• 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.

EL84 push pull idea

mark_titano

Member
2004-10-19 10:10 pm
Hi guys.
I'm trying to design a DC coupled el84 push pull.
I started from this:

http://www.wdehaan.demon.nl/mono/mono/el84dc.html

I would like to use a Gary Pimm's self bias ccs in the long tailed pair and a fed shunt regulator in the power supply ( to feed the long tailed pair). I've changed the working point of the driver stage, just to have much more current through the ecc88s. I think that, even if the power stage work in class AB, the grid current should not influence the driver stage in this way, however not so much like in the way it could if the current in every 88 would be only 1.8mA... At the start up, the self bias ccs doesn't start immediately to work. I think that I must delay the power for the el84s (or during power up the EL84's will have +255 volts applied to the grid they will be turned on hard).

I found a circuit that maybe should work here:
http://www.audiodesignguide.com/my/new300bxls2004/new300bxls.html

The relay datasheet:
http://chdist35.distrelec.com/distr...A1B25F9C1256DC700384C89/$FILE/~7559109.pdf

What do you think about this solution?

Some other thoughts: I'd like to replace the two res. on the cathode of the el84 with two current sources. In this manner the idle current will be set and stable even if the tubes are not matched. Changing the working point of the driver stage I'll be able to have a fine tunig of the bias. But what kind of ccs? Not a self biased ( too complicated for this purpose...the ccs will work only in dc)...not a sigle lm317( too high voltage).
Maybe a single mj340 or a cascode...

I'm thinking to try a different tube in the driver stage...maybe the 6h30.

Any comment or advice will be welcome :)

Mark
 

Attachments

  • 84 push pull.gif
    84 push pull.gif
    13.8 KB · Views: 843
mark_titano said:
Some other thoughts: I'd like to replace the two res. on the cathode of the el84 with two current sources. In this manner the idle current will be set and stable even if the tubes are not matched. Changing the working point of the driver stage I'll be able to have a fine tunig of the bias. But what kind of ccs? Not a self biased ( too complicated for this purpose...the ccs will work only in dc)...not a sigle lm317( too high voltage).
Maybe a single mj340 or a cascode...

Mark

An uprated version of the one shown here?: http://www.diyaudio.com/forums/showthread.php?postid=659393#post659393

Sheldon
 

rdf

Member
2004-06-21 8:04 am
big smoke
Will bias stability be an issue with direct coupling? The EL84's are right up against the 12 watt (305 volts X ~38 ma) limit and slightly past the 300 volt max. I've only had the opportunity to play with modern production versions but my favourite is one:

http://www.jj-electronic.sk/tube_el84.htm

One third the measured distortion in-circuit of an Electoharmonix and far more robust.
 

mark_titano

Member
2004-10-19 10:10 pm
Hi rdf.

I think that the stability is not an issue when heaters, ccs and everything else are stable. The problems is the start up. Maybe I'm wrong but in the el84EH datasheet the maximum power is 14W. But probably you are right ...what do you suggest as good working point?
 

rdf

Member
2004-06-21 8:04 am
big smoke
That's interesting. The JJ sheet specs max dissipation at 12 watts, as does the 1966 RCA Receiving Tube manual.

My amp isn't directly comparable to yours. It's a direct coupled pentode-mode SE which uses the EL84's cathode voltage as feedback to the EF86 driver screen for DC stability. The outputs run ~255 VDC plate-to-cathode and around 40 ma cathode current, or just over 10 watts. The latter was chosen by monitoring THD and intermod while varying the cathode resistor. THD rose and intermod fell with decreasing standing current. I eyeballed a compromise between the two and haven't been tempted to revisit the choice, though I might in the future. For paranoia's sake the outputs are dressed in Pearl Coolers.

Simulations showed this circuit too has the potential to punish the '84s on cold start. Since the supply has B+ to burn I employed a 5U4G rectifier to slow things down. The bonus is a hypermegacool glowing bottle up front to create the impression of a really powerful 3 watts.
 

kevinkr

Administrator
Paid Member
You might consider using a B+ delay relay or a slow warm up tube rectifier to allow the filaments in the ECC88 and 6BQ5's to warm up sufficiently before B+ is applied - this should prevent wildly high voltages from appearing at the grids of your output tubes.

I don't think the zeners are really necessary on the cathodes of the output tubes. Let the tubes find their optimum operating point.

Kevin
 

mark_titano

Member
2004-10-19 10:10 pm
rdf, is the topology of your amp something like this one ?

http://www.wdehaan.demon.nl/images/elb4cir2.jpg


If you look at the power supply schematics you can see the delay circuit. I don't like very much the reley on the HT but I don't really want to use two transformers for driver and power stage.
The great problem is the relay...please look at the datasheet link on first post and post your thoughts.

The zener on the cathode are there because "...when a class-AB cathode-biased amplifier sees a prolonged burst of signal, the output tubes' conduction can easily increase fivefold over the nominal idle current setting. This increase in cathode current will slowly charge up the bypass capacitor, which will effectively increase the negative bias voltage, as the grid will become more negative relative to the cathode. This new bias voltage will slowly collapse over time, but until then the amplifier will distort greatly as the output tubes might even be completely cut off after the large signal burst...The zener’s break voltage is purposely set higher than the bias voltage, so that when the amplifier leaves class-A, the zener limits the maximum voltage the capacitor can charge up to, which prevents under biasing the amplifier after a large current swing."

Look here:

http://www.tubecad.com/2005/April/blog0043.htm


I'm thinking to give a dual power supply to the driver (like +/-200V), then, finding a working point that guarantee the right negative potential for the ecc88 anode, I will be able to connect the 84s cathodes directly to gnd.

Mark
 
mark_titano said:
I'm thinking to give a dual power supply to the driver (like +/-200V), then, finding a working point that guarantee the right negative potential for the ecc88 anode, I will be able to connect the 84s cathodes directly to gnd.
Mark


If I understand correctly, to do what you propose you would need a bias voltage of minus 11 on the 84 grids. Which also means -11 at the driver plates, which means the driver grids are at -100 or more. To do that, you would have to cap couple the input and provide a high quality negative bias supply for the driver grids. Doable, but why do it that way? I would think it easier just to put coupling caps between the stages and have the negative bias supply on the output tubes if you want the cathodes grounded. Then you can put the operating voltage and bias of the drivers anywhere you want. But it's something completely different than the amp you started with. And you'd have to come up with a way to balance the output tubes, if you want that feature.

Sheldon
 

rdf

Member
2004-06-21 8:04 am
big smoke
If I read it correctly the danger is that with the driver tubes presenting a high impedance to the supply on initial start the plate voltage has the potential, due to the very low driver supply capacitance required by the regulator tubes, of coming up too fast and driving the outputs grids extremely over-positive for a short period. Is that correct?

The options which come to my limited mind, none you might consider acceptable, are:

- Replace the positive driver power supply with an ~8.4k ohm resistor and a lot of capacitance to force a slow ramp up. The -50 tail supply will come up quickly. With enough uF the positive supply will come up well after, forcing the output grids to ramp from the negative rail to the proper operating point instead of from the positive down. Your woofers might not appreciated the ramp up period however.
- Replace the relay with a switch and do it manually if it's the added complication you don't like.

Re: the relay or switch, I'ld consider splitting the 500 uF supply cap in two, say 250/250 (or even 50 uf by the rectifier and the remainder after), with a high value resistor between the two cap anodes and use the switch/relay to bypass the resistor. This provides a soft start for the output plates, allowing them to slowly ramp to a median plate voltage of your choice and warm a bit before getting hit with the shock of the full B+. This step start technique worked very well on some of the old 50 kWatt tube-type AM transmitters I maintained in the distant past.

BTW, my el84 couldn't look more different than your circuit, or be much older ;) :

http://vintageradio.me.uk/amplifier/mullard33_2.jpg

Hope something there helps!
 

mark_titano

Member
2004-10-19 10:10 pm
Thakyou rdf for the support.

The ccs start in 20sec. circa. If you look at Gary Pimm schematic you can see C5. The greater is C5 value slower the ccs turns itself on. This is a good thing for the heaters warm up time but if the ccs don't work and there is no current in the circuit the 84 grids see something like 255V...not so good. I've just the 500uF JJ capacitors. So I'll use them. However I can add a C-R- in front of that caps. I'm thinking that if I'd change the working point of the 84s, I could use two secondary, one for this power supply and one for the driver. To provide 340V (so 340-85V=255V. Negative bias -8V->77V on the driver stage) I can use a Vac between 250Vac and 260Vac. This is good because I could use a relay directly on the secondary.

But you are right...maybe the simplest solution is a switch. I'm just afraid because my sister often uses my amps :hot:

I've find this:

http://sound.westhost.com/project39.htm

I'll read it before to leave the idea of a relay.

Mark
 

rdf

Member
2004-06-21 8:04 am
big smoke
Ah, got it now Mark. Your concern is failure modes. I dealt with it by fusing the B+ supply for "ultimate protection".

I played with a TD relay before going the tube rectifier route. Didn't care for the complication and the tight packaging my amp chassis required. My route, the lazy man's way, was to buy an inexpensive Omron knock-off from a local electronics hobby shop (http://www.leeselectronic.com/). It's not on the site but I recall it being about $30 CDN. Good luck!
 

mark_titano

Member
2004-10-19 10:10 pm
Well not only failure...my explanation was not so clear. I'm afraid for the possibility that the EL84's warm up first. In this case they turned on hard because, if the driver stage is not warm, the voltage seen in the control grid is the voltage in the driver supply: 255V. Gary Pimm has suggested to use neon bulbs to limit how far positive the grids can be pulled. Look here

http://www.diyaudio.com/forums/showthread.php?threadid=29502

Neon bulbs have much lower and linear capacitance than zeners, this is the reason why he suggested them.

If the supply of the EL84's start (you are right, softly is better) when the driver stage in completely warm, there is no problem and no neon bulbs are needed.
So...two secondary and a delay reley is the solution ;)

Mark
 

rdf

Member
2004-06-21 8:04 am
big smoke
Referential difficulties! We've been talking different CCS's. You meant the one in the longtailed pair. Now it makes sense. It needs B+ to conduct, until that happens the driver plates go sky high. The revised options are:

- relay
- new little sister

I'm curious to hear how it all works out for you.
 
rdf said:
You meant the one in the longtailed pair. Now it makes sense. It needs B+ to conduct, until that happens the driver plates go sky high. The revised options are:
- relay
- new little sister


Yes.
Well there are several time races. I think that the ideal start up sequence should be:

1. heaters
2.driver power stage
3.negative ref for ccs in the long tailed pair
4.supply for the 84s

Uff...I have to work with the delay time due to C5 and find different settings for the two ccs, I think. And use a relay...I'm thinking to use a coupling cap ehehehe :smash:

My sister is not little...she loves music and this is a bigger issue :D

Mark
 
mark_titano said:



Yes.
Well there are several time races. I think that the ideal start up sequence should be:

1. heaters
2.driver power stage
3.negative ref for ccs in the long tailed pair
4.supply for the 84s

Uff...I have to work with the delay time due to C5 and find different settings for the two ccs, I think. And use a relay...I'm thinking to use a coupling cap ehehehe :smash:

My sister is not little...she loves music and this is a bigger issue :D

Mark

Or add a tube and make a hybrid bridge with an indirectly heated rectifier.

Sheldon