Good morning.
I've complete the test bench for my 6n3 buffer.
An externally hosted image should be here but it was not working when we last tested it.
The thermo ionic part in this case is a GE GL-5670.
1st: It's active; when disconnected the power it will take a around 10 seconds or more to the sound level and quality began to drop.
2nd: As predicted it have a large attenuation factor in passive mode.
I've only check at low level to confirm it's working.
In few hours I can evaluate better; with both valves: 6n3 and GL-5670.
Sleep well.
Good afternoon.
First tests reveal that 80V are better than 55V which were the "500mA 12 V AC" transformer can "make" the multiplier deliver with its 10,8V AC insufficiency.
After change for a 15V AC 1A transformer 80V are obtained. The heater DC regulator needs more heat sink, but the sound gets "tubby" with warm bass, a more shut-in midband and treble. We lost some impact and rhythm, but after all it seems to be not a big improvement over an OPA1612 stage with heavy Elna Silmic II fitering post regulators and with heavy Nichicon KW filtering pre regulators.
Further audition is needed, there is midband information that seems to be muffled with the tube stage.
With the solid state gear hot, turns to be a hard choice.
Future will tell.
Hope remains.
I was going to use this piece as a tube preamp for a TDA1554Q based 2.1 desktop speaker setup. Test results You've posted are very good to read for me as I didn't want the mid/hi freq sound to be very sharp and warm bass is what I was looking for.
Only 15V transformer makes this stuff harder as I already got a 2x12V supply.
With this supply I've got also a choice of using 12AX7 based pre and I'm hoping I'll get some warmness in whole range. And this is going to be the first thing to try.
Edit:
And it's good to read that this Chinese stuff isn't a hoax
I was going to use this piece as a tube preamp for a TDA1554Q based 2.1 desktop speaker setup. Test results You've posted are very good to read for me as I didn't want the mid/hi freq sound to be very sharp and warm bass is what I was looking for.
Only 15V transformer makes this stuff harder as I already got a 2x12V supply.
With this supply I've got also a choice of using 12AX7 based pre and I'm hoping I'll get some warmness in whole range. And this is going to be the first thing to try.
Edit:
And it's good to read that this Chinese stuff isn't a hoax
Hi.
Blue LED is useful because it is powered by the 6.3V DC Heater supply line and we still see the "orange" glow at the "side" of the glass.
The Red LED is powered by the AC line with a diode and a 1kOhm resistor.
I reckon that the Nichicon Fine Gold (330uF/50V) at the voltage quadruple (Two Greinacher cells) must have more burning time, as the Elna Silmic II (4.7uF/50V) at the input and output for DC decoupling.
Better sockets for interconnectors also help, but for now I just take as it is...
And this is the payoff for the 15V AC/1Amp Transformer.
An externally hosted image should be here but it was not working when we last tested it.
Awesome cooler, a "must" because the 7806 is "sodomized" with near 20V DC, being 25V DC the maximum allowed. With a near 0,35 A current and an output of 6.3V DC, some 4,5W of heat should be lost with decent efficiency by the heat sink.
All is designed to be cheap, you could maybe do better with higher voltage if power supply caps can handle the voltage. You could also try simple jfet CCS instead of that resistor... BTW, tear out that blue led at the bottom it masks nice tube heater glow...
Ok, "that resistor", a 12kOhms 1% unit that I believe can be replaced by a 1N5301 or 1N5302 jFET current limiters. The current flow in idle mode is 1,46mA at 80V DC supply.
About the audio quality, I don't know yet, but there are some variations of the 6n3/GD-5670 theme. I will start the 4x6n3 project with the new board I received, which I believe its a Aikido SRPP. Its a voltage gain stage and a buffer at the output. There is a constant current source provided by a triode coupled to each stage instead of a current regulator diode. Each tube have a current regulator triode for the other triode which work as a voltage gain or current gain. The 2 tubes option should be the simple version of the same voltage/current gain.
A lot of runs I have to do before change anything about voltage plate value.
Hi People.
Finally I finish the prototype tests for the "Phase 2" and "Phase 3" of 6N3 buffer.
Phase 2 consists in a CCS implementation with 2SK30A FET or a dedicated CCS diode.
Phase 2 already with a higher plate voltage PSU transformer. The Heater power line is separate now with a dedicated secondary voltage.
An externally hosted image should be here but it was not working when we last tested it.
Phase 3 with the same CCS theme based on the 2SK30A.
About 3.8mA cathode current rise the HD to high.
Now the new voltage regulators circuit are implemented and working well with a higher plate voltage, around 85,5V, and a good"Buffer input voltage" for the regulators do their job "well".
Smoother sound is noticed at the "start-up" of this toy.
An externally hosted image should be here but it was not working when we last tested it.
Final Phase 3 implementation.
Diode CCS is preferred since HD is around -50db (0,32%) at 3,5V RMS input (9.9V PP) 20Hz to 20kHz.A much lower HD value than obtained with the 2SK30 CCS; but its a matter of current since the FETs drain about 3,8mA and the diodes are configured to deliver a smaller current. The PCB space is small so this solution is "better".
An externally hosted image should be here but it was not working when we last tested it.
Signal attenuation is very low, around -0,4db, stereo separation is about 40db at 20kHz and frequency response is flat from 10Hz to 100kHZ.
An externally hosted image should be here but it was not working when we last tested it.
Poor specs for digital age but the sound is very gutsy...
...so I don´t give a "bit" about the poor HD and stereo separation.
For 2V RMS digital output HD is even better, and the output holds a 10kOhms load.
The tube is a 6N3 with a "M" grade mark. A "J" military grade is even better, specially at HD values. The GL-5670 works also very well or even better than the "J" grade 6n3.
Cheers
Hi.
It can be something like this...
An externally hosted image should be here but it was not working when we last tested it.
After you remove the old parts, you get a lot of holes and PCB tracks to adopt the new solution.
With a 20V AC input you get +50V and -50V at the regulators inputs. You can use 63V capacitors and 25V AC input to get around +/- 60V and adopt something like ±52V (±100V) or a bit ,more to the tube plate and cathode.
Study the tube datasheet to choose the cathode current and supply voltage.
Check the data sheets of the regulators too.
An externally hosted image should be here but it was not working when we last tested it.
Last edited:
The circuit in question is what is know as a self-biased cathode follower. The "bias" resistor is the 470R resistor that is in series with the cathode. It establishes Vgk through the 470K resistor. That resistor sets the input impedance. Since a slightly attenuated version of the signal appears on the end attached to the 470R bias resistor (the signal on the cathode is in phase with the input) there is very little AC voltage difference and very little signal current will flow through this resistor. This resistor is "bootstrapped" and its effective impedance to the signal is more like 4.7M. So, there IS a very high impedance between the input and the output.
Also, if you replace the 12K load resistor with a CCS (recommended) that is set to a higher current this will change the bias voltage since the voltage across the 470R bias resistor will change. You should determine your new OP on the set of plate curves and see what value of grid bias is required and calculate a new value for the bias resistor.
Hi John.
Thanks for the advice.
Actually the V between the Grid and the Cathode is a bit on the high side but less than 1V.
I can replace the 470 ohms resistor for 374 ohms and the Vgc goes back near 0,5V which was the value for the 60V supply, if I can remember correctly.
With 3,8mA we get a Vgc=2V and a lot of harmonic distortion for such a low Plate supply voltage.
If you have more suggestions, please be welcome.
With 0.5V grid to cathode that doesn't allow for much input voltage swing before you start drawing grid current and the resultant distortion. I would suggest downloading the plate curves for the tube in question and see if an optimum OP can be found. Since the plate voltage is already fixed you can play with different values of grid voltage and plate current on the gragh. With a CCS load the load line will be for all practical purposes horizontal at the current sourced by the current source. Just note that if you are trying to drive a capacitive load the load line will be an ellipse and the circuit will have to source enough current to charge the capacitance.
Hi John.
I admit that I'm not resourceful to fully understand the biasing of the tube, since the optimum OP I observe is at very higher voltage supply (around 170V and 10mA). So, something with my understanding of the tube biasing is very crude...
For such a low plate Voltage to get a linear zone for the OP, which current we should operate a 6n3 or a gl-5670?
Happy New Year.
