Hi All,
Great forum you have here!! Just learning all things tube audio, so discovering your place here is quite timely.
Now, I did some extensive searching on Constant current source information on this site. A great wealth of info came up concerning different design ideas etc. etc.
However, I am looking for some information that will break it right down and tell me why you would use one, and most importantly, how exactly it works, the effects on tube operation, benefits and drawbacks etc.
Any inod out there would be very much appreciated. Thank you :0)
Great forum you have here!! Just learning all things tube audio, so discovering your place here is quite timely.
Now, I did some extensive searching on Constant current source information on this site. A great wealth of info came up concerning different design ideas etc. etc.
However, I am looking for some information that will break it right down and tell me why you would use one, and most importantly, how exactly it works, the effects on tube operation, benefits and drawbacks etc.
Any inod out there would be very much appreciated. Thank you :0)
Others will correct me if I am wrong, but here is my understanding of why a CCS benefits a circuit.
In normal operation the current through a valve varies with the voltage out. This means that you have two shifting variables. By applying a CCS you fix at least one of those variables (the Current), which allows the valves voltage amplification to be more predictable- which equates to been more linear. This also means that the two halves of the output waveform have more equal current rather than one side having more current than the other. This would be especially noticable as large bass passages losing definition and sounding "flabby"- one of the main faults of SE designs.
All in all this adds up to the CCS loaded valve behaving more linearly and with a much reduced distortion. A Constant Current load is the best load in all situations !!!
Hope that helps.
Shoog
In normal operation the current through a valve varies with the voltage out. This means that you have two shifting variables. By applying a CCS you fix at least one of those variables (the Current), which allows the valves voltage amplification to be more predictable- which equates to been more linear. This also means that the two halves of the output waveform have more equal current rather than one side having more current than the other. This would be especially noticable as large bass passages losing definition and sounding "flabby"- one of the main faults of SE designs.
All in all this adds up to the CCS loaded valve behaving more linearly and with a much reduced distortion. A Constant Current load is the best load in all situations !!!
Hope that helps.
Shoog
Having written that it occurs to me that the whole situation is entirely more complex.
When we say that a CCS delivers constant current, this is only true for DC. This is good because DC sets the Bias point, which effectively gives us fixed bias. However at AC the CCS delivers a variable current, which complements the variable current of the loaded tube (they effectively enter a PP arrangement at AC). This then effectively delivers a more Constant current to the load. Since the loaded valve is the dominent contributer in terms of voltage output, and since it is operating at a fixed bias, the whole arrangement is very linear. The secondary role of the CCS is to give current to the load as needed, it has very little effect on the output voltage.
With a resistive load, as the current through the resistor varies, its voltage drop varies, which shifts the valves operating point, which in turn varies the current through the valve. This tends to intoduce none linearatites (ie distortion). For most situations this will be marginal but say we have a low value for the load resistor it will be more pronounced.
Shoog
When we say that a CCS delivers constant current, this is only true for DC. This is good because DC sets the Bias point, which effectively gives us fixed bias. However at AC the CCS delivers a variable current, which complements the variable current of the loaded tube (they effectively enter a PP arrangement at AC). This then effectively delivers a more Constant current to the load. Since the loaded valve is the dominent contributer in terms of voltage output, and since it is operating at a fixed bias, the whole arrangement is very linear. The secondary role of the CCS is to give current to the load as needed, it has very little effect on the output voltage.
With a resistive load, as the current through the resistor varies, its voltage drop varies, which shifts the valves operating point, which in turn varies the current through the valve. This tends to intoduce none linearatites (ie distortion). For most situations this will be marginal but say we have a low value for the load resistor it will be more pronounced.
Shoog
I've understood it the opposite way, i.e., the current is as constant as possible over as wide a frequency range as possible. That way. the tube is operating with essentially a horizontal load line. When you look at triode characteristics, it's evident that a horizontal load line avoids the portion where the curves start bunching up, so the distortion minimizes.
Another way to look at it is to consider the triode as being a Thevenin source with a voltage generator equal to mu times Vgk and a source resistance equal to rp. Now, rp varies with current much more so than mu, so running a CCS minimizes changes in rp. And because the CCS has such a high source impedance, the voltage divider formed by rp and the plate load resistor (in this case, close to infinite) reduces to unity.
In a differential amp tail, the AC constant current characteristic ensures balance, since AC current must be equal and opposite in the two plate load resistors.
SY you are correct about AC current been constant. Forgive my confusion, i'am trying to think this through in my head. Essentally my main point is that the two halves of the waveform have equal current delivery, which is different to the standard SE behaviour where current varies over the whole wave cycle. In standard SE as the valve approaches cut off (max input voltage) current falls to zero. In the CCS situation the current to the load remains constant at cut off because the CCS picks up the slack.
I had one more though.
The reason that the CCS contibutes little to the output voltage waveform is because it functions as a cathode follower with 100% NFB and so it follows the output valves voltage waveform very closely.
Its a complex concept in its details, but essentially its essence is that it makes the output behave more linearly.
A output pentode cannot be loaded with a CCS because it would be making a CCS load a CCS. Any slight inbalance will send it into a meltdown with big instability issues.
Shoog
I had one more though.
The reason that the CCS contibutes little to the output voltage waveform is because it functions as a cathode follower with 100% NFB and so it follows the output valves voltage waveform very closely.
Its a complex concept in its details, but essentially its essence is that it makes the output behave more linearly.
A output pentode cannot be loaded with a CCS because it would be making a CCS load a CCS. Any slight inbalance will send it into a meltdown with big instability issues.
Shoog
Which is why there exists feedback.
Y'all need to sit down with solid state sometime.. variation of 5x in gain in the same part (depending on current and temperature) oughta learn 'ya!
If you fix the plate voltage (as with a voltage follower-biased CCS), you can get gain in the 2000 range from a "Mu-stage" style pentode stage. Heh, even more if you used a negative resistance load (to cancel the pentode's plate resistance: since gain (mu) = Rp * Gm).
Tim
Y'all need to sit down with solid state sometime.. variation of 5x in gain in the same part (depending on current and temperature) oughta learn 'ya!
If you fix the plate voltage (as with a voltage follower-biased CCS), you can get gain in the 2000 range from a "Mu-stage" style pentode stage. Heh, even more if you used a negative resistance load (to cancel the pentode's plate resistance: since gain (mu) = Rp * Gm).
Tim
In the ideal "textbook" world a triode tube is a voltage controlled voltage operated device. It has voltage gain "Mu". The perfect load for an ideal triode is an infinite load for the DC and AC cases. Unfortunately our tubes were not made in a textbook, and they are not perfect and there is no such thing as an infinite load. We can operate a triode with a CCS plate load. This will offer a very high resistance for the DC case. The tube will still see the AC impedance of the following stage (or the OPT) as its AC load. For a low level circuit we can buffer this load with a cathode follower to allow the tube to see a higher impedance for the AC case. What does this buy us. In the perfect world it would allow the tubes gain to approach the Mu (ideal gain) of the tube and it lowers the tubes distortion. In the real world it works wonders on some tubes and offers minimal benefit for some others. I have been experimenting with CCS loaded triodes Buffered by a mosfet follower. The 5842 triode REALLY likes this. The stage gain is 40 almost up to the tubes Mu of 44. Distortion is about 1/10 of the distortion of a resistive load. Other triodes don't gain as much as the 5842. I am in the process of trying to find a tube that works like the 5842, but is cheap.
In the ideal textbook world a pentode is a voltage controlled current device. It has transconductance. The ideal load would be a constant voltage (vertical load line). In order to get any power out of the stage a load line is drawn to optimize the circuit to its intended load. With a CCS the load line is horizontal. The changing grid voltages attempts to change the plate current, which is fixed by the CCS. In the real world the plate voltage would vary and the tube would exhibit some gain. This would not be close to an optimun load for a pentode, so this configuration is not normally used.
In the ideal textbook world a pentode is a voltage controlled current device. It has transconductance. The ideal load would be a constant voltage (vertical load line). In order to get any power out of the stage a load line is drawn to optimize the circuit to its intended load. With a CCS the load line is horizontal. The changing grid voltages attempts to change the plate current, which is fixed by the CCS. In the real world the plate voltage would vary and the tube would exhibit some gain. This would not be close to an optimun load for a pentode, so this configuration is not normally used.
"A CCS as a pentode load works "
In the situation I tried it I had a pentode CCS load ontop of a pentode, I wasn't happy with the results. I suppose a more ideal CCS with a much higher resistance than the loaded pentode would be OK. If the CCS and the loaded pentode had similar gains but slightly mismatched current then this would create the problems I suggested.
When discussing the actual functioning of a CCS I think there is a big difference between an ideal constant current source with resistance in the Meg ohms, and the tube based CCS's which only approach true CCS operation from a great distance. My experience of CCS's would be more of the later, and this can better be described as a SEPP arrangement. With a power pentode CCS the AC impedence might only be 100Kohm or less.
Shoog
In the situation I tried it I had a pentode CCS load ontop of a pentode, I wasn't happy with the results. I suppose a more ideal CCS with a much higher resistance than the loaded pentode would be OK. If the CCS and the loaded pentode had similar gains but slightly mismatched current then this would create the problems I suggested.
When discussing the actual functioning of a CCS I think there is a big difference between an ideal constant current source with resistance in the Meg ohms, and the tube based CCS's which only approach true CCS operation from a great distance. My experience of CCS's would be more of the later, and this can better be described as a SEPP arrangement. With a power pentode CCS the AC impedence might only be 100Kohm or less.
Shoog
Shoog, with the right implementation, you can easily get multimeg source impedances from a pentode CCS. There's a couple of nice examples in Morgan Jones's book.
Personally, I never use pentodes- solid state works just fine for voltage or current sources and doesn't have heater complications- but pentodes do not lack for performance.
Personally, I never use pentodes- solid state works just fine for voltage or current sources and doesn't have heater complications- but pentodes do not lack for performance.
I use solid state CCS's for signal tubes. There was a thread where several people debated the use of a CCS load for a triode wired KT-88. Two working circuits resulted from that thread. One was pure SS and one (mine) was a hybrid. When a circuit needs to dissipate 30 watts at several hundred volts, I prefer a tube.
I have used the IXYS 10M45 chip as the plate load for a 45 tube with good results. I have had bad luck with anything bigger. When the chip blows, it puts the entire power supply across your tube. They generally don't like this! I am still waiting for the new 900 volt IXYS chips to show up at DigiKey. They might be useful for a bigger SS design.
http://www.diyaudio.com/forums/showthread.php?s=&threadid=67437&highlight=
I have used the IXYS 10M45 chip as the plate load for a 45 tube with good results. I have had bad luck with anything bigger. When the chip blows, it puts the entire power supply across your tube. They generally don't like this! I am still waiting for the new 900 volt IXYS chips to show up at DigiKey. They might be useful for a bigger SS design.
http://www.diyaudio.com/forums/showthread.php?s=&threadid=67437&highlight=
I did a lot of searching for suitable Pentode CCS's and even tried one of Gary Pims designs (the basic pentode CCS). This failed to work because the TT21(KT88) I used didn't draw any grid current and Garys design requires a few mA of grid current to work. I built it and tried it and had to scrap it.
I ended up with a very basic arrangement with 2K3 current sensing resistor on the cathode which with a the KT88's gain of 12 (I think) gives a puny load.
Still it seems to work satisfactorily, and delivers stable DC which is the main thing. At this stage if I wanted to improve things it would probably mean starting from scratch with different valves.
I am toying with the idea of trying UL at 50% to drag a bit more power output.
Shoog
I ended up with a very basic arrangement with 2K3 current sensing resistor on the cathode which with a the KT88's gain of 12 (I think) gives a puny load.
Still it seems to work satisfactorily, and delivers stable DC which is the main thing. At this stage if I wanted to improve things it would probably mean starting from scratch with different valves.
I am toying with the idea of trying UL at 50% to drag a bit more power output.
Shoog
For solid state CCS's I use the IXYS 10M45 chip available from DigiKey. It is good up to 450 volts and 100mA at 40 watts max dissipation. It is a simple circuit that works well enough to be used as a plate load for many triodes. If you use this add a 1K resistor in series with the gate lead. This "stopper" resistor eliminates oscillations that cause random parts death.
Many use the LM317 voltage regulator wired as a CCS. Also a simple circuit that works well in the cathode circuit of triodes and pentodes. This chip is only good to 35 volts but can handle up to an amp.
My hybrid design uses an IXYS chip in the cathode circuit of a triode wired pentode. This simple circuit works well, although I am sure that one of Gary Pimm's CCS designs would blow it away. I have never tried one of his due to the complicated circuit. I will put my hybrid circuit on my web site with the next update.
http://www.tubelab.com/
http://www.pacifier.com/~gpimm/
Many use the LM317 voltage regulator wired as a CCS. Also a simple circuit that works well in the cathode circuit of triodes and pentodes. This chip is only good to 35 volts but can handle up to an amp.
My hybrid design uses an IXYS chip in the cathode circuit of a triode wired pentode. This simple circuit works well, although I am sure that one of Gary Pimm's CCS designs would blow it away. I have never tried one of his due to the complicated circuit. I will put my hybrid circuit on my web site with the next update.
http://www.tubelab.com/
http://www.pacifier.com/~gpimm/
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.