28-06-2020, 10:56 AM
(This post was last modified: 28-06-2020, 10:57 AM by Mike Watterson.)
I'd need a schematic to see what you are doing.
Note ANY amplifier or oscillator that's non-linear or driven out of a normal linear region will work as a mixer. However there are sometimes more efficient ways to use a device.
So OF COURSE it will work, if it oscillates at all. It's not taking advantage of the 1j42a other than low voltage (Shouldn't have more than 18V HT max).
After all a regular valve, rod tube transistor or FET wired as an oscillator will also always act as mixer if you also feed in RF at the grid/base/gate. The disadvantage is that a lot of the local oscillator leaks out on the RF in and it's not got good conversion gain. So a transistor tends to be wired as a common base mode oscillator (oscillator input/feed back is emitter) and the RF fed to the base. A dual gate FET is not at all like the dual grid 1j37b or 1j42a as it behaves a bit like two FETs in series, there is some isolation of signal between the gates.
Compare a single gate FET and a dual Gate FET.
This is also why g2 was used as the Local Oscillator input on Pentode Mixers before pentagrid/heptode/octode mixer oscillators developed, or Triode Hexodes.
Now a dual gate FET doesn't work the same way as a 1j42a or 1j37b. All the Rod Pentodes actually have a grid plate either side of the filament. So as those becomes more negative the dual beam of electrons emerging at both sides is "pinched" and eventually extinguished.
So someone thought (his name is documented), what if instead of internally tying the two g1 plates together we brought out an extra pin?
Thus the plates are like the X or Y plates on an electrostatic scope:
1) Move the voltage the same on both and it's like a normal rod tube.
2) Have a differential voltage on the g1a and g1b and it's like half of a 7360 beam tube. https://www.radiomuseum.org/tubes/tube_7360.html
So if you tied the two anodes together on a 7360, and drove the deflection plates suitably, it would act like a doubler or rectifier as any positive OR negative DIFFERENCE in the signal on the deflection plates puts the beam on to either anode. The 1j37b or 1j42a has the opposite characteristic.
If you have the same voltage (any -Ve to 0V where there is still anode current) on g1a and g1b, then it behaves normally.
If there is a differential voltage on g1a or g1b, then the two beams are deflected and less electrons fall on the anode plates, which are more cunningly shaped on the 1j42a, the 1j37b is similar to regular rod tubes. They used g1 because in a rod tube those are already a pair of plates very close to the filament. The g2 and g3 are electrostatic focussing, pairs of rods, so they could have split the internal connection on g2 and/or g3. But g2 and g3 are just pairs of rods and further from the centre of the beam, so any deflection would have needed a much higher differential voltage.
So while connecting up a 1j42a in ANY oscillating circuit can work as a mixer (and using a separate triodised one as a LO and driving g2 works better), the Russians never ever used it that way. A separate oscillator is always better as otherwise the RF input or strong out of the IF band signals can cause FM and thus intermodulation of the desired IF. Only cheap simple transistor superhets use a mixer/oscillator. Better ones always, like good valve communication receivers, used a separate mixer and oscillator.
Actually the disadvantage of the Pentagrid/Heptode/Octode etc mixer osc compared with a triode-pentode or triode-hexode (or separate valves) is that while either will work in the actual circuit of the other, the heptode or octode leaks LO out the RF in and also suffers from intermodulation due to a strong RF signal causing FM on the oscillator. It's why having a second tuned RF amp was often done on USA MW only domestic sets, not just for more sensitivity. Blocking LO leakage on the aerial is the main reason for an RF preamp using EF80 (almost no gain) or 1/2 of ECC85 etc on VHF.
So the Russians only used ONE circuit idea with the 1j37b or 1j42a at an RF front end.
1) Separate oscillator always.
2) Use a balanced transformer for the RF in. The LO is fed to a centre tap, so if carefully designed and laid out, there is minimal LO leakage out of the RF in.
Note if you look at battery valve DC90 or DF97 (always triodised) VHF mixer/oscillators (and often single transistor VHF mixer/oscillators) you'll see it is quite complex transformer arrangement and service information (see Philips Annette or Colette 1954 to 1958) warns of the tricky alignment to null out the LO being emitted on the aerial. It's easily received on a second set at 10.7 MHz offset from tuned frequency.
So the LO signal is equal on g1a and g2a. But with no LO, the RF would effectively be frequency doubled and somewhat rectified. There would be little actual RF at the output compared to driving both grids with the same signal. Obviously it's so non-linear that it will multiply the common mode signal (the Local Oscillator from a separate triodised oscillator).
Thus the output on the anode:
You could feed RF via capacitors to both grids and then also feed the LO to only either but only one grid. Or vice versa if you wanted a mixer / osc, but how to stop both signals driving both grids almost equally without series resistors? You'd need a transformer. So it makes sense that the transformer is tuned on the primary for RF and the LO is fed to a centre tap. Obviously with a ferrite rod aerial you could simply have a centre tap for the LO, but then you need a floating RF tuning cap, or double gang with the two fixed vanes to either end of the ferrite rod. Then you'd want a triple gang to tune the LO, so a primary on the rod to tune and then a centre tapped secondary on the same rod to feed g1a and g1b.
Almost ANY arrangement will work as an osc/mixer. However the best way so as to not either radiate the LO and have avoid strong signals doing FM intermod and appearing on the IF is the separate oscillator (only another 11 mA filament) and balanced input for RF with a centre tap.
I'll leave it as an exercise as to how you do it at 38 kHz with two valves and have an FM multiplex decoder. Grundig did manage a single valve decoder. You need a 19 kHz pilot tone filter and it is fed to the 38 kHz LO, in a fashion to cause frequency lock. Since the 1j42 nicely frequency doubles, you just need weak feedback at 38 kHz for the LO. The other 1j42a is fed with high pass filter via a balanced transformer fed at centre tap with the 38 kHz LO and a low pass filter on the output gives L-R.
Note ANY amplifier or oscillator that's non-linear or driven out of a normal linear region will work as a mixer. However there are sometimes more efficient ways to use a device.
So OF COURSE it will work, if it oscillates at all. It's not taking advantage of the 1j42a other than low voltage (Shouldn't have more than 18V HT max).
After all a regular valve, rod tube transistor or FET wired as an oscillator will also always act as mixer if you also feed in RF at the grid/base/gate. The disadvantage is that a lot of the local oscillator leaks out on the RF in and it's not got good conversion gain. So a transistor tends to be wired as a common base mode oscillator (oscillator input/feed back is emitter) and the RF fed to the base. A dual gate FET is not at all like the dual grid 1j37b or 1j42a as it behaves a bit like two FETs in series, there is some isolation of signal between the gates.
Compare a single gate FET and a dual Gate FET.
This is also why g2 was used as the Local Oscillator input on Pentode Mixers before pentagrid/heptode/octode mixer oscillators developed, or Triode Hexodes.
Now a dual gate FET doesn't work the same way as a 1j42a or 1j37b. All the Rod Pentodes actually have a grid plate either side of the filament. So as those becomes more negative the dual beam of electrons emerging at both sides is "pinched" and eventually extinguished.
So someone thought (his name is documented), what if instead of internally tying the two g1 plates together we brought out an extra pin?
Thus the plates are like the X or Y plates on an electrostatic scope:
1) Move the voltage the same on both and it's like a normal rod tube.
2) Have a differential voltage on the g1a and g1b and it's like half of a 7360 beam tube. https://www.radiomuseum.org/tubes/tube_7360.html
So if you tied the two anodes together on a 7360, and drove the deflection plates suitably, it would act like a doubler or rectifier as any positive OR negative DIFFERENCE in the signal on the deflection plates puts the beam on to either anode. The 1j37b or 1j42a has the opposite characteristic.
If you have the same voltage (any -Ve to 0V where there is still anode current) on g1a and g1b, then it behaves normally.
If there is a differential voltage on g1a or g1b, then the two beams are deflected and less electrons fall on the anode plates, which are more cunningly shaped on the 1j42a, the 1j37b is similar to regular rod tubes. They used g1 because in a rod tube those are already a pair of plates very close to the filament. The g2 and g3 are electrostatic focussing, pairs of rods, so they could have split the internal connection on g2 and/or g3. But g2 and g3 are just pairs of rods and further from the centre of the beam, so any deflection would have needed a much higher differential voltage.
So while connecting up a 1j42a in ANY oscillating circuit can work as a mixer (and using a separate triodised one as a LO and driving g2 works better), the Russians never ever used it that way. A separate oscillator is always better as otherwise the RF input or strong out of the IF band signals can cause FM and thus intermodulation of the desired IF. Only cheap simple transistor superhets use a mixer/oscillator. Better ones always, like good valve communication receivers, used a separate mixer and oscillator.
Actually the disadvantage of the Pentagrid/Heptode/Octode etc mixer osc compared with a triode-pentode or triode-hexode (or separate valves) is that while either will work in the actual circuit of the other, the heptode or octode leaks LO out the RF in and also suffers from intermodulation due to a strong RF signal causing FM on the oscillator. It's why having a second tuned RF amp was often done on USA MW only domestic sets, not just for more sensitivity. Blocking LO leakage on the aerial is the main reason for an RF preamp using EF80 (almost no gain) or 1/2 of ECC85 etc on VHF.
So the Russians only used ONE circuit idea with the 1j37b or 1j42a at an RF front end.
1) Separate oscillator always.
2) Use a balanced transformer for the RF in. The LO is fed to a centre tap, so if carefully designed and laid out, there is minimal LO leakage out of the RF in.
Note if you look at battery valve DC90 or DF97 (always triodised) VHF mixer/oscillators (and often single transistor VHF mixer/oscillators) you'll see it is quite complex transformer arrangement and service information (see Philips Annette or Colette 1954 to 1958) warns of the tricky alignment to null out the LO being emitted on the aerial. It's easily received on a second set at 10.7 MHz offset from tuned frequency.
So the LO signal is equal on g1a and g2a. But with no LO, the RF would effectively be frequency doubled and somewhat rectified. There would be little actual RF at the output compared to driving both grids with the same signal. Obviously it's so non-linear that it will multiply the common mode signal (the Local Oscillator from a separate triodised oscillator).
Thus the output on the anode:
- Some of any RF input.
- Lots of the L.O. especially when there is no RF.
- Lots of RF - LO
- Lots of LO + RF
- Lots of LO - RF
- Some DC component of the AM of the RF
- RF x2
You could feed RF via capacitors to both grids and then also feed the LO to only either but only one grid. Or vice versa if you wanted a mixer / osc, but how to stop both signals driving both grids almost equally without series resistors? You'd need a transformer. So it makes sense that the transformer is tuned on the primary for RF and the LO is fed to a centre tap. Obviously with a ferrite rod aerial you could simply have a centre tap for the LO, but then you need a floating RF tuning cap, or double gang with the two fixed vanes to either end of the ferrite rod. Then you'd want a triple gang to tune the LO, so a primary on the rod to tune and then a centre tapped secondary on the same rod to feed g1a and g1b.
Almost ANY arrangement will work as an osc/mixer. However the best way so as to not either radiate the LO and have avoid strong signals doing FM intermod and appearing on the IF is the separate oscillator (only another 11 mA filament) and balanced input for RF with a centre tap.
I'll leave it as an exercise as to how you do it at 38 kHz with two valves and have an FM multiplex decoder. Grundig did manage a single valve decoder. You need a 19 kHz pilot tone filter and it is fed to the 38 kHz LO, in a fashion to cause frequency lock. Since the 1j42 nicely frequency doubles, you just need weak feedback at 38 kHz for the LO. The other 1j42a is fed with high pass filter via a balanced transformer fed at centre tap with the 38 kHz LO and a low pass filter on the output gives L-R.
