I have started my TSE build and wish to start a discussion on the operating conditions for the input stage of the TSE. I’m trying to understand the fundament theory behind the design. I am new to tubes and my electronics knowledge is very rusty to say the least.
As I understand it the Constant Current Source (CCS) does just that. It delivers a constant source of current which is set via the valve of R16. But should I add in any other resistance to ground i.e. R10? The CCS datasheet shows about 8ma out for an Rk of 330 ohms. This seems low as the 5842 datasheet lists an average plate current (Ia) at 25ma. R17 is to help stabilize the CCS. I’d like to be able to draw a load line for the operating parameters of the 5842 tube. Which I think will be a horizontal line.
Drawing
View attachment TSE Input Stage0001.PDF
The 5842 datasheet lists an average plate voltage of 150V. On the diagram it says that the R9 Trim Pot is used to adjust the Plate voltage. How does that work?
Any help getting me going in the right direction is very much appreciated.
Steve
As I understand it the Constant Current Source (CCS) does just that. It delivers a constant source of current which is set via the valve of R16. But should I add in any other resistance to ground i.e. R10? The CCS datasheet shows about 8ma out for an Rk of 330 ohms. This seems low as the 5842 datasheet lists an average plate current (Ia) at 25ma. R17 is to help stabilize the CCS. I’d like to be able to draw a load line for the operating parameters of the 5842 tube. Which I think will be a horizontal line.
Drawing
View attachment TSE Input Stage0001.PDF
The 5842 datasheet lists an average plate voltage of 150V. On the diagram it says that the R9 Trim Pot is used to adjust the Plate voltage. How does that work?
Any help getting me going in the right direction is very much appreciated.
Steve
If you randomly click around enough, you'll find George's page on what he calls "power drive". It's the combination of a CCS loaded tube voltage gain stage driving a source-follower FET current gain stage. It is the primary unique aspect of the TSE design.
PowerDrive
He even wrote a cookbook on how to setup the circuit in your own amp design:
PDcookbook
PowerDrive
He even wrote a cookbook on how to setup the circuit in your own amp design:
PDcookbook
Tom, thanks for the reply. I have purchased Morgan Jones’s 4th edition of “Valve Amplifiers” and I‘m wading through it. Starting out, I haven’t found the book all that clear, at least for me. It will probably get better once I get my mind around some of the basic concepts of valves as they pertain to audio amplifiers. My electronics classes in college were 32 years ago, and I forgot most of that a long time ago.
I’m better at learning by example, especially if I I can hold it in my hand. That’s why I’m building the TSE.
Do you have a feel for the Ia current value? and how it's set up?
Steve
I’m better at learning by example, especially if I I can hold it in my hand. That’s why I’m building the TSE.
Do you have a feel for the Ia current value? and how it's set up?
Steve
Hi Russ,
I have poked and clicked around. A lot. Some people (think spouse) might think too much.
From the Tube Lab Website
"1) The ideal load for a triode is a constant current. The constant current source IC (10M45) provides this at DC. The 91 ohm resistor sets the current (30 mA) "
So 91 ohms gives me a 30ma current (Ia) which more or less matches the too small graph on the datasheet I have for the CCS. OK I get that. Now if I use the 332 ohms(R16 ) from the TSE schematic and the too small graph for the CCS, I come up with 8ma - 9ma. Is that right? Seems small.
Steve
I have poked and clicked around. A lot. Some people (think spouse) might think too much.
From the Tube Lab Website
"1) The ideal load for a triode is a constant current. The constant current source IC (10M45) provides this at DC. The 91 ohm resistor sets the current (30 mA) "
So 91 ohms gives me a 30ma current (Ia) which more or less matches the too small graph on the datasheet I have for the CCS. OK I get that. Now if I use the 332 ohms(R16 ) from the TSE schematic and the too small graph for the CCS, I come up with 8ma - 9ma. Is that right? Seems small.
Steve
A triode works best with an infinite load resistance. The usual plate load resistor is a compromise and results in the triode producing about half the gain that it is capable of with increased distortion. A typical triode would produce more gain and very low distortion if the load was increased. The CCS circuit will allow a constant current to flow through it, yet exhibit a very high AC impedance. The 10M45 chip is about 1 megohm at all audio frequencies.
R16 sets the 10M45 current at about 8 mA. R17 keeps the 10M45 from oscillating in the RF region. We need a lot of drive voltage to push a 300B tube to full output, 80 to 100 volts peak to peak.
The data sheet value for the 5842 tube does indicate a typical tube current of 18 to 33mA, if the tube is used in its intended application, a wide bandwidth high frequency or IF amplifier. We want to use it for a high voltage gain audio amplifier that produces 100 volts P-P of output. This means we must figure out what conditions it needs for this application ourselves. I tested a box full of tubes and found that about 8 -10 ma worked the best for this application. We dont need, or want 10's of MHz of bandwidth.
Just simply stuffing 8 mA of current into the plate of a tube will not make an amplifier. The tube must have a bias voltage applied between the grid and cathode to set its operating conditions. We forced the tube current to be 10 mA with the CCS, but we must apply a bias voltage to set the plate voltage. Testing revealed that the best plate voltage for 100 volts P-P of output was about 175 volts. To get this we should need the grid to be about 5 volts more negative than the cathode. It is desirable to operate the grid at zero volts so that a capacitor is not needed in series with the input, so the cathode must be about 5 volts positive. We can do this by putting a resistor in series with the cathode.
I found no published curves for the 5842, so I tested my box full of tubes and found that they are all over the place. To allow for the wide range of bias voltages required for typical 5842 tubes, I added an adjustment pot in series with the cathode resistor. It can be adjusted for 175 volts on the 5842 plate, or tweaked for minimum distortion if measuring equipment is available.
R16 sets the 10M45 current at about 8 mA. R17 keeps the 10M45 from oscillating in the RF region. We need a lot of drive voltage to push a 300B tube to full output, 80 to 100 volts peak to peak.
The data sheet value for the 5842 tube does indicate a typical tube current of 18 to 33mA, if the tube is used in its intended application, a wide bandwidth high frequency or IF amplifier. We want to use it for a high voltage gain audio amplifier that produces 100 volts P-P of output. This means we must figure out what conditions it needs for this application ourselves. I tested a box full of tubes and found that about 8 -10 ma worked the best for this application. We dont need, or want 10's of MHz of bandwidth.
Just simply stuffing 8 mA of current into the plate of a tube will not make an amplifier. The tube must have a bias voltage applied between the grid and cathode to set its operating conditions. We forced the tube current to be 10 mA with the CCS, but we must apply a bias voltage to set the plate voltage. Testing revealed that the best plate voltage for 100 volts P-P of output was about 175 volts. To get this we should need the grid to be about 5 volts more negative than the cathode. It is desirable to operate the grid at zero volts so that a capacitor is not needed in series with the input, so the cathode must be about 5 volts positive. We can do this by putting a resistor in series with the cathode.
I found no published curves for the 5842, so I tested my box full of tubes and found that they are all over the place. To allow for the wide range of bias voltages required for typical 5842 tubes, I added an adjustment pot in series with the cathode resistor. It can be adjusted for 175 volts on the 5842 plate, or tweaked for minimum distortion if measuring equipment is available.
Hi George and thank you for the reply and sharing your wisdom. I love the methods you have used to derive your amps.
I’m a sucker for doing things out of the box. I starting looking into tubes about 2 months ago and my Edcors showed up today! But I can’t just build it, no I have to understand it, completely.
I have a datasheet for a 5842 and as you must know doesn’t even show a -5vdc grid voltage on the plate chart!
So your testing showed for the desired 100v p-p output that Ia of 8-10ma and a Vgc of -5vdc is needed.
The R16 sets the 10M45 to source 8-10ma (Ia). And R10 (adj with R9) with 8-10ma is approx 5.0vdc ( .010a * 500ohms = 5vdc)
Our tube is on and running were you want it.
Tomorrows post will be the FET stage. Stay tuned.
Thanks
Steve
I’m a sucker for doing things out of the box. I starting looking into tubes about 2 months ago and my Edcors showed up today! But I can’t just build it, no I have to understand it, completely.
I have a datasheet for a 5842 and as you must know doesn’t even show a -5vdc grid voltage on the plate chart!
So your testing showed for the desired 100v p-p output that Ia of 8-10ma and a Vgc of -5vdc is needed.
The R16 sets the 10M45 to source 8-10ma (Ia). And R10 (adj with R9) with 8-10ma is approx 5.0vdc ( .010a * 500ohms = 5vdc)
Our tube is on and running were you want it.
Tomorrows post will be the FET stage. Stay tuned.
Thanks
Steve
Thanks George!
Well, it looks like I'll be hoisting both of my TSE's back on the bench, and firing up Rightmark and the sound card........I'll use my attenuator box thingy and proceed carefully to preserve the remaining functional sound card channel.
Should be interesting......
Well, it looks like I'll be hoisting both of my TSE's back on the bench, and firing up Rightmark and the sound card........I'll use my attenuator box thingy and proceed carefully to preserve the remaining functional sound card channel.
Should be interesting......
Apologies to the OP as this is a bit off-topic.
I've been meaning to ask you about your TSEs Boywonder. My next project is a TSE of one flavor another. I have the board already and will be bringing some James transformers back to the US with me when I leave Korea this summer. The plan was originally to build a 300B, but as I'm also planning a large house-shaking PP octal amp after that (how's that Universal Driver board coming George
), I've been toying with the idea of a 45 TSE and some smaller efficient speakers. Anything you can share on the merits of the two? (If I remember you have a 300B and a 45?)Looking over what has been posted so far I still have a few questions.
I’d like to know the Amplification (Av) of the input stage.
Amplification (Av) = ∆Va/∆Vg
Anybody know what a typical signal voltage peek to peek that is riding on Vg is?
Thanks Steve
I’d like to know the Amplification (Av) of the input stage.
Amplification (Av) = ∆Va/∆Vg
Do we need the 80 - 100v PP after the input stage (Va) or after the “complete” power drive stage i.e. at the source of the MOSFET and the grid of the Output tube?
Anybody know what a typical signal voltage peek to peek that is riding on Vg is?
Thanks Steve
Apologies to the OP as this is a bit off-topic.
I've been meaning to ask you about your TSEs Boywonder. My next project is a TSE of one flavor another. I have the board already and will be bringing some James transformers back to the US with me when I leave Korea this summer. The plan was originally to build a 300B, but as I'm also planning a large house-shaking PP octal amp after that (how's that Universal Driver board coming George
You can just start your own thread...
I have one TSE that can drive both. It's still on the breadboard and has been neglected for a while now, but it is setup to drive a variety of output tubes including the 45 and 300B. I like the 45 setup the best so far. It can drive by 98dB speakers quite well and it gives detail unlike any other amplifier I have used to date.
Steve,
The answer is: both. The FET stage is a source follower. That means it has a voltage gain of about 1. Its purpose is to provide current gain to feed the grid of the output tube when it strays out of pure class A.
Here is my take on the source followers from what I've learned so far, which may or may not be totally correct:
As Russ mentioned, they are a unity gain buffer stage between the 5842's and the output tubes. This greatly reduces the need for the driver tube to source current to the output tubes.
Typically high voltage gain drivers have high plate R (so high output impedance) have a limited ability to provide the current that the output tubes require to drive their Miller capacitance. Triodes typically have relatively higher Miller C than tetrodes and pentodes since they lack the additional elements (grids/screens) used to reduce Miller C. Low mu drivers can source some current but have limited voltage swing. You can't have both since voltage x current =power and the small tubes max power spec will be exceeded (no free lunch...)
The mosfet follower provides a high input impedance for the driver tube so very little current is required from the driver tube. The mosfet follower has low output impedance to easily drive the output tube's Miller C.
The power supply for the mosfets needs to have low output impedance since this current is what drives the Milller C of the output tubes.
The negative voltage for the follower has to be at least as negative as the bias voltage plus some headroom. I'm not exactly clear on the positive voltage for the follower but if you are running A2 (or AB2 in PP designs) you need it to drive the output tube's grid positive.
So, in summary, the driver tube provides the voltage swing and the mosfet follower provides the current.
As Russ mentioned, they are a unity gain buffer stage between the 5842's and the output tubes. This greatly reduces the need for the driver tube to source current to the output tubes.
Typically high voltage gain drivers have high plate R (so high output impedance) have a limited ability to provide the current that the output tubes require to drive their Miller capacitance. Triodes typically have relatively higher Miller C than tetrodes and pentodes since they lack the additional elements (grids/screens) used to reduce Miller C. Low mu drivers can source some current but have limited voltage swing. You can't have both since voltage x current =power and the small tubes max power spec will be exceeded (no free lunch...)
The mosfet follower provides a high input impedance for the driver tube so very little current is required from the driver tube. The mosfet follower has low output impedance to easily drive the output tube's Miller C.
The power supply for the mosfets needs to have low output impedance since this current is what drives the Milller C of the output tubes.
The negative voltage for the follower has to be at least as negative as the bias voltage plus some headroom. I'm not exactly clear on the positive voltage for the follower but if you are running A2 (or AB2 in PP designs) you need it to drive the output tube's grid positive.
So, in summary, the driver tube provides the voltage swing and the mosfet follower provides the current.
Apologies to the OP as this is a bit off-topic.
I've been meaning to ask you about your TSEs Boywonder. My next project is a TSE of one flavor another. I have the board already and will be bringing some James transformers back to the US with me when I leave Korea this summer. The plan was originally to build a 300B, but as I'm also planning a large house-shaking PP octal amp after that (how's that Universal Driver board coming George
I like them both. If you've got the James 6123's (90ma IIRC) they are plenty big enough for 300B's, although you could also run 45's. James also has a smaller output transformer that could be interesting with 45's, especially if bi amping.
My speakers are TB-1772's in half Chang cabinets (95db) with a separate home brew isobaric woofer/plate amp. My listening room is somewhat large (vaulted ceilings, connected to other rooms) and the 45 amp may be a bit marginal at louder volumes, then again, maybe not........I mostly listen to acoustic stuff. The imaging with either of these amps is great in my setup.
The 300B TSE has been my main amp although recently I've been using the 45 amp since I just got it finished and dialed in. The 300B amp has vintage tube sockets that work fine, the 45 amp has new Chinese sockets that needed their contacts massaged with a small screwdriver to be happy.
Slightly off-topic, but here goes.....
I put my 300B TSE on the bench yesterday and connected up my sound card and attenuator box thingy in an effort to tweak the 5842 anode voltage and look at the distortion spectra.
Two observations: My two soundcards (both 24/96) differ greatly in their performance (duh!). My Azuntech Xplosion card has a least a 1 db rolloff between 10K and 20K on loopback testing. My M-audio firewire card is flat past 20K when looped back, and appears to have slightly better noise performance.
Anyway, I adjusted the 5842 anode voltage by twisting the pot and looking at the voltage with a DC voltmeter and observed the distortion using Rightmark Audio Analyzer software. I adjusted the anode voltage between about 180VDC and 115VDC. The higher harmonics (4th and higher) began to climb at voltages below around 130-135V. Going lower than 135V, the second and third also started to rise a little. Raising the voltage above 150-155V or so, nothing moved much. I had similar results on both channels.
I then set the anode voltages back to 175VDC and buttoned everything up.
Interesting.
I wish my electronics re education was a little further along, so that I might be able to understand this some what better.
I have tried to model this stage in LTSice but I’ve not been able to find a model for the CCS IXYS 10M45S. I did find one for a 10M90s but I can’t get it to run. I have substituted in a DN2540 but it’s not the same.
Thanks
Steve
Are you saying the distortion was about the same from 150vdc to 175vdc?
I wish my electronics re education was a little further along, so that I might be able to understand this some what better.
I have tried to model this stage in LTSice but I’ve not been able to find a model for the CCS IXYS 10M45S. I did find one for a 10M90s but I can’t get it to run. I have substituted in a DN2540 but it’s not the same.
Thanks
Steve
Sorry, no. I suppose I could have captured some screen shots......
If I get motivated, I'll do the same test with my 45 TSE.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.