create a parallel tuned circuit at a few MHz. The inductance of an electro due to only to connecting leads and its ength - just like other caps. .............. Phil |
#12
|
|||
|
|||
Hi RATs!
SERIOUS engineering? No such thing. Some of us get paid big bucks, but, what each of us realizes about the interaction between us and the Universe is special and unique. None of us enjoys a superior position. We each seek our truth as best we can. If you think real engineers are not subject to the realities of the marketplace, get serious Happy Ears! Al Alan J. Marcy Phoenix, AZ PWC/mystic/Earhead |
#13
|
|||
|
|||
Although Joe Curcio makes some valid points about power supplies, and how the quality of power alone can contribute to the sound of amplifiers feeding off the buss, his assertion that the DC needs to be actively regulated from DC to 1 Mhz sure seems to be marketing ploy by his company to sell his technology in the form of kits. I would imagine it's pretty hard to make a living in this business with so many "experts" touting their version of this or that. Get a MBA type as a partner, and all sorts of assertions with a weak technical basis could emerge. I don't doubt that Joe's technology works, but the assertions may be bent a bit in order to put food on the table. I have a pile of old issues of "Glass Audio" in my "reading room", and Joe has many articles written by (or for) him and his circuits, and inventions. To many he is highly regarded in the industry. Patrick, Thanks for giving me another perspective to ponder. You make good sense suggesting high R-C time constants, and zener shunt regulators where necessary. It is ironic however that one of your circuits (it was for a phono preamp) on your web page uses a active regulator. This circuit of yours inspired me to design my +170 volt active fet source follower. Yours uses a TO-3 case beefy bjt, and the base drive needs dictated the "R" component in the base r-c circuit such that the capacitors used had to be of high value (and physically big). With the fet, I was able to increase the "R" value to 150K, and the resultant need for the capacitor was only a few microfarads. Adding constant current feed to the zeners also helped tremendously since my dc feed had high AC ripple. The constant current feed provided decent line voltage regulation whereas before with a resistor dropping the voltage to the zeners, a change in line voltage was reflected in a change in zener diode current, and therefore the voltage (Vz) was changing as well. The degree of ripple reduction with constant current feed to the zeners increased by about 20 db (from 40 to 60 db). I have a project on the bench (Ham radio) where I need to provide 150 v dc to the screens of a pair of 6DQ6's running in a class AB1 grid driven RF linear amp. I just might try using three 50 volt 5 watt zeners in series as a shunt regulator being fed off a 300v buss with a series resistor. The one thing I don't like about HV zeners is the positive temp coefficient where Vz rises with temperature. That is why I used a 6.2 volt zener in my screen regulator, and the b-e junction of the error amp transistor almost perfectly offsets the zener drift with changing ambient temperature (so long as the two are thermally bonded together). I'll let you know how the zener shunt regulator goes. Might end up with a OD3/VR150! Regards, and thanks to all for their comments, and suggestions. Jim Candela |
#14
|
|||
|
|||
"Jim Candela" Although Joe Curcio makes some valid points about power supplies, and how the quality of power alone can contribute to the sound of amplifiers feeding off the buss, his assertion that the DC needs to be actively regulated from DC to 1 Mhz sure seems to be marketing ploy by his company to sell his technology in the form of kits. ** What phenomenal insight the Candle Man has !! A veritable Solomon among a throng of desperate audiophools has come to enlighten them and lead them out of the darkness of their ignorance. How can we praise you enough Mr Candela ??? No-one here can hold a candle to his light - not any of the Turneroids. The light that shines so brightly, out his hairy arse hole, like a carbon arc lamp up to the heavens. Just think how much **** has passed out that fabulous orifice - a staggering amount. But think - even more has passed out his mouth. Amasing. ............ Phil |
#15
|
|||
|
|||
Phil Allison wrote:
"Robert Casey" Regular electrolytic filter caps have a fair amount of inductance, ** That stupid myth is not still going is it ??? Test one and see how wrong you are. You obviously know sweet FA about RF. Let's think about an electrolytic - Hmm, a big roll of aluminium foil. Two layers with some goop in between. But you reckon it has zero inductance. What planet are you from. Just so you know, every single component you will ever use has capacitance, inductance and resistance. There are no, repeat no, perfect components in the real world. A 1 mm wire has an inductance of ~ 1 nH/mm of length, so even the hook up wire has significant inductance at 1 MHz. As an example, perhaps you would like to consider the impedance of a typical 1/2 W, 1 Mohm, metal oxide resistor at 1 MHz and 31 MHz? 1MHz 92000 -j289025 Ohms 31 MHz 101.3 -j10064 Ohms Yes, that is 100 Ohms resistive impedance. Not quite what's written on the tin, eh? These figures assume 14 nH of inductance and 0.5 pF of capacitance, typical values, due to the geometry of construction. best Andy Cowley |
#16
|
|||
|
|||
Hi Jim,
I need to provide 150 v dc to the screens of a pair of 6DQ6's [...] The one thing I don't like about HV zeners [...] Might end up with a OD3/VR150! Shunt regulators like voltage regulator tubes and zeners don't make much sense nowadays, except you insist on doing it "the old way" and like to deal with their drawbacks. Why not use a VB408 regulator IC (up to 40mA, costs ~1 Euro over here) with much better regulation, stability, protection and whatelse? I admit it doesn't glow nicely like a voltage regulator tube, but a zener chain doesn't either. The datasheet is he http://www.ndh.net/home/schlangen/misc/vb408.pdf Just beef it up with a HV transistor if you need more current, like (scroll down a bit for the schemo): http://home.t-online.de/home/0689839...ernetzteil.htm Couldn't get easier - so, more time left to worry about other things probably more important and/or interesting than how to regulate screen voltage ... Tom -- The large print giveth, the small print taketh away. - Tom Waits |
#17
|
|||
|
|||
"Tom Schlangen" wrote in message ... Hi Jim, Why not use a VB408 regulator IC (up to 40mA, costs ~1 Euro over here) with much better regulation, stability, protection and whatelse? Tom, thanks for the reply suggesting that ST component. The data sheet, and application note seemed to have a disagreement. The data sheet says that Vin-Vout = 30 volts maximum, although Vin can be 400volts. This means that For my 150v need, I cannot apply any more than 180 volts to vin (my source is a unregulated 250-300v). The application note shows 0-250 volt outputs continually adjusted with a fixed value vin. Am I missing something? It often seems that the off the shelf solution ends up more complex, and compromised than starting from scratch, and building it yourself..That said, in my Sherwood amp, the -24 volt solution was nicely done with a ST 7924. Regards, Jim |
#18
|
|||
|
|||
"Tom Schlangen" wrote in message ... Hello Patrick, Relying on large value electros is to me a valid way to make low impedance power supplies. Sure it is, I agree. But, as an article from Thorsten Loesch suggests, there seems to be more than only low impedance to a good PSU than only low impedance. E.g. Mr. Loesch says, that the PSU is part of the signal path (at least for the SE amp circuits he is discussing in his article) and therefore additional considerations have to be made. I freely admit that his article is way ahead of the "problems" level (and circuits level) I am dealing with, so I cannot say anything on his considerations, except he made them. The article I am refering to is he http://www.fortunecity.com/rivendell...heory/SEAmplif iertheory.html and the PSU related parts within it begin roughly 2/3 down from the start. Maybe this article is of interest to you and other RATs according to the subject of this thread. Tom -- Live is too short to be taken seriously. - Oscar Wilde just few words from always humble Choky: -Gainclone. Conditio sine qua non for Gainclone is smallest amount of capacitance in power supply as you can manage. less micros-more speed more micros-slower recovery. man must be his own Tube Garden to find exact amount and topology for own amp. -- -- -- .................................................. ........................ Choky Prodanovic Aleksandar YU "don't use force, "don't use force, use a larger hammer" use a larger tube - Choky and IST" - ZM .................................................. ........................... |
#19
|
|||
|
|||
Tom-Series stabs are economic,nothing more.
for low loads there is nothing better than Shunt reg,SS or Tube-use what you want. try it ,I'll be surprised if you than say different. and -don't skip fun part ........ ) -- -- -- .................................................. ........................ Choky Prodanovic Aleksandar YU "don't use force, "don't use force, use a larger hammer" use a larger tube - Choky and IST" - ZM .................................................. ........................... "Tom Schlangen" wrote in message ... Hi Jim, I need to provide 150 v dc to the screens of a pair of 6DQ6's [...] The one thing I don't like about HV zeners [...] Might end up with a OD3/VR150! Shunt regulators like voltage regulator tubes and zeners don't make much sense nowadays, except you insist on doing it "the old way" and like to deal with their drawbacks. Why not use a VB408 regulator IC (up to 40mA, costs ~1 Euro over here) with much better regulation, stability, protection and whatelse? I admit it doesn't glow nicely like a voltage regulator tube, but a zener chain doesn't either. The datasheet is he http://www.ndh.net/home/schlangen/misc/vb408.pdf Just beef it up with a HV transistor if you need more current, like (scroll down a bit for the schemo): http://home.t-online.de/home/0689839...teil/experimen tiernetzteil.htm Couldn't get easier - so, more time left to worry about other things probably more important and/or interesting than how to regulate screen voltage ... Tom -- The large print giveth, the small print taketh away. - Tom Waits |
#20
|
|||
|
|||
"Tim Williams" said:
Overkill is a buzz word used by bean counters to justify cutting a penny from the PS build cost I figured you'd say something like that. Any serious engineering is a total troll for you. ;-) Sadly, Al is right. At the end of the game, everything gets decided by bean counters. I've seen many desillusioned engineers deserting large companies because they couldn't design along the technical rules, but along financial rules only. That's why I chose to have a non-audio related job in the industry, and to do the tube stuff as a hobby. [Yes, Vacuum Audio Consultancy is a hobby]. -- Sander deWaal Vacuum Audio Consultancy |
#21
|
|||
|
|||
"Jim Candela" said:
Might end up with a OD3/VR150! Sexy, but noisy. -- Sander deWaal Vacuum Audio Consultancy |
#22
|
|||
|
|||
Jim -look for TL783 CKC
-- -- -- .................................................. ........................ Choky Prodanovic Aleksandar YU "don't use force, "don't use force, use a larger hammer" use a larger tube - Choky and IST" - ZM .................................................. ........................... "Jim Candela" wrote in message om... "Tom Schlangen" wrote in message ... Hi Jim, Why not use a VB408 regulator IC (up to 40mA, costs ~1 Euro over here) with much better regulation, stability, protection and whatelse? Tom, thanks for the reply suggesting that ST component. The data sheet, and application note seemed to have a disagreement. The data sheet says that Vin-Vout = 30 volts maximum, although Vin can be 400volts. This means that For my 150v need, I cannot apply any more than 180 volts to vin (my source is a unregulated 250-300v). The application note shows 0-250 volt outputs continually adjusted with a fixed value vin. Am I missing something? It often seems that the off the shelf solution ends up more complex, and compromised than starting from scratch, and building it yourself..That said, in my Sherwood amp, the -24 volt solution was nicely done with a ST 7924. Regards, Jim |
#23
|
|||
|
|||
Hi Choky,
Tom-Series stabs are economic,nothing more. for low loads there is nothing better than Shunt reg,SS or Tube-use what you want. try it ,I'll be surprised if you than say different. I did try, as you might know from my articles over on Greggggs board regarding my EL86PP project which needed reg'd screens voltage regulation. To wind it down, the VB408 IC was the perfect drop-in solution after having fiddled too long with zener chains etc. My GOT, why should I worry about zener noise, optimum current equilibrium between the shunt the circuit to supply and keeping dissipation power within specs by micro-soldering tiny copper rectangles to the wires of the zeners and whatelse? Been there, done that, worked thru it, learned from it (that's why I did it discrete and shunt way in the first place) and *then* changed to the no-fuss, point & shoot & forget 4E4 IC-based solution. and -don't skip fun part ........ ) I didn't :-) since trying out voltage reg tubes and zeners _was_ interesting and fun to learn from. Please note the "_was_" in the previous sentence. So much fun and interesting stuff with tubes still ahead, I have more interesting stuff to wade through and learn from than dealing with zener noise in pass reg circuits for screens :-) Tom -- A consultant is a man who knows 40 ways to make love, but doesn't know any women. |
#24
|
|||
|
|||
Hi Choky,
-Gainclone. What is this? I will ask Google, sure. Conditio sine qua non for Gainclone is smallest amount of capacitance in power supply as you can manage. Okay, but I have to learn about what "Gainclone" is, and its benefits (and drawbacks) first ... man must be his own Tube Garden to find exact amount and topology for own amp. I wholeheartly agree to that. But - before that, one has to wade through lots of experiments to have a base of "Erkenntnis" to chose from oneself. My base of experimets is very small so far, and widening only slowly. I found some personal preferences already, but they might be falsified by my next project. But, probably this is just the "fun" with DIY ... Tom -- The large print giveth, the small print taketh away. - Tom Waits |
#25
|
|||
|
|||
"Tom Schlangen" wrote in message ... Hi Choky, Tom-Series stabs are economic,nothing more. for low loads there is nothing better than Shunt reg,SS or Tube-use what you want. try it ,I'll be surprised if you than say different. I did try, as you might know from my articles over on Greggggs board regarding my EL86PP project which needed reg'd screens voltage regulation. To wind it down, the VB408 IC was the perfect drop-in solution after having fiddled too long with zener chains etc. My GOT, why should I worry about zener noise, optimum current equilibrium between the shunt the circuit to supply and keeping dissipation power within specs by micro-soldering tiny copper rectangles to the wires of the zeners and whatelse? Been there, done that, worked thru it, learned from it (that's why I did it discrete and shunt way in the first place) and *then* changed to the no-fuss, point & shoot & forget 4E4 IC-based solution. and -don't skip fun part ........ ) I didn't :-) since trying out voltage reg tubes and zeners _was_ interesting and fun to learn from. Please note the "_was_" in the previous sentence. So much fun and interesting stuff with tubes still ahead, I have more interesting stuff to wade through and learn from than dealing with zener noise in pass reg circuits for screens :-) Tom -- A consultant is a man who knows 40 ways to make love, but doesn't know any women. hehe,entshuldigung Tom; mebbe I didn't catch fact that you talk primary about your exact experience with G2 supplies ,even if I spotted everything at Grrrrrrrrrreeeeeeeegggggggg's; so-to make clearer what I have to say-when sort of higher Iload or just kind of sub-supply is in question,I agree 100% with you (and at least 50% of world). but,I have mostly on my mind most critical aspect of Stabs implementation-use in lover level stages,as phono preamps ,line preamps ,input stages of power amp etc. there you will find (hear) biggest differentiation between various topologies of supply. for my Golden Ears, from worst to best : 4.series stab with SS 3.series stab with tubes 2.no stab,just variants of C and L and R 1.Shunt stab ,preferably tube (preferably passive) ,or even SS regarding Gainclone-go to www.diyaudio.com ,look for "chip amps" forum. one of best forums ,with smallest amount of snake oil . ps-I made one of my amps with TL783 as series stab for input and LTP stages. it works fine and its cheap. -- -- -- .................................................. ........................ Choky Prodanovic Aleksandar YU "don't use force, "don't use force, use a larger hammer" use a larger tube - Choky and IST" - ZM .................................................. ........................... |
#26
|
|||
|
|||
"Andy Cowley" Phil Allison wrote: "Robert Casey" Regular electrolytic filter caps have a fair amount of inductance, ** That stupid myth is not still going is it ??? Test one and see how wrong you are. You obviously know sweet FA about RF. ** We have a stinking ** pommy smartarse ** here folks - this is gonna be fun !! Let's think about an electrolytic - Hmm, a big roll of aluminium foil. Two layers with some goop in between. But you reckon it has zero inductance. What planet are you from. ** Where did I say "zero inductance " - pommy smartarse ??? Just so you know, every single component you will ever use has capacitance, inductance and resistance. There are no, repeat no, perfect components in the real world. ** Where did I say otherwise - pommy smartarse ??? A 1 mm wire has an inductance of ~ 1 nH/mm of length, so even the hook up wire has significant inductance at 1 MHz. ** Whether it is "significant" depends on the situation - pommy smartarse. As an example, perhaps you would like to consider the impedance of a typical 1/2 W, 1 Mohm, metal oxide resistor at 1 MHz and 31 MHz? ** Got anything to do with my post or electro capacitor inductance - pommy smartarse ??? Not one tiny thing. So go get ****ed - pommy smartarse . BTW It is most * impolite * even for a *pommy smartarse* to snip info from a post that indicates the author already knew all about the ****ing issue : " The inductance of an electro due to only to connecting leads and its length - just like other caps." ............... Phil |
#27
|
|||
|
|||
Jim Candela wrote: Although Joe Curcio makes some valid points about power supplies, and how the quality of power alone can contribute to the sound of amplifiers feeding off the buss, his assertion that the DC needs to be actively regulated from DC to 1 Mhz sure seems to be marketing ploy by his company to sell his technology in the form of kits. I would imagine it's pretty hard to make a living in this business with so many "experts" touting their version of this or that. Get a MBA type as a partner, and all sorts of assertions with a weak technical basis could emerge. I don't doubt that Joe's technology works, but the assertions may be bent a bit in order to put food on the table. I have a pile of old issues of "Glass Audio" in my "reading room", and Joe has many articles written by (or for) him and his circuits, and inventions. To many he is highly regarded in the industry. Patrick, Thanks for giving me another perspective to ponder. You make good sense suggesting high R-C time constants, and zener shunt regulators where necessary. It is ironic however that one of your circuits (it was for a phono preamp) on your web page uses a active regulator. I used to have an SS active reg for the supply of by 10 tube preamp, but not any more, and I found CLCRCRC was just as effective, using plenty of C. And the phone stage in the revised amp has a fet to drive a grounded grid 6EJ7 ( cascode ) and the rail cap is a 470 uF. I don't even have any any zeners. When I look at the CRO trace at its most sensitive setting, with the gain turned up, there is SFA LF jitter. Its actually more stable than it use to be with the regulator. This circuit of yours inspired me to design my +170 volt active fet source follower. Yours uses a TO-3 case beefy bjt, and the base drive needs dictated the "R" component in the base r-c circuit such that the capacitors used had to be of high value (and physically big). The follower type of regulator seems to me to be the simplest of series regs. Its effectiveness depends on a well stabilised base voltage. So the hfe of the darlington arrangement of two transistors needs to be high, to give the base input circuit some high impedance, to make it easy to keep the base at a 0V signal potential. I have used 10k, 22 uF, 10k , 22 uF as an RCRC smoothing circuit for the base. In a preamp, zener referencing of the base reg voltage is not necessary, since the voltage value varies so little. So an R to ground from the base is enough to trim the output voltage down a bit to where one wants it. 22uF x 450v are small and cheap. With the fet, I was able to increase the "R" value to 150K, and the resultant need for the capacitor was only a few microfarads. Adding constant current feed to the zeners also helped tremendously since my dc feed had high AC ripple. That's the traditional way to get zener regged voltage to be stable. The constant current feed provided decent line voltage regulation whereas before with a resistor dropping the voltage to the zeners, a change in line voltage was reflected in a change in zener diode current, and therefore the voltage (Vz) was changing as well. The degree of ripple reduction with constant current feed to the zeners increased by about 20 db (from 40 to 60 db). In the preamp when I used to have a reg, I also had RC filters to each stage, starred off fro the regged B+, using 4.7k and 47 uF. That gave unmeasurable levels of ripple. I have a project on the bench (Ham radio) where I need to provide 150 v dc to the screens of a pair of 6DQ6's running in a class AB1 grid driven RF linear amp. I just might try using three 50 volt 5 watt zeners in series as a shunt regulator being fed off a 300v buss with a series resistor. Use a big cap in shunt with the zeners, they should be protected from transients. And they have a little noise one needs to shunt. The one thing I don't like about HV zeners is the positive temp coefficient where Vz rises with temperature. They can be made more effective if the zeners are taken to the base of a HV transistor, whose emitter is at 0V and collector is to the B+ you want to shunt reg. The transistor will take higher current easily. That is why I used a 6.2 volt zener in my screen regulator, and the b-e junction of the error amp transistor almost perfectly offsets the zener drift with changing ambient temperature (so long as the two are thermally bonded together). I'll let you know how the zener shunt regulator goes. Might end up with a OD3/VR150! I have never got around to using tube shunt regulators. I am lazy....... I have found HV mosfet reg devices to be quite unreliable. The slightest abuse, and they die, but with BU208 and the right protection diodes, one can short the output with a cscrew driver, blat, blat, and they survive like a tube. Regards, and thanks to all for their comments, and suggestions. Jim Candela Patrick Turner. |
#28
|
|||
|
|||
Andy Cowley schrieb:
As an example, perhaps you would like to consider the impedance of a typical 1/2 W, 1 Mohm, metal oxide resistor at 1 MHz and 31 MHz? 1MHz 92000 -j289025 Ohms 31 MHz 101.3 -j10064 Ohms Yes, that is 100 Ohms resistive impedance. Not quite what's written on the tin, eh? These figures assume 14 nH of inductance and 0.5 pF of capacitance, typical values, due to the geometry of construction. that is nonsense, the real part of a resistor does not dissappear at RF and impedance is built by RLC Peter |
#29
|
|||
|
|||
Andy Cowley wrote: Phil Allison wrote: "Robert Casey" Regular electrolytic filter caps have a fair amount of inductance, ** That stupid myth is not still going is it ??? Test one and see how wrong you are. You obviously know sweet FA about RF. Let's think about an electrolytic - Hmm, a big roll of aluminium foil. Two layers with some goop in between. But you reckon it has zero inductance. What planet are you from. But the electrolytics made now for SMPS are far less inductive than they used to be, and they act as a capacitor up to a higher F than they used to. I have never pulled one apart, but I assume they have multi lead connections to the foils; but it isn't necessary yo know how they are built, the specs show they are far less inductive than they used to be. Just so you know, every single component you will ever use has capacitance, inductance and resistance. There are no, repeat no, perfect components in the real world. A 1 mm wire has an inductance of ~ 1 nH/mm of length, so even the hook up wire has significant inductance at 1 MHz. As an example, perhaps you would like to consider the impedance of a typical 1/2 W, 1 Mohm, metal oxide resistor at 1 MHz and 31 MHz? 1MHz 92000 -j289025 Ohms 31 MHz 101.3 -j10064 Ohms Yes, that is 100 Ohms resistive impedance. Not quite what's written on the tin, eh? These figures assume 14 nH of inductance and 0.5 pF of capacitance, typical values, due to the geometry of construction. I built a 2 x 300 watt power amp using mosfets with a total of 100,000 uF on each rail. And to make sure the rails were bypassed effectively up to many MHz, I had two large Sprague 100,000 uF caps, with leads about 200mm long, then fuses to the devices, but close to the drain connections, I have 1,000 uF, and then plastic 0.47 uF, and finally individual 0.01 uF at each drain to 0V. Its wasn't entirely easy to make sure bypassing was effective to HF. I also have stopper resistors and ferrite beads on the gate circuits, and 33pF caps from drains to each gate; there are 6 devices per channel, and finally, the circuit began to work without any parasitic oscillations, or with a 100 kHz full power square wave test signal. Patrick Turner. best Andy Cowley |
#30
|
|||
|
|||
"Patrick Turner" But the electrolytics made now for SMPS are far less inductive than they used to be, and they act as a capacitor up to a higher F than they used to. ** The Turneroid has confused low ESR with low inductance. TheTurnereoid is permanently and terminally confused. I have never pulled one apart, but I assume they have multi lead connections to the foils; but it isn't necessary yo know how they are built, the specs show they are far less inductive than they used to be. ** Which specs ? The ESR figures ??? They have been improved by use of high conductance electrolytes. ......... Phil |
#31
|
|||
|
|||
"Choky" wrote in message ... Jim -look for TL783 CKC I did, and for a maximum of 125 vin-vout this is a good device. Should be ways to increase the Vin safely so long as vout never gets shorted to ground...Then I looked a little further. This one has a vin-vout rating of 450 volts. Check out the Supertex LR8: http://www.mouser.com/index.cfm?hand...496 7127&Ne=3 or if it doesn't come up: http://www.supertex.com/pdf/datasheets/LR8.pdf Add a emitter follower to the output for current boost, and now we're talking. Maybe add a series "R" on the emitter follower collector to add current limiting. The LR8 sure looks simple, and for very low cost. Hmmm. Jim |
#32
|
|||
|
|||
"Jim Candela" schreef in bericht
om... "Tom Schlangen" wrote in message ... Hi Jim, Why not use a VB408 regulator IC (up to 40mA, costs ~1 Euro over here) with much better regulation, stability, protection and whatelse? Tom, thanks for the reply suggesting that ST component. The data sheet, and application note seemed to have a disagreement. The data sheet says that Vin-Vout = 30 volts maximum, although Vin can be 400volts. This means that For my 150v need, I cannot apply any more than 180 volts to vin (my source is a unregulated 250-300v). The application note shows 0-250 volt outputs continually adjusted with a fixed value vin. Am I missing something? Vin-Vout = 30 Volts maximum DROPOUT voltage as I read it. Confusing datasheet tough, I admit. Rgds, Jan. It often seems that the off the shelf solution ends up more complex, and compromised than starting from scratch, and building it yourself..That said, in my Sherwood amp, the -24 volt solution was nicely done with a ST 7924. Regards, Jim |
#33
|
|||
|
|||
Jim Candela wrote: "Choky" wrote in message ... Jim -look for TL783 CKC I did, and for a maximum of 125 vin-vout this is a good device. Should be ways to increase the Vin safely so long as vout never gets shorted to ground...Then I looked a little further. This one has a vin-vout rating of 450 volts. Check out the Supertex LR8: http://www.mouser.com/index.cfm?hand...496 7127&Ne=3 or if it doesn't come up: http://www.supertex.com/pdf/datasheets/LR8.pdf Add a emitter follower to the output for current boost, and now we're talking. Maybe add a series "R" on the emitter follower collector to add current limiting. The LR8 sure looks simple, and for very low cost. Hmmm. Jim Good regs should be able to be shorted at their output indefinately, or repeatedly, with a screw driver from output to 0V, or from input to 0V. Otherwise, don't use SS regulators. I don't mind RCRC filters; a short should only blow up a cheap resistor, and never damage a choke, opt, tube, or any SS device. Patrick Turner. |
#34
|
|||
|
|||
"Sander deWaal" wrote in message
news Sadly, Al is right. At the end of the game, everything gets decided by bean counters. But us here don't underdesign for cost. Impossible to do with a class A amp! If it had an underrated PSU it would burn out in short time indeed. Anyone can put things together and make it work, but it takes an engineer to remove the useless things and still leave it barely standing. ;-) Tim -- "That's for the courts to decide." - Homer Simpson Website @ http://webpages.charter.net/dawill/tmoranwms |
#35
|
|||
|
|||
On Fri, 6 Feb 2004 11:07:22 +1100, "Phil Allison" wrote:
"Andy Cowley" Phil Allison wrote: "Robert Casey" Regular electrolytic filter caps have a fair amount of inductance, ** That stupid myth is not still going is it ??? Test one and see how wrong you are. You obviously know sweet FA about RF. ** We have a stinking ** pommy smartarse ** here folks - this is gonna be fun !! racist Pill.... Let's think about an electrolytic - Hmm, a big roll of aluminium foil. Two layers with some goop in between. But you reckon it has zero inductance. What planet are you from. ** Where did I say "zero inductance " - pommy smartarse ??? racist Pill.... Just so you know, every single component you will ever use has capacitance, inductance and resistance. There are no, repeat no, perfect components in the real world. ** Where did I say otherwise - pommy smartarse ??? racist Pill.... A 1 mm wire has an inductance of ~ 1 nH/mm of length, so even the hook up wire has significant inductance at 1 MHz. ** Whether it is "significant" depends on the situation - pommy smartarse. racist Pill.... As an example, perhaps you would like to consider the impedance of a typical 1/2 W, 1 Mohm, metal oxide resistor at 1 MHz and 31 MHz? ** Got anything to do with my post or electro capacitor inductance - pommy smartarse ??? racist Pill.... Not one tiny thing. So go get ****ed - pommy smartarse . racist Pill.... BTW It is most * impolite * even for a *pommy smartarse* to snip info racist Pill.... from a post that indicates the author already knew all about the ****ing issue : " The inductance of an electro due to only to connecting leads and its length - just like other caps." .............. Racist Phil |
#36
|
|||
|
|||
"Bob Flint" racist Pill.... ** ROTFLMAO. This lunatic thinks musicians are a race on AGA. He also thinks that ****wits are a race. Now he thinks "pommy smartarses" are a race as well. ROTFLMAO ............ Phil |
#37
|
|||
|
|||
Peter Völpel wrote:
Andy Cowley schrieb: As an example, perhaps you would like to consider the impedance of a typical 1/2 W, 1 Mohm, metal oxide resistor at 1 MHz and 31 MHz? 1MHz 92000 -j289025 Ohms 31 MHz 101.3 -j10064 Ohms Yes, that is 100 Ohms resistive impedance. Not quite what's written on the tin, eh? These figures assume 14 nH of inductance and 0.5 pF of capacitance, typical values, due to the geometry of construction. that is nonsense, the real part of a resistor does not dissappear at RF and impedance is built by RLC Peter Dear Peter, No it's not nonsense. It may be counter intuitive but the fact that there is a capacitive impedance in parallel with the 'real' resistive element and an inductive impedance in series _does_ give the results shown at the frequencies given. I gave the assumed conditions of 14 nH and 0.5 pF so do the maths yourself if you still have doubts. best Andy Cowley |
#38
|
|||
|
|||
Bob Flint wrote:
On Fri, 6 Feb 2004 11:07:22 +1100, "Phil Allison" wrote: "Andy Cowley" Phil Allison wrote: "Robert Casey" Regular electrolytic filter caps have a fair amount of inductance, ** That stupid myth is not still going is it ??? Test one and see how wrong you are. You obviously know sweet FA about RF. ** We have a stinking ** pommy smartarse ** here folks - this is gonna be fun !! Phil's colonial inferiority complex has obviously got the better of him again. What can one expect from a citizen of a nation of under-arm bowlers? ;-) best Andy Cowley |
Reply |
Thread Tools | |
Display Modes | |
|
|
Similar Threads | ||||
Thread | Forum | |||
FA: Vintage Neve Console Modules, Panels, Power Supplies + much more | Pro Audio | |||
here is how firewire ports fail | Pro Audio | |||
Damping Material Question | Car Audio | |||
FS: SOUNDSTREAM CLOSEOUTS AND MORE!! | Car Audio |