With transformer coupled input, the input voltage needed to get to clipping is 6~6.5 V RMS. With capacitor coupled input, the voltage required will be 3 dB less - so roughly 4.5 V RMS. A CD player provides about 2 V RMS.
With my 87 dB/W*m efficient speakers, the amp can be driven to ear splitting SPLs in a 16 m^2 living room using a CD player. To get to clipping with a CD player input, you'll need a preamp with a gain of about 10 dB (3x). This with transformer coupled input.
With a cap coupled input, you need roughly 7 dB of gain to get to clipping.
~Tom
With my 87 dB/W*m efficient speakers, the amp can be driven to ear splitting SPLs in a 16 m^2 living room using a CD player. To get to clipping with a CD player input, you'll need a preamp with a gain of about 10 dB (3x). This with transformer coupled input.
With a cap coupled input, you need roughly 7 dB of gain to get to clipping.
~Tom
Last edited:
I am lost here. Why would you run the CF at 1/3 the Ia as the driver stage? Doesn't that completely defeat the purpose of the CF, or atleast most of it? CF does give you lower Zout, but its also suppose to be the driver of current into the load. If you bias at 1/3 the driver stage, you have only 1/3 of the current into the load to deliver. Usually not what most designers strive for.
Driving a 300B is quite tricky - hence, the need for a cathode follower in the first place. However, the 4.5 mA biasing the cathode follower in my design is plenty.
Actually, you want the current in the cathode follower to be low as this results in lower distortion of the overall amp. This may seem a bit counter-intuitive, but described quite well by Morgan Jones in "Valve Amplifiers", 4th ed. The THD of the cathode follower is quite low to start with but increased by the lower bias current. As the harmonics produced by the cathode follower are 180 degrees out of phase with those produced by the preceding common cathode amplifier stage, the harmonics end up canceling, thus, lowering the THD of the overall amplifier.
The final design was the result of much simulation to get close followed by experimentation to minimize the THD of the overall amplifier.
~Tom
300B Amplifier - Power Supply Schematic & BOM
Folks,
The Schematic & Bill of Materials .pdf for the power supply board is attached.
I have identified one minor issue in the bias supply. With the values shown in the schematic and my less than optimal prototype setup, the Q1 tends to oscillate. Currently, I have fixed this by adding a 10 nF, 50 V ceramic cap connected from the gate to source (pin 1 to pin 3) on Q1. A more permanent solution will probably be a ferrite bead placed on the gate lead of Q1. Something like a Bourns FB43-110 or FB73-110 will probably work.
Symptoms of the oscillation include the output voltage changing dramatically as the tubes in the amplifier warm up and the load on the bias regulator increases.
As stated, it is likely that this oscillation is caused by the two feet of wire I have connecting Q1 of the power supply board to the load. But I think in the interest of reliability, I'll recommend adding the ferrite bead. I'll need to verify the fix first, though.
Under proper working conditions, the output voltage of the bias supply will change a few 100s of mV from no load to full load (~10 mA).
~Tom
Folks,
The Schematic & Bill of Materials .pdf for the power supply board is attached.
I have identified one minor issue in the bias supply. With the values shown in the schematic and my less than optimal prototype setup, the Q1 tends to oscillate. Currently, I have fixed this by adding a 10 nF, 50 V ceramic cap connected from the gate to source (pin 1 to pin 3) on Q1. A more permanent solution will probably be a ferrite bead placed on the gate lead of Q1. Something like a Bourns FB43-110 or FB73-110 will probably work.
Symptoms of the oscillation include the output voltage changing dramatically as the tubes in the amplifier warm up and the load on the bias regulator increases.
As stated, it is likely that this oscillation is caused by the two feet of wire I have connecting Q1 of the power supply board to the load. But I think in the interest of reliability, I'll recommend adding the ferrite bead. I'll need to verify the fix first, though.
Under proper working conditions, the output voltage of the bias supply will change a few 100s of mV from no load to full load (~10 mA).
~Tom
Last edited:
Mostly classic rock (Dire Straits, Mark Knopfler, Pink Floyd, etc.) but also more vocal-ish music like Nora Jones, Brandi Carlile, and Celine Dion and jazz (Stan Getz). Every now and then, I'll spin a classical CD. Love Mozart's Requiem.
My default test albums include (in no particular order):
- Dire Straits, "Brothers in Arms"
- Dire Straits, "On Every Street"
- Celine Dion, "Falling Into You"
- Eric Clapton, "Unplugged"
- Mark Knopfler & Chet Atkins, "Neck & Neck"
- Pink Floyd, "The Dark Side of the Moon"
- Pink Floyd, "The Wall"
- Pink Floyd, "The Final Cut"
- Sting, "Mercury Falling"
- Simon Preston, "Bach Organ Works", specifically, Tocata & Fuge in D minor, BWV 565
- Stan Getz, "Cafe Montmartre"
A well designed amp will play all kinds of music well. I don't buy into the idea that you need an amp designed for a particular genre. Feel free to agree or disagree, but that's my philosophy.
~Tom
My default test albums include (in no particular order):
- Dire Straits, "Brothers in Arms"
- Dire Straits, "On Every Street"
- Celine Dion, "Falling Into You"
- Eric Clapton, "Unplugged"
- Mark Knopfler & Chet Atkins, "Neck & Neck"
- Pink Floyd, "The Dark Side of the Moon"
- Pink Floyd, "The Wall"
- Pink Floyd, "The Final Cut"
- Sting, "Mercury Falling"
- Simon Preston, "Bach Organ Works", specifically, Tocata & Fuge in D minor, BWV 565
- Stan Getz, "Cafe Montmartre"
A well designed amp will play all kinds of music well. I don't buy into the idea that you need an amp designed for a particular genre. Feel free to agree or disagree, but that's my philosophy.
~Tom
300B Amp - Driver Board BOM
Bill of materials for the 300B Amp Driver Board is attached.
Of the three options (6N6P, ECC99, and 12BH7A) the 6N6P and ECC99 options are well tested. I tried the 12BH7A as well (using two 6.2 V zeners in series for the cathode bias and 10 mA anode current) but ended up preferring the 6N6P. I didn't play as much with the 12BH7A, but I think I arrived at a pretty good operating point. If you prefer a more colored sound, I suggest trying the 12BH7A. I prefer a cleaner sound, hence, would choose the 6N6P (or ECC99).
Note that to change from 6N6P to ECC99 or 12BH7A, the options OPT1A,B and OPT2A,B need to change as follows:
- When using 6N6P, connect OPT1B with a wire jumper. Connect OPT2B with a wire jumper.
- When using ECC99 or 12BH7A, connect OPT1A with a wire jumper. Connect OPT2A with a wire jumper.
These options change the heater connections. The heater voltage is 6.3 V for all options. After assembly, you can verify that you've got the options set correctly by measuring the heater voltages at pins 4, 5, and 9. Consult with the data sheet for the tube of your choice to ensure that you've got it connected correctly.
~Tom
Bill of materials for the 300B Amp Driver Board is attached.
Of the three options (6N6P, ECC99, and 12BH7A) the 6N6P and ECC99 options are well tested. I tried the 12BH7A as well (using two 6.2 V zeners in series for the cathode bias and 10 mA anode current) but ended up preferring the 6N6P. I didn't play as much with the 12BH7A, but I think I arrived at a pretty good operating point. If you prefer a more colored sound, I suggest trying the 12BH7A. I prefer a cleaner sound, hence, would choose the 6N6P (or ECC99).
Note that to change from 6N6P to ECC99 or 12BH7A, the options OPT1A,B and OPT2A,B need to change as follows:
- When using 6N6P, connect OPT1B with a wire jumper. Connect OPT2B with a wire jumper.
- When using ECC99 or 12BH7A, connect OPT1A with a wire jumper. Connect OPT2A with a wire jumper.
These options change the heater connections. The heater voltage is 6.3 V for all options. After assembly, you can verify that you've got the options set correctly by measuring the heater voltages at pins 4, 5, and 9. Consult with the data sheet for the tube of your choice to ensure that you've got it connected correctly.
~Tom
Last edited:
I'm tempted to give this project a go because I still have a pair of 300BXLS and some 6N6P's. Any output transformer you favor for this amp? I'm trying to figure out the cost for the whole amp. For the moment I'm busy to see if all the parts are also available at Mouser. Hate Digi-Key, way to expensive (customs and s..t)
Regards
Regards
The parts are pretty generic and should be available world-wide. I would think Farnell in Europe would be able to help you out. The only tricky part is the LT3092 as it is available in many different packages. You want the SOT-223 version.
I've used the Edcor CXSE25-8-5k and Electra-Print 300B, 100 mA, 5 kOhm primary, 4 & 8 ohm secondary, copper wire with this amp. The Edcors are quite nice. At about $80/each they're incredible bang for the buck. I went with the Electra-Print transformers because I needed the dual secondary and wanted to try Jack's transformers. I am very pleased with the transformers I got. I like the tone and the detail in both the highs and lows. The Edcor is spec'ed to 50 Hz, the E-P 25 Hz. So you get an extra octave... But the E-P is also about 3x the cost of the Edcor. Tradeoffs, tradeoffs...
~Tom
I've used the Edcor CXSE25-8-5k and Electra-Print 300B, 100 mA, 5 kOhm primary, 4 & 8 ohm secondary, copper wire with this amp. The Edcors are quite nice. At about $80/each they're incredible bang for the buck. I went with the Electra-Print transformers because I needed the dual secondary and wanted to try Jack's transformers. I am very pleased with the transformers I got. I like the tone and the detail in both the highs and lows. The Edcor is spec'ed to 50 Hz, the E-P 25 Hz. So you get an extra octave... But the E-P is also about 3x the cost of the Edcor. Tradeoffs, tradeoffs...
~Tom
It's amazing what a 15-cent ferrite bead can do for you... I have ironed the last kink out of the power supply.
To ensure stability of the bias regulator, Q1, a Bourns FB43-100 ferrite bead should be added to the gate lead of Q1. That's Digikey P/N M8702-ND.
If a ferrite bead isn't available in your local market, use a 10 nF, 50 V ceramic capacitor from Q1 gate to Q1 source (pins 1 and 3).
I've updated the schematic and BOM to reflect this.
~Tom
To ensure stability of the bias regulator, Q1, a Bourns FB43-100 ferrite bead should be added to the gate lead of Q1. That's Digikey P/N M8702-ND.
If a ferrite bead isn't available in your local market, use a 10 nF, 50 V ceramic capacitor from Q1 gate to Q1 source (pins 1 and 3).
I've updated the schematic and BOM to reflect this.
~Tom
Tom, can you post some more pictures of your latest and greatest versions? I just checked the datasheets of the Bourns FB-43-100-RC ferrite bead and the Fairchild FQP3P50 transistor. It seems that the bead will only fit over the lower part of the transistors leg and I didn't think that part was exposed [for the bead to go over it]. I have a few other such build oriented questions on my mind that would probably be resolved by looking at pictures from other angles. No need to rush into action. I am busy with other stuff and can't start my build quite yet. Thinking about it nonetheless.
Q1 is in a TO-220 package and it's sitting on a little heat sink. I purposely designed the footprint for the Q1-heatsink combo such that you bend the leads on Q1 just after they go from wide to skinny (i.e. you bend the skinny part to form a 90 degree angle).
I measured a couple of TO-220 packages. When adding the thickness of the heat sink, there ends up being 4.6 mm from the bottom of the transistor lead to the top of the PCB. This is enough for the ferrite bead. And as the bend on the lead is on the skinny part, the ferrite drapes over just fine. The ferrite sits vertically between the PCB and the flat part of the TO-220 lead.
I hope this makes sense. I'll follow up with pictures in a few days. Thanks for the questions, by the way. It's important to me that people are able to assemble the amp successfully. Your questions will help me improve the overall quality of my project documentation. Thanks.
~Tom
I measured a couple of TO-220 packages. When adding the thickness of the heat sink, there ends up being 4.6 mm from the bottom of the transistor lead to the top of the PCB. This is enough for the ferrite bead. And as the bend on the lead is on the skinny part, the ferrite drapes over just fine. The ferrite sits vertically between the PCB and the flat part of the TO-220 lead.
I hope this makes sense. I'll follow up with pictures in a few days. Thanks for the questions, by the way. It's important to me that people are able to assemble the amp successfully. Your questions will help me improve the overall quality of my project documentation. Thanks.
~Tom
Last edited:
The ClassicTone 40-18069 power transformer has two 15 V windings, each rated for 2 A. Recall, I'm using my Universal Filament Regulators to power the 300B and 6N6P filaments. The Universal Filament Regulators are switching regulators and they are very efficient (~90 %). So Power In = 1.1 * Power Out. Give or take...
Hence, each 300B filament regulator only draws 1.4*5*1.1 = 7.7 W from the transformer. 7.7 * 2 = 15.4 W. The winding is rated for 15 * 2 = 30 W, so there's plenty of margin.
The worst case condition on the 6.3 V supply is when you're using ECC99s as they draw about 1.8 A total. So 6.3 * 1.8 * 1.1 = 12.5 W drawn from the transformer. Still... Pretty far below 30 W.
Note, that the RMS current drawn by the rectifier is quite a bit higher than that drawn by the load. This is due to the way full-wave rectifiers work. They draw the current in pulses so the peak current is quite high. I ran extensive simulations on all the supplies to ensure that the transformer windings were not overloaded. I have verified the validity of my sims in my prototype. The transformer is running at around 80 % of full capacity. It gets lukewarm (maybe 35~40 deg C). That's still well within the range of normal.
Of course, if you're using linear regulators, you won't be able to use the ClassicTone 40-18069 transformer -- at least not to supply the filament supplies. That would also result in some ungodly amount of power dissipated in the filament regulators. That's why I went with switching regulators in the first place. If you insist on using linear regulators, you're probably better off using a separate transformer for the filament supplies. Something like an 2 x 8 V, 50 VA tranny should work, but you'll have to do your own design there.
~Tom
Hence, each 300B filament regulator only draws 1.4*5*1.1 = 7.7 W from the transformer. 7.7 * 2 = 15.4 W. The winding is rated for 15 * 2 = 30 W, so there's plenty of margin.
The worst case condition on the 6.3 V supply is when you're using ECC99s as they draw about 1.8 A total. So 6.3 * 1.8 * 1.1 = 12.5 W drawn from the transformer. Still... Pretty far below 30 W.
Note, that the RMS current drawn by the rectifier is quite a bit higher than that drawn by the load. This is due to the way full-wave rectifiers work. They draw the current in pulses so the peak current is quite high. I ran extensive simulations on all the supplies to ensure that the transformer windings were not overloaded. I have verified the validity of my sims in my prototype. The transformer is running at around 80 % of full capacity. It gets lukewarm (maybe 35~40 deg C). That's still well within the range of normal.
Of course, if you're using linear regulators, you won't be able to use the ClassicTone 40-18069 transformer -- at least not to supply the filament supplies. That would also result in some ungodly amount of power dissipated in the filament regulators. That's why I went with switching regulators in the first place. If you insist on using linear regulators, you're probably better off using a separate transformer for the filament supplies. Something like an 2 x 8 V, 50 VA tranny should work, but you'll have to do your own design there.
~Tom
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.