I found this site:
Link
It shows an active-load amplifier.
It is simply an SRPP output stage.
I never saw an output stage like this one before, so my question is:
Is this one a good solution?
I have 4 KT88, I could realise an amplifer wih this topology, but why nobody has already built it?
My idea was to build an SRPP input stage using two 12AX7 (paralleled) and then drive with it an SRPP output stage made with KT88.
It seems to be a good solution: a perfectly balanced PP output (if you make the right calculation...), more (twice) power than SE, pure class A, more speed and dynamics (at least this is the theory).
Why it is not used at all?
What are your feelings?
Ciao,
Giovanni
Link
It shows an active-load amplifier.
It is simply an SRPP output stage.
I never saw an output stage like this one before, so my question is:
Is this one a good solution?
I have 4 KT88, I could realise an amplifer wih this topology, but why nobody has already built it?
My idea was to build an SRPP input stage using two 12AX7 (paralleled) and then drive with it an SRPP output stage made with KT88.
It seems to be a good solution: a perfectly balanced PP output (if you make the right calculation...), more (twice) power than SE, pure class A, more speed and dynamics (at least this is the theory).
Why it is not used at all?
What are your feelings?
Ciao,
Giovanni
I saw several schematics for amps loke this about 5 years ago. It seems that this fad never took off. I built one like this, and tried several variations. The basic concepts are on my page:
http://www.tubelab.com/SEoutput.htm
I did all of these experiments about 5 or 6 years ago, and unfortunately did not document them well. I remember that the EL34 was the best tube that I tried and that 6AS7's were the worst due to bias drift. I did not try KT88's but some of them are known to drift over time. You will need to make at least one of the cathode resistors adjustable, and adjust it for symetrical clipping. If this changes too much after a few hours of operation, try different tubes.
I have since found that replacing the top tube with a CCS IC works the best for me. That winds up being a CCS loaded active parafeed amp which wastes a lot of power but sounds good. I then began experimenting with hybrid tube - transistor output stages and never returned to finish up the SRPP output.
http://www.tubelab.com/SuperTubeSE.htm
Since drawing those schematics, I discovered the IXYS 10M45 CCS IC. That makes Constant Current Sources easy. Just make sure that you use a gate stopper resistor (1K) otherwise they will ocsillate!
I am still experimenting with this concept and will put the results on the web site when done, but the recent hurricane has set all Tubelab experimentation back for months.
I was beginning to re - visit the Totem Pole output topology since Triode Electronics came out with a driver transformer with 4 independent secondaries. I want to build an H bridge output stage using tubes.
http://www.tubelab.com/TotemPole.htm
http://www.tubelab.com/SEoutput.htm
I did all of these experiments about 5 or 6 years ago, and unfortunately did not document them well. I remember that the EL34 was the best tube that I tried and that 6AS7's were the worst due to bias drift. I did not try KT88's but some of them are known to drift over time. You will need to make at least one of the cathode resistors adjustable, and adjust it for symetrical clipping. If this changes too much after a few hours of operation, try different tubes.
I have since found that replacing the top tube with a CCS IC works the best for me. That winds up being a CCS loaded active parafeed amp which wastes a lot of power but sounds good. I then began experimenting with hybrid tube - transistor output stages and never returned to finish up the SRPP output.
http://www.tubelab.com/SuperTubeSE.htm
Since drawing those schematics, I discovered the IXYS 10M45 CCS IC. That makes Constant Current Sources easy. Just make sure that you use a gate stopper resistor (1K) otherwise they will ocsillate!
I am still experimenting with this concept and will put the results on the web site when done, but the recent hurricane has set all Tubelab experimentation back for months.
I was beginning to re - visit the Totem Pole output topology since Triode Electronics came out with a driver transformer with 4 independent secondaries. I want to build an H bridge output stage using tubes.
http://www.tubelab.com/TotemPole.htm
Konnichiwa,
Depends. I have been messing with active load solutions on papaer for a while.
The key drawback is that you need nearly double the +B to choke loaded parallelfeed or series feed.
Here a little trick for you.
Combine a KT-88 and the IXYS CCS Tubelab mentions (replace the cathode R of the upper SRPP device with the CCS). This way dissipation in the Solid State CCS is limited, most is dissipated in the KT88. You can take the output from the lower Valve Anode (basically traditional SE parallelfeed) or from the upper KT-88 Cathode for basically a Power Mu-Follower.
Here some more active load SE Amp's for you:
http://www.studioerosbarone.it/diego/MAD.htm
http://home.pacifier.com/~gpimm/schematics.htm
Sayonara
croccodillo said:
Depends. I have been messing with active load solutions on papaer for a while.
The key drawback is that you need nearly double the +B to choke loaded parallelfeed or series feed.
Here a little trick for you.
Combine a KT-88 and the IXYS CCS Tubelab mentions (replace the cathode R of the upper SRPP device with the CCS). This way dissipation in the Solid State CCS is limited, most is dissipated in the KT88. You can take the output from the lower Valve Anode (basically traditional SE parallelfeed) or from the upper KT-88 Cathode for basically a Power Mu-Follower.
Here some more active load SE Amp's for you:
http://www.studioerosbarone.it/diego/MAD.htm
http://home.pacifier.com/~gpimm/schematics.htm
Sayonara
As I type this, I'm listening to an amp inspired by Peter Millett's design you linked to. I ended up using a pair of 12e1s both triode connected.
more details here
Have been listening to various different versions of this circuit for over a year and not bored with it yet which is fairly unusual for me.
Worth experimenting with I reckon
more details here
Have been listening to various different versions of this circuit for over a year and not bored with it yet which is fairly unusual for me.
Worth experimenting with I reckon
You mean something like THIS ?
I cannot find the IXYS CCS here in Italy, but I can realize a good discrete CCS (not the one in the schematic, perhaps the Gary Pimm's one).
The idea is really good, the KT88 will handle all the job, controlled in a precise way by the CCS. I like it!
With the value of R2 in the schematic the current through the KT88s is 100mA, more than enough to have 8-10W output power.
My question is: how can I bias correctly the lower tube? Do I have to bias it with a negative voltage and, then, use a trimmer to regulate its anode voltage to 350V (half the power supply voltage)?
Again: A KT88 connected in this way could be driven by a SRPP made with two 12AX (I have lot of them...) or is it better to use two EL84 as driver (I have some EL84 and some ECL84 tube).
The lower KT-88 can be biased as shown with negative grid voltage, or it could be done with a cathode resistor. Either one should work. You can start by adjusting the plate voltage to half supply, but if you have a scope, adjust for symetrical clipping when slightly overdriven.
First set up the CCS to the desired operating current. I usually put a current meter in place of the bottom tube (plate to cathode) and set the current. Then install the bottom tube. The cathode resistor (or grid voltage) sets the plate voltage on the bottom triode when fed by a CCS.
If you use a CCS on the input tube you will need to raise the supply voltage to allow for the needed voltage across the CCS.
Which driver tube to use? That is your call, I am sure both would work.
I also think that this concept is a good idea, and I would like to experiment further, but I am not able to experiment right now, so let us all know how it works!
Are you using Eagle for your schematics? If so, where did you get your Symbols for KT-88 and EL-84, do they have PCB sockets?
Also the value for the grid resistor for the KT-88 is too high. They will go into a runaway condition if this resistor is too high. Check the data sheet for the brand tube that you are using. Some of the Chinese ones don't like much over 75K.
First set up the CCS to the desired operating current. I usually put a current meter in place of the bottom tube (plate to cathode) and set the current. Then install the bottom tube. The cathode resistor (or grid voltage) sets the plate voltage on the bottom triode when fed by a CCS.
If you use a CCS on the input tube you will need to raise the supply voltage to allow for the needed voltage across the CCS.
Which driver tube to use? That is your call, I am sure both would work.
I also think that this concept is a good idea, and I would like to experiment further, but I am not able to experiment right now, so let us all know how it works!
Are you using Eagle for your schematics? If so, where did you get your Symbols for KT-88 and EL-84, do they have PCB sockets?
Also the value for the grid resistor for the KT-88 is too high. They will go into a runaway condition if this resistor is too high. Check the data sheet for the brand tube that you are using. Some of the Chinese ones don't like much over 75K.
Thanks, Tubelab (may I call you George?) it sound strange to me to call people with nicknames...
Well, I visited your site, I've seen the damages done by the hurricane... Really impressive!!!
All my best wishes for the next future, I hope you (and all of your compatriots ) can solve all the related problems soon.
Yes, I'm using Eagle, the symbols are from an add-in library you can find under the "Download->libraries" section of the CadSoft site,
HERE
The library is, of course, "tubes.zip",
THIS
In the library you can find also PCB sockets, yes.
HERE a screenshot of the library from Eagle.
I'm still collecting all the material I need to build the amplifier, I miss the two output transformers (my previous intention was to build a KT88 PSE amp with suitable O/T, but now I need to change tipology of O/T) and power transformer.
In short term, I miss all the iron.
Ciao,
Giovanni
Well, I visited your site, I've seen the damages done by the hurricane... Really impressive!!!
All my best wishes for the next future, I hope you (and all of your compatriots ) can solve all the related problems soon.
Yes, I'm using Eagle, the symbols are from an add-in library you can find under the "Download->libraries" section of the CadSoft site,
HERE
The library is, of course, "tubes.zip",
THIS
In the library you can find also PCB sockets, yes.
HERE a screenshot of the library from Eagle.
I'm still collecting all the material I need to build the amplifier, I miss the two output transformers (my previous intention was to build a KT88 PSE amp with suitable O/T, but now I need to change tipology of O/T) and power transformer.
In short term, I miss all the iron.
Ciao,
Giovanni
Konnichiwa,
Yes, something like that.
I would make the following changes:
1) R7 = 68....100KOhm (tubelab remarked on that too).
2) R6 = 1K
3) Remove C3
4) Bypass R1 with a suitable capacitor
5) Use the EL84 as Pentode, Ra = 12KOhm, Rg2 adjusted for 150V screengrid and 250V Anode with a R1 = 470R, big film capacitor on screengrid.
6) For kicks use a 300B as lower valve, not a KT88.
7) Try taking the output both from the Top KT88 Cathode or the Bottom KT88 (or 300B) Anode.
The one in the picture will not be all THAT bad either.
As Tubelab says, first set the current, then set the lower valve anode voltage. Done.
KT88 (or even 300B) needs a good deal of swing, better use the EL84 as pentode or use a 6SL7/ECC83 SRPP or something.
Sayonara
croccodillo said:
Yes, something like that.
I would make the following changes:
1) R7 = 68....100KOhm (tubelab remarked on that too).
2) R6 = 1K
3) Remove C3
4) Bypass R1 with a suitable capacitor
5) Use the EL84 as Pentode, Ra = 12KOhm, Rg2 adjusted for 150V screengrid and 250V Anode with a R1 = 470R, big film capacitor on screengrid.
6) For kicks use a 300B as lower valve, not a KT88.
7) Try taking the output both from the Top KT88 Cathode or the Bottom KT88 (or 300B) Anode.
croccodillo said:
The one in the picture will not be all THAT bad either.
croccodillo said:
As Tubelab says, first set the current, then set the lower valve anode voltage. Done.
croccodillo said:
KT88 (or even 300B) needs a good deal of swing, better use the EL84 as pentode or use a 6SL7/ECC83 SRPP or something.
Sayonara
Giovanni:
Yes, you can call me George. Thanks for the info on Eagle. I downloaded that library when it first appeared (at least 2 years ago) and never looked back. It has been updated with many more tubes since I downloaded it. I got the new version.
As to the hurricane damage, we finally got our electricity restored this weekend after two weeks in the dark. There are still many people with no power. We were not able to go to work for 1 week, but my project deadlines did not change, so now I go to work for at least 10 hours per day, so no time for tubes. I will get back to the experiments in a month or two.
Any progress on the SMPS?
I would agree with the parts value changes proposed by Kuei Yang Wang (numbers 1 - 4), but use a KT-88 if you have them. After you have everything working, then you can experiment with the 300B. If you need good 300B's for a reasonable price, the new Chinese ones made by Shuguang (marketed by many companies) work pretty good. Driver in pentode mode? Depends on how much signal you have.
Yes, you can call me George. Thanks for the info on Eagle. I downloaded that library when it first appeared (at least 2 years ago) and never looked back. It has been updated with many more tubes since I downloaded it. I got the new version.
As to the hurricane damage, we finally got our electricity restored this weekend after two weeks in the dark. There are still many people with no power. We were not able to go to work for 1 week, but my project deadlines did not change, so now I go to work for at least 10 hours per day, so no time for tubes. I will get back to the experiments in a month or two.
Any progress on the SMPS?
I would agree with the parts value changes proposed by Kuei Yang Wang (numbers 1 - 4), but use a KT-88 if you have them. After you have everything working, then you can experiment with the 300B. If you need good 300B's for a reasonable price, the new Chinese ones made by Shuguang (marketed by many companies) work pretty good. Driver in pentode mode? Depends on how much signal you have.
Thanks for all your help.
I'm slightly begin to understand vacuum tubes...
George,
The preliminary schematic for the SMPS is ready, the magnetic core is ready with all the windings done, but I have no time at all to build the circuit.
Just to show it, THIS is the schematic.
It is really simple (keep it simple as you can).
THe core is an EPCOS ETD44/N67 core, and with the given oscillator frequency (35Khz) and the given topology (half bridge) it can output a theoretical continuous power of 700W!!!
Of course we do not need so much power: the mosfets are rated for a 300W continuous power (not so much powerful Mosfets -as IRF830A are- means they are easier to drive).
The output voltage, with the ratio reported, is of about 500V.
Higher voltage can easily be obtained with more turns on the secondary windong.
The core must be provided with an air gap (0.5 - 1 mm.), in manner to avoid histeresis shift of the core if drivers are not exactly balanced.
The C3 capacitor is used to block DC component in the core, in manner to make it more stable (Please don't ask me to explain it: it is too complicated in your language for me...).
The Jumper SJ1 can switch between 115V (when closed) and 230V main supply.
The SG3525 IC is used as the main oscillator/PWM controller; its PWM outputs are reported to IC1, an high-voltage Mosfet driver used to minimize the turn-on/off time of Mosfets.
The feedback is taken from the opto-coupler (they are quite linear in a certain range of current), driven by a Mosfet.
This past of the circuit is the one I'm not sure of: I never tried a feedback made in this way (my past SMPSs have current feedback), so it will need some serious tests to make it work fine.
I hope to build it in the next year (yes, year...).
Due to the lack of time, my idea is to build my tube amplifier with a traditional power supply (I could buy some cheap surplus transformers), and to exchange it when the SMPS will be ready.
Ciao,
Giovanni
I'm slightly begin to understand vacuum tubes...
George,
The preliminary schematic for the SMPS is ready, the magnetic core is ready with all the windings done, but I have no time at all to build the circuit.
Just to show it, THIS is the schematic.
It is really simple (keep it simple as you can).
THe core is an EPCOS ETD44/N67 core, and with the given oscillator frequency (35Khz) and the given topology (half bridge) it can output a theoretical continuous power of 700W!!!
Of course we do not need so much power: the mosfets are rated for a 300W continuous power (not so much powerful Mosfets -as IRF830A are- means they are easier to drive).
The output voltage, with the ratio reported, is of about 500V.
Higher voltage can easily be obtained with more turns on the secondary windong.
The core must be provided with an air gap (0.5 - 1 mm.), in manner to avoid histeresis shift of the core if drivers are not exactly balanced.
The C3 capacitor is used to block DC component in the core, in manner to make it more stable (Please don't ask me to explain it: it is too complicated in your language for me...).
The Jumper SJ1 can switch between 115V (when closed) and 230V main supply.
The SG3525 IC is used as the main oscillator/PWM controller; its PWM outputs are reported to IC1, an high-voltage Mosfet driver used to minimize the turn-on/off time of Mosfets.
The feedback is taken from the opto-coupler (they are quite linear in a certain range of current), driven by a Mosfet.
This past of the circuit is the one I'm not sure of: I never tried a feedback made in this way (my past SMPSs have current feedback), so it will need some serious tests to make it work fine.
I hope to build it in the next year (yes, year...).
Due to the lack of time, my idea is to build my tube amplifier with a traditional power supply (I could buy some cheap surplus transformers), and to exchange it when the SMPS will be ready.
Ciao,
Giovanni
Looking at the schematic, which should be, in your opinion, the right supply voltage?
Was thinking something about 700V@100mA: with this value you should be able to drive the two KT88 about in the middle of their curve in triode connection (I hope you can understand what I mean, my language is not so correct).
Looking at the curve (this) we should be in the middle of the linear zone.
Using such a voltage and such a current (100mA), how can I calculate the output power?
Ciao,
Giovanni
Was thinking something about 700V@100mA: with this value you should be able to drive the two KT88 about in the middle of their curve in triode connection (I hope you can understand what I mean, my language is not so correct).
Looking at the curve (this) we should be in the middle of the linear zone.
Using such a voltage and such a current (100mA), how can I calculate the output power?
Ciao,
Giovanni
Konnichiwa,
First work out the lower valves operating conditions the normal way (load line).
Find the highest positive anodeswing point you want to have without clipping.
Then consider the SS CCS and it's minimum operation voltage (say 5V for the IXYS CCS).
Look at the valves curves and find a point where the current is the value have chosen and the gridbias equals your minimum voltage.
Read off the anode voltage, add the bias voltage and add to that derived earlier as the maximum positive swing for the lower valve. This gives your +B.
If using self bias for the lower valve add the selfbias voltage and if using a direct coupled driver stage add that stages anode voltage.
Here an example for what I call "optimised loop" amplifier (note, theoretical/experiomental ONLY)...
In the simulator (with Valve models based on real triode curves from Sofia) it suggests 8W RMS @ 3% 2nd HD with a textbook behaviour for the other harmonics.
Frequency response prior to the output transformer is pretty much DC to light. The Sowter 8995 would probably the ticket for this circuit, they are 76 £ each, so 112 Euro or 132 $ US each, plus taxes which in the EU is 17.5% and shipping (I not normally recommend Sowter BTW).
The FET/Battery CCS can be substituted by a IXYS CCS (obviously) beware of thermal drift in the J-Fet CCS!
Sayonara
croccodillo said:
First work out the lower valves operating conditions the normal way (load line).
Find the highest positive anodeswing point you want to have without clipping.
Then consider the SS CCS and it's minimum operation voltage (say 5V for the IXYS CCS).
Look at the valves curves and find a point where the current is the value have chosen and the gridbias equals your minimum voltage.
Read off the anode voltage, add the bias voltage and add to that derived earlier as the maximum positive swing for the lower valve. This gives your +B.
If using self bias for the lower valve add the selfbias voltage and if using a direct coupled driver stage add that stages anode voltage.
Here an example for what I call "optimised loop" amplifier (note, theoretical/experiomental ONLY)...
An externally hosted image should be here but it was not working when we last tested it.
In the simulator (with Valve models based on real triode curves from Sofia) it suggests 8W RMS @ 3% 2nd HD with a textbook behaviour for the other harmonics.
Frequency response prior to the output transformer is pretty much DC to light. The Sowter 8995 would probably the ticket for this circuit, they are 76 £ each, so 112 Euro or 132 $ US each, plus taxes which in the EU is 17.5% and shipping (I not normally recommend Sowter BTW).
The FET/Battery CCS can be substituted by a IXYS CCS (obviously) beware of thermal drift in the J-Fet CCS!
Sayonara
Let's see if I've understood...
I traced some load curves on the above graph.
I have a pair of very good toroidal transformer with a primary impedance of 2600Ohm (they are not audio equipment but industrial components, at least good as audio iron), I will use them for the first tests.
Following your suggestion:
Fixed the operating point of the lower tube at 350V, 100mA (blue point).
The load curve is the magenta one; it means that I can have some more that 400V swing and a current variation from a minimum of 20 to a maximum of 170 mA.
The load current, then, will vary from 0 (no signal) to a maximum of 150 mA (maximum swing).
That mean that, with a load of 2600 Ohm I can have a power of (Ohm's law?):
P=R * I^2 = 2300 * 0,150^2 = 50W
This should be the peak to peak power, the rms power should be:
50/2.86 = 17,5 W.
is it correct?
Let's say that the maximum voltage at maximum swing is 560V.
Let's use 5 Volt as minimum CCS voltage.
From the curve the point where we have 100 mA at such a minimum grid voltage is 150V.
So,
560V (max voltage@max swing) +
5V (min CCS voltage) +
150V (operating point@min grid voltage) =
-------------------------------------------------------
715V B+
Not so far from what I've proposed in the previous post... I'm lucky.
I do prefer to have a separate bias negative voltage for the lower tube, so the above voltage is my B+, and another bias voltage will be needed.
Your comments, please?
I'm beginning to have a lot of fun with tubes!!!!
An externally hosted image should be here but it was not working when we last tested it.
I traced some load curves on the above graph.
I have a pair of very good toroidal transformer with a primary impedance of 2600Ohm (they are not audio equipment but industrial components, at least good as audio iron), I will use them for the first tests.
Following your suggestion:
Fixed the operating point of the lower tube at 350V, 100mA (blue point).
The load curve is the magenta one; it means that I can have some more that 400V swing and a current variation from a minimum of 20 to a maximum of 170 mA.
The load current, then, will vary from 0 (no signal) to a maximum of 150 mA (maximum swing).
That mean that, with a load of 2600 Ohm I can have a power of (Ohm's law?):
P=R * I^2 = 2300 * 0,150^2 = 50W
This should be the peak to peak power, the rms power should be:
50/2.86 = 17,5 W.
is it correct?
Let's say that the maximum voltage at maximum swing is 560V.
Let's use 5 Volt as minimum CCS voltage.
From the curve the point where we have 100 mA at such a minimum grid voltage is 150V.
So,
560V (max voltage@max swing) +
5V (min CCS voltage) +
150V (operating point@min grid voltage) =
-------------------------------------------------------
715V B+
Not so far from what I've proposed in the previous post... I'm lucky.
I do prefer to have a separate bias negative voltage for the lower tube, so the above voltage is my B+, and another bias voltage will be needed.
Your comments, please?
I'm beginning to have a lot of fun with tubes!!!!
I think that your power output expectations are a bit optimistic. This is the way that I would approach this circuit. First I would consider the static (DC) conditions at the signal extremes. This is to make sure that you won't run out of voltage and clip unexpectedly.
The bottom tube has its cathode grounded so the grid voltage is not limited to the minimum CCS voltage, it could even be driven positive if desired. From my experience triode wired pentodes don't like to have the grid driven positive. DHT's (300B, 845, etc) can be operated in this manner (class A2) if you have a good driver. So for the peak positive going extreme we will take the grid of the bottom KT-88 to zero volts. The distortion would begin to rise if driven above this value. At a fixed current of 100 mA (from your curves) it looks like the tube will want 100 volts across it. This is the minimum voltage across the tube when driven by a positive going peak.
The quiescent point was set arbitrarily at 350 volts, 100 mA. This gives a bias voltage of about -30 volts. Since the amp is designed with symetrical sine waves we must assume that the maximum negative excursion at the grid would be -60 volts. This implies a 60 volt peak to peak drive signal centered around - 30 volts. From the load line this would indicate a maximum plate excursion of 500 volts. This is the maximum voltage across the tube when driven by a negative going peak. This gives a peak to peak value of 400 volts. This gives a power output of about 8 watts RMS. At this point the top tube will need 150 volts plus 5 volts for the CCS giving a total B+ requirement of 655 volts.
Since you have already considered a 700 volt supply voltage, You might try raising the operating point of the bottom tube to about 400 volts. You will have to lower the operating current to stay within the dissipation limits for the tube. Redraw the load line and repeat the above steps. Avoid minimum tube currents much below 10 mA, they become non-linear. After a few itterations you might be able to find a point that gives about 10 watts.
I find that it becomes easier to just build it and adjust for best sound since there is a sweet spot for every tube, and you won't find it on a set of curves!
I agree with the simple linear power supply first. Get all to work and then build your SMPS. I looked at your circuit and understand most of it. I may not know exactly why C3 is required but I know that the switch fets have a short unhappy life without it.
The bottom tube has its cathode grounded so the grid voltage is not limited to the minimum CCS voltage, it could even be driven positive if desired. From my experience triode wired pentodes don't like to have the grid driven positive. DHT's (300B, 845, etc) can be operated in this manner (class A2) if you have a good driver. So for the peak positive going extreme we will take the grid of the bottom KT-88 to zero volts. The distortion would begin to rise if driven above this value. At a fixed current of 100 mA (from your curves) it looks like the tube will want 100 volts across it. This is the minimum voltage across the tube when driven by a positive going peak.
The quiescent point was set arbitrarily at 350 volts, 100 mA. This gives a bias voltage of about -30 volts. Since the amp is designed with symetrical sine waves we must assume that the maximum negative excursion at the grid would be -60 volts. This implies a 60 volt peak to peak drive signal centered around - 30 volts. From the load line this would indicate a maximum plate excursion of 500 volts. This is the maximum voltage across the tube when driven by a negative going peak. This gives a peak to peak value of 400 volts. This gives a power output of about 8 watts RMS. At this point the top tube will need 150 volts plus 5 volts for the CCS giving a total B+ requirement of 655 volts.
Since you have already considered a 700 volt supply voltage, You might try raising the operating point of the bottom tube to about 400 volts. You will have to lower the operating current to stay within the dissipation limits for the tube. Redraw the load line and repeat the above steps. Avoid minimum tube currents much below 10 mA, they become non-linear. After a few itterations you might be able to find a point that gives about 10 watts.
I find that it becomes easier to just build it and adjust for best sound since there is a sweet spot for every tube, and you won't find it on a set of curves!
I agree with the simple linear power supply first. Get all to work and then build your SMPS. I looked at your circuit and understand most of it. I may not know exactly why C3 is required but I know that the switch fets have a short unhappy life without it.
I built just such an amp.
I started out with the "RH807" partial feedback amp design, using an 807 as the output tube. I had 700V to use up so I put in a TT21 (KT88 with a top cap) at the top as a CCS. I used a resistor to set the CCS and then placed another in the cathode with a 1K value (I think) to stiffen the CCS action. The TT21 is running very lean at about 250V across it, but it seems happy enough. Took the output from the anode of the 807. Sounds very nice and works great. Can't claim to have optimised it much. Tried a Gary Pimm Pentode CCS, but the voltage across the TT21 collapsed.
My understanding is that taking the ouput as a Mu- stage may lead to problems, because a SRPP likes to see a constant fixed load, and most speakers aren't.
Certainly worth a try if you have plenty of volts to use up.
Shoog
I started out with the "RH807" partial feedback amp design, using an 807 as the output tube. I had 700V to use up so I put in a TT21 (KT88 with a top cap) at the top as a CCS. I used a resistor to set the CCS and then placed another in the cathode with a 1K value (I think) to stiffen the CCS action. The TT21 is running very lean at about 250V across it, but it seems happy enough. Took the output from the anode of the 807. Sounds very nice and works great. Can't claim to have optimised it much. Tried a Gary Pimm Pentode CCS, but the voltage across the TT21 collapsed.
My understanding is that taking the ouput as a Mu- stage may lead to problems, because a SRPP likes to see a constant fixed load, and most speakers aren't.
Certainly worth a try if you have plenty of volts to use up.
Shoog
A CCS inherently has a high PSRR by its very nature. If a CCS (of any topology) exhibited a perfectly constant current, it would be totally immune to power supply variations. Unfortunately nothing is perfect. When I was developing the PowerDrive circuit, I tested several different CCS topologies and settled on the IXYS 10M45 chip as the best compromize between performance and simplicity.
Of the tube and hybrid topologies that I tried, those that used a true pentode worked the best. The curves of a pentode are closer than a triode to a CCS. I did not try the topology that has been proposed in this thread. I will do so as soon as I can get my lab operational, it became a storage shed during the hurricane. I am thinking 845 on the bottom, 813 on the top, 2KV of B+. If it didn't blow up it would sound awesome.
If I had to guess, I would think that you would see just as much variation between two different triodes, as between a triode and a triode strapped pentode.
Of the tube and hybrid topologies that I tried, those that used a true pentode worked the best. The curves of a pentode are closer than a triode to a CCS. I did not try the topology that has been proposed in this thread. I will do so as soon as I can get my lab operational, it became a storage shed during the hurricane. I am thinking 845 on the bottom, 813 on the top, 2KV of B+. If it didn't blow up it would sound awesome.
If I had to guess, I would think that you would see just as much variation between two different triodes, as between a triode and a triode strapped pentode.
I have a different idea for the CCS:
I have some LM833 ics that I can use for this project: these are differential op-amp specifically designed for audio (THIS is its datasheet).
They have extremely low noise and distorsion, an extremely wide bandwidth and an extreme wide dynamic range.
If I can design a little driver circuit between them and a mosfet (I'm thinking at a class-A little driver) I should be able to have a perfect CCS, with almost no distorsion at all.
Of course my idea is to use a floating power supply for each CCS: the LM833 works at its best at a dual voltage of 15V, so I'll have to provide a little transformer and a stabilizer circuit for each CCS.
But we'll have only two CCS per each stereo amplifier, so the expense will be not high at all.
Ciao,
Giovanni
I have some LM833 ics that I can use for this project: these are differential op-amp specifically designed for audio (THIS is its datasheet).
They have extremely low noise and distorsion, an extremely wide bandwidth and an extreme wide dynamic range.
If I can design a little driver circuit between them and a mosfet (I'm thinking at a class-A little driver) I should be able to have a perfect CCS, with almost no distorsion at all.
Of course my idea is to use a floating power supply for each CCS: the LM833 works at its best at a dual voltage of 15V, so I'll have to provide a little transformer and a stabilizer circuit for each CCS.
But we'll have only two CCS per each stereo amplifier, so the expense will be not high at all.
Ciao,
Giovanni
A brand new CCS
Here we are.
This is the circuit I will try to realise.
TR1 is a tiny transformer, about 3-4VA.
Unfortunately we will need a separate supply for each CCS.
But, due the small size and the small cost of the trafo there are no problems at all.
With the given values of R2, R3, R4 and R5 the CCS current can be set from a minimum of 40mA to a maximum of 150 mA using R5 trimmer.
The CCS can work as it is (without tubes), but with the limitation of the maximum voltage across Q1 (Vceo=100V).
Using transistors with higher Vceo could raise the maximum operating voltage across LPS3 and LPS4 up to 600-800V (Remember to remove the D2 zener).
The R1-C5 network is used to soft-start the CCS: the C5 capacitor will be charged up to 5 Volts in about 1 minute and an half, increasing in the same way the output current of the CCS, that will start from 0 to the desidered value.
Remove the R1 resistor to have immediate nominal output current.
The whole thing can be mounted onto a really small PCB.
Here we are.
This is the circuit I will try to realise.
TR1 is a tiny transformer, about 3-4VA.
Unfortunately we will need a separate supply for each CCS.
But, due the small size and the small cost of the trafo there are no problems at all.
With the given values of R2, R3, R4 and R5 the CCS current can be set from a minimum of 40mA to a maximum of 150 mA using R5 trimmer.
The CCS can work as it is (without tubes), but with the limitation of the maximum voltage across Q1 (Vceo=100V).
Using transistors with higher Vceo could raise the maximum operating voltage across LPS3 and LPS4 up to 600-800V (Remember to remove the D2 zener).
The R1-C5 network is used to soft-start the CCS: the C5 capacitor will be charged up to 5 Volts in about 1 minute and an half, increasing in the same way the output current of the CCS, that will start from 0 to the desidered value.
Remove the R1 resistor to have immediate nominal output current.
The whole thing can be mounted onto a really small PCB.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.