• 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.

holy cow, this sounds nice

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Again a EL84PP amp in triode using the famous zener string arrangement. See for zener schematic: http://www.t-linespeakers.org/tubes/triode-trick.html

This amp sounds better than the average 300B amp. I am using 50 year old OPT's (35-80000 Hz, within 1 dB). Choke-input with Chinese 5U4G tube rectifier. It is a revelation.

I first tried UL but triode is better. I woulds like to try pentode mode with local feedback. Any ideas how to accomplish this ?
 

Attachments

  • el84pp_jh2.jpg
    el84pp_jh2.jpg
    62.3 KB · Views: 1,261
Jaap,
An interesting design.

The "Secret Weapon" which lowers screen voltage on the EL84 will lower the gain of the EL84 and therefore make them perform better in triode mode. A nice trick. Lower gain pentodes (and beam tetrodes) perform better when triode strapped which is why 6V6 is a better candidate for triode strapping than EL84.

However I'm NOT that impressed with the diff amp tail current source. It looks like it was designed specifically for low compliance voltage not for good current source performance. On first glance I thought - hey!! thats sophisticated, must be good. On analysis its very simple and suffers the usual limitations of a single transistor (T4) current source in that AC impedance will be limited by Early effect, although it may be good enough due to very low anode loads and high current operation. The other three transistors are just wired as diodes and simply set bias point and prevent T4 from being saturated (Baker Clamp style). If you have enough compliance voltage or can organise a small negative rail then a LED referenced cascode current source would give better performance which would be noticed as better detail in the sound.

Thanks for posting your schematic. It always good to see what others are up to with their designs and hopefully you get some useful feedback.

Cheers,
Ian
 
OK, Lowering screen voltage lowers gain and increases linearity.

1. My Quicksilver V4 amp runs the KT-88 screens at 300V (through 120 Ohms/tube) and the trans CT at 600V. The screens do not have a percentage of plate signal on them, does this affect linearity?

Any effect on damping factor?

2. In simulations a transistor current source in a 12AU7 diff pair makes a good phase splitter, but adding a few ohms directly to each cathode seems to reduce 2nd harmonic distortion. Comments?
 
Current Sources and all that.

Compliance Volts Issues
With the "-ve" input of the diff amp grounded the common cathode connection will have a signal on it equal to half of Vin.

When you superimpose that Vin/2 on the cathode DC voltage, the swing MUST not be so large as to cause the current source transistor to either saturate or turn off. That Voltage "headroom" is what I call the "compliance" voltage. When the triodes are run at heavy current like in this design there will be very little DC bias or "compliance" voltage. Thats why you usually see the current source returned to a small -ve rail rather than 0V. When a transistor saturates the collector voltage can drop below the base voltage and there is a significant recovery time associated with the transistor recovering from saturation. In this design T5 which is wired as a diode prevents the collector from dropping below 1 diode drop above the base keeping T4 from saturating and hence avoiding the recovery time issue. This is called a Baker Clamp. Its used a lot in older switch mode power supplies which use trabsistors as the switching device.

Cathode Resistors
In any diff amp there will be 3rd harmonic distortion due to device non-linearity and second harmonic distortion due mostly to imbalance between the devices. A perfectly balanced diff amp will have practically no 2nd harmonic distortion. Small resistors in each cathode (or emitter in the case of an SS diff amp) degenerate the gm of the devices and force better balance and hence lower 2nd harmonic distortion. Large amounts of degeneration can case the sound to be "dry" and a bit "sterile". I generally use resistors of 1/10th of 1/gm in the cathode on each side BUT I have also used a pot with the ends to the cathodes and the wiper to the current source. Then adjust the pot for equal anode voltages thus balancing the diff amp exactly AT DC. What this is doing is degenerating the gm of the two sides by different amounts until they are equal (for DC anyway).

There are some CAVEATS though.
In diff amps using large currents each side there WILL be some grid current and so it is important to try to match the resistance from grid to 0V on each side such that the DC grid voltages will be the same. That often means that the -ve input (especially if not connected for feedback) will need a reasonable value of rsistor to 0V and that will then need to be shunted by a good capacitor for a good AC signal ground.
ALSO I haven't quite got my head around if degenerating gm (via the pot arrangement) until the 2 sides are equal at DC will mean that they remain equal at all AC frequencies.

On the single transistor current source
AC impedance will never be greater than approx 300 KOhms due to Early effect and can be significantly less than this with bad design. Keep Vref as high as possible for best results. If you need more than that then the cascode current source is the answer BUT it will almost certainly need to be returned to a negative rail rather than the 0V line because of compliance voltage issues.

I'm fairly sure the above is all correct BUT if someone thinks I'm wrong then please say so - we don't want to lead anyone astray.

Cheers,
Ian
 
I'm going to stick my neck out and guess that DC operating point and AC gain can not both be balanced by one pot. Now given that the dual triodes are in the same envelope, and the current is identical, they will largely match. I think a gain imbalance in the phase splitter could cause a DC offset in the output transformer primary winding (well maybe not, I'll digest this some more), my guess is that gain balance is more important (but probably closely related to DC operating point).

The trouble with all this is that audiophiles as a rule are concerned with miniscule effects. After all any reasonable amplifier design will reproduce recognizable music it's tiny details like this that make the difference between acceptable and great amplifiers.

I agree compliance is most important, for my simulation I ran grids at DC gnd, cathodes ended up around -2V and I ran the current source down to the output bias supply of -40V. (A related question since the cathodes can never go positive by more than a volt or two, will a 60V transistor be safe here?)

You didn't answer my question about screen grids, maybe I oversimplified the problem?
 
"1. My Quicksilver V4 amp runs the KT-88 screens at 300V (through 120 Ohms/tube) and the trans CT at 600V. The screens do not have a percentage of plate signal on them, does this affect linearity?"

Unless you're setting up for ultralinear, the screens should be at AC ground and have no signal appearing there. Stiff screen supplies improve linearity. That's why I included electronic regulation for screen volts in an 807 amp design I did.

"Any effect on damping factor?"

Local feedback at the finals lowers output impedance and helps damping.

"2. In simulations a transistor current source in a 12AU7 diff pair makes a good phase splitter, but adding a few ohms directly to each cathode seems to reduce 2nd harmonic distortion. Comments?"

12AU7s: YUCK! Try 6SN7s or 6CG7s instead. Save the 12AU7s for oscillators or small RF amplifiers/frequency multipliers or that VT version of a Pentium IV :D Keep them far away from any quality audio design.

"I first tried UL but triode is better. I woulds like to try pentode mode with local feedback. Any ideas how to accomplish this ?"

Take a LQQK at gingertube's "Baby Huey" for a local feedback scheme that would also work with your design since both of these use a LTP phase splitter to drive the finals.
 
Ah, yes! This is a form of boosted triode. They make fine little amps :)

12AU7s: YUCK! Try 6SN7s or 6CG7s instead. Save the 12AU7s for oscillators or small RF amplifiers/frequency multipliers or that VT version of a Pentium IV :D Keep them far away from any quality audio design.

Unless GNFB is implamented, a 12AU7's shortcomings can be quite noticeable. (psssst, try a 12AV7 :cool: )
 
Ex-Moderator
Joined 2003
I won't say it's a typo (because it's handwritten) but that 500 Ohm variable resistor should be 50 Ohm. Otherwise, the current through it will cause so much voltage drop across the resistor that the CCS won't work.

Another point is that the upper transistor is likely to dissipate about 0.6W - it would probably be a good idea to replace the 2N3904 with something a bit beefier. I recently ordered some transistors and discovered why gingertube uses MJE340 in low voltage circuits - it's the same price and performance as a low voltage transistor. Use an MJE340.

ALSO I haven't quite got my head around if degenerating gm (via the pot arrangement) until the 2 sides are equal at DC will mean that they remain equal at all AC frequencies.

Provided that the anode load resistors are perfectly matched, the only way that there can be an AC imbalance at the anodes is if some signal current is lost to ground by the CCS. Otherwise, the signal current simply swings backwards and forwards between the two anode loads. Adding the resistor in the cathode circuit simply allows the necessary input voltage imbalance to occur that compensates for valves with unmatched parameters.
 
12AU7 & 6CG7

The best 12AU7 type is Brimar 13D5 & that is the only type I can live with in one application.

Close to the 12AU7 parameters are Amperex PQ 6085. Slightly better than Brimar 13D5, but draws twice the filament current. Make sure the tube has gold pins & states PQ (Premium Quality). Another version exists without the PQ & is different sounding.

As for 6CG7 the best are made by Raytheon & these are rare. True Raytheon have tall flat black plates & a center shield. These are found in the red-white-blue boxes. Most Raytheon found on eBay & at tube dealers were made in Japan in a white & red box- they sound poor.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.