EL34 Pentode in Push-Pull!

[ray_moth]

what schematic do youuse for your EL34 Class A Triode. Can I switch this to Class AB1 (16W) easily so I cna have a low power and high power option? I think this might be the way to go for my tweeter amp.

I know little of tweeter amps but I thought less power would be required than for bass or middle frequencies?

I don't use class A, only AB1. The circuit I use is an all-balanced one consisting of 6SL7 LTP splitter with 6AU6 (pentode) CCS in the tail; 6SN7 differential driver; EL34 triode-connected PP into 7k:8ohm p-p load (ideally, for full power, should probably be 5k p-p but 7k seems OK). The B+ is a little under 400v and bias is fixed. OP tubes run at about 55mA quiescent current. This is not class A and, I think, could not be biased to true class A with such a high B+.

If you want class A, you need much lower B+ (about 300v or less) and higher quiescent current (95mA or more). Might sound better with shared cathode bias provided by CCS, rather than fixed bias.

my understanding.

a) some of the stridency of teh Pentode will not heard as the woofer is rolled off.

b) Using the EL34 in Class A (Triode or Pentode) permits less stable grid voltages

c) since the speaker in question is only 2 woofers connected in parallel (no crossover) damping factor of a Pentode mode might not be so critical.

a) I agree.

b) I don't understand what you mean. With class A I would be inclined to use cathode bias, though. I would have easily enough B+ for this, since class A needs lower plate-cathode voltage (and higher current) anyway.

c) I disagree,In general, damping factor is more important at lower frequencies. Higher DF means tighter bass, lower DF sounds "boomy" but that depends also on your speakers. Lower efficiency speakers tend to need higher DF.

 

[navin]

I know little of tweeter amps but I thought less power would be required than for bass or middle frequencies?

I don't use class A, only AB1. ...If you want class A, you need much lower B+ (about 300v or less) and higher quiescent current (95mA or more). Might sound better with shared cathode bias provided by CCS, rather than fixed bias.

b) I don't understand what you mean. With class A I would be inclined to use cathode bias, though. I would have easily enough B+ for this, since class A needs lower plate-cathode voltage (and higher current) anyway.

c) I disagree,In general, damping factor is more important at lower frequencies. Higher DF means tighter bass, lower DF sounds "boomy" but that depends also on your speakers. Lower efficiency speakers tend to need higher DF.

Thank you for all your help.

I expect to make the woofer and tweeter amps almost the same - just call them tweeter and woofer amps to diffrentiate between them.

Oh yes I did forget that. Damping factor is very important in the bass. Thanks. My speakers are nothing unusual. 2 18cm woofers giving about 87-88db after baffle step. My tweeter is about 3 db more sensitive.

Also since you use Class AB1 you will get about double the power I am expecting. This leads me to another question (sorry if this seems stupid).

For the woofer amp (16W) one which is preferable EL 34 Triode Class AB1 PP or EL 34 Triode Class A PPP? I understand if you have not used EL 34 Class A PPP so may not be in a position to offer a comphrehenisve comparision betweeen these two but either would be preferable to EL34 Class A Pentode; right?

 

[ray_moth]

For the woofer amp (16W) one which is preferable EL 34 Triode Class AB1 PP or EL 34 Triode Class A PPP? I understand if you have not used EL 34 Class A PPP so may not be in a position to offer a comphrehenisve comparision betweeen these two but either would be preferable to EL34 Class A Pentode; right?

You'd probably get slightly less power from triode class A PPP than from triode class AB1. There are other important factors to consider, however.

A problem with class A is that the tubes need to be run close to their dissipation limit, in order to get full power. In fact, when delivering full power, the OP tubes will actually run cooler than when at idle. This is because, at idle, all of the energy consumed from the power supply is dissipated as heat in the EL34 tubes. As you turn up the volume, a progressively higher proportion of it is converted to mechanical energy in the speakers. This makes sense, since in class A the average plate current is virtually constant regardless of the output power being delivered to the speakers.

PPP class A will run hotter still, since there will be twice as many output tubes. It will also cost more to build, for that reason, and will require an OP transformer that can tolerate a high primary current.

In triode class AB1, the EL34s are biased closer to cutoff and, in consequence, run much cooler until more output power is required, when they start to take more current (and get hotter). The extent of this change and th epoint at which it occurs depends on the bias point. So, in general, a class AB1 amp runs cooler than a class A amp.

Pentode mode is not bad and some people would probably recommend it for the woofer amplifier, anyway, because it will need considerably more power than the tweeter amp. If you use enough NFB, you can get a good damping factor (UL mode would need less NFB to get the same DF.) Pentodes are reputed to have an 'authority' that triodes lack and I think pentodes (or UL) might do a very good job for a bass amp for rock music, for instance. Triodes tend to be sweeter in the higher frequencies, where pentodes can sound harsher.

To get a reasonable balance of power, then, your best solution could be a pentode (or UL) woofer amp and a triode tweeter amp. I would think that class AB1 would be best for both.

 

[navin]

...To get a reasonable balance of power, then, your best solution could be a pentode (or UL) woofer amp and a triode tweeter amp. I would think that class AB1 would be best for both.

I read you post twice just to see if there was some part i could disagree with 🙂 just kidding.

I thank you a lot for this fantastic advice.

Pentode EL34 PP AB1 for the woofer will give me about 30W and Triode EL34 PP AB1 for the tweeter will give me 15W. More than what I expected.

Now the difficult part. ou described your schematic in an earlier post (for the PP AB1 Triode) does a diagram of this exist somewhee on the web? So far the one ckt that looks interesting is the one by Allen Heath called the PP-1C from this site (http://www.vacuumstate.com/). I hear that a PP-2C is also in the works. Dave (Planet 10) posted it on another thread.

Now to get a Preamp and Crossover design! 🙂

 

[anatech]

Hi Ray,
Most of what you say I agree with, but I have one amp that is UL where the highs are incredibly smooth. Better than anything. I will bet the transformer has something to do with this.

Anyhow, it kills any triode unit I have so far. Could be I used polystyrene caps where possible, ant the feedback compensation. That would mean the circuit has a lot to do with it.

-Chris

 

[ray_moth]

Schematic attached, e3xcl. power supply. I hope it's readable but, anyway, I'll be pleased to answer any questions.

 

[ray_moth]

Hi Chris,

Sure, I agree, the quality of the OP trannies has everything to do with it. If those are mediocre, then it's pretty hard to achieve good sound!

Other components are important too. In my circuit, I believe the PSU caps probably don't affect the sound because all stages are balanced so there should be little or no signal going through the electrolytics. However, coupling caps can make a difference.

 

[anatech]

Hi Ray,
That looks like an interesting circuit. I might build it with a SS current sink. I see you are partially cathode bias and fixed bias. Best of both worlds?

Also, I see -15V for the grid bias circuit? I can't see the value clearly.

-Chris

 

[ray_moth]

I see you are partially cathode bias and fixed bias. Best of both worlds? Also, I see -15V for the grid bias circuit? I can't see the value clearly.

It's fully fixed bias. The 10 ohm resistors in the EL34 cathodes are just there to measure the quiescent current (50mA = 0.5v across the resistor).

I was afraid it might not be legible. The main negative rail voltage is -115, a bit less for the CCS because there's some decoupling.

 

[navin]

Hi Ray,
That looks like an interesting circuit. I might build it with a SS current sink.

SS sink is also what i have been considering. only my design skills are not capable of this so if you do come up with a ckt with SS current sink please feel free to send me a copy! 🙂

I was afraid it might not be legible. The main negative rail voltage is -115, a bit less for the CCS because there's some decoupling.

It was quite legible. the voltage rails look like -110V on the 6AU6, -350V, -115V and -500V on the OPT. Are R13 and R15 10 ohms? what wattage?

 

[pengboon]

Some questions...

Hi Ray_mote, i like your (or TL's idea) of the feedback done in this way! 🙂 It seems like a really nice amp to build!

The -350 and -500 should be a +350 and +500? And at the signal, max 500mV? How can the sensitivity be reduced, when say I have a preamp that already has an output of 3.5V (from CDP of 350mV)? IS your first resistor after the signal a 220K?

Is there also a way to implement adjustable feedback? I guess not that easy because the feedback loop is also the B+ for your 6SL7s right?

Sorry for the noob question, thanks for helping! 🙂

 

[navin]

Re: Some questions...

Is there also a way to implement adjustable feedback? I guess not that easy because the feedback loop is also the B+ for your 6SL7s right? )

the feedback resistor controlls the ammount of feedback so i assume bu adjusting that one can control the ammount of feedback.

 

[pengboon]

Re: Some questions...

Hi Ray_mote, i like your (or TL's idea) of the feedback done in this way! 🙂 It seems like a really nice amp to build!

The -350 and -500 should be a +350 and +500? And at the signal, max 500mV? How can the sensitivity be reduced, when say I have a preamp that already has an output of 3.5V (from CDP of 350mV)? IS your first resistor after the signal a 220K?

Is there also a way to implement adjustable feedback? I guess not that easy because the feedback loop is also the B+ for your 6SL7s right?

Sorry for the noob question, thanks for helping! 🙂

Sorry for the typo. Should be Ray_moth. I could not spot the typo so late at night (or should I say early in the morning)...

 

[ray_moth]

The -350 and -500 should be a +350 and +500? And at the signal, max 500mV? How can the sensitivity be reduced, when say I have a preamp that already has an output of 3.5V (from CDP of 350mV)? IS your first resistor after the signal a 220K?

You're right about the voltages, of c ourse - problem of legibility. Sensitivity can be reduced by increasing the value of the resistor between the cathodes of the 6SL7.

Also, as you say, the input grid resistor is 220k and could be made lower but should not be increased, because the Miller capacitance of the 6SL7 will cause treble roll-off. Most people would probably say it should be 100k or less; however, I have a 250k log volume control before it, so I'm limited in the value I can use here. I wouldn't recommend a volume pot without a "back-up" grid resistor, because if the pot wiper doesn't always make perfect contact, it will upset the bias and cause noise.

Is there also a way to implement adjustable feedback? I guess not that easy because the feedback loop is also the B+ for your 6SL7s right?

Not really. the main plate load of each 6SL7 triode is a 120k resistor. You could reduce the 2.2Meg cross-coupled NFB resistors to, say, 1Meg without causing instability or upsetting the DC conditions too much. I used to use that value but then decided to reduce the NFB until it gave just enough damping for my speakers.

 

[navin]

Not really. the main plate load of each 6SL7 triode is a 120k resistor. You could reduce the 2.2Meg cross-coupled NFB resistors to, say, 1Meg without causing instability or upsetting the DC conditions too much. I used to use that value but then decided to reduce the NFB until it gave just enough damping for my speakers.

What if R38/39 (2.2M) were replaced by a 1Meg resistor and 1M pot in series? Wont that make NFB variable?

 

[pengboon]

You're right about the voltages, of c ourse - problem of legibility. Sensitivity can be reduced by increasing the value of the resistor between the cathodes of the 6SL7.

Also, as you say, the input grid resistor is 220k and could be made lower but should not be increased, because the Miller capacitance of the 6SL7 will cause treble roll-off. Most people would probably say it should be 100k or less; however, I have a 250k log volume control before it, so I'm limited in the value I can use here. I wouldn't recommend a volume pot without a "back-up" grid resistor, because if the pot wiper doesn't always make perfect contact, it will upset the bias and cause noise.

Not really. the main plate load of each 6SL7 triode is a 120k resistor. You could reduce the 2.2Meg cross-coupled NFB resistors to, say, 1Meg without causing instability or upsetting the DC conditions too much. I used to use that value but then decided to reduce the NFB until it gave just enough damping for my speakers.

Thanks Ray_moth.

Can you explain the purpose of the one resistor between the cathodes of the 6SL7 (it looks like R30 4.7M?)

How about if a 100k log vol pot was used? I guess a 470k would do the trick nicely?

I noticed your grid stoppers for the 6SL7 and 6SN7 are 10k and 2.2k? Any reason why since I see most folks use 1k or less?

I also tried (but could not find) any info about the anode to anode resistance of OPT of EL34 in triode mode, any info on this? I only saw Lundahl's paper which states that it should be higher than the 6.6k of a UL connection.

I guess having a good quality pot in the FB loop enables some adjutability. In this subject on resistor quality, I guess all the signal path components should be of a high quality? i.e. R25,26,34,35,16,19,7,8,38,39 and C19,4,5,2,3.

How does the amp sound like? Any improvements down the road like using CCS biasing (if it improves at all...)? What kind of PT should I be looking at if I wanted to build this?

There are a couple of 6SL7-6SN7-EL34 schematics, but yours seems the most interesting, which tubes are you using?

thanks for helping this noob! 🙂

 

[ray_moth]

Can you explain the purpose of the one resistor between the cathodes of the 6SL7 (it looks like R30 4.7M?)

It's a 4.7k resistor (legibility problem again!) It moderates the effect of the individual 1.5k cathode resistors and is a convenient way to control gain. Originally, I didn't use it but that resulted in too little gain. Rather than reduce both of the 1.5k resistors, I used this parallel resistor.

How about if a 100k log vol pot was used? I guess a 470k would do the trick nicely?

Yes, that would be fine. The only reason I used a 250k log pot was because I didn't have anything smaller.

I noticed your grid stoppers for the 6SL7 and 6SN7 are 10k and 2.2k? Any reason why since I see most folks use 1k or less?

I didn't have any scientific reason for my choice of stoppers. I just used values I'd seen in other circuits. I notice that a stopper of 10k or bigger is quite common for the input stage. (I must admit, I don't really know why!)

I also tried (but could not find) any info about the anode to anode resistance of OPT of EL34 in triode mode, any info on this? I only saw Lundahl's paper which states that it should be higher than the 6.6k of a UL connection.

Mullard suggests 5k P-P for EL34 PP triode mode. I use 7k p-p, because that's what I had available. The OPT was actually intended for EL34 pentode PP, @3.5k p-p, but I have my 8 ohm speakers on the 4 ohm tap, giving me 7k. That is OK for triode and results in slightly less power and less distortion. 5k p-p would be better.

I guess having a good quality pot in the FB loop enables some adjutability. In this subject on resistor quality, I guess all the signal path components should be of a high quality? i.e. R25,26,34,35,16,19,7,8,38,39 and C19,4,5,2,3.

I'd be careful of putting pots in the NFB loops - a voltage drop of 150v is involved. A better idea to make NFB variable might be to use a 10k pot between the cathodes of the 6SL7 in each channel. The quality (and physical size) of resistors is very important. I like metal film in the signal path adn wirewound in the power supply. If you can, use minimum 1 watt resistors everywhere, except in places where more than 0.5w dissipation is called for. Resistor rating should be at least 2x the expected dissipation

How does the amp sound like? Any improvements down the road like using CCS biasing (if it improves at all...)? What kind of PT should I be looking at if I wanted to build this?

It sounds good to me, with absolutely no audible noise. I have Yamaha 3-way speakers, with nominal 89dB sensitivity. Better and more efficient speakers night show up some noise, for all I know. However, this all-balanced design provides good isolation from power supply hum/noise.

My PT is just about big enough tol supply the two channels. It's an Amplimo toroidal design, giving 340v @700mA, 6.3v @6.8A and 50v @100mA. I use a voltage doubler to get the required -106v neg. supply.

I use a supplementary filament transformer (Amplimo toroid again) to get sufficient current for all the heaters, which includes 3.8A for a pair of paralleled 6D22S TV damper diodes acting as a B+ slow-start device (in the negative line, which avoids nasty high voltages on the top caps). My B+ rectifier is actually an SS bridge of 4 x MUR4100E ultra-fast recovery diodes. B- uses 2 x MUR4100E in a voltage-doubler with 220uF caps.

There are a couple of 6SL7-6SN7-EL34 schematics, but yours seems the most interesting, which tubes are you using?

I like it better than other designs I've seen, too. I'm using Sovtek 6SL7s (reputed to be good), Electroharmonix 6SN7s (reputed to be very good) and a pair of Sovtek EL34s plus a pair of Svetlana EL34s (not "Winged C", unfortunately) , which are just 'OK'. (If I could afford to spend more on better tubes, especially the EL34s, I would.)

thanks for helping this noob!

Glad to be able to help.

Regards, Ray

 

[pengboon]

Mullard suggests 5k P-P for EL34 PP triode mode. I use 7k p-p, because that's what I had available. The OPT was actually intended for EL34 pentode PP, @3.5k p-p, but I have my 8 ohm speakers on the 4 ohm tap, giving me 7k. That is OK for triode and results in slightly less power and less distortion. 5k p-p would be better.

Claus Byrth's paper mentioned that EL34 in triode would require 8-10k, quite different from Mullard... 😕

My PT is just about big enough tol supply the two channels. It's an Amplimo toroidal design, giving 340v @700mA, 6.3v @6.8A and 50v @100mA. I use a voltage doubler to get the required -106v neg. supply.

Then how did you get -115V? 😕

I use a supplementary filament transformer (Amplimo toroid again) to get sufficient current for all the heaters, which includes 3.8A for a pair of paralleled 6D22S TV damper diodes acting as a B+ slow-start device (in the negative line, which avoids nasty high voltages on the top caps). My B+ rectifier is actually an SS bridge of 4 x MUR4100E ultra-fast recovery diodes. B- uses 2 x MUR4100E in a voltage-doubler with 220uF caps.

I can do the trick by manually switching on the B+ with another switch 1 minute or 2 after the filaments right?

Glad to be able to help.

Thanks again Ray, you've been most helpful! 😀

 

[pengboon]

My PT is just about big enough tol supply the two channels. It's an Amplimo toroidal design, giving 340v @700mA, 6.3v @6.8A and 50v @100mA. I use a voltage doubler to get the required -106v neg. supply.

I use a supplementary filament transformer (Amplimo toroid again) to get sufficient current for all the heaters, which includes 3.8A for a pair of paralleled 6D22S TV damper diodes acting as a B+ slow-start device (in the negative line, which avoids nasty high voltages on the top caps). My B+ rectifier is actually an SS bridge of 4 x MUR4100E ultra-fast recovery diodes. B- uses 2 x MUR4100E in a voltage-doubler with 220uF caps.

Hi Ray_Moth,

Sorry to bug you again. For the filament supplies, I am guessing that you use the main PT for the EL34s, and the supplementary one for the 6SL7 and 6SN7 right? Where do you connect the ground of the supply to?

Another question is how much current are your sinking from the 6SL7 and 6SN7? A SS CCS would do just fine here right? How about the cathodes of the 6SN7, would it be better to have CCS here too or not necessary?

thanks again!

 

[ray_moth]

You guess correctly. Ground is connected to a centre tap, created from a pair of 1k resistors, in the case of the 2 windings used for EL34s and for 6SN7/6SL7s. The third winding, used for 6D22Ss and 6AU6s, is connected to the negative -105v rail because the 6D22S can tolerate up to 600v heater to cathode but the 6AU6 can only tolerate 100v.

6SN7s are drawing 3.2mA per triode, 6SL7s 0.9mA per triode. The cathode current of each 6AU6 is about 3mA, because of the screen current.

SS would be fine for the CCS. You could then use a much lower negative rail voltage, of course. I'd be inclined to use a MOSFET, e.g. IRF820.

I don't think a CCS for the 6SN7 would give any real improvement but it wouldn't hurt to try it.