For various follower-type power amps, there is a need in preamp with large enough voltage swing and a voltage gain of more than 20dB. This can be solved by a simplified Zen9-like stage, with a Radio Frequency j-FET transistor at its input. I have put together a prototype, as shown in the picture, and made some measurements.
Available voltage swing is nearly 80V p-t-p. Idle current is 220mA, output impedance (no NFB) is [ 270 Ohms, phase shift=-0,53deg at 1kHz] , and [ 270Ohms, phase shift=-3,6deg at 100kHz]. Voltage gain with a connected load of 1,3kOhms is 12,2 (22dB). Square wave response at 70V p-t-p, 100kHz, 1k3 load is shown in the foto. Larger size curve belongs to output signal, smaller – to input. Slew rate = [ U(0,9)-U(0,1) ] / dT = 380V/uS at the front edge, and 212V/uS at the decay. Trying to use the NFB as it realized in Zen9, with 33kOhms and 330kOhms resistors, makes time response of the stage much worse. I guess I should take more difficult way for adjusting the gains of the channels. Also I plan a decent shunt PS for this stage.
Available voltage swing is nearly 80V p-t-p. Idle current is 220mA, output impedance (no NFB) is [ 270 Ohms, phase shift=-0,53deg at 1kHz] , and [ 270Ohms, phase shift=-3,6deg at 100kHz]. Voltage gain with a connected load of 1,3kOhms is 12,2 (22dB). Square wave response at 70V p-t-p, 100kHz, 1k3 load is shown in the foto. Larger size curve belongs to output signal, smaller – to input. Slew rate = [ U(0,9)-U(0,1) ] / dT = 380V/uS at the front edge, and 212V/uS at the decay. Trying to use the NFB as it realized in Zen9, with 33kOhms and 330kOhms resistors, makes time response of the stage much worse. I guess I should take more difficult way for adjusting the gains of the channels. Also I plan a decent shunt PS for this stage.
Attachments
KP903A and KP926A are russian parts. KP903A is an analog of USA CP650, in a screw-like case, even more convenient. There are a lot of sources for them in Moscow, something 1...2 USD/pcs. In USA CP650 as I googled costs 17 USD. KP926A are power SIT-transistors, something similar to unobtanium japaneese 2SK60, but quality is much worse than 2SK60 (big gate capacitance, big leakage etc.). They could be sourced in Moscow for 6...8 USD/pcs. But I would much prefer 2SK60. As I understand, Nelson Pass even ordered special development of power SITs according to his specification.
Thanks a lot Nelson for this hint, since I was going to mess with device matching etc. We can get two profits from this simple step - decreasing out impedance and gain matching.
Great preamp
Pity that those devices are not available outside of Russia. I would like to build with them a differential version as a voltage gain stage for a circlotron power buffer with laterals.
Hi, Juma
I could try to get such parts and then mail to you. Please tell me a quantity. A problem is that they can not be matched easily, espesially KP926A, for differential operation.
Thanks Vladimir for your kind offer. Right now we are working here (in Serbia) on establishing the "Russian connection" - people are interested in tubes and power Ge transistors too. If we fail, I'll be free to contact you 
When it comes to matching, I suppose some source degeneration and SuSy differential feedback can do the trick.
When it comes to matching, I suppose some source degeneration and SuSy differential feedback can do the trick.
Hey Vladimir,
The actíve devices looks very interesting, do you have any datasheets that isn´t in russian?
How do you get away with two CCS´s on top of each other? A cascode is also to be looked upon as one? Normally it isn´t so easy to deal with as the DC-voltage at the cascode output could be all over the place.
The actíve devices looks very interesting, do you have any datasheets that isn´t in russian
?How do you get away with two CCS´s on top of each other? A cascode is also to be looked upon as one? Normally it isn´t so easy to deal with as the DC-voltage at the cascode output could be all over the place.
Hey Vladimir,
The actíve devices looks very interesting, do you have any datasheets that isn´t in russian
How do you get away with two CCS´s on top of each other? A cascode is also to be looked upon as one? Normally it isn´t so easy to deal with as the DC-voltage at the cascode output could be all over the place.
Hello, Lars
I can only enclose the CP650 datasheet, that is similar to KP903.
As of KP926, it is a high-voltage (450V, 16A) SIT transistor designed 20 years ago, when it was considered to be good for switch power converters. It has triode-like Id vs Vds, quite a big Idss (I guess 2500pF), I dissipated 50W of heat on them during measurements. They like high voltages. Not many of them have big enough Id(Ugs=0) nearly 1A at Uds=10V, some of them could be even closed at Ugs=0. For several amps Id, they must be operated as BJT transistors with positive Ugs. One russian designer (Konstantin Musatoff) used them with 300V power supply, low currents and output transformer.
In this preamp schematics there is only one CCS, it is in fact a simplified Zen9. The top KP926A acts as a CCS only conditionally, its properties something between a simple resistor and a real CCS. If interesting, I can prepare and enclose its Id vs Uds in the CCS configuration.
Last edited:
Hello Vladimir, I am trying to learn about your preamplifier circuit... my understanding is that the max input is limited to 1.6V peak by the gate-to-source bias. Then 1.6V x 12.2 (gain) x 2 (for peak to peak) = 39V peak to peak. So I do not understand how you produce an output of 70V peak to peak? I am a beginner with design... so is there a simple answer for me?
Thank you,
J.
Hello, Jen
The Ugs=1,6V indicated in the schematics is a design estimate, actual precise value should be measured, what I still have to do.
Inspite of this, the CP 650 j-fet has something near 8V pinch-off gate-to-source voltage, and also allows operation with positive gate bias. Total acceptable swing of Ugs approaches 10V. This can explain the reported behaviour of the schematics.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.