Home |
Search |
Today's Posts |
#1
Posted to rec.audio.tubes
|
|||
|
|||
JJ ECC99 for CCDA
Hello All!
I want to build a tube linestage based on the design I found yesterday called CCDA. This is the link of this circuit which I saw from the www.tubecad.com http://www.tubecad.com/2009/03/blog0161.htm http://www.tubecad.com/2009/04/blog0162.htm They also present a regulator called Feed-forward shunt regulator, is this regulator is a must for CCDA circuitry. Can I build a power supply with big reservoir along with gas tube regulator instead of using FFSR in the power supply? Will it be a good choice by putting ECC99 in the CCDA? Thanks! |
#3
Posted to rec.audio.tubes
|
|||
|
|||
JJ ECC99 for CCDA
flipper wrote:
On Thu, 30 Apr 2009 10:54:49 +0100, Ian Bell wrote: wrote: Hello All! I want to build a tube linestage based on the design I found yesterday called CCDA. This is the link of this circuit which I saw from the www.tubecad.com http://www.tubecad.com/2009/03/blog0161.htm http://www.tubecad.com/2009/04/blog0162.htm They also present a regulator called Feed-forward shunt regulator, is this regulator is a must for CCDA circuitry. Can I build a power supply with big reservoir along with gas tube regulator instead of using FFSR in the power supply? The whole point of the CCDA seems to be it draws a constant current from the supply and hence Quote:
the current draw is constant a straightforward supply with a big reservoir would be OK. However, note the following: 1. For good ripple reduction (always a must for preamps) you are better off with several RC decoupling stages than a single one. For example if you use a 10K resistor and 100uF capacitor you will get better ripple reduction if you replace it with five sets of 2K resistors and 20uF capacitors. Total resistance and capacitance is the same but ripple reduction is hugely improved. 2. IMHO the CCDA concept is a waste of time. In class A circuits with RC decoupling the current draw from the power supply already IS constant no matter what circuit topology is used. No it isn't. Power supply filtering has no impact on what the *circuit* current demand is. I never said it did. Moment to moment current variation Which means it isn't constant despite you just saying it was.. The circuit downstream of the last, i.e. local, RC decoupling has moment to moment changes. If the final RC filter (i.e. local) is sufficient to be effective at reducing ripple it acts as a local charge store to provide for moment to moment circuit currents. The current flowing from the supply is virtually constant. needs are provided by the C in the RC decoupling. Except it's on the outboard side of all that hum filtering. What are you talking about. The CCDA circuit reference has R9 and C4 local decoupling. You can, of course, make the last filter cap arbitrarily large but it's moot with CCDA because current demand is constant. But any RC filter also present already has the same effect. One of the benefits is reduced crosstalk because there's, theoretically, '0' signal ripple on the supply. Also one of the benefits of proper decoupling. Cheers Ian |
#4
Posted to rec.audio.tubes
|
|||
|
|||
JJ ECC99 for CCDA
flipper wrote:
On Thu, 30 Apr 2009 15:50:53 +0100, Ian Bell wrote: flipper wrote: On Thu, 30 Apr 2009 10:54:49 +0100, Ian Bell wrote: wrote: Hello All! I want to build a tube linestage based on the design I found yesterday called CCDA. This is the link of this circuit which I saw from the www.tubecad.com http://www.tubecad.com/2009/03/blog0161.htm http://www.tubecad.com/2009/04/blog0162.htm They also present a regulator called Feed-forward shunt regulator, is this regulator is a must for CCDA circuitry. Can I build a power supply with big reservoir along with gas tube regulator instead of using FFSR in the power supply? The whole point of the CCDA seems to be it draws a constant current from the supply and hence Quote:
the current draw is constant a straightforward supply with a big reservoir would be OK. However, note the following: 1. For good ripple reduction (always a must for preamps) you are better off with several RC decoupling stages than a single one. For example if you use a 10K resistor and 100uF capacitor you will get better ripple reduction if you replace it with five sets of 2K resistors and 20uF capacitors. Total resistance and capacitance is the same but ripple reduction is hugely improved. 2. IMHO the CCDA concept is a waste of time. In class A circuits with RC decoupling the current draw from the power supply already IS constant no matter what circuit topology is used. No it isn't. Power supply filtering has no impact on what the *circuit* current demand is. I never said it did. May not be what you 'meant' but it is what you said. Unless you're trying to play semantics by arguing 'some' RC filers are part of the power supply while 'other' RC power filters aren't. Sounds to me like you are the one who is playing semantics. Moment to moment current variation Which means it isn't constant despite you just saying it was.. The circuit downstream of the last, i.e. local, RC decoupling has moment to moment changes. Exactly, it isn't constant. If the final RC filter (i.e. local) is sufficient to be effective at reducing ripple it acts as a local charge store to provide for moment to moment circuit currents. That's what I said with "You can, of course, make the last filter cap arbitrarily large." That's the point you are missing. There's no need to make the last filter cap, by which I presume you mean the one AT the preamp, arbitrarily large. The current flowing from the supply is virtually constant. As I said, the effect of non constant current can be made arbitrarily small with arbitrarily large filtering. As I said, normal local RC decoupling values are more than sufficient to achieve this effect. There is no need to use an arbitrarily large cap. needs are provided by the C in the RC decoupling. Except it's on the outboard side of all that hum filtering. What are you talking about. I'm talking about your multi-stage RC filter suggestion. That is part of the power supply itself, not the preamp. The CCDA circuit reference has R9 and C4 local decoupling. Yep. So? So, if the current is constant why have the decoupling?? You can, of course, make the last filter cap arbitrarily large but it's moot with CCDA because current demand is constant. But any RC filter also present already has the same effect. An RC filter can only approximate the effect as you make it arbitrarily large. How many times do I have to say this, there is no need to make it arbitrarily large. In real life component tolerances will also prevent the CCDA from being a 'perfect 0'. So the current will not be constant. One of the benefits is reduced crosstalk because there's, theoretically, '0' signal ripple on the supply. Also one of the benefits of proper decoupling. 'Proper' is dependant on circuit needs and the needs are less with a constant current draw. Proper decoupling PROVIDES a near constant current draw, at least as good as a CCDA with normal tube tolerances. Look, no one said there's only 'one way' to do it. The point was a constant current draw eases the PS design and what does it cost past some clever thinking? Yes, that is the point.Exactly how has the power supply design for this preamp been eased. What circuit simplifications are possible in the PSU because of using a CCDA?? Cheers Ian Cheers Ian |
#5
Posted to rec.audio.tubes
|
|||
|
|||
JJ ECC99 for CCDA
Ian Bell wrote: flipper wrote: On Thu, 30 Apr 2009 15:50:53 +0100, Ian Bell wrote: flipper wrote: On Thu, 30 Apr 2009 10:54:49 +0100, Ian Bell wrote: wrote: Hello All! I want to build a tube linestage based on the design I found yesterday called CCDA. This is the link of this circuit which I saw from the www.tubecad.com http://www.tubecad.com/2009/03/blog0161.htm http://www.tubecad.com/2009/04/blog0162.htm They also present a regulator called Feed-forward shunt regulator, is this regulator is a must for CCDA circuitry. Can I build a power supply with big reservoir along with gas tube regulator instead of using FFSR in the power supply? The whole point of the CCDA seems to be it draws a constant current from the supply and hence Quote:
the current draw is constant a straightforward supply with a big reservoir would be OK. However, note the following: 1. For good ripple reduction (always a must for preamps) you are better off with several RC decoupling stages than a single one. For example if you use a 10K resistor and 100uF capacitor you will get better ripple reduction if you replace it with five sets of 2K resistors and 20uF capacitors. Total resistance and capacitance is the same but ripple reduction is hugely improved. 2. IMHO the CCDA concept is a waste of time. In class A circuits with RC decoupling the current draw from the power supply already IS constant no matter what circuit topology is used. No it isn't. Power supply filtering has no impact on what the *circuit* current demand is. I never said it did. May not be what you 'meant' but it is what you said. Unless you're trying to play semantics by arguing 'some' RC filers are part of the power supply while 'other' RC power filters aren't. Sounds to me like you are the one who is playing semantics. Moment to moment current variation Which means it isn't constant despite you just saying it was.. The circuit downstream of the last, i.e. local, RC decoupling has moment to moment changes. Exactly, it isn't constant. If the final RC filter (i.e. local) is sufficient to be effective at reducing ripple it acts as a local charge store to provide for moment to moment circuit currents. That's what I said with "You can, of course, make the last filter cap arbitrarily large." That's the point you are missing. There's no need to make the last filter cap, by which I presume you mean the one AT the preamp, arbitrarily large. The current flowing from the supply is virtually constant. As I said, the effect of non constant current can be made arbitrarily small with arbitrarily large filtering. As I said, normal local RC decoupling values are more than sufficient to achieve this effect. There is no need to use an arbitrarily large cap. needs are provided by the C in the RC decoupling. Except it's on the outboard side of all that hum filtering. What are you talking about. I'm talking about your multi-stage RC filter suggestion. That is part of the power supply itself, not the preamp. The CCDA circuit reference has R9 and C4 local decoupling. Yep. So? So, if the current is constant why have the decoupling?? You can, of course, make the last filter cap arbitrarily large but it's moot with CCDA because current demand is constant. But any RC filter also present already has the same effect. An RC filter can only approximate the effect as you make it arbitrarily large. How many times do I have to say this, there is no need to make it arbitrarily large. In real life component tolerances will also prevent the CCDA from being a 'perfect 0'. So the current will not be constant. One of the benefits is reduced crosstalk because there's, theoretically, '0' signal ripple on the supply. Also one of the benefits of proper decoupling. 'Proper' is dependant on circuit needs and the needs are less with a constant current draw. Proper decoupling PROVIDES a near constant current draw, at least as good as a CCDA with normal tube tolerances. Look, no one said there's only 'one way' to do it. The point was a constant current draw eases the PS design and what does it cost past some clever thinking? Yes, that is the point.Exactly how has the power supply design for this preamp been eased. What circuit simplifications are possible in the PSU because of using a CCDA?? I come late to this thread. Power supplies for the preamp stages shown don't need to be elaborate and any old CRCRCRC filter using Si diodes and 470uF electros will do as long as the hum at the final C 0.1mV, and the RC filters have C at least 470uF and R at least 100 ohms. So a supply which produces b The use of 3 filter RC sections 100ohms +470uF will give a 100Hz attenuation of 1/27,000 approximately. If Ia for 2 channels = 40mA, the Vripple is 0.2Vrms approx at C4, and will be lees than 0.1mV at C3. With only 4vdc across each 100 ohms, the B+ at C1 could be as low as +215V, but then if a given PT makes more than +265V, the R can be increased to still give 200V at C4. A plastic cap should also be used to bypass the final C4 electro, 0.47uF is OK. But I looked at the page at http://www.tubecad.com/2009/03/blog0161.htm The use of 6SN7 halves each with Ia = 10mA and Ea = 100V is not the right way to build a preamp IMHO. The RL are 10k only, which is only slightly more than Ra, and to get a 1/2 6SN7 to produce low THD/IMD and to sound its best the RL should be far higher than 10k for both the gain tube and CF. Therefore the use of CC Source for anode dc supply and the cathode CC Sink would be wise. There is utterly no need for Ia = 10mA; 6 mA is plenty, and signal will be clean. With a CCS to V1 anode, there is no Ia current change and only a voltage change. Therefore the use of an unbypassed Rk will not reduce gain which will be close to 20, or the µ of the tube because there is no local current FB, so it doesn't matter if you bypass the Rk or not. Anyone who can detect sonic differences can send their reports to the Society Of Doubtful Audiophiles, SODA, asap for official verification. Some slight variation to the CCDA preamp to allow for real world applications is at http://www.turneraudio.com.au/Line-preamp-2003.htm Patrick Turner. Cheers Ian Cheers Ian |
#6
Posted to rec.audio.tubes
|
|||
|
|||
JJ ECC99 for CCDA
On Fri, 01 May 2009 09:53:25 GMT, Patrick Turner
wrote: Some slight variation to the CCDA preamp to allow for real world applications is at http://www.turneraudio.com.au/Line-preamp-2003.htm Patrick Turner. Patrick, in that article you say that the most people are likely to want out of this would be 0.25V RMS, yet it demands power supplies of +250V, +125V and -125V. Are you entirely sure you have thought this through as well as you might? Also I wonder about your assumptions about the anode loading. The current source is not buffered, so the Early Effect comes into play. I reckon that for the MJE350 you are seeing an effective anode load purely in that transistor of about 50kohms (about 150k for the MJE340 in the cathode follower). So not only is the current source in the anode doing very little by way of constant current, but that in the following cathode is largely shunted by an indeterminate load resistance in whatever follows, so is also essentially redundant. d |
#7
Posted to rec.audio.tubes
|
|||
|
|||
JJ ECC99 for CCDA
Patrick Turner wrote:
Therefore the use of CC Source for anode dc supply and the cathode CC Sink would be wise. There is utterly no need for Ia = 10mA; 6 mA is plenty, and signal will be clean. With a CCS to V1 anode, there is no Ia current change and only a voltage change. Therefore the use of an unbypassed Rk will not reduce gain which will be close to 20, or the µ of the tube because there is no local current FB, so it doesn't matter if you bypass the Rk or not. Anyone who can detect sonic differences can send their reports to the Society Of Doubtful Audiophiles, SODA, asap for official verification. Except its not true (I made the same assumption in the past) at AC the valve is still seeing a load of 40k ( 50k pot || 120k feedback || CCS ) so the current through the cathode resistor will still vary with signal, so there will be local degeneration on the cathode, so reduction in gain and increase in anode resistance. -- Nick |
#8
Posted to rec.audio.tubes
|
|||
|
|||
JJ ECC99 for CCDA
flipper wrote:
On Fri, 01 May 2009 09:58:29 +0100, Ian Bell wrote: flipper wrote: On Thu, 30 Apr 2009 15:50:53 +0100, Ian Bell wrote: flipper wrote: On Thu, 30 Apr 2009 10:54:49 +0100, Ian Bell wrote: wrote: Hello All! I want to build a tube linestage based on the design I found yesterday called CCDA. This is the link of this circuit which I saw from the www.tubecad.com http://www.tubecad.com/2009/03/blog0161.htm http://www.tubecad.com/2009/04/blog0162.htm They also present a regulator called Feed-forward shunt regulator, is this regulator is a must for CCDA circuitry. Can I build a power supply with big reservoir along with gas tube regulator instead of using FFSR in the power supply? The whole point of the CCDA seems to be it draws a constant current from the supply and hence Quote:
the current draw is constant a straightforward supply with a big reservoir would be OK. However, note the following: 1. For good ripple reduction (always a must for preamps) you are better off with several RC decoupling stages than a single one. For example if you use a 10K resistor and 100uF capacitor you will get better ripple reduction if you replace it with five sets of 2K resistors and 20uF capacitors. Total resistance and capacitance is the same but ripple reduction is hugely improved. 2. IMHO the CCDA concept is a waste of time. In class A circuits with RC decoupling the current draw from the power supply already IS constant no matter what circuit topology is used. No it isn't. Power supply filtering has no impact on what the *circuit* current demand is. I never said it did. May not be what you 'meant' but it is what you said. Unless you're trying to play semantics by arguing 'some' RC filers are part of the power supply while 'other' RC power filters aren't. Sounds to me like you are the one who is playing semantics. No, power supply filtering is power supply filtering. Moment to moment current variation Which means it isn't constant despite you just saying it was.. The circuit downstream of the last, i.e. local, RC decoupling has moment to moment changes. Exactly, it isn't constant. If the final RC filter (i.e. local) is sufficient to be effective at reducing ripple it acts as a local charge store to provide for moment to moment circuit currents. That's what I said with "You can, of course, make the last filter cap arbitrarily large." That's the point you are missing. There's no need to make the last filter cap, by which I presume you mean the one AT the preamp, arbitrarily large. For heaven's sake, it has to be SOME value. How do you pick it? By the amount ripple, whether AC hum or signal induced. you're willing to live with. The less ripple you want the more filtering you need. You can make it as large as you like and, without a spec here, that's arbitrarily large, depending on how much filtering you want. "Arbitrarily large:" pick a ripple spec ---- pick filter large enough to achieve it. The current flowing from the supply is virtually constant. As I said, the effect of non constant current can be made arbitrarily small with arbitrarily large filtering. As I said, normal local RC decoupling values are more than sufficient to achieve this effect. Whatever the heck 'normal' means it's less if there's a constant current draw. There is no need to use an arbitrarily large cap. 'Normal' is 'arbitrarily large'. needs are provided by the C in the RC decoupling. Except it's on the outboard side of all that hum filtering. What are you talking about. I'm talking about your multi-stage RC filter suggestion. That is part of the power supply itself, not the preamp. As I thought. Playing semantics. Power supply filtering is power supply filtering. The CCDA circuit reference has R9 and C4 local decoupling. Yep. So? So, if the current is constant why have the decoupling?? Ask the designer. You can, of course, make the last filter cap arbitrarily large but it's moot with CCDA because current demand is constant. But any RC filter also present already has the same effect. An RC filter can only approximate the effect as you make it arbitrarily large. How many times do I have to say this, there is no need to make it arbitrarily large. You can say it all you like but it's still nothing more than you bitching about the choice of words. The fact of the matter is it needs to be larger when current isn't constant. In real life component tolerances will also prevent the CCDA from being a 'perfect 0'. So the current will not be constant. That's right. Nothing is perfect. Which might be a good reason to have a nominal filter as well. On the other hand, a thousand times less current variation (arbitrarily selected) might be insignificant enough. One of the benefits is reduced crosstalk because there's, theoretically, '0' signal ripple on the supply. Also one of the benefits of proper decoupling. 'Proper' is dependant on circuit needs and the needs are less with a constant current draw. Proper decoupling PROVIDES a near constant current draw, Filtering doesn't CHANGE the circuit's current draw ONE WHIT at least as good as a CCDA with normal tube tolerances. If you make the filtering arbitrarily large enough, yes. Just makes it a simpler to deal with if the current ripple is zero, or close to it. Look, no one said there's only 'one way' to do it. The point was a constant current draw eases the PS design and what does it cost past some clever thinking? Yes, that is the point.Exactly how has the power supply design for this preamp been eased. What circuit simplifications are possible in the PSU because of using a CCDA?? Less filtering needed and it comes for FREE, past expending a bit of mental effort. There is no less filtering. Having a constant current draw does not reduce the the requirements to reduce ripple. I see no simplification of the power supply. Cheers ian |
#9
Posted to rec.audio.tubes
|
|||
|
|||
JJ ECC99 for CCDA
Ian Bell wrote:
flipper wrote: On Fri, 01 May 2009 09:58:29 +0100, Ian Bell wrote: flipper wrote: On Thu, 30 Apr 2009 15:50:53 +0100, Ian Bell wrote: flipper wrote: On Thu, 30 Apr 2009 10:54:49 +0100, Ian Bell wrote: wrote: Hello All! I want to build a tube linestage based on the design I found yesterday called CCDA. This is the link of this circuit which I saw from the www.tubecad.com http://www.tubecad.com/2009/03/blog0161.htm http://www.tubecad.com/2009/04/blog0162.htm They also present a regulator called Feed-forward shunt regulator, is this regulator is a must for CCDA circuitry. Can I build a power supply with big reservoir along with gas tube regulator instead of using FFSR in the power supply? The whole point of the CCDA seems to be it draws a constant current from the supply and hence Quote:
straightforward supply with a big reservoir would be OK. However, note the following: 1. For good ripple reduction (always a must for preamps) you are better off with several RC decoupling stages than a single one. For example if you use a 10K resistor and 100uF capacitor you will get better ripple reduction if you replace it with five sets of 2K resistors and 20uF capacitors. Total resistance and capacitance is the same but ripple reduction is hugely improved. 2. IMHO the CCDA concept is a waste of time. In class A circuits with RC decoupling the current draw from the power supply already IS constant no matter what circuit topology is used. No it isn't. Power supply filtering has no impact on what the *circuit* current demand is. I never said it did. May not be what you 'meant' but it is what you said. Unless you're trying to play semantics by arguing 'some' RC filers are part of the power supply while 'other' RC power filters aren't. Sounds to me like you are the one who is playing semantics. No, power supply filtering is power supply filtering. Moment to moment current variation Which means it isn't constant despite you just saying it was.. The circuit downstream of the last, i.e. local, RC decoupling has moment to moment changes. Exactly, it isn't constant. If the final RC filter (i.e. local) is sufficient to be effective at reducing ripple it acts as a local charge store to provide for moment to moment circuit currents. That's what I said with "You can, of course, make the last filter cap arbitrarily large." That's the point you are missing. There's no need to make the last filter cap, by which I presume you mean the one AT the preamp, arbitrarily large. For heaven's sake, it has to be SOME value. How do you pick it? By the amount ripple, whether AC hum or signal induced. you're willing to live with. The less ripple you want the more filtering you need. You can make it as large as you like and, without a spec here, that's arbitrarily large, depending on how much filtering you want. "Arbitrarily large:" pick a ripple spec ---- pick filter large enough to achieve it. The current flowing from the supply is virtually constant. As I said, the effect of non constant current can be made arbitrarily small with arbitrarily large filtering. As I said, normal local RC decoupling values are more than sufficient to achieve this effect. Whatever the heck 'normal' means it's less if there's a constant current draw. There is no need to use an arbitrarily large cap. 'Normal' is 'arbitrarily large'. needs are provided by the C in the RC decoupling. Except it's on the outboard side of all that hum filtering. What are you talking about. I'm talking about your multi-stage RC filter suggestion. That is part of the power supply itself, not the preamp. As I thought. Playing semantics. Power supply filtering is power supply filtering. The CCDA circuit reference has R9 and C4 local decoupling. Yep. So? So, if the current is constant why have the decoupling?? Ask the designer. You can, of course, make the last filter cap arbitrarily large but it's moot with CCDA because current demand is constant. But any RC filter also present already has the same effect. An RC filter can only approximate the effect as you make it arbitrarily large. How many times do I have to say this, there is no need to make it arbitrarily large. You can say it all you like but it's still nothing more than you bitching about the choice of words. The fact of the matter is it needs to be larger when current isn't constant. In real life component tolerances will also prevent the CCDA from being a 'perfect 0'. So the current will not be constant. That's right. Nothing is perfect. Which might be a good reason to have a nominal filter as well. On the other hand, a thousand times less current variation (arbitrarily selected) might be insignificant enough. One of the benefits is reduced crosstalk because there's, theoretically, '0' signal ripple on the supply. Also one of the benefits of proper decoupling. 'Proper' is dependant on circuit needs and the needs are less with a constant current draw. Proper decoupling PROVIDES a near constant current draw, Filtering doesn't CHANGE the circuit's current draw ONE WHIT at least as good as a CCDA with normal tube tolerances. If you make the filtering arbitrarily large enough, yes. Just makes it a simpler to deal with if the current ripple is zero, or close to it. Look, no one said there's only 'one way' to do it. The point was a constant current draw eases the PS design and what does it cost past some clever thinking? Yes, that is the point.Exactly how has the power supply design for this preamp been eased. What circuit simplifications are possible in the PSU because of using a CCDA?? Less filtering needed and it comes for FREE, past expending a bit of mental effort. There is no less filtering. Having a constant current draw does not reduce the the requirements to reduce ripple. I see no simplification of the power supply. Cheers ian Are you sure, wouldn't supply rail signal also cancel out between the two stages? -- Nick |
#10
Posted to rec.audio.tubes
|
|||
|
|||
JJ ECC99 for CCDA
Nick Gorham wrote:
Are you sure, wouldn't supply rail signal also cancel out between the two stages? Thinking a bit more, no it wouldn't. -- Nick |
#11
Posted to rec.audio.tubes
|
|||
|
|||
JJ ECC99 for CCDA
Nick Gorham wrote:
Ian Bell wrote: flipper wrote: On Fri, 01 May 2009 09:58:29 +0100, Ian Bell wrote: flipper wrote: On Thu, 30 Apr 2009 15:50:53 +0100, Ian Bell wrote: flipper wrote: On Thu, 30 Apr 2009 10:54:49 +0100, Ian Bell wrote: wrote: Hello All! I want to build a tube linestage based on the design I found yesterday called CCDA. This is the link of this circuit which I saw from the www.tubecad.com http://www.tubecad.com/2009/03/blog0161.htm http://www.tubecad.com/2009/04/blog0162.htm They also present a regulator called Feed-forward shunt regulator, is this regulator is a must for CCDA circuitry. Can I build a power supply with big reservoir along with gas tube regulator instead of using FFSR in the power supply? The whole point of the CCDA seems to be it draws a constant current from the supply and hence Quote:
straightforward supply with a big reservoir would be OK. However, note the following: 1. For good ripple reduction (always a must for preamps) you are better off with several RC decoupling stages than a single one. For example if you use a 10K resistor and 100uF capacitor you will get better ripple reduction if you replace it with five sets of 2K resistors and 20uF capacitors. Total resistance and capacitance is the same but ripple reduction is hugely improved. 2. IMHO the CCDA concept is a waste of time. In class A circuits with RC decoupling the current draw from the power supply already IS constant no matter what circuit topology is used. No it isn't. Power supply filtering has no impact on what the *circuit* current demand is. I never said it did. May not be what you 'meant' but it is what you said. Unless you're trying to play semantics by arguing 'some' RC filers are part of the power supply while 'other' RC power filters aren't. Sounds to me like you are the one who is playing semantics. No, power supply filtering is power supply filtering. Moment to moment current variation Which means it isn't constant despite you just saying it was.. The circuit downstream of the last, i.e. local, RC decoupling has moment to moment changes. Exactly, it isn't constant. If the final RC filter (i.e. local) is sufficient to be effective at reducing ripple it acts as a local charge store to provide for moment to moment circuit currents. That's what I said with "You can, of course, make the last filter cap arbitrarily large." That's the point you are missing. There's no need to make the last filter cap, by which I presume you mean the one AT the preamp, arbitrarily large. For heaven's sake, it has to be SOME value. How do you pick it? By the amount ripple, whether AC hum or signal induced. you're willing to live with. The less ripple you want the more filtering you need. You can make it as large as you like and, without a spec here, that's arbitrarily large, depending on how much filtering you want. "Arbitrarily large:" pick a ripple spec ---- pick filter large enough to achieve it. The current flowing from the supply is virtually constant. As I said, the effect of non constant current can be made arbitrarily small with arbitrarily large filtering. As I said, normal local RC decoupling values are more than sufficient to achieve this effect. Whatever the heck 'normal' means it's less if there's a constant current draw. There is no need to use an arbitrarily large cap. 'Normal' is 'arbitrarily large'. needs are provided by the C in the RC decoupling. Except it's on the outboard side of all that hum filtering. What are you talking about. I'm talking about your multi-stage RC filter suggestion. That is part of the power supply itself, not the preamp. As I thought. Playing semantics. Power supply filtering is power supply filtering. The CCDA circuit reference has R9 and C4 local decoupling. Yep. So? So, if the current is constant why have the decoupling?? Ask the designer. You can, of course, make the last filter cap arbitrarily large but it's moot with CCDA because current demand is constant. But any RC filter also present already has the same effect. An RC filter can only approximate the effect as you make it arbitrarily large. How many times do I have to say this, there is no need to make it arbitrarily large. You can say it all you like but it's still nothing more than you bitching about the choice of words. The fact of the matter is it needs to be larger when current isn't constant. In real life component tolerances will also prevent the CCDA from being a 'perfect 0'. So the current will not be constant. That's right. Nothing is perfect. Which might be a good reason to have a nominal filter as well. On the other hand, a thousand times less current variation (arbitrarily selected) might be insignificant enough. One of the benefits is reduced crosstalk because there's, theoretically, '0' signal ripple on the supply. Also one of the benefits of proper decoupling. 'Proper' is dependant on circuit needs and the needs are less with a constant current draw. Proper decoupling PROVIDES a near constant current draw, Filtering doesn't CHANGE the circuit's current draw ONE WHIT at least as good as a CCDA with normal tube tolerances. If you make the filtering arbitrarily large enough, yes. Just makes it a simpler to deal with if the current ripple is zero, or close to it. Look, no one said there's only 'one way' to do it. The point was a constant current draw eases the PS design and what does it cost past some clever thinking? Yes, that is the point.Exactly how has the power supply design for this preamp been eased. What circuit simplifications are possible in the PSU because of using a CCDA?? Less filtering needed and it comes for FREE, past expending a bit of mental effort. There is no less filtering. Having a constant current draw does not reduce the the requirements to reduce ripple. I see no simplification of the power supply. Cheers ian Are you sure, wouldn't supply rail signal also cancel out between the two stages? That's the idea behind the CCDA but you still need to reduce mains ripple the dame a before. Cheers ian |
#12
Posted to rec.audio.tubes
|
|||
|
|||
JJ ECC99 for CCDA
Nick Gorham wrote:
Nick Gorham wrote: Are you sure, wouldn't supply rail signal also cancel out between the two stages? Thinking a bit more, no it wouldn't. LOL Ian |
#13
Posted to rec.audio.tubes
|
|||
|
|||
JJ ECC99 for CCDA
flipper wrote:
On Fri, 01 May 2009 16:06:49 +0100, Ian Bell wrote: flipper wrote: On Fri, 01 May 2009 09:58:29 +0100, Ian Bell wrote: flipper wrote: On Thu, 30 Apr 2009 15:50:53 +0100, Ian Bell wrote: flipper wrote: On Thu, 30 Apr 2009 10:54:49 +0100, Ian Bell wrote: wrote: Hello All! I want to build a tube linestage based on the design I found yesterday called CCDA. This is the link of this circuit which I saw from the www.tubecad.com http://www.tubecad.com/2009/03/blog0161.htm http://www.tubecad.com/2009/04/blog0162.htm They also present a regulator called Feed-forward shunt regulator, is this regulator is a must for CCDA circuitry. Can I build a power supply with big reservoir along with gas tube regulator instead of using FFSR in the power supply? The whole point of the CCDA seems to be it draws a constant current from the supply and hence Quote:
the current draw is constant a straightforward supply with a big reservoir would be OK. However, note the following: 1. For good ripple reduction (always a must for preamps) you are better off with several RC decoupling stages than a single one. For example if you use a 10K resistor and 100uF capacitor you will get better ripple reduction if you replace it with five sets of 2K resistors and 20uF capacitors. Total resistance and capacitance is the same but ripple reduction is hugely improved. 2. IMHO the CCDA concept is a waste of time. In class A circuits with RC decoupling the current draw from the power supply already IS constant no matter what circuit topology is used. No it isn't. Power supply filtering has no impact on what the *circuit* current demand is. I never said it did. May not be what you 'meant' but it is what you said. Unless you're trying to play semantics by arguing 'some' RC filers are part of the power supply while 'other' RC power filters aren't. Sounds to me like you are the one who is playing semantics. No, power supply filtering is power supply filtering. Moment to moment current variation Which means it isn't constant despite you just saying it was.. The circuit downstream of the last, i.e. local, RC decoupling has moment to moment changes. Exactly, it isn't constant. If the final RC filter (i.e. local) is sufficient to be effective at reducing ripple it acts as a local charge store to provide for moment to moment circuit currents. That's what I said with "You can, of course, make the last filter cap arbitrarily large." That's the point you are missing. There's no need to make the last filter cap, by which I presume you mean the one AT the preamp, arbitrarily large. For heaven's sake, it has to be SOME value. How do you pick it? By the amount ripple, whether AC hum or signal induced. you're willing to live with. The less ripple you want the more filtering you need. You can make it as large as you like and, without a spec here, that's arbitrarily large, depending on how much filtering you want. "Arbitrarily large:" pick a ripple spec ---- pick filter large enough to achieve it. The current flowing from the supply is virtually constant. As I said, the effect of non constant current can be made arbitrarily small with arbitrarily large filtering. As I said, normal local RC decoupling values are more than sufficient to achieve this effect. Whatever the heck 'normal' means it's less if there's a constant current draw. There is no need to use an arbitrarily large cap. 'Normal' is 'arbitrarily large'. needs are provided by the C in the RC decoupling. Except it's on the outboard side of all that hum filtering. What are you talking about. I'm talking about your multi-stage RC filter suggestion. That is part of the power supply itself, not the preamp. As I thought. Playing semantics. Power supply filtering is power supply filtering. The CCDA circuit reference has R9 and C4 local decoupling. Yep. So? So, if the current is constant why have the decoupling?? Ask the designer. You can, of course, make the last filter cap arbitrarily large but it's moot with CCDA because current demand is constant. But any RC filter also present already has the same effect. An RC filter can only approximate the effect as you make it arbitrarily large. How many times do I have to say this, there is no need to make it arbitrarily large. You can say it all you like but it's still nothing more than you bitching about the choice of words. The fact of the matter is it needs to be larger when current isn't constant. In real life component tolerances will also prevent the CCDA from being a 'perfect 0'. So the current will not be constant. That's right. Nothing is perfect. Which might be a good reason to have a nominal filter as well. On the other hand, a thousand times less current variation (arbitrarily selected) might be insignificant enough. One of the benefits is reduced crosstalk because there's, theoretically, '0' signal ripple on the supply. Also one of the benefits of proper decoupling. 'Proper' is dependant on circuit needs and the needs are less with a constant current draw. Proper decoupling PROVIDES a near constant current draw, Filtering doesn't CHANGE the circuit's current draw ONE WHIT at least as good as a CCDA with normal tube tolerances. If you make the filtering arbitrarily large enough, yes. Just makes it a simpler to deal with if the current ripple is zero, or close to it. Look, no one said there's only 'one way' to do it. The point was a constant current draw eases the PS design and what does it cost past some clever thinking? Yes, that is the point.Exactly how has the power supply design for this preamp been eased. What circuit simplifications are possible in the PSU because of using a CCDA?? Less filtering needed and it comes for FREE, past expending a bit of mental effort. There is no less filtering. Having a constant current draw does not reduce the the requirements to reduce ripple. You just love the heck out of semantic word games. The "requirement to reduce ripple" is based on what you pick as a design maximum. Having no signal current is 0 ripple, which I suspect is less than your 'requirement'. Ah, now I see where we are misunderstanding each other. The 'requirement to reduce ripple' that I am talking about is mains hum ripple on the HT. You seem to think I mean signal current ripple - I don't. I see no simplification of the power supply. If you don't see that reducing 0 to 0 is simpler then that's your problem. Look, CCDA or not, there is a need to reduce mains ripple to an acceptable level. That is what primarily determines the values of PSU smoothing and down stream RC decoupling components. The decoupling RC at the pre-amp are part of this and as a by product of reducing HT hum to acceptable levels they also smooth out signal ripple currents. Hence my original comment that IMHO the CCDA ia waste of time. Cheers Ian Cheers ian |
#14
Posted to rec.audio.tubes
|
|||
|
|||
JJ ECC99 for CCDA
Don Pearce wrote: On Fri, 01 May 2009 09:53:25 GMT, Patrick Turner wrote: Some slight variation to the CCDA preamp to allow for real world applications is at http://www.turneraudio.com.au/Line-preamp-2003.htm Patrick Turner. Patrick, in that article you say that the most people are likely to want out of this would be 0.25V RMS, yet it demands power supplies of +250V, +125V and -125V. Are you entirely sure you have thought this through as well as you might? The tubes concerned with loads as shown work best with the voltages shown. By "best" I mean with low THD and a normal comfortable dynamic cieling of 40dB above 0.25Vrms. On rarely ever uses the ceiling provided by tubes but that's the nature of tube use. But please don't let me hold you up if you insist on using an opamp operating from +/- 10Vdc. Just don't insist on me having to use an opamp, or on me having to use tubes with +/-12V supplies. Also I wonder about your assumptions about the anode loading. The current source is not buffered, so the Early Effect comes into play. I reckon that for the MJE350 you are seeing an effective anode load purely in that transistor of about 50kohms (about 150k for the MJE340 in the cathode follower). So not only is the current source in the anode doing very little by way of constant current, but that in the following cathode is largely shunted by an indeterminate load resistance in whatever follows, so is also essentially redundant. If you'd done your home work, you'd have built a sample of my circuit and discovered perhaps to your amazement that indeed the effective collector resistance of the many applications of my single BJT CCS is extremely high, and far more than the 50k you cite. Build the circuit, then measure the voltage change across the emitter resistance and hence emitter current change for a given large voltage signal at the anode/collector. Then divide collector voltage change by colector current change to find the effective collector resistance. You will find it is a huge figure like a pentode with large value cathode resistance. People keep lumbering me with criticisms that the Early Effect stops the BJT as I have it from having a high enough effective collector resistance that makes it into what can be considered to be a CCS. They don't offer me any proof that the Early Effect will rain on my parade. I've used similar CCS in LTPs and as long as the anode loads are within 1% of having the same resistance, so are the two anode voltages. In a typical 6SN7 with say 47k per anode, the ONLY way to get virtually equal anode + and - voltages is if the CCS common cathode tail has many megohms of resistance. Patrick Turner. d |
#15
Posted to rec.audio.tubes
|
|||
|
|||
JJ ECC99 for CCDA
Ian Bell wrote: flipper wrote: On Fri, 01 May 2009 09:58:29 +0100, Ian Bell wrote: flipper wrote: On Thu, 30 Apr 2009 15:50:53 +0100, Ian Bell wrote: flipper wrote: On Thu, 30 Apr 2009 10:54:49 +0100, Ian Bell wrote: wrote: Hello All! Snip.... Look, no one said there's only 'one way' to do it. The point was a constant current draw eases the PS design and what does it cost past some clever thinking? Yes, that is the point.Exactly how has the power supply design for this preamp been eased. What circuit simplifications are possible in the PSU because of using a CCDA?? Less filtering needed and it comes for FREE, past expending a bit of mental effort. There is no less filtering. Having a constant current draw does not reduce the the requirements to reduce ripple. I see no simplification of the power supply. Where you have a CCS operating from a noisy B+ rail, then the low Ra of a triode acts to form a resistance divider with the CCS. Because the low Ra is so much lower than the CCS actual R value, noise in the B+ rail cannot affect triode gain stage. But where a following CF buffer stage had its anode fed with noisy B+ rail, some of the noise will appear at the cathode output. So B+ filtering always needs to be good. Patrick Turner. Cheers ian |
#16
Posted to rec.audio.tubes
|
|||
|
|||
JJ ECC99 for CCDA
Nick Gorham wrote: Patrick Turner wrote: Therefore the use of CC Source for anode dc supply and the cathode CC Sink would be wise. There is utterly no need for Ia = 10mA; 6 mA is plenty, and signal will be clean. With a CCS to V1 anode, there is no Ia current change and only a voltage change. Therefore the use of an unbypassed Rk will not reduce gain which will be close to 20, or the µ of the tube because there is no local current FB, so it doesn't matter if you bypass the Rk or not. Anyone who can detect sonic differences can send their reports to the Society Of Doubtful Audiophiles, SODA, asap for official verification. Except its not true (I made the same assumption in the past) at AC the valve is still seeing a load of 40k ( 50k pot || 120k feedback || CCS ) so the current through the cathode resistor will still vary with signal, so there will be local degeneration on the cathode, so reduction in gain and increase in anode resistance. OK, in many of my circuits there is a CCS to feed the wanted idle dc flow to the triode. This avoids a resistance. To get from Ea = +125V to the B+ of +250V, and at say 6mA, the R if used would have to be about 20k. In addition to this there may well be a gain pot and a FB shunt resistance who's total R is say 40k, and 40k AC load in parallel with 20k is about 13.3k, and only about 1.3Ra, a load value which will always produce much more THD/IMD than if you had a total anode load of 40k, which will be about 4Ra. Have a look at my page on another preamp, http://www.turneraudio.com.au/preamp...rated-2006.htm The µ-follower stages shown have the voltage gain produced by the bottom triodes at very low THD because the effective anode circuit is kept as high as possible. And yet another... http://www.turneraudio.com.au/preamp...e+psu-2005.htm This one uses a CCS feed to a normal common cathode but also with shunt NFB and a gain pot. THD is extremely low because the anode load is high relative to Ra, and there is shunt NFB. Gain is only 4x, or +12dB, so a maximum 1.4Vrms CD player signal can be fed in and it comes out at 5.6Vrms maximum and still at low THD. Having the gain tube before the gain pot means there tube noise is also attenuated and SNR is excellent. Patrick Turner. -- Nick |
#17
Posted to rec.audio.tubes
|
|||
|
|||
JJ ECC99 for CCDA
On Sun, 03 May 2009 04:59:59 GMT, Patrick Turner
wrote: Don Pearce wrote: On Fri, 01 May 2009 09:53:25 GMT, Patrick Turner wrote: Some slight variation to the CCDA preamp to allow for real world applications is at http://www.turneraudio.com.au/Line-preamp-2003.htm Patrick Turner. Patrick, in that article you say that the most people are likely to want out of this would be 0.25V RMS, yet it demands power supplies of +250V, +125V and -125V. Are you entirely sure you have thought this through as well as you might? The tubes concerned with loads as shown work best with the voltages shown. By "best" I mean with low THD and a normal comfortable dynamic cieling of 40dB above 0.25Vrms. On rarely ever uses the ceiling provided by tubes but that's the nature of tube use. But please don't let me hold you up if you insist on using an opamp operating from +/- 10Vdc. Just don't insist on me having to use an opamp, or on me having to use tubes with +/-12V supplies. Wouldn't dream of it, you already have everything you need right there. The pair of BJTs in there in the signal path (and yes, they really are despite being configured as loads), why not take out that superfluous pair of valves, and do the job just with the trannies. Those valves really are there simply as something glowing to look at as you have vastly more gain than you need without them. As for signal headroom, you need to look at that in system terms. You say you have 40dB - what on earth for? Do you think the following amplifier is going to accept 250V of signal with any kind of equanimity? There is no point taking headroom in any stage beyond the clipping point of a subsequent stage - all you are doing is misaligning successive dynamic ranges to the detriment of both distortion and noise floor. Also I wonder about your assumptions about the anode loading. The current source is not buffered, so the Early Effect comes into play. I reckon that for the MJE350 you are seeing an effective anode load purely in that transistor of about 50kohms (about 150k for the MJE340 in the cathode follower). So not only is the current source in the anode doing very little by way of constant current, but that in the following cathode is largely shunted by an indeterminate load resistance in whatever follows, so is also essentially redundant. If you'd done your home work, you'd have built a sample of my circuit and discovered perhaps to your amazement that indeed the effective collector resistance of the many applications of my single BJT CCS is extremely high, and far more than the 50k you cite. Build the circuit, then measure the voltage change across the emitter resistance and hence emitter current change for a given large voltage signal at the anode/collector. Then divide collector voltage change by colector current change to find the effective collector resistance. You will find it is a huge figure like a pentode with large value cathode resistance. People keep lumbering me with criticisms that the Early Effect stops the BJT as I have it from having a high enough effective collector resistance that makes it into what can be considered to be a CCS. They don't offer me any proof that the Early Effect will rain on my parade. I've used similar CCS in LTPs and as long as the anode loads are within 1% of having the same resistance, so are the two anode voltages. In a typical 6SN7 with say 47k per anode, the ONLY way to get virtually equal anode + and - voltages is if the CCS common cathode tail has many megohms of resistance. We don't even need to worry about Early effect - you have described in the text how the current source is completely slugged by resistors hanging across it (to the tune of 40k). And of course there should be no AC loading on the B+ line - you use local decoupling to take care of that. d |
#18
Posted to rec.audio.tubes
|
|||
|
|||
JJ ECC99 for CCDA
Patrick Turner wrote:
Nick Gorham wrote: Patrick Turner wrote: Therefore the use of CC Source for anode dc supply and the cathode CC Sink would be wise. There is utterly no need for Ia = 10mA; 6 mA is plenty, and signal will be clean. With a CCS to V1 anode, there is no Ia current change and only a voltage change. Therefore the use of an unbypassed Rk will not reduce gain which will be close to 20, or the µ of the tube because there is no local current FB, so it doesn't matter if you bypass the Rk or not. Anyone who can detect sonic differences can send their reports to the Society Of Doubtful Audiophiles, SODA, asap for official verification. Except its not true (I made the same assumption in the past) at AC the valve is still seeing a load of 40k ( 50k pot || 120k feedback || CCS ) so the current through the cathode resistor will still vary with signal, so there will be local degeneration on the cathode, so reduction in gain and increase in anode resistance. OK, in many of my circuits there is a CCS to feed the wanted idle dc flow to the triode. This avoids a resistance. To get from Ea = +125V to the B+ of +250V, and at say 6mA, the R if used would have to be about 20k. In addition to this there may well be a gain pot and a FB shunt resistance who's total R is say 40k, and 40k AC load in parallel with 20k is about 13.3k, and only about 1.3Ra, a load value which will always produce much more THD/IMD than if you had a total anode load of 40k, which will be about 4Ra. Yep thats all good. But I was just referring to your statement that the cathode resistor doesn't require bypassing because of the CCS in the anode load. I made exactly this mistake myself assuming the CCS would hold the signal current constant so a bypass was not needed. In fact the signal current through the cathode resistor (in this case) is just the same as if you used aprox 40k anode load. As to if you want to bypass or not is another question, I was just making people aware that if its not bypassed gain will be reduced, and effective anode resistance will be increased. -- Nick |
#19
Posted to rec.audio.tubes
|
|||
|
|||
JJ ECC99 for CCDA
flipper wrote:
major snippage There will be 'some' signal ripple rejection from the last filter but, under the best of conditions, the preceding 'hum' ripple stages are irrelevant. Agreed that preceding hum ripple stages are nor relevant to this discussion. I.E. If they have no droop then the last filter's droop (from signal) is the 'max' but anything less than 'perfect' in the preceding stages simply aggravates the matter. The problem can be highlighted by considering a 120Hz signal tone. If it took 4 RC stages on the supply side to get 120Hz hum to acceptable levels then only 1 cap on the signal side might not be enough and the situation only gets worse as we go on down to 20Hz. That is the key question. If the triodes are matched to 10% then the signal current will only cancel to 10% of its original value. Let's be generous and assume the triodes match to within 5% that will reduce the signal current in the HT by 26dB, theoretically at all frequencies. I have just simulated a 6SN7 CC stage with a 33K anode load, 1K cathode resistor and final stage decoupling consisting of a modest 10uF cap with a 4K7 series resistor. I checked the ac current in the anode resistor and the 4K7 decoupling resistor at 120Hz and it is 30dB below the current in the anode resistor. So at 120Hz it looks as good as the CCDA. Yes it will be worse at lower frequencies. At 20Hz the signal ripple is just 15dB down. At higher frequencies it is of course better. At 1KHz it is nearly 50dB down. Now, granted, that's not rigorous, because it depends on impedance, current draw, and other factors, but, the point is, simply having enough 'on the front' to handle hum doesn't necessarily mean you're walking in high cotton on the signal side. Perhaps not, but the simulation above indicates that in the vast majority of cases it will be more than enough to beat the CCDA hands down. Cheers Ian It's also worth looking at his whole design because that's not the only 'enhancement' he used. In particular, he used an active feed forward ripple reducer to, again, save on caps. That means, in the simplest case, there's hardly any output reservoir cap at all. No wonder he added 'just one more', I would've added one anyway just to make sure any potential MOSFET noise didn't get through. Now, whether anyone can 'hear the difference' I don't know but I've noticed that Broskie is almost obsessed with PSRR in not only this one but his Akido amp, which gets rave reviews, so just maybe it's worth considering. Cheers Ian Cheers ian |
#20
Posted to rec.audio.tubes
|
|||
|
|||
JJ ECC99 for CCDA
Don Pearce wrote: On Sun, 03 May 2009 04:59:59 GMT, Patrick Turner wrote: Don Pearce wrote: On Fri, 01 May 2009 09:53:25 GMT, Patrick Turner wrote: Some slight variation to the CCDA preamp to allow for real world applications is at http://www.turneraudio.com.au/Line-preamp-2003.htm Patrick Turner. Patrick, in that article you say that the most people are likely to want out of this would be 0.25V RMS, yet it demands power supplies of +250V, +125V and -125V. Are you entirely sure you have thought this through as well as you might? The tubes concerned with loads as shown work best with the voltages shown. By "best" I mean with low THD and a normal comfortable dynamic cieling of 40dB above 0.25Vrms. On rarely ever uses the ceiling provided by tubes but that's the nature of tube use. But please don't let me hold you up if you insist on using an opamp operating from +/- 10Vdc. Just don't insist on me having to use an opamp, or on me having to use tubes with +/-12V supplies. Wouldn't dream of it, you already have everything you need right there. The pair of BJTs in there in the signal path (and yes, they really are despite being configured as loads), why not take out that superfluous pair of valves, and do the job just with the trannies. Those valves really are there simply as something glowing to look at as you have vastly more gain than you need without them. I doubt you really belong on this news group. You seem like the guy who tries to convince those who like to saila around the bay that they hsould furl their sails, and chop down their mast, and use a ****ing diesel engine. I've been accused of using transistors as part of a hybrid circuit by quite a few clowns before your sort came along. The use of a transistor CCS simply avoids having to use a pentode CCS or high value choke of some other type of high ac resistance dc supply. because the bjt CCS **is a a virtual CCS** the bjt contribution to distortion artifacts in the signal or any possible deleterious coloration of the signal is **impossible**. Indded, the bjts act as very willing slaves to enable what the tubes want to do, ie, amplify a line level signal with minimum distortion for such a low amount of applied NFB. YOU could easily devise whatever BJT contraption you wish, and moor around the bay if you wish, but I am happy to rely on the wind, and free of the smell and noise of a diesel engine. In essence, I'll do it my way thanks. As for signal headroom, you need to look at that in system terms. You say you have 40dB - what on earth for? Do you think the following amplifier is going to accept 250V of signal with any kind of equanimity? There is no point taking headroom in any stage beyond the clipping point of a subsequent stage - all you are doing is misaligning successive dynamic ranges to the detriment of both distortion and noise floor. Bull****. If I have to make enough voltage to make an insensitive power amp clip, such as williamson, 2Vrms, or Quad-II, 1.4Vrms, then the preamp fidelity could happily benefit by having at least 20dB headroom above the voltages just quoted, ie able to easily make 20Vrms before the preamp clips. Most tube preamps can make 50Vrms at clipping with say 5% THD, mainly 2H, without any loop NFB, typical with 6SN7, 12AU7 etc. With loop FB the THD is reduced to maybe 2% with gain at 1/4 of the open loop gain at 50Vrms. Therefore since we might be feeding a preamp with a CD signal of 1.4vrms max, then Vo = 5.6Vrms and THD will be perhaps 0.2%. But most intelligent people wouldn't use any gain stage between CD player and their power amp, but only whe they have to use a low level input signal typical of older gear, maybe 200mV, and the the preamp output is only 0.8Vrms and THD = 0.13% max. The lower the B+ voltages, the lower the dynamic range and the higher is the THD for the same output voltage. Intelligent ppl wanting to use a CD player won't need any gain, but will only need a buffer stage with unity gain after the gain attenuator, so the gain stage can be bypassed with CD player signal fed directly to gain attenuator. A CF buffer is a nice sounding solution. Also I wonder about your assumptions about the anode loading. The current source is not buffered, so the Early Effect comes into play. I reckon that for the MJE350 you are seeing an effective anode load purely in that transistor of about 50kohms (about 150k for the MJE340 in the cathode follower). So not only is the current source in the anode doing very little by way of constant current, but that in the following cathode is largely shunted by an indeterminate load resistance in whatever follows, so is also essentially redundant. If you'd done your home work, you'd have built a sample of my circuit and discovered perhaps to your amazement that indeed the effective collector resistance of the many applications of my single BJT CCS is extremely high, and far more than the 50k you cite. Build the circuit, then measure the voltage change across the emitter resistance and hence emitter current change for a given large voltage signal at the anode/collector. Then divide collector voltage change by colector current change to find the effective collector resistance. You will find it is a huge figure like a pentode with large value cathode resistance. People keep lumbering me with criticisms that the Early Effect stops the BJT as I have it from having a high enough effective collector resistance that makes it into what can be considered to be a CCS. They don't offer me any proof that the Early Effect will rain on my parade. I've used similar CCS in LTPs and as long as the anode loads are within 1% of having the same resistance, so are the two anode voltages. In a typical 6SN7 with say 47k per anode, the ONLY way to get virtually equal anode + and - voltages is if the CCS common cathode tail has many megohms of resistance. We don't even need to worry about Early effect - you have described in the text how the current source is completely slugged by resistors hanging across it (to the tune of 40k). And of course there should be no AC loading on the B+ line - you use local decoupling to take care of that. All resistance loading of triodes is essentially evil because of the increasing distortion as the load value becomes lower. CCS loading or loading where RL 20Ra is the ideal. But ideals cannot always be met. I will always try to have loads on mediu, µ triodes to be 4Ra or higher. People using triodes for line level stages to get the best sound will do well to have the highest load possible. If you examine the anode Ra curves for triodes, you'll see that loads approximating close to a horizontal load line give excellent natural linearity, regardless of the Ia. A 6SN7 triode will thus give excellent linearity anywhere between 3mA and 10mA. Pentodes could also be used with a CCS like I have shown but then the gain ( and distortion ) becomes enormous, and way to high for any line preamp, so then the shunt FB network with similar R divider values used for the triode will then limit gain to the same mild +12dB that is wanted. It means open loop gain of perhaps 300 is reduced to 4, a drop by a factor of 1/75, and 7% open loop THD at 50Vrms is reduced to 0.12% approx, and at 5Vrms output perhaps its only 0.01%. Whether the pentode with a shunt FB network sounds better than the triode is a moot point, but such things were used in much pro gear of the 1950s and 60s. There is nothing at all to stop you using a Ziclai connected pair of bjts for a gain stage to do the same job as a pair of triodes. Be my guest, and if it sounds as well as the triodes, they you're lucky. There must be 101 ways of building a line stage preamp. I have my reasons for doing the tube stages my way as shown at my website. I have also built solid state amps, and I'm sure the nit-pickers of this world will vehemently disagree with my ways of doing things. I am not moved by their rantings, and will only take any notice if they quit the bull**** and set up a website to illustrate their principles backed up with a large number of tested examples of their work. Patrick Turner. d |
#21
Posted to rec.audio.tubes
|
|||
|
|||
JJ ECC99 for CCDA
Nick Gorham wrote: Patrick Turner wrote: Nick Gorham wrote: Patrick Turner wrote: Therefore the use of CC Source for anode dc supply and the cathode CC Sink would be wise. There is utterly no need for Ia = 10mA; 6 mA is plenty, and signal will be clean. With a CCS to V1 anode, there is no Ia current change and only a voltage change. Therefore the use of an unbypassed Rk will not reduce gain which will be close to 20, or the µ of the tube because there is no local current FB, so it doesn't matter if you bypass the Rk or not. Anyone who can detect sonic differences can send their reports to the Society Of Doubtful Audiophiles, SODA, asap for official verification. Except its not true (I made the same assumption in the past) at AC the valve is still seeing a load of 40k ( 50k pot || 120k feedback || CCS ) so the current through the cathode resistor will still vary with signal, so there will be local degeneration on the cathode, so reduction in gain and increase in anode resistance. OK, in many of my circuits there is a CCS to feed the wanted idle dc flow to the triode. This avoids a resistance. To get from Ea = +125V to the B+ of +250V, and at say 6mA, the R if used would have to be about 20k. In addition to this there may well be a gain pot and a FB shunt resistance who's total R is say 40k, and 40k AC load in parallel with 20k is about 13.3k, and only about 1.3Ra, a load value which will always produce much more THD/IMD than if you had a total anode load of 40k, which will be about 4Ra. Yep thats all good. But I was just referring to your statement that the cathode resistor doesn't require bypassing because of the CCS in the anode load. I made exactly this mistake myself assuming the CCS would hold the signal current constant so a bypass was not needed. In fact the signal current through the cathode resistor (in this case) is just the same as if you used aprox 40k anode load. As to if you want to bypass or not is another question, I was just making people aware that if its not bypassed gain will be reduced, and effective anode resistance will be increased. OK, fair enough. But in a case of a µ-follower stage where the bottom gain triode has a very high RL, or in the case of a gain stage + CF buffer where the gain stage as a CCS, then the cathode signal current in an Rk is very low so gain approaches the µ of the triode. Even with a 12AU7 with unbypased Rk you get gain = 15x including the CF. This is too much for most preamps, so then you have to reduce it with a shunt NFB network which does not load down the gain stage, but which itself isn't too high an impedance path. With 12AU7, and 40k RL, open gain max with bypassing is about 12, and unbypassed its about 8. One reason to bypass the Rk is to keep the effective Rout as low as possible. With an unbypassed Rk, Rout effectively becomes RL in parallel with (Ra + {[µ+1] x Rk}). For best bandwidth you want Rout as low as possible, so bypassing isn't the evil nelly its made out to be. Patrick Turner. -- Nick |
#22
Posted to rec.audio.tubes
|
|||
|
|||
JJ ECC99 for CCDA
On Sun, 03 May 2009 15:20:45 GMT, Patrick Turner
wrote: Patrick Turner. Patrick, having had a proper look at the circuit and its function, I can see that what you have designed is a chindogu.It will still function perfectly if you remove every component apart from the sockets, switch and pot - all you need to do is turn the pot ever so slightly more clockwise than would otherwise be the case. d |
#23
Posted to rec.audio.tubes
|
|||
|
|||
JJ ECC99 for CCDA
Don Pearce wrote: On Sun, 03 May 2009 15:20:45 GMT, Patrick Turner wrote: Patrick Turner. Patrick, having had a proper look at the circuit and its function, I can see that what you have designed is a chindogu.It will still function perfectly if you remove every component apart from the sockets, switch and pot - all you need to do is turn the pot ever so slightly more clockwise than would otherwise be the case. I think some gem of wisdom is lost on me. I am not fully equipped to understand everything, such as what a "chindogu" is. Maybe its a special American way of saying something is especially badly designed. I intend to keep my mast, and my sails, and to not instal a bigger diesel engine to "pull better lookin chicks". Patrick Turner. d |
#24
Posted to rec.audio.tubes
|
|||
|
|||
JJ ECC99 for CCDA
On Tue, 05 May 2009 11:05:41 GMT, Patrick Turner
wrote: Don Pearce wrote: On Sun, 03 May 2009 15:20:45 GMT, Patrick Turner wrote: Patrick Turner. Patrick, having had a proper look at the circuit and its function, I can see that what you have designed is a chindogu.It will still function perfectly if you remove every component apart from the sockets, switch and pot - all you need to do is turn the pot ever so slightly more clockwise than would otherwise be the case. I think some gem of wisdom is lost on me. I am not fully equipped to understand everything, such as what a "chindogu" is. Maybe its a special American way of saying something is especially badly designed. No it is a Japanese thing. What it amounts to is a completely pointless invention - look it up, there are some very amusing ones. Your preamp falls nicely into that category, because everything inside it is entirely pointless, since there is already enough signal going in. As I say above, all that is actually needed is a bunch of sockets, a switch and a pot - job done. I intend to keep my mast, and my sails, and to not instal a bigger diesel engine to "pull better lookin chicks". But that is precisely what you have done - all those sexy glowing tubes, and all. The diesel engine analogy is actually rather apt - hot and noisy. d |
#25
Posted to rec.audio.tubes
|
|||
|
|||
JJ ECC99 for CCDA
Don Pearce wrote: On Tue, 05 May 2009 11:05:41 GMT, Patrick Turner wrote: Don Pearce wrote: On Sun, 03 May 2009 15:20:45 GMT, Patrick Turner wrote: Patrick Turner. Patrick, having had a proper look at the circuit and its function, I can see that what you have designed is a chindogu.It will still function perfectly if you remove every component apart from the sockets, switch and pot - all you need to do is turn the pot ever so slightly more clockwise than would otherwise be the case. I think some gem of wisdom is lost on me. I am not fully equipped to understand everything, such as what a "chindogu" is. Maybe its a special American way of saying something is especially badly designed. No it is a Japanese thing. What it amounts to is a completely pointless invention - look it up, there are some very amusing ones. Your preamp falls nicely into that category, because everything inside it is entirely pointless, since there is already enough signal going in. As I say above, all that is actually needed is a bunch of sockets, a switch and a pot - job done. I intend to keep my mast, and my sails, and to not instal a bigger diesel engine to "pull better lookin chicks". But that is precisely what you have done - all those sexy glowing tubes, and all. The diesel engine analogy is actually rather apt - hot and noisy. Ah, your post is so elegantly dismissive of so much. Were we to apply your philosophy to the world we could shoot 3/4 of everyone and all they have made, because they are merely suplerflous, and their need for possessions is all just irrational greed and stupidity thus ruining a good world. We might begin at least with the bankers and wayward financiers of recent times. Then for some after dinner sport we could atom bomb the backyard of Pakistan and save having to go over and dig out Osama and his ratbag mates from the limestone caves. Do you really need a wife to cook your dinner, and darn your socks? It'd be so much easier to abreviate any need for her complex and boring perceptions about houskeeping and surely you should rationalise the amount of time you have to waste listening to her small talk about grocery prices. I kinda think you'd be on your own pretty soon if you consistently applied your ideas, and tried to run for President. Meanwhile, I have said to many people that all they need to get a signal from a CD player to a speaker is an integrated amplifier or perhaps a little box with a 50k pot and source select switch inside, and a pair of power amps. This all works fine for some folks until they have a low level source, insensitive power amps and insensitive speakers. Then gain is needed. No matter what I say, people contact me to build them a preamp and often the reason is because they just want to try something with tubes. That 2003 preamp with grey anodized case gave the owner a fine item which he said sounded fabulous. Meanwhile, I'd suggest that whenever we look at the high-end audio gear products made in the US, we see 40dB more chingdogu than what one might see in my amps, if you insist on examinations and reformations. IMHO, the best way to listen to music is to go where real musicians and singers are performing without any electronic aids whatsover. Take your dear lady wife and friends and enjoy the outing. Try to stop yourself making dumb comments about what they are wearing and their hairstyles. Patrick Turner. d |
#26
Posted to rec.audio.tubes
|
|||
|
|||
JJ ECC99 for CCDA
On Tue, 05 May 2009 15:22:33 GMT, Patrick Turner
wrote: Don Pearce wrote: On Tue, 05 May 2009 11:05:41 GMT, Patrick Turner wrote: Don Pearce wrote: On Sun, 03 May 2009 15:20:45 GMT, Patrick Turner wrote: Patrick Turner. Patrick, having had a proper look at the circuit and its function, I can see that what you have designed is a chindogu.It will still function perfectly if you remove every component apart from the sockets, switch and pot - all you need to do is turn the pot ever so slightly more clockwise than would otherwise be the case. I think some gem of wisdom is lost on me. I am not fully equipped to understand everything, such as what a "chindogu" is. Maybe its a special American way of saying something is especially badly designed. No it is a Japanese thing. What it amounts to is a completely pointless invention - look it up, there are some very amusing ones. Your preamp falls nicely into that category, because everything inside it is entirely pointless, since there is already enough signal going in. As I say above, all that is actually needed is a bunch of sockets, a switch and a pot - job done. I intend to keep my mast, and my sails, and to not instal a bigger diesel engine to "pull better lookin chicks". But that is precisely what you have done - all those sexy glowing tubes, and all. The diesel engine analogy is actually rather apt - hot and noisy. Ah, your post is so elegantly dismissive of so much. Were we to apply your philosophy to the world we could shoot 3/4 of everyone and all they have made, because they are merely suplerflous, and their need for possessions is all just irrational greed and stupidity thus ruining a good world. We might begin at least with the bankers and wayward financiers of recent times. Then for some after dinner sport we could atom bomb the backyard of Pakistan and save having to go over and dig out Osama and his ratbag mates from the limestone caves. Ever come across William Heath Robinson?, or Rube Goldberg in the States? Not sure if there is an Aussie equivalent, but I guess there probably is. Do you really need a wife to cook your dinner, and darn your socks? It'd be so much easier to abreviate any need for her complex and boring perceptions about houskeeping and surely you should rationalise the amount of time you have to waste listening to her small talk about grocery prices. I kinda think you'd be on your own pretty soon if you consistently applied your ideas, and tried to run for President. We don't have one of those here. We have an idiot Prime Minister who claimed that he had designed boom and bust out of the British economy. Meanwhile, I have said to many people that all they need to get a signal from a CD player to a speaker is an integrated amplifier or perhaps a little box with a 50k pot and source select switch inside, and a pair of power amps. This all works fine for some folks until they have a low level source, insensitive power amps and insensitive speakers. Then gain is needed. 2dB? I think probably insufficient. In such a case, of course, a preamp would be needed - but that really isn't what we are talking about here, is it? No matter what I say, people contact me to build them a preamp and often the reason is because they just want to try something with tubes. That 2003 preamp with grey anodized case gave the owner a fine item which he said sounded fabulous. Meanwhile, I'd suggest that whenever we look at the high-end audio gear products made in the US, we see 40dB more chingdogu than what one might see in my amps, if you insist on examinations and reformations. IMHO, the best way to listen to music is to go where real musicians and singers are performing without any electronic aids whatsover. I do that about once a week. I'll be seeing Jeff Beck at the Albert Hall in August. Would you call his amplification an electronic aid, or a necessary part of the instrument? That is a great and necessary use of valves, I think. Take your dear lady wife and friends and enjoy the outing. Try to stop yourself making dumb comments about what they are wearing and their hairstyles. Patrick Turner. I have not the slightest idea what any of that was about. d |
#27
Posted to rec.audio.tubes
|
|||
|
|||
JJ ECC99 for CCDA
Don Pearce wrote: On Tue, 05 May 2009 15:22:33 GMT, Patrick Turner wrote: Don Pearce wrote: On Tue, 05 May 2009 11:05:41 GMT, Patrick Turner wrote: Don Pearce wrote: On Sun, 03 May 2009 15:20:45 GMT, Patrick Turner wrote: Patrick Turner. Patrick, having had a proper look at the circuit and its function, I can see that what you have designed is a chindogu.It will still function perfectly if you remove every component apart from the sockets, switch and pot - all you need to do is turn the pot ever so slightly more clockwise than would otherwise be the case. I think some gem of wisdom is lost on me. I am not fully equipped to understand everything, such as what a "chindogu" is. Maybe its a special American way of saying something is especially badly designed. No it is a Japanese thing. What it amounts to is a completely pointless invention - look it up, there are some very amusing ones. Your preamp falls nicely into that category, because everything inside it is entirely pointless, since there is already enough signal going in. As I say above, all that is actually needed is a bunch of sockets, a switch and a pot - job done. I intend to keep my mast, and my sails, and to not instal a bigger diesel engine to "pull better lookin chicks". But that is precisely what you have done - all those sexy glowing tubes, and all. The diesel engine analogy is actually rather apt - hot and noisy. Ah, your post is so elegantly dismissive of so much. Were we to apply your philosophy to the world we could shoot 3/4 of everyone and all they have made, because they are merely suplerflous, and their need for possessions is all just irrational greed and stupidity thus ruining a good world. We might begin at least with the bankers and wayward financiers of recent times. Then for some after dinner sport we could atom bomb the backyard of Pakistan and save having to go over and dig out Osama and his ratbag mates from the limestone caves. Ever come across William Heath Robinson?, or Rube Goldberg in the States? Not sure if there is an Aussie equivalent, but I guess there probably is. Do you really need a wife to cook your dinner, and darn your socks? It'd be so much easier to abreviate any need for her complex and boring perceptions about houskeeping and surely you should rationalise the amount of time you have to waste listening to her small talk about grocery prices. I kinda think you'd be on your own pretty soon if you consistently applied your ideas, and tried to run for President. We don't have one of those here. We have an idiot Prime Minister who claimed that he had designed boom and bust out of the British economy. Meanwhile, I have said to many people that all they need to get a signal from a CD player to a speaker is an integrated amplifier or perhaps a little box with a 50k pot and source select switch inside, and a pair of power amps. This all works fine for some folks until they have a low level source, insensitive power amps and insensitive speakers. Then gain is needed. 2dB? I think probably insufficient. In such a case, of course, a preamp would be needed - but that really isn't what we are talking about here, is it? No matter what I say, people contact me to build them a preamp and often the reason is because they just want to try something with tubes. That 2003 preamp with grey anodized case gave the owner a fine item which he said sounded fabulous. Meanwhile, I'd suggest that whenever we look at the high-end audio gear products made in the US, we see 40dB more chingdogu than what one might see in my amps, if you insist on examinations and reformations. IMHO, the best way to listen to music is to go where real musicians and singers are performing without any electronic aids whatsover. I do that about once a week. I'll be seeing Jeff Beck at the Albert Hall in August. Would you call his amplification an electronic aid, or a necessary part of the instrument? That is a great and necessary use of valves, I think. When I have gone to a local venue here when the Jazz teachers sometimes give a concert, there is the inevitable bass player with bass amp and speakers. It usually sounds boomy and awful, not because tubes are used, but for other acoustics reasons. A Jazz guitar needs the reverb and amp distortions, and because F is higher there are less acoustics bothers. But one lady who teaches Harp in the classical music School gave a concert with a whole bunch of digital processing gear and she played along to herself, and even stopped to put a bow through the harp to play a string with a bow and the whole effect was utterly unique, extremely tuneful and musical and unlike the kids who battle to play anything beyond 3 chords or notes on an electric guitar, she was just fabulous, both hands and both feet furiuously working everything workable. The gear she used couldn't be used if tubed because it'd have to be the size of a shipping container. Take your dear lady wife and friends and enjoy the outing. Try to stop yourself making dumb comments about what they are wearing and their hairstyles. Patrick Turner. I have not the slightest idea what any of that was about. Never mind. Patrick Turner. d |
#28
Posted to rec.audio.tubes
|
|||
|
|||
JJ ECC99 for CCDA
"Patrick Turner" wrote in message ... If you'd done your home work, you'd have built a sample of my circuit and discovered perhaps to your amazement that indeed the effective collector resistance of the many applications of my single BJT CCS is extremely high, and far more than the 50k you cite. Build the circuit, then measure the voltage change across the emitter resistance and hence emitter current change for a given large voltage signal at the anode/collector. Then divide collector voltage change by colector current change to find the effective collector resistance. You will find it is a huge figure like a pentode with large value cathode resistance. The whole point is that even if the emitter current is constant, the collector current is not, because beta is changing. Typically for a small transistors collector impedance is about 1Mohm per 1mA. For example, for 5mA output impedance will be 200Kohm. For a pentode such load would be considered quite low, but for a triode -- like infinite. Contributing to this effect of finite collector impedance is thermal processes. When Ic increases, die temperature goes up, beta goes up (up to 2% per degree), base current down and consequently collector current up. This effect causes virtually inductve (!) component in the output collector impedance... However, we shall not be carried away. We are not talking precision op-amps, but just crude audio amps. Who cares after all. You are right Partick. You can simply ignore the Early effect and thermal beta-modulation effect, and collector capacitance modulation effect... Again we are not looking into 0.0001% IMD domain, but 0.1...1%. |
#29
Posted to rec.audio.tubes
|
|||
|
|||
JJ ECC99 for CCDA
Alex wrote: "Patrick Turner" wrote in message ... If you'd done your home work, you'd have built a sample of my circuit and discovered perhaps to your amazement that indeed the effective collector resistance of the many applications of my single BJT CCS is extremely high, and far more than the 50k you cite. Build the circuit, then measure the voltage change across the emitter resistance and hence emitter current change for a given large voltage signal at the anode/collector. Then divide collector voltage change by colector current change to find the effective collector resistance. You will find it is a huge figure like a pentode with large value cathode resistance. The whole point is that even if the emitter current is constant, the collector current is not, because beta is changing. OK, so if Ec changes, then you will indeed get some change in Ie, and a small change in Ee, and thus a small change in Ib, thus a change in Eb if Eb is not held at a constant potential with the 0V end of Re. Its all a negligible effect if Re is a high enough value, say over 1k. I suggest you set up any of the CCS i have detailed at my website and that you have a 1k series R between anode and collector, and measure the V change across this 1k. In a typical LTP with common cathode = 12mA through an MJE340 CCS, and with maybe 6V signal at the common cathodes, then with 1k between kk and c the signal voltage across the 1k is below 1mV, and if it was 1mV, then you have a signal current of 1uA, and because there is a 6V signal at the kk, then the efective current sink resistance would be 6 / 1uA, or 6 megohms, and a high enough resistance to be labelled a constant current sink. Typically for a small transistors collector impedance is about 1Mohm per 1mA. For example, for 5mA output impedance will be 200Kohm. For a pentode such load would be considered quite low, but for a triode -- like infinite. The use of an unbypassed Re 1k vastly raises the effective collector resistance, so gain of the pentode approaches µ, maybe a thousand or more, and mainly limited by whatever small amount of shunt C seen by the anode. Contributing to this effect of finite collector impedance is thermal processes. When Ic increases, die temperature goes up, beta goes up (up to 2% per degree), base current down and consequently collector current up. This effect causes virtually inductve (!) component in the output collector impedance... No such effects are seen in my circuits. Where I have an MJE350 whose collector feeds a triode anode, I have a resistance divider between B+ and 0V controlling the bjt base voltage. This keeps the effective Rc very high down to DC. But if the divider works between anode and B+, and the R on the divider betwwn base and B+ is cap bypassed, then the collector resistance is still extremely high for most audio F. But as F falls the bypass cap goes open and the effective DC collector resistance becomes [ R2/ ( R1 + R2 )] x Re. So if R2 was 220k between anode and base, and R1 was 22k between base and B+, and Re = 1k, then effective Rc = approx 11 x 1k, ie, 11k. This is an extremely useful feature so that a pentode's anode dc voltage then tends to stay far more stable when signal voltage amplitude changes. OK, in such a case at audio F the anode load is 220k in parallel with the very high Rc because the base to B+ cap can be an electro with excellent bypassing except at DC. So this situation mimics the use of a choke with high inductance and high dc wire resistance but with very little shunt capacitance and no iron caused hysterisis distortions. If you had a pentode with a resistance load only with 150V across this load and Ia = 5mA, then RL = 30k. And if gm was say 4mA/V, gain = 120 approximately. If you had a maximum load value of 220k of my CCS, then gain would be about 800, for the same Ia. Rout would be the Ra of the pentode in parallel with 220k and rather uselessly too high. So a direct CF can be used to reduce it to say 600 ohms. Then shunt NFB divider using say 470k and 39k can be used to reduce gain from 800 to 10. If THD was 5% at 50V without NFB, then with the FB its becomes 10 x 5%/800 = 0.025%, and at 1V out it becomes theoretically less than 0.001%. If RL = 30k, THD would be higher than with RL = 220k, maybe 8% at 50Vout. If the gain reduction was only from 120 to 10, THD = 10x 8%/120 = 0.75%, and at 1Vo, THD = 0.015%, a much worse performance than with the CCS. The shunt FB in both cases would reduce the Rout to much less than 600 ohms and to maybe 60 ohms, so a resistance from the cathode to the outside world should be used, say 560 ohms, to set the Rout of the preamp, and prevent cable C affecting the performance. However, we shall not be carried away. We are not talking precision op-amps, but just crude audio amps. Who cares after all. You are right Partick. You can simply ignore the Early effect and thermal beta-modulation effect, and collector capacitance modulation effect... Again we are not looking into 0.0001% IMD domain, but 0.1...1%. See above. In a typical paralleled 6CG7 SET input stage with CCS load used with its base biasing divider between B+ and 0V, and a C coupled load of say 220k, THD at 20Vout 0.1%, or 0.01% at a typical 2V operation condition. This is 10dB better than if I were to have a 33k dc carrying resistance betwwn anodes and B+. I offer my customers excellent sound plus excellent measurements with gear that never has THD much above 0.04% at normal loud listeing levels which usually *average* only a watt. The 300watters make about 0.2% at 300W. At 3W where Vout is 1/10 the 300W level, the THD is 0.02% or less. I also use minimal amounts of NFB, 15dB total instead of the common 26dB used back in 1965 for many quality amps which could boast less than 0.1% at 30W. Trying to make a class A amp with ultra low THD is rather pointless, because the horribleness factor of THD/IMD is low because there are no switching distortions. On the other hand, PP SS amps with tiny amounts of class A need a lot of NFB. The RDH4 says that THD just becomes audible on a wide bandwidth system when it exceeds 0.5% of 2H. However, with higher harmonics the threshold diminishes. I assume 0.5% of 7H would render music horrible. In a typical tube PP amp, at a watt, maybe you have 0.02% 3H, 0.01% 2H, 0.005% of all other H. Patrick Turner. |
#30
Posted to rec.audio.tubes
|
|||
|
|||
JJ ECC99 for CCDA
Alex:
The whole point is that even if the emitter current is constant, the collector current is not, because beta is changing. Patrick: OK, so if Ec changes, then you will indeed get some change in Ie, and a small change in Ee, and thus a small change in Ib, thus a change in Eb if Eb is not held at a constant potential with the 0V end of Re. Its all a negligible effect if Re is a high enough value, say over 1k. I suggest you set up any of the CCS i have detailed at my website and that you have a 1k series R between anode and collector, and measure the V change across this 1k. In a typical LTP with common cathode = 12mA through an MJE340 CCS, and with maybe 6V signal at the common cathodes, then with 1k between kk and c the signal voltage across the 1k is below 1mV, and if it was 1mV, then you have a signal current of 1uA, and because there is a 6V signal at the kk, then the efective current sink resistance would be 6 / 1uA, or 6 megohms, and a high enough resistance to be labelled a constant current sink. Alex: This is wrong. The collector current is changing not (solely) because the emitter current changing, but in the first place because beta is changing. Beta is not constant if collector voltage is swinging. First of all collector voltage is directly modulating beta (Early effect). Secondly, collector voltage modulates collector power dissipation which in turn modulates die temperature which in turn modulates beta. As you know, beta rises with temperature so this temperature effect compounds with the Early effect. Thermal effect is more pronounced at low frequencies because of thermal time constant. If the emitter resistor is finite, then emitter voltage variation also starts to play its output impedance reduction game... All three effects add up. However, the above is more of academic interest. Who cares if a triode load is 200K or 6M? Its gain is close to mu anyway... And if the current source is in the long tail of a differential tube stage, then the unbalance potentially caused by a finite impedance of the current source is far less than the unbalance caused by 5% tolerance resistors used as loads at both plates. So though in theory you are wrong, for all practical purposes you are right. Regards, Alex Typically for a small transistors collector impedance is about 1Mohm per 1mA. For example, for 5mA output impedance will be 200Kohm. For a pentode such load would be considered quite low, but for a triode -- like infinite. The use of an unbypassed Re 1k vastly raises the effective collector resistance, so gain of the pentode approaches µ, maybe a thousand or more, and mainly limited by whatever small amount of shunt C seen by the anode. Contributing to this effect of finite collector impedance is thermal processes. When Ic increases, die temperature goes up, beta goes up (up to 2% per degree), base current down and consequently collector current up. This effect causes virtually inductve (!) component in the output collector impedance... No such effects are seen in my circuits. Where I have an MJE350 whose collector feeds a triode anode, I have a resistance divider between B+ and 0V controlling the bjt base voltage. This keeps the effective Rc very high down to DC. But if the divider works between anode and B+, and the R on the divider betwwn base and B+ is cap bypassed, then the collector resistance is still extremely high for most audio F. But as F falls the bypass cap goes open and the effective DC collector resistance becomes [ R2/ ( R1 + R2 )] x Re. So if R2 was 220k between anode and base, and R1 was 22k between base and B+, and Re = 1k, then effective Rc = approx 11 x 1k, ie, 11k. This is an extremely useful feature so that a pentode's anode dc voltage then tends to stay far more stable when signal voltage amplitude changes. OK, in such a case at audio F the anode load is 220k in parallel with the very high Rc because the base to B+ cap can be an electro with excellent bypassing except at DC. So this situation mimics the use of a choke with high inductance and high dc wire resistance but with very little shunt capacitance and no iron caused hysterisis distortions. If you had a pentode with a resistance load only with 150V across this load and Ia = 5mA, then RL = 30k. And if gm was say 4mA/V, gain = 120 approximately. If you had a maximum load value of 220k of my CCS, then gain would be about 800, for the same Ia. Rout would be the Ra of the pentode in parallel with 220k and rather uselessly too high. So a direct CF can be used to reduce it to say 600 ohms. Then shunt NFB divider using say 470k and 39k can be used to reduce gain from 800 to 10. If THD was 5% at 50V without NFB, then with the FB its becomes 10 x 5%/800 = 0.025%, and at 1V out it becomes theoretically less than 0.001%. If RL = 30k, THD would be higher than with RL = 220k, maybe 8% at 50Vout. If the gain reduction was only from 120 to 10, THD = 10x 8%/120 = 0.75%, and at 1Vo, THD = 0.015%, a much worse performance than with the CCS. The shunt FB in both cases would reduce the Rout to much less than 600 ohms and to maybe 60 ohms, so a resistance from the cathode to the outside world should be used, say 560 ohms, to set the Rout of the preamp, and prevent cable C affecting the performance. However, we shall not be carried away. We are not talking precision op-amps, but just crude audio amps. Who cares after all. You are right Partick. You can simply ignore the Early effect and thermal beta-modulation effect, and collector capacitance modulation effect... Again we are not looking into 0.0001% IMD domain, but 0.1...1%. See above. In a typical paralleled 6CG7 SET input stage with CCS load used with its base biasing divider between B+ and 0V, and a C coupled load of say 220k, THD at 20Vout 0.1%, or 0.01% at a typical 2V operation condition. This is 10dB better than if I were to have a 33k dc carrying resistance betwwn anodes and B+. I offer my customers excellent sound plus excellent measurements with gear that never has THD much above 0.04% at normal loud listeing levels which usually *average* only a watt. The 300watters make about 0.2% at 300W. At 3W where Vout is 1/10 the 300W level, the THD is 0.02% or less. I also use minimal amounts of NFB, 15dB total instead of the common 26dB used back in 1965 for many quality amps which could boast less than 0.1% at 30W. Trying to make a class A amp with ultra low THD is rather pointless, because the horribleness factor of THD/IMD is low because there are no switching distortions. On the other hand, PP SS amps with tiny amounts of class A need a lot of NFB. The RDH4 says that THD just becomes audible on a wide bandwidth system when it exceeds 0.5% of 2H. However, with higher harmonics the threshold diminishes. I assume 0.5% of 7H would render music horrible. In a typical tube PP amp, at a watt, maybe you have 0.02% 3H, 0.01% 2H, 0.005% of all other H. Patrick Turner. |
#31
Posted to rec.audio.tubes
|
|||
|
|||
JJ ECC99 for CCDA
Alex wrote: Alex: The whole point is that even if the emitter current is constant, the collector current is not, because beta is changing. Patrick: OK, so if Ec changes, then you will indeed get some change in Ie, and a small change in Ee, and thus a small change in Ib, thus a change in Eb if Eb is not held at a constant potential with the 0V end of Re. Its all a negligible effect if Re is a high enough value, say over 1k. I suggest you set up any of the CCS i have detailed at my website and that you have a 1k series R between anode and collector, and measure the V change across this 1k. In a typical LTP with common cathode = 12mA through an MJE340 CCS, and with maybe 6V signal at the common cathodes, then with 1k between kk and c the signal voltage across the 1k is below 1mV, and if it was 1mV, then you have a signal current of 1uA, and because there is a 6V signal at the kk, then the efective current sink resistance would be 6 / 1uA, or 6 megohms, and a high enough resistance to be labelled a constant current sink. Alex: This is wrong. The collector current is changing not (solely) because the emitter current changing, but in the first place because beta is changing. Beta is not constant if collector voltage is swinging. First of all collector voltage is directly modulating beta (Early effect). Secondly, collector voltage modulates collector power dissipation which in turn modulates die temperature which in turn modulates beta. As you know, beta rises with temperature so this temperature effect compounds with the Early effect. Thermal effect is more pronounced at low frequencies because of thermal time constant. If the emitter resistor is finite, then emitter voltage variation also starts to play its output impedance reduction game... All three effects add up. However, the above is more of academic interest. Who cares if a triode load is 200K or 6M? Its gain is close to mu anyway... Well smarty pants, at the end of the day after taking into account your Early effect, we might later see that the change in effective collector resistance in the CCS circuits which I have shown so well employed at my website may not add up to very much. With whatever early effect exists I still measure the effective Rc at over 1M easily, and it would not matter if the Rc' varied during a full 50Vrms wave from between 1M and 2M, because the Ra of the triode is so low by comparison. Could you tell me how the effective Rc varies? If it does, then would not that either tend to make THD worse or maybe better than a perfect CCS? Which way would it be? Even a slightly non linear but very high resistance dc feed will have a much lessening effect on tube THD caused by the presence of a damned resistor to convey dc to the anode. And if the current source is in the long tail of a differential tube stage, then the unbalance potentially caused by a finite impedance of the current source is far less than the unbalance caused by 5% tolerance resistors used as loads at both plates. So though in theory you are wrong, for all practical purposes you are right. I routinely see LTP balance being within 0.5% of the same amplitude at each anode. And this occurs if you have a 12AT7 on one side and 12AU7 on the other side. Balance is at least 99.5% determined by the closeness of anode resistances. What will change if the triodes on each side differ is the THD which will be mainly 2H of the same amplitude on each anode, but the fundememental signal tone levelo will be equal. Patrick Turner. Regards, Alex |
Reply |
Thread Tools | |
Display Modes | |
|
|
Similar Threads | ||||
Thread | Forum | |||
Atma-sphere MP-3 12AU7 to ECC99 mod | Vacuum Tubes |