Push Pull Output From A Single Bottle

[darryl]

There has been recent discussion of low powered push pull amplifiers, and tubenit has collated and archived useful circuits and design ideas. http://www.el34world.com/Forum/index.php?topic=12526.msg117936#new

This design is another such amplifier, but using a dual tetrode output valve. I have, in the past, attempted to use small signal dual triodes, such as the 12AU7 or 6SN7 as push-pull output valves, but always found the results disappointing. The valve chosen for this project is the Soviet-era dual tetrode GU-17, which is electrically equivalent to the QQE 03/12 and the 6360.

http://tubedata.milbert.com/sheets/018/g/GU17.pdf

http://tubezone.net/pdf/6360.pdf

The GU-17's are available at low cost from several of the online Russian and Ukrainian valve vendors. One useful feature of the GU-17 and its equivalents is a 12.6 volt centre-tapped heater, making it very easy to use a solo GU-17 with this amplifier's 12 volt heater supply.

The original idea when these valves were purchased was a self-split output stage, so a complete push-pull amplifier could be built using just two valves - a twin triode preamp valve and the GU-17. I decided to first build a more conventional amplifier to use as a benchmark for a future self-split design, and this amp is the result.

Using a dual output valve with a separate phase inverter in the circuit described here does not actually reduce the required number of valves - a pair of 6BM8's, for example, would produce the same result, but it's more fun to do something different.   

Enough talk, it's on with the construction. The chassis for this amplifier is the same one I have used for many low powered amplifiers - a diecast aluminium box similar in size to a Hammond 1550G. After drilling, filing and punching, the chassis was beaten into submission, and the valve sockets and protective grommets fitted.




Next were the tagstrips and the power transformer. The 50VA toroidal power transformer is a type used for several previous amplifiers. It has four 12VAC 1A secondaries - one is used to power the heaters, while the other three are connected in series to supply the B+ voltage, using a voltage multiplier.




The heater wiring was run hard down on the chassis from the transformer to the valve sockets, but also to tags, where hum balancing resistors would be fitted. The IEC power inlet socket and the power switch were installed on the back panel, 240VAC wiring to the power transformer was installed and exposed terminals sheathed with heatshrink tubing. An earth ( or ground ) lead was run from the IEC connector to a firmly attached lug on the nearest tagstrip.



To be continued . . . 

[tubenit]

Sounds interesting!  THANKS for sharing your experiment with us. Look forward to the results.

With respect, Tubenit

[birt]

that's a very interesting idea. looking forward to the results!

[John]

Yeah, what they said!

[Heinz]

I have made experiments with a QQE03/12 in a self-split configuration. It works but has some drawbacks compared to the more "standard" designs. In order to get a reasonable output power you need a pretty high signal level to drive the tetrodes and a fairly large "tail" resistor to elevate the cathodes to the required voltage. In addition, you have to provide a stable G2 voltage which is considerably lower than the plate voltage. The data sheets give the values. This adds some extra circuitry to your power supply. However, depending on your circuit, when using a voltage multiplier you could tap the multiplier and use an intermediate voltage (e.g. 2/3 of the plate voltage). This requires additional filtering.

On the other hand, I have also made experiments with 6BM8s (or better 16A8/PCL82) and found that the triodes in all the tubes I have (and I have quite a lot since they are so versatile) are very microphonic. Using them in a phase splitter is no problem at all, but using them in a first stage did not work for me. There are other "unknown" tubes that might work as first stages in a 2-bottle amp, e.g. ECF80 or PCF802. They are widely available and inexpensive. Using an inter stage transformer instead of a phase splitter would allow you to run the output tubes in a true push-pull arrangement instead of the lossy self-split configuration. You can use a regular small 115V/2x24V power transformer as an inter stage transformer. I have tried that in an experiment and can confirm that it drives the output tubes. However, I have not used this in an amp design, so I have no idea how it sounds.

[darryl]

I have made experiments with a QQE03/12 in a self-split configuration.

And given us plenty of good ideas to experiment with in future builds. 



This is the current circuit of the "Brisk 4" amplifier. There is also an .sch file attached.





Continuing with the build details:

The output transformer was installed, along with "roll bars" to offer some protection for the valves.





Most of the power supply components were installed next - including the four electrolytic capacitors and four diodes which make up the voltage quadrupler power supply. The calculated B+ from this arrangement was 36 * 1.4 * 4 = 202 volts DC. This B+ voltage was well below the permissible 300 volts on the plates, but it did match the 200 volt limit on the screens. This restricted the amplifier's output, but simplified the power supply, as a separate screen supply was not required.

The front panel LED was fitted and powered from the heater supply, through a dropping resistor. The hum balance resistors were installed, and their junction connected to the output valve's common cathode. This was to raise the heaters above ground potential and so minimise hum. A short piece of tagstrip was attached to the back panel for possible inclusion of a conjunctive filter across the output transformer primary.



More to come . . . 

[PRR]

> you need a pretty high signal level to drive the tetrodes

Double the signal you need to each G1 with push-pull drive. Maybe a bit more for self-split losses.

> a fairly large "tail" resistor

Yes, the self-bias resistor is roughly 1/gm. For good self-split it must be larger than 1/gm.

> you have to provide a stable G2 voltage which is considerably lower than the plate voltage.

That depends on the Mu(g2) of the tubes. "Audio power" pentodes are designed to Mu near 10. This works well for audio with Vg2 nearly Vplate. MAX-current (sweep) tubes must be designed for lower Mu(g2), near 4, which leads to Vg2 much lower than a good Vplate. This also gives lower loss at high current, but that's a minor thing for audio, and not worth the trouble of extra G2 supply (unless it just falls in your lap, as with a multiplier supply).

[Heinz]

I really like your power supply. It is simple and elegant and uses standard components. Just to give you an idea about the multiplier tap, here's the ps that I designed for the QEE03/12 test. It was designed around those 21V transformers because I got a huge pile of them for free. I designed it using SwitcherCAD so I could run simulations before I sat down to actually build it. It operates the tube according to the data sheet with 300V plate and roughly 200V G2 voltage for maximum output power. The B2+ and B3+ voltages can of course be adjusted by using different resistor value.

The main disadvantage of the design is its size and the large number of components. However, it is pretty inexpensive if you can re-use parts from your junk box.

[kagliostro]

Hi Darryl

I like your amps 

When you have more to show, please post it

Kagliostro

[darryl]

When you have more to show, please post it

I'd better get on with it then!   

Components associated with the output stage and phase splitter were installed next. A 2.2kΩ 5 watt wire wound resistor was used as a common screen dropping resistor. A 47Ω carbon film resistor was then used as a screen stopper, to reduce the possibility of spurious oscillation. Grid stopper resistors were also used, their relatively large value of 47kΩ being to prevent both instability and blocking distortion. A conjunctive filter was installed across the output transformer primary.





Next were the front panel controls, and components associated with the preamp valve. The preamp circuit and layout were identical to that used in several other of my amplifiers. Keeping the preamp stages identical will allow the unusual output valve in this design to be compared to the more conventional valve types in other amps. A footswitch socket on the back panel was connected in parallel with one section of the front panel gain switch.



More to come . . . 

[Madison]

Holy cow darryl.
You do fantastic work!
Can't wait to see the end result.

[darryl]

Moving along:

Once the under-chassis wiring was complete, adhesive labels were applied to the exterior, and a set of valves was plugged in:





The output stage uses cathode bias, so to seek an appropriate cathode resistor value, a 1k resistor was temporarily installed. The voltage drop across it was measured so the cathode current and valve dissipation could be calculated. The cathode resistor value was approximately halved, and halved again, until the bias was estimated to be appropriate. The final resistance was 235Ω, two 470Ω 1W resistors in parallel.

The voltage drop across the 235Ω cathode resistor was 13VDC, so applying Mr Ohm's law, the cathode current was 55mA. The cathode-to-plate voltage was 212 - 13 = 199VDC, so the total plate+screen dissipation was .055 * 199 = 10.9 watts. The dissipation per plate was therefore about 5.5 watts. The data sheet quotes a maximum dissipation of 7 watts per plate, so 5.5 watts should be about right.

The output transformer used in this amplifier is 8k plate-to-plate, with a single 8Ω secondary. The output voltage was measured at the onset of clipping, with 4, 8 and 16 ohm loads, and the output power calculated.

          Load                  Output Measured                 Calculated Power

           4Ω                             3.4VAC                              2.9W
           8Ω                             5.9VAC                              4.3W
          16Ω                            7.8VAC                              3.8W

So, with the "optimum" 8 ohm speaker load, this amplifier is able to deliver four watts. This is not a lot of power, but that was not the objective of this project. As Heinz suggested in an earlier post, running the GU-17 with the permitted 300 volts on the plates would achieve a greater output, but at the cost of greater power supply complexity.

I don't play guitar, so over the coming weeks I'll get this amplifier out to some of my guitarist friends to find out how it sounds.

 

[tubenit]

Truly impressive!  Great job!  Thanks for sharing your creativity.

If later you can post sound clips or video, that'd be fantastic.

With respect, Tubenit

[Madison]

Man, that is nice nice nice!

My dumb question: why do you wire all the grounds of your terminal strips together?
The chassis isn't sufficient?

Peace.

[Heinz]

I think you may be able to increase output power a bit by stabilizing the screen voltage of the output tube. Adding a cap (e.g. 16uF) between ground and the junction of the 2.2k and the 47R screen resistors should do the trick. As it is right now the screen grid current will reduce the screen voltage through the voltage drop in the 2.2k resistor. This will in turn act as negative feedback.

[darryl]

...why do you wire all the grounds of your terminal strips together?

[Translation note: I'll use Australian terminology - "earth" rather than "ground".  ]

The tagstrips are riveted in place, and a riveted lug can never be trusted to make a reliable earth connection. In each row of tagstrips there is one earth lug that is attached to the chassis with a bolt, nut and two shakeproof washers. The riveted lugs are then joined to this reliable earth point.

This construction method does not follow the usual "star" configuration, but its background hum level is still very low. While ever this method works, I'll keep using it. 

This will in turn act as negative feedback.

Most of my amplifier designs use a common screen dropping resistor which is not bypassed with an electrolytic capacitor. Each screen then has an individual 47Ω "stopper" resistor. The common resistor is usually larger than the typical data sheet value, to protect the screens when the amplifier is overdriven - as a guitar amplifier always will be.

Leaving the common resistor unbypassed is deliberate - the small amount of negative feedback generated does help to reduce crossover distortion in the output stage.

[billcreller]

Outstanding job on that ! 

[birt]

i have another single bottle PP output stage for you. ECLL800. it was used in radios and might be hard to find but its a good sounding tube. (it's something like one triode and 2 EL95 pentodes in one tube)

[gcwills]

I have made experiments with a QQE03/12 in a self-split configuration. It works but has some drawbacks compared to the more "standard" designs. In order to get a reasonable output power you need a pretty high signal level to drive the tetrodes and a fairly large "tail" resistor to elevate the cathodes to the required voltage. In addition, you have to provide a stable G2 voltage which is considerably lower than the plate voltage. The data sheets give the values. This adds some extra circuitry to your power supply. However, depending on your circuit, when using a voltage multiplier you could tap the multiplier and use an intermediate voltage (e.g. 2/3 of the plate voltage). This requires additional filtering.

On the other hand, I have also made experiments with 6BM8s (or better 16A8/PCL82) and found that the triodes in all the tubes I have (and I have quite a lot since they are so versatile) are very microphonic. Using them in a phase splitter is no problem at all, but using them in a first stage did not work for me. There are other "unknown" tubes that might work as first stages in a 2-bottle amp, e.g. ECF80 or PCF802. They are widely available and inexpensive. Using an inter stage transformer instead of a phase splitter would allow you to run the output tubes in a true push-pull arrangement instead of the lossy self-split configuration. You can use a regular small 115V/2x24V power transformer as an inter stage transformer. I have tried that in an experiment and can confirm that it drives the output tubes. However, I have not used this in an amp design, so I have no idea how it sounds.

You may want to check out my amp design here http://www.guitargear.net.au/discussion/index.php?topic=25648.0  for an amp that I built some time ago using a QQE03/12 driven by a 6BL8 (ECF80)

[kagliostro]

Those ecll800 are very precious tubes nowadays

http://search.store.yahoo.net/thetubestore/cgi-bin/nsearch?catalog=thetubestore&query=ecll800&.autodone=http%3A%2F%2Fthetubestore.com%2F

in an other way a full PP amp is also possible with only two tubes

http://www.solorb.com/elect/musiccirc/squirrelmonkey1/ (this one is SE using only one tube)

and they aren't too expensive, only you must find the adequate socket (I've find with no much difficulties at € 2.5 each, not very cheap but affordable)

http://search.store.yahoo.net/thetubestore/cgi-bin/nsearch?unique=f94a2&catalog=thetubestore

this is a layout for the SE amp I've draw sometime ago



pay attention to the adapted for 220v PS, I've on the desk near all the parts but still now not build it because I'm missing an adequate OT

@ Gcwills

which is the primary impedance of the OT in your QQE03/12 ?

Thanks

Kagliostro

[Heinz]

gcwills,

thanks for the link. That is a very cool amp. I love those minimalistic designs. Does the ECF80 have any microphony issues?

[gcwills]

My amp used a M1115 speaker line transformer available here in Australia - in this application it reflects an 8K plate to plate load to the output valves.

Re ECF80 microphony, my amp is not a high gain amp and there is no microphony evident