| Home |
| Search |
| Today's Posts |
|
#1
|
|||
|
|||
When is it necessary to cap couple line stages?
Hi Group.
I've noticed in a few designs that when cascading a grounded cathode stage into a CPI that the coupling cap can be dropped. I realize that it blocks DC from one stage to another and can also filter AC. Obviously I'd love to leave out the cap. Is there a simple logical rule of thumb when you can or cannot leave the cap out between two line stages? Thanks alot guys. Wessel |
|
#2
|
|||
|
|||
|
On 24 Apr 2005 03:23:13 -0700, "Wessel Dirksen" wrote:
Hi Group. I've noticed in a few designs that when cascading a grounded cathode stage into a CPI that the coupling cap can be dropped. I realize that it blocks DC from one stage to another and can also filter AC. Obviously I'd love to leave out the cap. Is there a simple logical rule of thumb when you can or cannot leave the cap out between two line stages? Thanks alot guys. Wessel If the voltage is the same at the two points you want to couple, you can leave out the cap... generally... If the voltages are close but not equal, you need to see if connecting them will cause excess current, or will force the circuit to compensate. Usually with a line stage you have only about 1ma of current to worry about, and hundreds of thousands of ohms to play with, so it isn't too critical. You can always test for frequency response and phase and noise between the 2 topologies... |
|
#3
|
|||
|
|||
|
Bob wrote in message . ..
On 24 Apr 2005 03:23:13 -0700, "Wessel Dirksen" wrote: Hi Group. I've noticed in a few designs that when cascading a grounded cathode stage into a CPI that the coupling cap can be dropped. I realize that it blocks DC from one stage to another and can also filter AC. Obviously I'd love to leave out the cap. Is there a simple logical rule of thumb when you can or cannot leave the cap out between two line stages? Thanks alot guys. Wessel If the voltage is the same at the two points you want to couple, you can leave out the cap... generally... If the voltages are close but not equal, you need to see if connecting them will cause excess current, or will force the circuit to compensate. Usually with a line stage you have only about 1ma of current to worry about, and hundreds of thousands of ohms to play with, so it isn't too critical. You can always test for frequency response and phase and noise between the 2 topologies... Thanks for the response Bob, I was thinking in in the same lines. I tried this the other day with a current sourced, non bypassed with 5db ONF grounded cathode stage with an 12AX7 (gain = 90 (40dB) with no ONF, Zout 220K(ouch)) going into a CPI with both triodes of a ECC81 in parallel. (Rk and Ra = 14k) Both tubes get a 450V main which shouldn't overcook anything. For both the V(cath to plate) is close to 170V (+/- 5v). For the both grid bias voltage calculates out to about -1.5V (+/- 0.1) The first stage measures 150 volts DC at Ra. Anyway, I fired it up with no cap in between and there was a strange sounding pretty loud periodic sound, it has a higher than 50hz fundamental main frequency (I'm guessing about 200 hz, I should measure the freq, but didn't yet)It doesn't seem to osscilate out of control but is unacceptable. When I put a 0.22 cap between it this went away and sounded really quite good but still not quite enough gain. My goal is to eliminate one of the 2 series grounded cathode stages. I realize that this may be unrealistic but I'm stubborn and learning. Any ideas? Wessel |
|
#4
|
|||
|
|||
|
|
|
#5
|
|||
|
|||
|
Bob wrote in message . ..
On 26 Apr 2005 04:13:41 -0700, (Wessel Dirksen) wrote: Bob wrote in message . .. On 24 Apr 2005 03:23:13 -0700, "Wessel Dirksen" wrote: Hi Group. I've noticed in a few designs that when cascading a grounded cathode stage into a CPI that the coupling cap can be dropped. I realize that it blocks DC from one stage to another and can also filter AC. Obviously I'd love to leave out the cap. Is there a simple logical rule of thumb when you can or cannot leave the cap out between two line stages? Thanks alot guys. Wessel If the voltage is the same at the two points you want to couple, you can leave out the cap... generally... If the voltages are close but not equal, you need to see if connecting them will cause excess current, or will force the circuit to compensate. Usually with a line stage you have only about 1ma of current to worry about, and hundreds of thousands of ohms to play with, so it isn't too critical. You can always test for frequency response and phase and noise between the 2 topologies... Thanks for the response Bob, I was thinking in in the same lines. I tried this the other day with a current sourced, non bypassed with 5db ONF grounded cathode stage with an 12AX7 (gain = 90 (40dB) with no ONF, Zout 220K(ouch)) going into a CPI with both triodes of a ECC81 in parallel. (Rk and Ra = 14k) Both tubes get a 450V main which shouldn't overcook anything. For both the V(cath to plate) is close to 170V (+/- 5v). For the both grid bias voltage calculates out to about -1.5V (+/- 0.1) The first stage measures 150 volts DC at Ra. Anyway, I fired it up with no cap in between and there was a strange sounding pretty loud periodic sound, it has a higher than 50hz fundamental main frequency (I'm guessing about 200 hz, I should measure the freq, but didn't yet)It doesn't seem to osscilate out of control but is unacceptable. When I put a 0.22 cap between it this went away and sounded really quite good but still not quite enough gain. My goal is to eliminate one of the 2 series grounded cathode stages. I realize that this may be unrealistic but I'm stubborn and learning. Any ideas? Wessel Well... first I would try it without the feedback loop... remember that a cap adds some phase shift... so the lack of a cap will change your overall loop... What I described was without the global feedback. I added it later once the cap was in there I don't like feedback in tube amps... but that's just me...  I like a little (5db instad of the 15 or so you often hear about), so we see to have the same general taste. I prefer the "none" sound but a little tightens things up a bit with very little loss in imaging. Did you monitor the supply in case you have a de-coupling problem? I'm going to have plead ignorance on that one. Could you explain? Is there voltage across the coupling cap you put? Well DC no. When you measure the DC voltage on the grounded cath side I'm getting 150 volts, on the other side nothing. BTW I don't think your gain problem will be solved by removing the coupling cap! I know. I guess I didn't make myself clear. I want some OFB but wish to experiment with only one gain stage and a phase splitter driving 4x El34's. So I hope to have one gain stage replace two of them in series. Since I want to have feedback, bypassing Rk can't happen. I have enough gain +/- 48 dB with a single ECC82 triode and ECC81 single triode both grounded cathode in series with each other then into the CPI. I'm trying to get one of the gain stages removed by using a 12AX7 instead. I'm getting about 40 dB of gain which just doesn't cut it. I'm going to try a mu follower next but this should offer no better gain, just a better Zout. It sounds good. But I want to ditch the extra cap in between. I'm now believing I won't get anymore gain out of it this way. Try increasing the plate load R and is the first stage cathode bypassed? Maybe increase the cathode resistor and bypass it? I need a schemagic! Well difficult to draw. I partially explained it above. Everything gets 450v with a lot of buffer capacitance. The constant current allows 0.8ma through the 12AX7 gain stage where Ra = 120K and Rk = 1.8K non bypassed. This goes into an ECC81 CPI where both triodes are parallel. Ra and Rk are 14K. This feeds an output stage where 2 parallel El34's per side are in triode mode. Caps are 0.047 and RL = 220K (So I guess that is 0.1 micF and 110K to the CPI) THe stop resistors are 2.7K. The output is biased to 60mA. The original recipe recommended 50mA but 60 mA kicks butt. Does that make any sense. Thanks for the input. I just hope to be able to run it like this with no cap in between the gain and CPI stages. I'm afraid I'll have to live with the just sub par gain. Wessel |
|
#7
|
|||
|
|||
|
Bob wrote in message . ..
On 27 Apr 2005 15:47:00 -0700, (Wessel Dirksen) wrote: Bob wrote in message . .. On 26 Apr 2005 04:13:41 -0700, (Wessel Dirksen) wrote: Bob wrote in message . .. On 24 Apr 2005 03:23:13 -0700, "Wessel Dirksen" wrote: Hi Group. I've noticed in a few designs that when cascading a grounded cathode stage into a CPI that the coupling cap can be dropped. I realize that it blocks DC from one stage to another and can also filter AC. Obviously I'd love to leave out the cap. Is there a simple logical rule of thumb when you can or cannot leave the cap out between two line stages? Thanks alot guys. Wessel If the voltage is the same at the two points you want to couple, you can leave out the cap... generally... If the voltages are close but not equal, you need to see if connecting them will cause excess current, or will force the circuit to compensate. Usually with a line stage you have only about 1ma of current to worry about, and hundreds of thousands of ohms to play with, so it isn't too critical. You can always test for frequency response and phase and noise between the 2 topologies... Thanks for the response Bob, I was thinking in in the same lines. I tried this the other day with a current sourced, non bypassed with 5db ONF grounded cathode stage with an 12AX7 (gain = 90 (40dB) with no ONF, Zout 220K(ouch)) going into a CPI with both triodes of a ECC81 in parallel. (Rk and Ra = 14k) Both tubes get a 450V main which shouldn't overcook anything. For both the V(cath to plate) is close to 170V (+/- 5v). For the both grid bias voltage calculates out to about -1.5V (+/- 0.1) The first stage measures 150 volts DC at Ra. Anyway, I fired it up with no cap in between and there was a strange sounding pretty loud periodic sound, it has a higher than 50hz fundamental main frequency (I'm guessing about 200 hz, I should measure the freq, but didn't yet)It doesn't seem to osscilate out of control but is unacceptable. When I put a 0.22 cap between it this went away and sounded really quite good but still not quite enough gain. My goal is to eliminate one of the 2 series grounded cathode stages. I realize that this may be unrealistic but I'm stubborn and learning. Any ideas? Wessel Well... first I would try it without the feedback loop... remember that a cap adds some phase shift... so the lack of a cap will change your overall loop... What I described was without the global feedback. I added it later once the cap was in there I don't like feedback in tube amps... but that's just me... I like a little (5db instad of the 15 or so you often hear about), so we see to have the same general taste. I prefer the "none" sound but a little tightens things up a bit with very little loss in imaging. Did you monitor the supply in case you have a de-coupling problem? I'm going to have plead ignorance on that one. Could you explain? I don't know how you wired the power, but most people use a series resistor and capacitor to ground "pi" filter supply.... sometimes this causes feedback in the stages due to a lack of capacity and 'current sharing', and the voltage on the supply will vary with the 'thumping' of the bad stages. I wanted to know if you looked at the supply when you had the problem to see if the voltage was varying. Ok, I see what you mean. The PS is straight forward I think. Full wave bridge to 450V DC. A small value but high power resistor of 10 ohm after the rectifier, then 1600 micF of ballast capacitance. With the original config described above, there is a further tap off resistance to get the CPI supply to 350V with an extra 220micF balast, then a tap again off this tap to get 250V also with an extra 220micF balast. Now with this current experiment, because the gain stage uses one triode as a current source, it wants more voltage, so I've made everything (output + line stage) run off 450V with the 1600micF (the other 2 220micF are too low in power to run on 450V) I can try to measure if there any substantial fluctuations. It's always been very quiet, also quiet with the coupling cap there. You could always try another supply line to see if that helps... it has for me. I always use a 'spider' positive! Could you explain what you mean by 'spider' positive? I love to hear about new stuff I don't know. Is there voltage across the coupling cap you put? Well DC no. When you measure the DC voltage on the grounded cath side I'm getting 150 volts, on the other side nothing. Are you saying you get 150 to ground on one side of the cap and zero on the other?? This is 150 across the cap! That would be bad if you short the cap out! Velly velly bad! Well, yeah that is what is happening. One side of the cap to ground is 150V DC, the other nada. If you remove (short) the cap you get the weird sound but it still works. Why bad? BTW I don't think your gain problem will be solved by removing the coupling cap! I know. I guess I didn't make myself clear. I want some OFB but wish to experiment with only one gain stage and a phase splitter driving 4x El34's. So I hope to have one gain stage replace two of them in series. Since I want to have feedback, bypassing Rk can't happen. I have enough gain +/- 48 dB with a single ECC82 triode and ECC81 single triode both grounded cathode in series with each other then into the CPI. I'm trying to get one of the gain stages removed by using a 12AX7 instead. I'm getting about 40 dB of gain which just doesn't cut it. I'm going to try a mu follower next but this should offer no better gain, just a better Zout. It sounds good. But I want to ditch the extra cap in between. I'm now believing I won't get anymore gain out of it this way. Try increasing the plate load R and is the first stage cathode bypassed? Maybe increase the cathode resistor and bypass it? I need a schemagic! Well difficult to draw. I partially explained it above. Everything gets 450v with a lot of buffer capacitance. The constant current How are you getting this 'constant current' ?? sounds dicey... allows 0.8ma through the 12AX7 gain stage where Ra = 120K and Rk = 1.8K non bypassed. I think you can get more gain from a 12AX7 using a 2.7k cathode resistor... that's what I use sometimes... and try a bigger plate load too... 220k or 330 I'll try that option. I'm using a stategy for grounded cathode I found in an older article on TUBECAD journal. There he recommends a sweet spot. Your method is something I have also seen before. It sounds good that way? see what happens... This goes into an ECC81 CPI where both triodes are parallel. Ra and Rk are 14K. Any reason to use this topology? I use a diff/amp config here, it gives me both extra gain and a place to add feedback - even if I don't use it! And you already have the 2 triodes... That is indeed where I'll go next. You probably mean local feedback too, I imagine. What I am doing is experimenting with all kinds of concepts. This is an attempt to have only 2 line stages, and thus hopefully only one series cap before the power tubes. But is it wise to phase split the input of only 1 or so volts? If this can be done, then I could run dual differential gain stages and still keep this minimum approach alive. This feeds an output stage where 2 parallel El34's per side are in triode mode. Caps are 0.047 and RL = 220K (So I guess that is 0.1 micF and 110K to the CPI) THe stop resistors are 2.7K. The output is biased to 60mA. The original recipe recommended 50mA but 60 mA kicks butt. So this is class AB push pull output? You must have a weird output tranny! Yes AB PP, I think pretty standard Plintron OPT though. Raa of 2700 ohm. Does that make any sense. Thanks for the input. I just hope to be able to run it like this with no cap in between the gain and CPI stages. I'm afraid I'll have to live with the just sub par gain. Or change the circuit... I just finished building a P-P EL84 18 watt amp that I'm VERY happy with! Cool! Yup, this is very fun stuff. I'm just getting started. Bob |
|
#8
|
|||
|
|||
|
On 29 Apr 2005 03:50:16 -0700, (Wessel Dirksen) wrote:
Bob wrote in message . .. On 27 Apr 2005 15:47:00 -0700, (Wessel Dirksen) wrote: Bob wrote in message . .. On 26 Apr 2005 04:13:41 -0700, (Wessel Dirksen) wrote: Bob wrote in message . .. On 24 Apr 2005 03:23:13 -0700, "Wessel Dirksen" wrote: Hi Group. I've noticed in a few designs that when cascading a grounded cathode stage into a CPI that the coupling cap can be dropped. I realize that it blocks DC from one stage to another and can also filter AC. Obviously I'd love to leave out the cap. Is there a simple logical rule of thumb when you can or cannot leave the cap out between two line stages? Thanks alot guys. Wessel If the voltage is the same at the two points you want to couple, you can leave out the cap... generally... If the voltages are close but not equal, you need to see if connecting them will cause excess current, or will force the circuit to compensate. Usually with a line stage you have only about 1ma of current to worry about, and hundreds of thousands of ohms to play with, so it isn't too critical. You can always test for frequency response and phase and noise between the 2 topologies... Thanks for the response Bob, I was thinking in in the same lines. I tried this the other day with a current sourced, non bypassed with 5db ONF grounded cathode stage with an 12AX7 (gain = 90 (40dB) with no ONF, Zout 220K(ouch)) going into a CPI with both triodes of a ECC81 in parallel. (Rk and Ra = 14k) Both tubes get a 450V main which shouldn't overcook anything. For both the V(cath to plate) is close to 170V (+/- 5v). For the both grid bias voltage calculates out to about -1.5V (+/- 0.1) The first stage measures 150 volts DC at Ra. Anyway, I fired it up with no cap in between and there was a strange sounding pretty loud periodic sound, it has a higher than 50hz fundamental main frequency (I'm guessing about 200 hz, I should measure the freq, but didn't yet)It doesn't seem to osscilate out of control but is unacceptable. When I put a 0.22 cap between it this went away and sounded really quite good but still not quite enough gain. My goal is to eliminate one of the 2 series grounded cathode stages. I realize that this may be unrealistic but I'm stubborn and learning. Any ideas? Wessel Well... first I would try it without the feedback loop... remember that a cap adds some phase shift... so the lack of a cap will change your overall loop... What I described was without the global feedback. I added it later once the cap was in there I don't like feedback in tube amps... but that's just me... I like a little (5db instad of the 15 or so you often hear about), so we see to have the same general taste. I prefer the "none" sound but a little tightens things up a bit with very little loss in imaging. Did you monitor the supply in case you have a de-coupling problem? I'm going to have plead ignorance on that one. Could you explain? I don't know how you wired the power, but most people use a series resistor and capacitor to ground "pi" filter supply.... sometimes this causes feedback in the stages due to a lack of capacity and 'current sharing', and the voltage on the supply will vary with the 'thumping' of the bad stages. I wanted to know if you looked at the supply when you had the problem to see if the voltage was varying. Ok, I see what you mean. The PS is straight forward I think. Full wave bridge to 450V DC. A small value but high power resistor of 10 ohm after the rectifier, then 1600 micF of ballast capacitance. With the original config described above, there is a further tap off resistance to get the CPI supply to 350V with an extra 220micF balast, then a tap again off this tap to get 250V also with an extra 220micF balast. Now with this current experiment, because the gain stage uses one triode as a current source, it wants more voltage, so I've made everything (output + line stage) run off 450V with the 1600micF (the other 2 220micF are too low in power to run on 450V) I can try to measure if there any substantial fluctuations. It's always been very quiet, also quiet with the coupling cap there. You could always try another supply line to see if that helps... it has for me. I always use a 'spider' positive! Could you explain what you mean by 'spider' positive? I love to hear about new stuff I don't know. I run all my positive wires direct to the cap, same as all negative wires run to the star ground... I don't daisy chain positive wires around so it looks like a spider... Is there voltage across the coupling cap you put? Well DC no. When you measure the DC voltage on the grounded cath side I'm getting 150 volts, on the other side nothing. Are you saying you get 150 to ground on one side of the cap and zero on the other?? This is 150 across the cap! That would be bad if you short the cap out! Velly velly bad! Well, yeah that is what is happening. One side of the cap to ground is 150V DC, the other nada. If you remove (short) the cap you get the weird sound but it still works. Why bad? You are short circuiting 150 volts?! You have to ask? What is the short current? BTW I don't think your gain problem will be solved by removing the coupling cap! I know. I guess I didn't make myself clear. I want some OFB but wish to experiment with only one gain stage and a phase splitter driving 4x El34's. So I hope to have one gain stage replace two of them in series. Since I want to have feedback, bypassing Rk can't happen. I have enough gain +/- 48 dB with a single ECC82 triode and ECC81 single triode both grounded cathode in series with each other then into the CPI. I'm trying to get one of the gain stages removed by using a 12AX7 instead. I'm getting about 40 dB of gain which just doesn't cut it. I'm going to try a mu follower next but this should offer no better gain, just a better Zout. It sounds good. But I want to ditch the extra cap in between. I'm now believing I won't get anymore gain out of it this way. Try increasing the plate load R and is the first stage cathode bypassed? Maybe increase the cathode resistor and bypass it? I need a schemagic! Well difficult to draw. I partially explained it above. Everything gets 450v with a lot of buffer capacitance. The constant current How are you getting this 'constant current' ?? sounds dicey... allows 0.8ma through the 12AX7 gain stage where Ra = 120K and Rk = 1.8K non bypassed. I think you can get more gain from a 12AX7 using a 2.7k cathode resistor... that's what I use sometimes... and try a bigger plate load too... 220k or 330 I'll try that option. I'm using a stategy for grounded cathode I found in an older article on TUBECAD journal. There he recommends a sweet spot. Your method is something I have also seen before. It sounds good that way? see what happens... This goes into an ECC81 CPI where both triodes are parallel. Ra and Rk are 14K. Any reason to use this topology? I use a diff/amp config here, it gives me both extra gain and a place to add feedback - even if I don't use it! And you already have the 2 triodes... That is indeed where I'll go next. You probably mean local feedback too, I imagine. What I am doing is experimenting with all kinds of concepts. This is an attempt to have only 2 line stages, and thus hopefully only one series cap before the power tubes. But is it wise to phase split the input of only 1 or so volts? If this can be done, then I could run dual differential gain stages and still keep this minimum approach alive. I don't see why not... lots of designs run multiple differential stages. This feeds an output stage where 2 parallel El34's per side are in triode mode. Caps are 0.047 and RL = 220K (So I guess that is 0.1 micF and 110K to the CPI) THe stop resistors are 2.7K. The output is biased to 60mA. The original recipe recommended 50mA but 60 mA kicks butt. So this is class AB push pull output? You must have a weird output tranny! Yes AB PP, I think pretty standard Plintron OPT though. Raa of 2700 ohm. Does that make any sense. Thanks for the input. I just hope to be able to run it like this with no cap in between the gain and CPI stages. I'm afraid I'll have to live with the just sub par gain. Or change the circuit... I just finished building a P-P EL84 18 watt amp that I'm VERY happy with! Cool! Yup, this is very fun stuff. I'm just getting started. Bob |
| Reply |
| Thread Tools | |
| Display Modes | |
|
|
Similar Threads
|
||||
| Thread | Forum | |||
| KISS 121 by Andre Jute | Vacuum Tubes | |||
| FS PMC LB1 Transmission Line Loudspeakers | Marketplace | |||
| Tube line and phono pre amp project >PING Patrick | Vacuum Tubes | |||
| Question about line drivers | Car Audio | |||
| Question about line drivers | Car Audio | |||
Linear Mode
