I'm using an old pair of Altec 15095A line input transformers wired for the 1:5 step-up ratio. These are the only input transformers I've ever used, so I haven't compared them to anything else yet. Manufacturer specs are perhaps not on par with modern units but I like them very much.
Attachments
Last edited:
A 1:5 step up ratio. The voltage steps up by 5 x.
But the impedance steps up my 5 x 5 = 25.
I could not read the small print of the data sheets.
But fore an example:
With a 100k grid resistor, Rg, the input impedance is 100k / 25 = 4 kOhms.
The preamp or CD player would need to have an output impedance of 400 Ohms to keep distortion down.
Sometimes, the free lunch of voltage increase comes with a price.
But the impedance steps up my 5 x 5 = 25.
I could not read the small print of the data sheets.
But fore an example:
With a 100k grid resistor, Rg, the input impedance is 100k / 25 = 4 kOhms.
The preamp or CD player would need to have an output impedance of 400 Ohms to keep distortion down.
Sometimes, the free lunch of voltage increase comes with a price.
J_Perkins,
You are correct, the DC return for the grid is the secondary DCR!
But . . .
The reason I suggested a 100k resistor, Rg, is it might partially swamp-out the distributed capacitance of the secondary.
That secondary's distributed capacitance is . . . 5 Times more capacitance, as it appears when reflected onto the Primary.
The input to the transformer has lots of capacitance across it.
I would venture a guess that the secondary distributed capacitance, is larger than the Miller Effect Capacitance of most high gain tubes (that the secondary has to drive).
Almost everything has some kind of tradeoff, or compromise.
That is why they get the big bucks $$$ for the very best input transformers . . .
It is not easy to minimize all the tradeoffs.
I had a free lunch today, but only because I am a member of an organization. And membership is not free.
You are correct, the DC return for the grid is the secondary DCR!
But . . .
The reason I suggested a 100k resistor, Rg, is it might partially swamp-out the distributed capacitance of the secondary.
That secondary's distributed capacitance is . . . 5 Times more capacitance, as it appears when reflected onto the Primary.
The input to the transformer has lots of capacitance across it.
I would venture a guess that the secondary distributed capacitance, is larger than the Miller Effect Capacitance of most high gain tubes (that the secondary has to drive).
Almost everything has some kind of tradeoff, or compromise.
That is why they get the big bucks $$$ for the very best input transformers . . .
It is not easy to minimize all the tradeoffs.
I had a free lunch today, but only because I am a member of an organization. And membership is not free.
Last edited:
Yes no free lunch.you want extra gain .you pay with extra capacitance.you need a low source z driver to do that. Transformers are gr8t devices.
To answer (or muddle) J_Perkins' question, the source is a roughly 2-year-old MacBook laptop that, according to the published documentation, has an onboard hardware DAC that can handle up to 96kHz and master quality data from streaming services like Tidal (which I use). The headphone output jack detects the impedance of the headphones or amp connected to it and automatically adjusts the output impedance accordingly. So, maybe it's being wicked smart about driving the circuit on the breadboard, because it sounds fantastic.
Drawing to follow...
Drawing to follow...
I finally set up a direct-coupled circuit with stacked power supplies and was able to do measurement and listening tests.
The bottom supply drives the 26 and is rectified by a Raytheon B-H cold cathode with a choke input, followed by two more LC stages and an RC stage with a dropping resistor.
The top supply drives the 2A3 and is rectified by a 5V4GA with a choke input and two more LC stages.
At first there was too much current running through the tubes, with the 26 also having too much voltage while the 2A3 did not have enough. Through calculation, trial, and error, I changed transformers, went "up" the rectifiers on the top half (5R4GY --> 5U4GB --> 5V4GA), and adjusted the dropping and bias resistors to arrive at a very good-sounding circuit.
The first listening test had the 26 running at a higher voltage and current, and the sound was a bit stringent. After increasing the dropping resistor in the final leg of the bottom PS, it's now running at about 150Vak/5mA (although the plate current could be higher -- I'm not entirely sure how to read the datasheet). I plan to bring it down to the 135V/5.5mA op point soon.
The 2A3 is grid-biased. At 250Vak and 45V between the grid and cathode, It should be running right at 60mA for the standard operating point of that tube.
This circuit sounds impressive: very full-bodied, delicate, open, natural. I need to give it more listening time and to pay attention to acoustic instrument timbre in order to form a better judgment (the family is asleep, so I can't let it rip). However, I think I might prefer this to the next runner-up, which was 26 choke-loaded and cap-coupled to 2A3.
Schematic and breadboard photo attached. The moody lava lamp aura inspired me to nickname the rectifiers Little Boy and Fat Man.
The bottom supply drives the 26 and is rectified by a Raytheon B-H cold cathode with a choke input, followed by two more LC stages and an RC stage with a dropping resistor.
The top supply drives the 2A3 and is rectified by a 5V4GA with a choke input and two more LC stages.
At first there was too much current running through the tubes, with the 26 also having too much voltage while the 2A3 did not have enough. Through calculation, trial, and error, I changed transformers, went "up" the rectifiers on the top half (5R4GY --> 5U4GB --> 5V4GA), and adjusted the dropping and bias resistors to arrive at a very good-sounding circuit.
The first listening test had the 26 running at a higher voltage and current, and the sound was a bit stringent. After increasing the dropping resistor in the final leg of the bottom PS, it's now running at about 150Vak/5mA (although the plate current could be higher -- I'm not entirely sure how to read the datasheet). I plan to bring it down to the 135V/5.5mA op point soon.
The 2A3 is grid-biased. At 250Vak and 45V between the grid and cathode, It should be running right at 60mA for the standard operating point of that tube.
This circuit sounds impressive: very full-bodied, delicate, open, natural. I need to give it more listening time and to pay attention to acoustic instrument timbre in order to form a better judgment (the family is asleep, so I can't let it rip). However, I think I might prefer this to the next runner-up, which was 26 choke-loaded and cap-coupled to 2A3.
Schematic and breadboard photo attached. The moody lava lamp aura inspired me to nickname the rectifiers Little Boy and Fat Man.
Attachments
While it's breadboarded you might consider trying battery bias on the 26s, which will eliminate the bypass cap. That's what I did on a similar amp, the Cockeyed Monkey, which I built last year. The batteries last as long as their shelf life. This might also allow you to simplify the filament supply. With battery grid bias the two 26s can be heated by a single supply, assuming the transformer can supply the necessary current.
FlaCharlie,
Nice!
You put the battery in series with the signal.
Whenever I used battery bias, it was always in series with the signal.
That is the right way to do it; the only current the battery ever sees is if there is grid current (for most topologies, the amplifier is clipping then, and I prefer to turn the volume down because I do not like the sound of clipping).
I did not have an input transformer. Instead, the wiper of a 50k volume control connected to the + of the battery, and the other end of the battery (-) connected to the grid stopper resistor (another way to do series connection of the battery).
Some prefer to put the signal to the top of Rg, and connect the battery to ground the bottom end of Rg.
The disadvantage of doing it that way: the battery is constantly charging and discharging because the signal voltage sends current from Rg.
$0.03
Nice!
You put the battery in series with the signal.
Whenever I used battery bias, it was always in series with the signal.
That is the right way to do it; the only current the battery ever sees is if there is grid current (for most topologies, the amplifier is clipping then, and I prefer to turn the volume down because I do not like the sound of clipping).
I did not have an input transformer. Instead, the wiper of a 50k volume control connected to the + of the battery, and the other end of the battery (-) connected to the grid stopper resistor (another way to do series connection of the battery).
Some prefer to put the signal to the top of Rg, and connect the battery to ground the bottom end of Rg.
The disadvantage of doing it that way: the battery is constantly charging and discharging because the signal voltage sends current from Rg.
$0.03
Thanks for all this input. I'll try battery bias one of these days soon. And will also place a 100K grid leak resistor on the 26 to hear the difference.
First, I wanted to tame that 25 down. I think I was running it at close to 150Vak and almost 7mA last night. It sounded fantastic but that's running it way too hard.
Today I added resistance in the last leg of the lower PS, upping it to 6.8K. That brought the 26 down to about 145V anode and 8.7V cathode, for an op point of about 137Vak/5.5mA (by the datasheet, once you add 0.75V for the internal filament bias offset).
That then required me to up the 2A3 grid bias resistance to 6.3K, in order to get 45V bias and keep that tube running at 250Vak/60mA.
Need to find a circuit drawing program so I don't have to do it on paper all the time, otherwise I'd offer a revised one now.
It sounds really marvelous. I'm going to do more critical listening, but if I were to build with the 26 driving 2A3, I'm pretty sure this is the circuit I would do.
Next stop will be the 10Y, for which I need to order plate chokes. I'm thinking Lundahl 1668, which offers 100H and 25mA current capacity.
First, I wanted to tame that 25 down. I think I was running it at close to 150Vak and almost 7mA last night. It sounded fantastic but that's running it way too hard.
Today I added resistance in the last leg of the lower PS, upping it to 6.8K. That brought the 26 down to about 145V anode and 8.7V cathode, for an op point of about 137Vak/5.5mA (by the datasheet, once you add 0.75V for the internal filament bias offset).
That then required me to up the 2A3 grid bias resistance to 6.3K, in order to get 45V bias and keep that tube running at 250Vak/60mA.
Need to find a circuit drawing program so I don't have to do it on paper all the time, otherwise I'd offer a revised one now.
It sounds really marvelous. I'm going to do more critical listening, but if I were to build with the 26 driving 2A3, I'm pretty sure this is the circuit I would do.
Next stop will be the 10Y, for which I need to order plate chokes. I'm thinking Lundahl 1668, which offers 100H and 25mA current capacity.
I've used both methods of battery grid bias. Here's another build I did using the battery in series, as you describe:
I was under the impression that the method in the previous post, with the battery positive grounded, doesn't produce a significant amount of charging / discharging so I'm using a normal battery.
I've also tried using battery bias on the cathode which I know requires the use of a rechargeable battery. I abandoned that method because the battery would run down if I didn't use the amp regularly. Removing it and recharging it in a charger was too inconvenient.
If you use a grid leak resistor on the 26 I assume that means you'd be eliminating the input / step up transformer. But if you did that wouldn't you need to add another tube stage in order to drive the 2A3? Or would there be some advantage to adding it between the secondary of the SUT and ground?
As you see, I'm using a similar operating point for the 26 in the Nuance: 139v / 4.9mA.
I use the Drawing portion of an ancient "all in one" Mac program called Appleworks to draw schematics. It's so old that it won't run on my "newer" Mac Mini, which is now almost 14 years old, so I have an even older Mac Mini that's only used to draw schematics. I was worried it might die so I picked up a spare Mini on eBay for $50 shipped. If I ever get a new computer maybe I'll look around for a modern solution but Appleworks is great, free, and easy to use.
FlaCharlie,
Good, You were right!
Your schematic in Post # 549 . . .
When the interstage transformer secondary has more voltage than the battery voltage, there are 2 conditions:
1. When the top of the secondary goes Positive, the grid draws grid current. But that makes the bottom of the secondary go negative, which mildly charges the battery during peak signals (and only when the volume is turned too high, and grid current occurs, and we hear distortion; turn the volume down, this is Hi Fi not Electric Guitar).
2. When the top of the secondary goes Negative, the grid is negative, no grid current, no discharging of the battery.
In that circuit, the battery will have essentially the same as shelf life.
Good, You were right!
Your schematic in Post # 549 . . .
When the interstage transformer secondary has more voltage than the battery voltage, there are 2 conditions:
1. When the top of the secondary goes Positive, the grid draws grid current. But that makes the bottom of the secondary go negative, which mildly charges the battery during peak signals (and only when the volume is turned too high, and grid current occurs, and we hear distortion; turn the volume down, this is Hi Fi not Electric Guitar).
2. When the top of the secondary goes Negative, the grid is negative, no grid current, no discharging of the battery.
In that circuit, the battery will have essentially the same as shelf life.
jdrouin ,
Well done., good to see.
There are large current charging pulses to the input of your filament DC supply.
You can model this - look there for improvement.
Well done., good to see.
There are large current charging pulses to the input of your filament DC supply.
You can model this - look there for improvement.
Yes I need to get higher inductance chokes for the DC filament supply. It does make a little hum right now but it’s not bad at all.
I’ve been so blown away by the performance of the direct coupled circuit with the Lowther full range drivers that I moved the breadboard downstairs to play the Altec Valencias in my main system. It didn’t sound quite as good as the Lowthers.
I also moved the amps that had been playing the Valencias — 300B mono blocks and a line preamp reverse-engineered from George Wright products — upstairs to try with the Lowthers, and that wasn’t as good either.
Interesting that amp/speaker matching involves a certain Je ne sais quoi.
Maybe the 2A3’s are a little underpowered for the Valencias, or the direct coupled topology doesn’t work as well with two-way speakers and a crossover network?
I’ll listen for another day or two and then move them back.
I ordered Lundahl anode chokes for 10Y tubes, to swap in for the 26. They won’t be here for another six weeks, so in the meantime I plan to develop a version of this amp with the 45 on output, since I have all necessary parts on hand. 26 should drive the 45 even better than the 2A3.
I’ve been so blown away by the performance of the direct coupled circuit with the Lowther full range drivers that I moved the breadboard downstairs to play the Altec Valencias in my main system. It didn’t sound quite as good as the Lowthers.
I also moved the amps that had been playing the Valencias — 300B mono blocks and a line preamp reverse-engineered from George Wright products — upstairs to try with the Lowthers, and that wasn’t as good either.
Interesting that amp/speaker matching involves a certain Je ne sais quoi.
Maybe the 2A3’s are a little underpowered for the Valencias, or the direct coupled topology doesn’t work as well with two-way speakers and a crossover network?
I’ll listen for another day or two and then move them back.
I ordered Lundahl anode chokes for 10Y tubes, to swap in for the 26. They won’t be here for another six weeks, so in the meantime I plan to develop a version of this amp with the 45 on output, since I have all necessary parts on hand. 26 should drive the 45 even better than the 2A3.
Question about measuring plate current in a grid-biased output tube in the direct-coupled topology posted in #548 above.
If I'm wanting to run the 2A3 tube at the standard 250Vak/60mA, which has the grid biased at 45V below the cathode, can I trust that if I get it operating at those voltages that it is in fact running at 60mA?
What if I achieve those voltage points, but the voltage drop measured across the DCR of the OPT primary suggests the current is higher? At 60mA and with an OPT primary DCR of 118 ohm, there should be a 7V drop across the primary winding (118 x 0.06 = 7.08).
When I was trying different resistor values to get the op point right, there were times when even though the tube was running at 250Vak at a -45V bias, the voltage drop across the OPT primary was anywhere from 8V - 10V, which is way too high. I eventually got it so that the voltages were as desired and the drop across the primary was 7V - 8V, settling at 7V after the amp had been running for an hour or two.
A corollary to that is, What if the tube is running at 250Vak, the bias voltage is less than desired -- say 40V -- but there is a 7V drop across the 118 ohm primary. Could I trust that it's running at 60mA plate current then?
I guess what I'm asking is should I trust the the OPT primary voltage drop measurement or the voltage points at the three electrodes of the tube (assuming they're at the desired values)?
If I'm wanting to run the 2A3 tube at the standard 250Vak/60mA, which has the grid biased at 45V below the cathode, can I trust that if I get it operating at those voltages that it is in fact running at 60mA?
What if I achieve those voltage points, but the voltage drop measured across the DCR of the OPT primary suggests the current is higher? At 60mA and with an OPT primary DCR of 118 ohm, there should be a 7V drop across the primary winding (118 x 0.06 = 7.08).
When I was trying different resistor values to get the op point right, there were times when even though the tube was running at 250Vak at a -45V bias, the voltage drop across the OPT primary was anywhere from 8V - 10V, which is way too high. I eventually got it so that the voltages were as desired and the drop across the primary was 7V - 8V, settling at 7V after the amp had been running for an hour or two.
A corollary to that is, What if the tube is running at 250Vak, the bias voltage is less than desired -- say 40V -- but there is a 7V drop across the 118 ohm primary. Could I trust that it's running at 60mA plate current then?
I guess what I'm asking is should I trust the the OPT primary voltage drop measurement or the voltage points at the three electrodes of the tube (assuming they're at the desired values)?
1. Current x DCR = Voltage Drop
Voltage drop / DCR = current
Always, forever, eternally, amen
2A3 tubes like many other tubes, have medium variations on quiescent operating voltages and current.
Self bias resistor, if you use one is: 45V / 60mA = 750 Ohms, the Classic 2A3 circuit.
I remember the 2A3 Baby Ongaku, it had 50mA, 300V plate to filament, and a 1k self bias resistor, for 50V self bias. Requires 350V B+, not including the output transformer voltage drop from 50mA x DCR.
2. Now you have discovered why I use Individual self bias resistors.
B+ voltage has to be more than 250V.
7V DCR drop + 250V plate to filament + 45V across the self bias resistor = 302V B+.
302V B+ is only steady at 302V . . . when power mains voltage is steady at the right voltage, and there is the DCR of the power transformer primary and secondary, (and the variable impedance of a tube rectifier if you use that) so B+ varies versus load current.
Now, you have to try several 2A3 tubes until you get fairly close to 7V across DCR, with the 45V bias you are using.
We are cutting a raw diamond, and all diamonds are different.
Note: When they heat up, The DCR of the power transformer windings vary, and so do the output transformer's primary DCR.
Another thing that can affect the tube quiescent voltages and currents . . . is the filament voltage. If that varies, so does the rest of the triode parameters. Not all filament supplies are constant.
Temperature of the filament secondaries for example.
Do not forget to measure your power mains.
Mine vary from 117VAC rms, to 123VAC rms, on a true rms meter.
Voltage drop / DCR = current
Always, forever, eternally, amen
2A3 tubes like many other tubes, have medium variations on quiescent operating voltages and current.
Self bias resistor, if you use one is: 45V / 60mA = 750 Ohms, the Classic 2A3 circuit.
I remember the 2A3 Baby Ongaku, it had 50mA, 300V plate to filament, and a 1k self bias resistor, for 50V self bias. Requires 350V B+, not including the output transformer voltage drop from 50mA x DCR.
2. Now you have discovered why I use Individual self bias resistors.
B+ voltage has to be more than 250V.
7V DCR drop + 250V plate to filament + 45V across the self bias resistor = 302V B+.
302V B+ is only steady at 302V . . . when power mains voltage is steady at the right voltage, and there is the DCR of the power transformer primary and secondary, (and the variable impedance of a tube rectifier if you use that) so B+ varies versus load current.
Now, you have to try several 2A3 tubes until you get fairly close to 7V across DCR, with the 45V bias you are using.
We are cutting a raw diamond, and all diamonds are different.
Note: When they heat up, The DCR of the power transformer windings vary, and so do the output transformer's primary DCR.
Another thing that can affect the tube quiescent voltages and currents . . . is the filament voltage. If that varies, so does the rest of the triode parameters. Not all filament supplies are constant.
Temperature of the filament secondaries for example.
Do not forget to measure your power mains.
Mine vary from 117VAC rms, to 123VAC rms, on a true rms meter.
Last edited:
Why not use 10R resistor between the power stage "ground" and 2A3 (virtual) cathode?
Virtual cathode: 47R resistors from each leg of filament pins.
Can you measure on 10R resistor 600mV if anode current is 60mA.
The "real" anode-cathode voltage (for example 250V) can be measured between the anode and virtual cathode.
euro21,
That is a pretty good idea.
But let me modify that so that there is a little less effect on the 2A3's performance.
You have two 47 Ohm resistors, coming from the 2A3 filament ends, and the 47 Ohm resistors joining to the top of a 10 Ohm resistor, and from to ground.
47 / 2 = 23.5 Ohms from filament ends to the 10 Ohm resistor. 23.5 + 10 = 33.5 Ohms.
The 2A3 plate resistance, rp, is 800 Ohms. The 2A3 u = 4.2
Un-bypassed filament resistor(s) to ground Increases the plate resistance, rp.
33.5 Ohms x u = 140.7 Ohms.
The new plate resistance is 800 Ohms + 140.7 Ohms = 940.7 Ohms (instead of the 800 Ohm specification).
Improving on that . . .
A 10 Ohm resistor from each filament end, and 10 Ohm current sense resistor.
10 Ohms / 2 = 5 Ohms
5 Ohms + 10 Ohma = 15 Ohms.
15 Ohms x u = 15 x 4.2 = 63 Ohms.
The new plate resistance, rp is 800 + 63 = 863 Ohms.
863 is a little better than 940.7
The 10 Ohm resistors from the filament ends are effectively in series with each other, and are across 2.5V, that is 2.5V / 20 Ohms = 0.125 Amps.
The 2A3 Gm is 5250 uMhos.
1 / 5250 uMhos is 190.5 Ohms.
Compare the 190.5 Ohms, versus the 33.5 Ohms, or the 15 Ohms, that connects to ground.
15 Ohms is only 8% of the 190.5 Ohms, a little less loss of gain versus the 33.5 Ohms to ground.
That is a pretty good idea.
But let me modify that so that there is a little less effect on the 2A3's performance.
You have two 47 Ohm resistors, coming from the 2A3 filament ends, and the 47 Ohm resistors joining to the top of a 10 Ohm resistor, and from to ground.
47 / 2 = 23.5 Ohms from filament ends to the 10 Ohm resistor. 23.5 + 10 = 33.5 Ohms.
The 2A3 plate resistance, rp, is 800 Ohms. The 2A3 u = 4.2
Un-bypassed filament resistor(s) to ground Increases the plate resistance, rp.
33.5 Ohms x u = 140.7 Ohms.
The new plate resistance is 800 Ohms + 140.7 Ohms = 940.7 Ohms (instead of the 800 Ohm specification).
Improving on that . . .
A 10 Ohm resistor from each filament end, and 10 Ohm current sense resistor.
10 Ohms / 2 = 5 Ohms
5 Ohms + 10 Ohma = 15 Ohms.
15 Ohms x u = 15 x 4.2 = 63 Ohms.
The new plate resistance, rp is 800 + 63 = 863 Ohms.
863 is a little better than 940.7
The 10 Ohm resistors from the filament ends are effectively in series with each other, and are across 2.5V, that is 2.5V / 20 Ohms = 0.125 Amps.
The 2A3 Gm is 5250 uMhos.
1 / 5250 uMhos is 190.5 Ohms.
Compare the 190.5 Ohms, versus the 33.5 Ohms, or the 15 Ohms, that connects to ground.
15 Ohms is only 8% of the 190.5 Ohms, a little less loss of gain versus the 33.5 Ohms to ground.