12-07-2018, 06:10 PM
A bit of a change with this post. I thought I would put it here as it does include a Russian Rod tube as well as a regular Russian miniature tube.
I have been thinking about interfacing valve circuits with transistor. Particularly a transistor audio output stage as that would easily avoid the use of an expensive output transformer, usually used with valve circuitry, transformerless output being tricky. The issue here is often matching the higher impedance output to lower input impedance of transistor stages.
Thinking reverse engineering with a difference I remembered an interesting circuit in my Fet Applications Handbook (see photos). This appears to sort impedance matching issues.
This circuit uses a fet input and transistor output. The design aims in the article (that I can more readily test and aim for) are a gain of 60 +/- 30%, O/P imp < 1.5kOhm, 1V O/P PtoP (1% distortion), with a supply of 45V.
My plan was to design a circuit using a 6j1b miniature pentode in place of the fet in one circuit. I also modified an existing circuit which used a 1j24b rod pentode to give the same result. For the valve circuits I planned to increase the supply voltage to 60V and use a (higher voltage) BC546 as the transistor.
I planned to use a 10kOhm resistor as load (to be much greater than output impedance) for testing gain. Input was sinewave at 1kHz. The article states bandwidth design, but I generally followed and did not test for at this point), Output impedance, although not easily tested without good equipment, I would likely get a reasonable approximation by lowering the load until the output wave form was reduce by half (not totally sure but it seemed to get figures in the right area – please let me know if that is a false assumption). I cannot easily test the distortion figure but looked for minute changes in the output waveform as the limit when measuring input and output limits. I did find lowering the output load resulted in distortion in the waveform so had to reduce the input level to a point where this did not occur. All signal voltages are P to P.
The designed circuits are shown in the photos.
With the 1j24b the results were gain 46 (input 100mV), output imp 650 Ohm (approx with I/P 50mV), max output level 14V (between 250 and 300mV I/P – exact figure difficult to measure) and current consumption 950uA.
With the 6j1b the results were gain 48 (input 100mV), Output imp < 500 Ohm (with input 50mV), max output level 14V (between 250 and 300mV input – exact figure difficult to measure) and current consumption 1.4mA.
I will say that perhaps the biggest thing to come out of this was that my test equipment is rather limited, but I am pleased that my results, in their limited subset of the original article specs, seem to come well into the range planned in the article perhaps apart from the supply current which is slightly over, but acceptable. This bodes well for future possible replacement of fets by valves (rather than the usual other way round) in the future. I plan to go further with these modules and see how useful they are.
Tracy
Tracy
I have been thinking about interfacing valve circuits with transistor. Particularly a transistor audio output stage as that would easily avoid the use of an expensive output transformer, usually used with valve circuitry, transformerless output being tricky. The issue here is often matching the higher impedance output to lower input impedance of transistor stages.
Thinking reverse engineering with a difference I remembered an interesting circuit in my Fet Applications Handbook (see photos). This appears to sort impedance matching issues.
This circuit uses a fet input and transistor output. The design aims in the article (that I can more readily test and aim for) are a gain of 60 +/- 30%, O/P imp < 1.5kOhm, 1V O/P PtoP (1% distortion), with a supply of 45V.
My plan was to design a circuit using a 6j1b miniature pentode in place of the fet in one circuit. I also modified an existing circuit which used a 1j24b rod pentode to give the same result. For the valve circuits I planned to increase the supply voltage to 60V and use a (higher voltage) BC546 as the transistor.
I planned to use a 10kOhm resistor as load (to be much greater than output impedance) for testing gain. Input was sinewave at 1kHz. The article states bandwidth design, but I generally followed and did not test for at this point), Output impedance, although not easily tested without good equipment, I would likely get a reasonable approximation by lowering the load until the output wave form was reduce by half (not totally sure but it seemed to get figures in the right area – please let me know if that is a false assumption). I cannot easily test the distortion figure but looked for minute changes in the output waveform as the limit when measuring input and output limits. I did find lowering the output load resulted in distortion in the waveform so had to reduce the input level to a point where this did not occur. All signal voltages are P to P.
The designed circuits are shown in the photos.
With the 1j24b the results were gain 46 (input 100mV), output imp 650 Ohm (approx with I/P 50mV), max output level 14V (between 250 and 300mV I/P – exact figure difficult to measure) and current consumption 950uA.
With the 6j1b the results were gain 48 (input 100mV), Output imp < 500 Ohm (with input 50mV), max output level 14V (between 250 and 300mV input – exact figure difficult to measure) and current consumption 1.4mA.
I will say that perhaps the biggest thing to come out of this was that my test equipment is rather limited, but I am pleased that my results, in their limited subset of the original article specs, seem to come well into the range planned in the article perhaps apart from the supply current which is slightly over, but acceptable. This bodes well for future possible replacement of fets by valves (rather than the usual other way round) in the future. I plan to go further with these modules and see how useful they are.
Tracy
Tracy
