Choke loaded VFET SET amp

[tomp]

Finally, success. David Berning and I have been working for several months now to produce a truly unique and interesting amp. It is a choke loaded Single Ended Transistor amp using two Yamaha 2SK77 VFET transistors I found in my parts bin. I have had them for about 40 years and forgot about them until recently. When I brought them over to Dave's we ran them on his curve tracer and lo and behold they looked similar to a 300B.

I will not go into all the details of the project but the driving consideration to the design was to protect the 2SK77s at all costs as they are no longer available. As a result we had a lot of false starts because on a slew of protection circuits that has a habit of turning the amp off at the wrong times. We will have a meeting in the fall to go into the amp in detail but we are both very busy with our schedules now. We're thinking of having a SET day will all SETs invited at Dave's house as all my speakers are tri-amped.

Just briefly, the amp puts out over 15 watts at 8 ohms, 17 watts at 6 ohms and over 11 watts at 4 ohms. Frequency is flat from 20KHz to almost 20 Hz with just a slight drop at 20 Hz. It is stable into any load from open circuit with a 1uf capacitor on the outputs to any resistive load. We listened to it compared to Dave's 845 SET and they are virtually indistinguishable. Surprising, the lowest frequencies, ie below 20 Hz were better with the 845, not what we expected but Dave said that is probably because the 2SK77s are higher impedance than the 845s through his ZOTL impedance converter used with the 845s and the output chokes like lower impedance.

I'm including three photos, the 2SK77s in comparison to a TO3 and TO220 transistor to give you an indication of the size of the beast, The underside of the amp showing the circuit board, and the completed amp. SInce this is a "retro" project I decided to paint it with a metal flake glitter paint called "cherry bomb" for effect.

Tom

Attachments

[Guy]

Well, this will be interesting to hear! Looks good, too!

[Roscoe Primrose]

Well, this will be interesting to hear! Looks good, too!

Indeed it will...

I've got 8 2SK82s, which ought to curve out very similarly, and be good for about half that much power. While they're long out of production, they are to some extent still available...

Roscoe

[tomp]

There are still some 2SK180 VFETs to be had on ebay for $85 each. Someone advertised them for $23 each but when I ordered them they said they quoted a price for a different item. The 2SK180 is very similar in the same mongo sized case and has higher gain and lower impedance which should make it work better if using choke loading. I don't know about stability with the higher gain.

[Pelliott321]

for my newbie self please explain choke loading

[tomp]

Instead of a push/pull arrangement of two active devices to produce both halves of the AC waveform, a single ended amplifier uses a constant current source to supply a current which runs through the active device. The active device is set so that with that current, the voltage at the active device is half of the voltage of which the constant current source is capable of operation into. Therefore by varying the impedance of the active device, the output voltage can swing up and down to produce the voltage to drive the speaker. It operates in class A mode.

For a low power circuit where efficiency is not important, a quasi constant current can be obtained by using a high voltage in series with a high value resistor. If that combo is in series with the active device a small change in the voltage across the active device has little effect on the "constant current" source.

For higher powers a common type of constant current source uses an active device such as a transistor to maintain a constant voltage across a series resistor. That combo is used in series with the active device and will provide the constant curent necessary for the impedance changes in the active device to produce the needed output signal to the speaker. The downside is that power dissipation is high because the source voltage driving the constant current source must be high enough to supply the peak voltage in one direction of the output signal. In quiescent or small signal conditions, the pass transistor is dissipating a lot of power.

For higher efficeincy, a choke can be used in place of the transistor type of constant current source. The characteristic of a choke of resisting any change of current flowing through it produces a constant current result. If you put a choke in series with the active device and then use the active device to set the required quiescent DC current, as you vary the impedance of the active device with the audio signal, the choke will maintain the quiescent current resulting in an increase or decrease of the voltage at the choke/active device junction to drive the speaker. The advantage of the choke is that the power loss can be much less that the transistor type of constant current source because the only voltage that appears across the choke is the drop caused by the DC resistance of the winding times the quiescent current. With proper design of the choke that resistance can be low.
The downside is the construction of the choke must be large enough to maintain that current level at the lowest output frequencies desired without saturation. The cost the resulting choke can be higher, but in the long run the lower power dissipation will more than make up for the additional cost. Remember that you not only have to pay for the power but also the AC to remove it during the summer. The amp is already a class A design which is very inefficient so it pays not to add any more power loss when it is not necessary.

Tom

[Stuart Polansky]

This is my best guess at this. I'm sure TomP, Dave B, Dave M, Bruce, Roscoe, et. al. will correct my errors!

In the text below, feel free to substitute "FET" for "tube".

As per the drawings attached, a typical grounded cathode gain stage most times uses a resistor as a plate load. Advantages are simplicity, compact construction and low cost. Drawbacks are higher distortion than other options due to a relatively low resistance versus plate resistance and load to be driven (typically). This gain stage is highly dependent upon load to be driven for it's performance. Installing a larger resistor to gain better linearity (that is a linear current swing +/- vs. grid voltage input) helps, but requires a higher voltage power supply along with it's own attendant problems.

Three other options generally exist: constant current source (a solid state circuit, or a vacuum tube in a follower mode (variations within)), choke loading or transformer loading. All three serve to increase the plate load to a higher impedance and get the tube to operate in a more linear fashion.

With a CCS, the input capacitance and impedance presented to the tube's plate are the important parameters. Instead of a plate resistor of, say 100k ohms, a CCS can "look" like literally millions of ohms to the tube. Drawing a load line on a set of plate load curves can quickly display the advantages of a very high plate load.

The transformer load put's the next stage's input load into the plate circuit of the gain tube; multiplied by the square of the turns ratio. The primary inductance of the transformer determines the low frequency cutoff. This method eliminates a coupling cap, but adds a coupling transformer. Which is worse? Depends.

Similar to the transformer, the choke load is a very high impedance load, and the low frequencies are limited by primary inductance.

Both the choke and transformer have huge advantages over any other loading:
1) Very low resistance (tens to a few hundreds of ohms) means that little B+ is dropped across the choke and almost all is preserved for the tube and voltage swing.
2) Chokes store energy. When the tube current is reduced in the negative going half of the input signal, the choke releases that energy and, in effect, B+ is nearly doubled.
3) Containing no active devices, a well made choke is quiet and transparent, sonically.

Choke and transformer loading is not commonly used for two main reasons: size and cost. If one seeks the finest in amplification, eventually you end up with something like Duelund caps. The cost and size of those caps can exceed that of a transformer. With a choke load, you TYPICALLY still need a DC blocking cap on the output, or a complex bipolar supply and output servo.In the case of the circuit shown of the single ended amp, the plate voltage of the first tube provides bias voltage for the second. The first tube is choke loaded, the second, transformer.

Note: USUALLY the choke load is returned to B+, not as shown in Epstein's schematic, a neat trick.

Hope this helps.

S

Attachments

grounded cathode.gif (5.32 KiB) Viewed 45485 times

[Pelliott321]

Thanks I believe I understand it!
Chokes are pretty neat devices

[Guy]

Chokes are pretty neat devices

And it's durn hard to make good ones!

[tomp]

According to our choke guru David Berning these chokes turned out quite well. I'll be doing an article on the amp for "AudioXpress" and all will be explained. I'm using LAUD today to run frequency response and distortion measurements.

Tom