Hi all!
After a rather long night I decided to put together a project I've been wanting to work on for a while: a hybrid tube headphone amp using a 832A tube. And.... it works!
I have an oscilloscope and a frequency generator to test it with, but I'm not sure how I would go about improving this amp, using those tools for testing, etc. Would someone know of a place I can get info on how to test a hybrid amp, and tuning it for best efficiency and sound?
Here is the schematic I made: http://i.imgur.com/AwjTAS0.png
Here it is driving a pair of 8Ohm speakers to pretty impressive levels: My Saturday Night Creation: A Vacuum Tube amplifier without the dangerous high voltages! - Imgur
Thanks!
Jarek
After a rather long night I decided to put together a project I've been wanting to work on for a while: a hybrid tube headphone amp using a 832A tube. And.... it works!
I have an oscilloscope and a frequency generator to test it with, but I'm not sure how I would go about improving this amp, using those tools for testing, etc. Would someone know of a place I can get info on how to test a hybrid amp, and tuning it for best efficiency and sound?
Here is the schematic I made: http://i.imgur.com/AwjTAS0.png
Here it is driving a pair of 8Ohm speakers to pretty impressive levels: My Saturday Night Creation: A Vacuum Tube amplifier without the dangerous high voltages! - Imgur
Thanks!
Jarek
Do the pentodes really share cathode resistor or is it just schematic error?
470uF output caps will remove some lower end frequencies for 8 ohm speakers but i guess this is not it's intended design and the speakers are too small to reproduce the bass anyway.
470uF output caps will remove some lower end frequencies for 8 ohm speakers but i guess this is not it's intended design and the speakers are too small to reproduce the bass anyway.
The Cathodes are tied internally inside the tube; They come out to a single pin, so I can't control them individually 🙁
I'm having some trouble selecting components which remove a minimum amount of frequency ranges. Do you know of a calculator or formula I can plug the expected impedence -and- RC filter parameters into to check which component values would only filter unwanted signals, or some design choices where filters become unnecessary?
I'm having some trouble selecting components which remove a minimum amount of frequency ranges. Do you know of a calculator or formula I can plug the expected impedence -and- RC filter parameters into to check which component values would only filter unwanted signals, or some design choices where filters become unnecessary?
Formula is simple, f = 1/(2*pi*R*C).
If you want response down to 20Hz for 8 ohm speakers, then the output cap should be at least 1,000 uF for a cutoff freq of 19.9 Hz. I would go with 2,200uF.
Any application store for your smartphone should have this RC filter calculator app available for download.
If you want response down to 20Hz for 8 ohm speakers, then the output cap should be at least 1,000 uF for a cutoff freq of 19.9 Hz. I would go with 2,200uF.
Any application store for your smartphone should have this RC filter calculator app available for download.
Well, that's a unique idea! I've never tried running an 832 tube on 12V B+...
How much current are you getting through the tube? And/or what DC voltage on the cathodes and plates?
I would use diodes or an LED for the cathode bias instead of a bypassed resistor - there will be less interaction between channels. Though with 1000uF bypass it's probably OK anyway.
For an 8 ohm load, 2200uF would be good, but if you're just driving headphones (32 ohms and up, usually) 470uF is plenty.
How much current are you getting through the tube? And/or what DC voltage on the cathodes and plates?
I would use diodes or an LED for the cathode bias instead of a bypassed resistor - there will be less interaction between channels. Though with 1000uF bypass it's probably OK anyway.
For an 8 ohm load, 2200uF would be good, but if you're just driving headphones (32 ohms and up, usually) 470uF is plenty.
Yup, I had that formula, which is how I came up with the current filter parameters, but I'm not clear on how the size of the capacitor correlates to the output impedance, like how to choose a huge 2200uF cap or a 470uF cap for 8ohm vs 32ohm load.
Kunaphil: VCC = 12VDC
Pmillet: My Bencthop power supply maxes out at 1A on its adjustable channel, which is what my ammeter is showing, so I'm gonna try to find a fixed 12V power supply that can handle a higher load. The datasheet says the heaters themselves consume 1.6A so I think I'll be good with a 2A?
It's only unique in that I replaced the 12AU7 tube in a popular hybrid design with this tube 🙂
What value diode/LED could I use for the cathode bias? I'm going to be doing a lot of experimentation with this amp to get ideal conditions, so I'd like to explore every possibility 🙂
6J7: Would that 1ohm (lots of watts?) resistor coming from each plate then be tied to the screen shared by both channels?
Are you asking about how a cap alters the filter frequency? Because to find the cap size, all you need to do is to plug in the numbers. Think of a cap as a resistor that reduces in value the higher the frequency that passes it. RC filter is simply a voltage divider. See C3 and R10 on your schematic. If you replace C3 with a resistor, they make a voltage divider right? The lower the freq that passes through C3, the lower the voltage output from that divider.
A green LED equals about 1.3 volts of voltage drop, just like a simple silicon diode will drop about 0.7 volts. So you need to measure the current voltage at the cathode and series enough LED to drop the same voltage value. Say that you measure 2.5 volts between ground and cathode (voltage on R1), then you only need to replace R1 with 2 green LEDs (2 x 1.3 =2.6 volts, close enough). Different color LEDs will have slightly different voltage drops.
No, what 6J7 meant was to have 1 ohm resistors for each of the plate (pin 8 and 9) before going to the 4K7 resistors (R5 and R6). Wattage doesn't have to be big because you'll only flow a few mA on that resistor. However, this is not really necessary because you can just simply measure the voltages on R5 and R6 to calculate its respective plate current (use ohm's law). Fixed bias means you connect the cathode directly to ground and control the plate currents by connecting some negative voltages on the grids. This isn't really necessary for your intended purpose because you want each plates to flow equal current value. So, cathode bias using LED or even bypassed-resistor like your current design is OK.
I understand the divider part, and I understand choosing a resistor to match the capacitor. I'm wondering why choosing, for example, a 1uF capacitor and getting a matching resistor to fit that calculation is not going to work if I have an 8ohm load. Basically, why was it recommended to go with a very high cap to begin with?
I measured the voltage across R1 to be 0.535V, guess that means I have to stick with a small diode, I'll search my part box for one. Do diode based solutions for cathode bias produce better sound, or are they just another alternative?
Gotcha, thanks 🙂
>>I'm wondering why choosing, for example, a 1uF capacitor and getting a matching resistor to fit that calculation is not going to work if I have an 8ohm load. Basically, why was it recommended to go with a very high cap to begin with?
Because your resistor is that 8 ohm load. Paralleling two resistors of very high value (say 100-1000ohm) and very low one (like 8ohm) will yield a resistor closer in value to the lower one.. so you have to calculate your cap value based on the assumption that R=8 ohm.
Because your resistor is that 8 ohm load. Paralleling two resistors of very high value (say 100-1000ohm) and very low one (like 8ohm) will yield a resistor closer in value to the lower one.. so you have to calculate your cap value based on the assumption that R=8 ohm.
Aaaaaaaahhhhhhhhhhh, now it makes sense. I have to think of the speaker's resistive load as a resistor parallel to the resistor to ground in the RC filter! Now I get it 🙂
Hmm something that doesn't add up for me here. Your .535V across the 2.7k cathode resistor gives ~100uA per plate less screen current which is most likely about 1uA or less. This equates to only .47V drop across the 4.7k plate resistors puting the plate at ~11.5V. Not sure how your getting voltage swing out that isn't completely distorted.
Ah, there was a mistake in the schematic. That resistor was only 270Ohm. That's what I get for making a schematic at 4AM 😛
Sorry for the double post, but I've got some strange behavior I don't understand:
I'm putting a 1KHz 0.5Vpp signal into the left channel, with quiet music on the right channel. There is a strange flat part on the bottom of the waveform. What could be causing that? I tried swapping out all my components, and tried biasing with a 1N4148 instead of a cap/resistor (which made my signal come out the right channel...), so I'm a bit stumped...
I'm putting a 1KHz 0.5Vpp signal into the left channel, with quiet music on the right channel. There is a strange flat part on the bottom of the waveform. What could be causing that? I tried swapping out all my components, and tried biasing with a 1N4148 instead of a cap/resistor (which made my signal come out the right channel...), so I'm a bit stumped...
Others with more experience may be able to pinpoint the cause of that clipping behavior by simply looking at your schematic. I can't but here are some things you can experiment with:
1. Reduce that 10M grid leak capacitor to 100-220K
2. Make the plate load resistor (4K7) adjustable. Replace with 10K pot and adjust as necessary
3. Make the cathode resistor adjustable. Replace with 1K pot and adjust as necessary.
Did you say your power supply can only handle 1A? Configured to run at 12V, the heater itself will consume about 800mA. Each LM317 is biased at 1.25V / 10R = 125mA, so both of them consume 250mA. This is already more than 1A.
1. Reduce that 10M grid leak capacitor to 100-220K
2. Make the plate load resistor (4K7) adjustable. Replace with 10K pot and adjust as necessary
3. Make the cathode resistor adjustable. Replace with 1K pot and adjust as necessary.
Did you say your power supply can only handle 1A? Configured to run at 12V, the heater itself will consume about 800mA. Each LM317 is biased at 1.25V / 10R = 125mA, so both of them consume 250mA. This is already more than 1A.
You would need a higher swing, that could be archieved elevating the cathode and anode resistors like ballpencil said to higher ones but this would reduce the standing current causing horrible distortions in the output, also there would be pretty huge current imbalance between chanels at so lowish voltages causing a chanel to distort earlier, you could try elevating the GU-32 supply voltage to at least 40 volts driving the output of the mosfet through a capacitor
Last edited:
OK, the 832A looks cool, but what's a RF high-voltage power-tube doing in a voltage amplifying circuit and even running at a low plate-supply way out of spec's?
I'm surprised you can make it work at all.
Signal-clipping however, isn't surprising given the conditions.
A question about the 832A and cousins if allowed:
The plate-pins, could they be used for conduction-cooling the plates when running this tube near or over max. spec's, i.e. Class AB2 ?
It wouldn't be too hard machining some alu- or copperfin's for this purpose.
rgds,
/tricomp
I'm surprised you can make it work at all.
Signal-clipping however, isn't surprising given the conditions.
A question about the 832A and cousins if allowed:
The plate-pins, could they be used for conduction-cooling the plates when running this tube near or over max. spec's, i.e. Class AB2 ?
It wouldn't be too hard machining some alu- or copperfin's for this purpose.
rgds,
/tricomp
- Status
- Not open for further replies.