• WARNING: Tube/Valve amplifiers use potentially LETHAL HIGH VOLTAGES.
    Building, troubleshooting and testing of these amplifiers should only be
    performed by someone who is thoroughly familiar with
    the safety precautions around high voltages.

Single Ended EL34 Amp

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
I was sitting in my quiet office the other day, when it dawned on me that a) it was very quiet because I don't have a stereo in there, and b) I have almost all of the parts to build a stereo which would rectify the situation.

So, I drew up a schematic using stuff I have around, and am wondering if there are any comments on changes I might consider. Basically, it is a single ended EL34 amp (though a pair of EL34's is the one thing I don't have, so that could change) with a 6SN7 as the driver. The 6SN7 has a CCS plate load, and I just put both sides in parallel for simplicity. The schematic is quite close to Tubelab's SimpleSE, but at a lower voltage and current. This is really just because the transformer and chokes I have are all rated at 150mA, and the caps are all 500V, so I didn't want to push it too far.

The tubes are biased via LEDs. I just drew one on the schematic for simplicity, but I would expect 2 or 3 for the 6SN7, and perhaps a 3x9 grid for the EL34.

I figure about 270V on the plate of the EL34, and the grid biased up about 18V. This gives 60mA of current. The OPTs are spec'd at 70mA, so this seemed a decent conservative operating point.

I also have other things around I could use as driver tubes (6J5, 6n1p, 6dj8, 6n6p, 6h30, 5687) but the 6SN7 seemed promising and aren't being used anywhere else.

For speakers, I am not sure what I'll use yet, but probably I'll just pick up some small boxes at parts express and a reasonably efficient full range driver from madisound. I have not thought that far in advance though.

Comments?
 

Attachments

  • el34.png
    el34.png
    60.9 KB · Views: 2,381
Well, you already know that I have built a few of these. I couldn't get enough gain from a 6SN7 to get to full output with a CD player input. A 6SL7 worked fine, but I wanted to use a tube that I had lots of, so I settled on the 12AT7. I have since discovered a few hundred 6SL7's in a big box.

I would loose the cap from the B+ side of the CCS to ground. The CCS provides plenty of isolation for B+ noise. For a detailed explanation read this thread:

http://www.diyaudio.com/forums/showthread.php?s=&threadid=94956

In fact with your B+ voltage you may not even need the resistor. I added it when I started approaching 500 volts. The CCS got too HOT!
 
I would be very very happy to help you use those pesky 6SL7's before they go to waste in the salt air.

Before everyone gets excited, all of the tubes that I got a few years ago were very poorly treated over the years. They are all loose unboxed tubes, and were stored for at least 10 years in a warehouse that had holes in the roof, in Central Florida. They were exposed to bird, and rat "substance" and a lot of rain. I have been going through boxes one at a time, over the past 4 years. After cleaning the "stuff" off, and throwing away the broken, or severely corroded tubes. I am yielding about 20 to 40% testable tubes. Most of the miniature tubes work. I have found issues with octals because the "stuff" has worked its way in between the base and the glass. This makes small signal tubes noisy, but can make flames on an output tube! Many have been painted over with some type of military clear lacquer. I have seen the same stuff on vintage military PC boards. It glows green under UV light.

Back on thread, I have experimented with LED bias on my SimpleSE. The input tube can be LED biased fairly easilly. I put Some pin sockets in place of the cathode cap (and left the resistor out). I then tested a bunch of LED's. I look for the right plate voltage, and minimum signal across the LED with a scope. The best ones came out of an old bag of LED's.

I tried a series - parallel string of LED's for cathode bias in the KT88 output tube (like the Red Light District amp). I had over 100 LEDs wired together and connected into one channel (I didn't have enough LED's) When I cranked the amp into clipping a few LED's got dim, then the output tube started glowing, followed by a few more fried LED's. I turned it off before things got ugly. I don't think that I had enough paralleled strings, and these were LED's that I have had for at least 15 years. They were not all glowing at the same brightness when I started the test. I think that you are going to need a lot of LED's, and they need to be somewhat matched. At least each string should drop the same amount of voltage.

I tried a 39 volt 5 watt zener diode for bias. It was a waste of $2. Sounded terrible.
 
tubelab.com said:
I tried a series - parallel string of LED's for cathode bias in the KT88 output tube (like the Red Light District amp). I had over 100 LEDs wired together and connected into one channel (I didn't have enough LED's) When I cranked the amp into clipping a few LED's got dim, then the output tube started glowing, followed by a few more fried LED's. I turned it off before things got ugly. I don't think that I had enough paralleled strings, and these were LED's that I have had for at least 15 years.

I have done 2 parallel strings of 3 before without issue. I figured here I'd need 9 in series to get the bias to 18V, and probably 4 in parallel for the 60mA. A few less than you used, but still a lot of little parts to solder together.

The chance of this amp ever going really loud, or into hard clipping, is pretty low, though. I tend to run anything I build pretty conservatively.



kevinkr said:
I'd recommend fixed bias in this application. Take a look at my pages for the 45 SE amplifier - driver circuit would suit the EL34 perfectly.

Interesting. Is C3 in the signal path? If it is, what is the advantage over cathode bias, and if not, why a Blackgate?

Part of the reason for the 6SN7 as opposed to a 6SL7 is that with both sides in parallel I can run the current high enough for LED bias. My understanding is that it needs something over 10mA for this to sound good. Otherwise, I'd use the 6J5's I have.
 
Administrator
Joined 2004
Paid Member
Hi Doug,
Actually C3 is not in the signal path, I just happened to have a blackgate that was pressed into service for that application. Any good electrolytic will do.

The advantage of fixed bias is that it eliminates any component in the cathode circuit to ground whether it be leds (ok) or large high voltage electrolytic caps or film caps (not so ok and quite audible.)

As several members have pointed out large arrays of leds can be difficult to deal with when significant voltages and currents are involved. Personally I doubt that a huge array of leds is going to outperform conventional fixed bias sonically in any case. In addition if you make the bias adjustable, and make provisions to measure the cathode current you can tweak the bias for the best sound by your definition. When I designed the cited 45SE amplifier I quickly found that the operating point that yielded the lowest distortion did not seem to sound the best, and I ended up at a significantly different point by the end of it all.

The advantage of the 6SL7 in this application is if you need voltage gain you have it - this can be advantageous if you want to use it with a passive line stage or even as an integrated. The 6SN7 will not produce full output power with older components that adhere to the old line level standard of 700mVrms.

One suggestion I would make would be to use a power transformer with a 50V bias tap, and derive your bias that way. (My scheme only works well with very robust film and foil caps - this circuit gave me a lot of trouble initially.)
 
I quickly found that the operating point that yielded the lowest distortion did not seem to sound the best, and I ended up at a significantly different point by the end of it all.

Kevin:

If you have access to an FFT analyzer and you have the time to experiment, see if the "best sound" happens to correlate with minimum 3rd harmonic distortion, not minimum THD. I have found a close correlation in several of my SE triode amps (DHT and IDHT) including a 45 amp. This does not always seem to be the case with SE UL or any P-P amp though.

I have tested a few amps and about a dozen diferent OPT's and I found a unusual discovery. I started with a low bias current and slowly increased it while watching the FFT analyzer. There is a point where the total harmonic distortion dips, then as the current is further increased the second harmonic rises and the third drops. There is a point where the third almost disapears and then rises as the current is further increased. This null in the 3H is very near the point where I tuned for best sound. The interesting finding is that this null in 3H is unique to each OPT and almost independent on the type of tube used! Some OPT's work best at a current level above the spec for some tubes. The $18 Edcors will show almost no 3H at 80mA. That leaves out 45's and a lot of other tubes. I don't know who would spring for 45's and then use $18 transformers though.

dsavitsk

I figured here I'd need 9 in series to get the bias to 18V, and probably 4 in parallel for the 60mA. A few less than you used, but still a lot of little parts to solder together.

I am currently running Electro Harmonix KT88EH's with 465 volts of B+ (about 400 across the tube) at about 90 mA. This takes about 39 volts of bias. I had 6 strings of 18 LED's each, temporarily tack soldered together. They didn't live long. I now use a 560 ohm 5 watt resistor, and a 1500 uF Electrolytic cap. The cap is switchable to ground, or the 8 ohm output tap. This gives 0 or 5 db of cathode feedback.

This was concieved as a beginners level amp that could be simply built and used. After selling a few dozen TubelabSE amps it became obvious that a lot of people have problems (or fear) with setting the bias, and remembering to check it when changing tubes. This is why I used cathode bias. The design can be built on a strict budget, or taken to the max at the builders discression.

This amp or its predecessor, both built with top quality components and all of the optional power supply parts (choke and motor RUN cap) has been auditioned by several people on several speaker systems, including some real expensive Lowther driven horns. Only positive comments so far.

I have built 5 of these amps, the latest uses Transcendar transformers (triode only). It is the best sounding (and looking) one so far. I have used several different output tubes including about 5 kinds of EL34's. Right now I like the KT88EH, I may change my mind next week though.
 

Attachments

  • glowingamp4lores_b.jpg
    glowingamp4lores_b.jpg
    73.2 KB · Views: 871
A followup question on fixed bias: would it work to use a couple of resistors from B+ to ground to make a voltage divider and use that to bias the grid? B+, with the two chokes, should be very quiet so it would not inject any noise. What am I missing?

As to the bias tap, I can get one of those Allied power transformers for $40 or so, but anything that is going to add a bias tap is going to be a lot more I think. I am trying to keep this is dirt cheap as I can, which, excepting parts I already have, is looking to be about $100. Any more than that, and I'll have to justify the project to myself, which, frankly I can't do. I've got a closet full of solid state amps that I simply choose not to use :)

-d
 
A followup question on fixed bias: would it work to use a couple of resistors from B+ to ground to make a voltage divider and use that to bias the grid? B+, with the two chokes, should be very quiet so it would not inject any noise. What am I missing?

you are missing the MINUS sign. Grid bias voltage must be negative. It is possible to use a second rectifier on your power transformer to generate a negative voltage, then use a resistive divider to reduce the voltage.

Look at the power supply for my TubelabSE amp. It uses the $37 Allied transformer (6K56VG) for power, generating about + 320 volts with a tube rectifier and - 200 with the diodes. This combination allows the bias to come up instantly, and the B+ after the 5AR4 warms up. I use the 6K7VG for my SimpleSE amps, but I get 430 to 450 volts with these guys. This puts 380 to 400 volts across the output tube with cathode bias (and voltage drop in the OPT).

http://www.tubelab.com/TubelabSE.htm
 
dsavitsk said:
A followup question on fixed bias: would it work to use a couple of resistors from B+ to ground to make a voltage divider and use that to bias the grid? B+, with the two chokes, should be very quiet so it would not inject any noise. What am I missing?

As to the bias tap, I can get one of those Allied power transformers for $40 or so, but anything that is going to add a bias tap is going to be a lot more I think. I am trying to keep this is dirt cheap as I can, which, excepting parts I already have, is looking to be about $100. Any more than that, and I'll have to justify the project to myself, which, frankly I can't do. I've got a closet full of solid state amps that I simply choose not to use :)

-d

You can make a bias supply very cheaply. Just take a small transformer (you need very little current), say 6-24V or so. Connect the secondary across your filament supply (backwards, if you will). If you have a 12V transformer and connect it to a 12V filament supply, you'll have 110V on the output. Rectify it, then use a string of LC's to drop the voltage to where you want it. Several LC's will give you a very quiet output. Got this idea from a poster here (forgot, but I think I mentioned it in my low power SE amp thread).

Tubelab (or one of you other clever folk), why not a current sensing circuit driving a servo type regulated voltage to have a self setting bias supply? Someone must have done this before.

Sheldon
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.