Mondo, my figure of a milliamp, or so, was for 1j24b. Since this one seems to be more like an output type of valve (the triode gain suggests this) , it may want a bit more current; after all, it's good for a bit over one watt.
I agree, this is what I thought. (Actually didn't read the 1J24B...)
I have run 10k anode load, and gain suffered somewhat, but I didn't spend much time with such a low load, so have few notes.
I'll have to try again.
I have run 10k anode load, and gain suffered somewhat, but I didn't spend much time with such a low load, so have few notes.
I'll have to try again.
Since they come out with a Voltage gain of 5 or 6 V/V, with diode self bias, that mu is might more than that, when triode wired.
So perhaps good for a line stage, but I dont think they have the "guts" for a headphone amp!
So perhaps good for a line stage, but I dont think they have the "guts" for a headphone amp!
I have a friend who has made a nice sounding DHT headphone amp by modifying a battery powered Bottlehead Quickie DHT preamp with constant current source he added a couple of inexpensive speco transformers.
The circuit which uses one 3S4 tube per channel its a very simple and nice sounding pre I have one myself but have not converted it to a headphone amp as yet.
The circuit which uses one 3S4 tube per channel its a very simple and nice sounding pre I have one myself but have not converted it to a headphone amp as yet.
Well, if you raise the anode voltage to about 120 V, the current to 8 mA and use output transformers, then it will surely work as a headphone amp. You could use mains toroids and capacitor couple to them (parafeed). I would definitely use a CCS rather than a resistor on the anode. You can use 2 stages to get enough gain.
Using what I have laying about, I rigged up a single BJT as CCS, using TIP47 (PNP Vce of 100V) and red LED, so I have retained the B+ 90V, to stay within the limits of the BJT.
Crude, but it seems to work OK.
Crude, but it seems to work OK.
If I were you, I would set up a differential pair and use your 250 V supply. Then you can vary the tail current whilst using appropriate anode resistors for that current; note what the bias is at 5, 6, 7 and 8 mA.
Of course you'll need a negative supply of about 10-15 V but any old NPNs will do in the tail.
Of course you'll need a negative supply of about 10-15 V but any old NPNs will do in the tail.
You mean a long tailed pair input stage with 2 valves?
Agreed a higher B+ would ease things somewhat WRT THD, but at the moment, without buying some transistors, I am at the limits of the silicon I have.
I could probably switch to a CC Sink for the cathode circuit, using what I have to hand.
Agreed a higher B+ would ease things somewhat WRT THD, but at the moment, without buying some transistors, I am at the limits of the silicon I have.
I could probably switch to a CC Sink for the cathode circuit, using what I have to hand.
That's why I suggested a LTP; the cathode CCS only needs low voltage transistors, BC547, BC337, anything really. Constraining Va, Ia and measuring the resultant bias, then you have your valve characterised.
I have to admit it would be the first differential stage, that I have built with Valves.
I'm struggling to see the advantage, it will take me a while...
I'd guess, for the same input signals and desired output differential signal, then gain required in each triode, is halved.
Thus THD would be reduced accordingly.
But...if I'm only using one of the differential outputs, then gain is halved & THD is the same as for a common cathode stage, with the same gain and swing.
Being as I have only just set up the CCS anode load, I might see how far I can get with that.
First tests, checking quiescent V and I, look encouraging.
Using BC237/557/107 for cathode CC sink should be easy enough to build too.
my scope has gone for repair, so FFTs will have to wait
I'm struggling to see the advantage, it will take me a while...
I'd guess, for the same input signals and desired output differential signal, then gain required in each triode, is halved.
Thus THD would be reduced accordingly.
But...if I'm only using one of the differential outputs, then gain is halved & THD is the same as for a common cathode stage, with the same gain and swing.
Being as I have only just set up the CCS anode load, I might see how far I can get with that.
First tests, checking quiescent V and I, look encouraging.
Using BC237/557/107 for cathode CC sink should be easy enough to build too.
my scope has gone for repair, so FFTs will have to wait
I'm suggesting the differential stage only as a way to easily characterise the valve (and match, if you like) since you don't have a set of triode curves for it. Careful, BC557 is PNP, I think.
Oh I see, I missed that interpretation...
I guess you could match a pair in such a way, balancing both phase outputs for amplitude - a good pair would cancel a bit of 2nd harmonic, I guess.
Or rather a good pair match would cancel almost all signal, when measured on 2 ch of a scope, with one channel inverted and added to the other.
Using CCS anode load and diode string for Vgk bias volts, seems to work alright to adjust plate voltage and current.
I could try using bypassed cathode resistor to set plate volts, but it almost doesnt seem worth the effort, for the more continuously variable range it would afford.
And yes, either BC557 or 237 is PNP, I can never recall from P/N alone.
I guess you could match a pair in such a way, balancing both phase outputs for amplitude - a good pair would cancel a bit of 2nd harmonic, I guess.
Or rather a good pair match would cancel almost all signal, when measured on 2 ch of a scope, with one channel inverted and added to the other.
Using CCS anode load and diode string for Vgk bias volts, seems to work alright to adjust plate voltage and current.
I could try using bypassed cathode resistor to set plate volts, but it almost doesnt seem worth the effort, for the more continuously variable range it would afford.
And yes, either BC557 or 237 is PNP, I can never recall from P/N alone.
Last edited:
Some magic happened.
I've been tinkering with the odd 30 mins here and there, and taking FFT for a looking at when I have time.
So I have plugged some numbers in Excel at last - eyeballing the figures and THD looks improved - but the calculated result is better than I had thought.
Shocked how much difference a cheap simple, far from ideal CCS anode load is working!
4-6mA quiescent current seems the sweet spot
Roughly speaking THD stays under 0.5% across almost all the operating range (literally, until I drive the valve into clipping the top half of the wave)
Harmonic spectrum is dominated by 3rd, at about -50dB across the span of output voltage.
All other harmonics are about -60dB down.
This is still open loop.
At 13V rms output (1.8V rms input) THD remains under 0.5%. 😱
This surprised me a little, being as at this point, the grid bias voltage is being exceeded by the signal voltage by quite a long way.
I've been tinkering with the odd 30 mins here and there, and taking FFT for a looking at when I have time.
So I have plugged some numbers in Excel at last - eyeballing the figures and THD looks improved - but the calculated result is better than I had thought.
Shocked how much difference a cheap simple, far from ideal CCS anode load is working!
4-6mA quiescent current seems the sweet spot
Roughly speaking THD stays under 0.5% across almost all the operating range (literally, until I drive the valve into clipping the top half of the wave)
Harmonic spectrum is dominated by 3rd, at about -50dB across the span of output voltage.
All other harmonics are about -60dB down.
This is still open loop.
At 13V rms output (1.8V rms input) THD remains under 0.5%. 😱
This surprised me a little, being as at this point, the grid bias voltage is being exceeded by the signal voltage by quite a long way.
Last edited:
I read somewhere that these rod valves don't produce significant grid current until grid to cathode voltage is more than a volt positive.
Well there is certainly some truth in that.
Driving about 5Vp-p into the grid, while Vgk is at 2.6V, means that I am in the region of >1V positive grid bias.
By the time I am clipping the top half cycle, I suspect the grid is 2 to 3V positive, and grid current clipping occurs.
Driving about 5Vp-p into the grid, while Vgk is at 2.6V, means that I am in the region of >1V positive grid bias.
By the time I am clipping the top half cycle, I suspect the grid is 2 to 3V positive, and grid current clipping occurs.
No, I mean 5V p-p, 1.8V RMS.
Enough drive voltage to take the grid above 1V positive.
I don't know if this is enough to cause grid currents or indeed what grid current may or may not be produced- I'm not measuring it.
Enough drive voltage to take the grid above 1V positive.
I don't know if this is enough to cause grid currents or indeed what grid current may or may not be produced- I'm not measuring it.
Today I set out to double check everything, and retest at the same operating points, to confirm all my measurements are repeatable.
Filament volts and B+ on.
Quiescent current 5mA & Vgk checked.
Then left for lunch to return after 20 minutes and check everything, take a FFT and recheck voltages.
On return I find that the quiescent current has remained solid 5mA, less than 1/20th mA change.
But anode volts have slowly drifted from the set point of 62V to about 70V.
I'm not sure if this is the valve stabilising, the diode string stabilising, the CCS adjust wirewound potentiometer or the gain of the BJT stabilising as everything comes up to a stable working temperature.
If I had to guess I'd say the BJT.
TIP47 isn't the most high bandwidth, high Hfe device, just what was immediately available.
For the currents and voltages I'm looking at, something like a Zetec E-line case PNP would work, 6mA looks to be just inside the SOAR curve for one device at reasonable voltages. Better Hfe and Vce too.
Worst news is that once warmed for 30 minutes like this, the maximum output of the CCS is reduced, where once I briefly got to 8mA I can now barely make 6mA. This is the reason I suspect the BJT.
Better news is that once warm, I can dial the bias current down, and the anode voltage responds, and remains stable.
So there is definitely something related to warm up stability and some hysteresis - I.e. x anode current, with y grid cathode voltage, does not give exactly the same result; less quiescent current, all else being as equal as I can make it, is required to drive the valve toward cutting off.
Clearly there is a gradient or curve to the DC loadline, close to horizontal but still wonky. That's the simple one BJT CCS I figure.
Filament volts and B+ on.
Quiescent current 5mA & Vgk checked.
Then left for lunch to return after 20 minutes and check everything, take a FFT and recheck voltages.
On return I find that the quiescent current has remained solid 5mA, less than 1/20th mA change.
But anode volts have slowly drifted from the set point of 62V to about 70V.
I'm not sure if this is the valve stabilising, the diode string stabilising, the CCS adjust wirewound potentiometer or the gain of the BJT stabilising as everything comes up to a stable working temperature.
If I had to guess I'd say the BJT.
TIP47 isn't the most high bandwidth, high Hfe device, just what was immediately available.
For the currents and voltages I'm looking at, something like a Zetec E-line case PNP would work, 6mA looks to be just inside the SOAR curve for one device at reasonable voltages. Better Hfe and Vce too.
Worst news is that once warmed for 30 minutes like this, the maximum output of the CCS is reduced, where once I briefly got to 8mA I can now barely make 6mA. This is the reason I suspect the BJT.
Better news is that once warm, I can dial the bias current down, and the anode voltage responds, and remains stable.
So there is definitely something related to warm up stability and some hysteresis - I.e. x anode current, with y grid cathode voltage, does not give exactly the same result; less quiescent current, all else being as equal as I can make it, is required to drive the valve toward cutting off.
Clearly there is a gradient or curve to the DC loadline, close to horizontal but still wonky. That's the simple one BJT CCS I figure.
I have noticed I have been stating incorrect part no.s for the BJT.
It's not TIP47.
I'm not sure why I wrote that (its NPN).
To correct myself, the BJT is TIP42C (which is PNP)
I've just bought a handful of 2SA2142 and ZTX558 to experiment with.
It's not TIP47.
I'm not sure why I wrote that (its NPN).
To correct myself, the BJT is TIP42C (which is PNP)
I've just bought a handful of 2SA2142 and ZTX558 to experiment with.
Well I figured out my wandering anode volts.
Filament volts.
Even a 50mV change is enough to move plate volts by about 5 Volts, if not more.
So my stabilised supply being used for filament, wound down near its minimum output, seems to be wandering around, ever so slightly.
So little that it's barely noticable, but enough to perturb the anode voltage, and fox me for a while!
Filament volts.
Even a 50mV change is enough to move plate volts by about 5 Volts, if not more.
So my stabilised supply being used for filament, wound down near its minimum output, seems to be wandering around, ever so slightly.
So little that it's barely noticable, but enough to perturb the anode voltage, and fox me for a while!