So if the Gilbert cell is simply a differential pair on top of a
constant current source, and you are advocating only the differential
pair, then I suppose you are advocating the equivalent of a Gilbert
cell with no current source.

Someone had mentioned that I should be using a silicon diode mixer, but that's
not the point... I want my designs to be all battery tubes (plate 25-60
volt), so that the gear can be portable and also withstand electromagnetic
pulse.

The 7360 and 6AR8 require too much plate voltage.

Now from what I understand, the passive double balanced mixer has the best
port isolation, which makes it superior to the Gilbert cell for avoiding
spurs. On the other hand, the Gilbert cell has conversion gain but is more
vulnerable to spurs. So I wonder if the better answer is to build a DBM in
glass, or the differential pair?

The Eternal Squire




(Avery Fineman) wrote in message ...
In article ,
(The Eternal Squire) writes:

Has anyone ever implemented a gilbert cell mixer using valves instead
of FETs?
I'm considering this instead of using the increasingly rare and costly
heptode mixer.

To do this, one needs a minimum of three triodes, the top pair
being (essentially) a differential amplifier, the bottom being a
configured constant-current source replacing a moderately-
high common cathode resistor for the differential pair.

That's a LOT of circuit work where a single dual triode could
(and has) work just as well. Connect it as a differential pair
and put the signal in one side, the LO in the other.

Any valve that runs its control grid into the positive region is
going to be operating in a non-linear region and will therefore
"mix" well enough to do some heterodyning.

The name "Gilbert cell" got there in later integrated circuit
times to describe a particular arrangement of BJT junctions
to do mixing or AGC actions. Valve circuitry had other
names and worked for decades as mixers quite will without
fancy names. :-)

In article ,
(The Eternal Squire) writes:

So if the Gilbert cell is simply a differential pair on top of a
constant current source, and you are advocating only the differential
pair, then I suppose you are advocating the equivalent of a Gilbert
cell with no current source.

I wan't "advocating" anything, just stating generalities.

Mixers MUST be non-linear in order to do the mixing.

Someone had mentioned that I should be using a silicon diode mixer, but that's
not the point... I want my designs to be all battery tubes (plate 25-60
volt), so that the gear can be portable and also withstand electromagnetic
pulse.

Okay, then use the 1R5 pentagrid and be done with it. That
worked fine for Motorola and Hallicrafters in the old days.

Lacking that humongous EMP simulator, I don't know how you
are going to check the EMP-withstanding qualities you want. :-)

The 7360 and 6AR8 require too much plate voltage.

I never mentioned those.

Now from what I understand, the passive double balanced mixer has the best
port isolation, which makes it superior to the Gilbert cell for avoiding
spurs. On the other hand, the Gilbert cell has conversion gain but is more
vulnerable to spurs. So I wonder if the better answer is to build a DBM in
glass, or the differential pair?

1. You've never outlined the necessity of the double-balance in a
mixer. The non-balanced type has worked fine in the original
WW2 "handie-talkie" and on into the BC-1000 VHF manpack
transceiver and lots of battery-operated consumer radios.
Unbalanced mixers were used in the Korean War era PRC-8
series using subminiature battery tubes. For both the Tx and
Rx sections. Also the PRC-6 handy-talky, also VHF.

2. A balanced mixer of any kind is not necessarily a relief from
spurious responses. The choice of frequencies to mix will do
that...for any mixer type. Note: The intermodulation products
are a different situation and depend on the characteristics of
the mixer.

3. I'm not convinced that "battery tubes" wil "withstand an EMP."
It's become an urban myth that "all solid-state electronics is
destroyed by EMP but tubes/valves miraculously survive."
Not absolutely true...but I can't quibble with urban myths so
I've just met the MIL STDs with attention to detail on the
probable EMP effects which then passed the EMP simulator.

4. Designing a circuit using battery powered, directly-heated
filaments as a differential pair is going to be difficult...unless you
have a separate "A" battery supply for that differential pair.
Since the cathodes ARE the filaments, not separate as in
indirectly-heated tubes, those cathode-filaments are going to
be elevated or, if run near common, will require a "B-" supply
for the long-tailed pair's large "cathode" resistor.

5. Battery packs are almost in the unobtanium category except
for the single, lower voltage variety. You could use DC-DC
converters but those are now all solid-state and that doesn't
meet the "EMP requirement." Electro-mechanical vibrators
could generate the higher B+ (or B-) but those are terribly
inefficient, short-lived, and get bulky with transformers that
must be at low AC frequencies. Primary batteries such as
the carbon-zinc variety don't last long, maybe several years
if kept very cold to slow down the internal chemistry...all those
being made 30 to 40 years ago are now NG.

6. You CAN use techniques for suppressing ESD (electrostatic
discharge) to protect from EMP effects, then go ahead and
work with solid-state devices with some assurance of
surviveability. But, you MUST know the EMP characteristics
and do a thorough design task analysis on every part. Anyone
using battery-filament tubes should do the same thing although
I haven't any idea if anyone has done that.

Avery Fineman wrote:

In article ,
(The Eternal Squire) writes:


-- snip --

4. Designing a circuit using battery powered, directly-heated
filaments as a differential pair is going to be difficult...unless you
have a separate "A" battery supply for that differential pair.
Since the cathodes ARE the filaments, not separate as in
indirectly-heated tubes, those cathode-filaments are going to
be elevated or, if run near common, will require a "B-" supply
for the long-tailed pair's large "cathode" resistor.

-- snip --


I agree with everything else, but at RF you can just make a
bifilar-wound filament choke -- you'll be at DC ground but you can
inject whatever signal you want. Since everything is so high impedance
and you want low power you can make it the final inductor in a matching
network, for that matter.

--

Tim Wescott
Wescott Design Services
http://www.wescottdesign.com

Avery Fineman wrote:

In article ,
(The Eternal Squire) writes:


-- snip --

4. Designing a circuit using battery powered, directly-heated
filaments as a differential pair is going to be difficult...unless you
have a separate "A" battery supply for that differential pair.
Since the cathodes ARE the filaments, not separate as in
indirectly-heated tubes, those cathode-filaments are going to
be elevated or, if run near common, will require a "B-" supply
for the long-tailed pair's large "cathode" resistor.

-- snip --


I agree with everything else, but at RF you can just make a
bifilar-wound filament choke -- you'll be at DC ground but you can
inject whatever signal you want. Since everything is so high impedance
and you want low power you can make it the final inductor in a matching
network, for that matter.

--

Tim Wescott
Wescott Design Services
http://www.wescottdesign.com

In article ,
(The Eternal Squire) writes:

So if the Gilbert cell is simply a differential pair on top of a
constant current source, and you are advocating only the differential
pair, then I suppose you are advocating the equivalent of a Gilbert
cell with no current source.

I wan't "advocating" anything, just stating generalities.

Mixers MUST be non-linear in order to do the mixing.

Someone had mentioned that I should be using a silicon diode mixer, but that's
not the point... I want my designs to be all battery tubes (plate 25-60
volt), so that the gear can be portable and also withstand electromagnetic
pulse.

Okay, then use the 1R5 pentagrid and be done with it. That
worked fine for Motorola and Hallicrafters in the old days.

Lacking that humongous EMP simulator, I don't know how you
are going to check the EMP-withstanding qualities you want. :-)

The 7360 and 6AR8 require too much plate voltage.

I never mentioned those.

Now from what I understand, the passive double balanced mixer has the best
port isolation, which makes it superior to the Gilbert cell for avoiding
spurs. On the other hand, the Gilbert cell has conversion gain but is more
vulnerable to spurs. So I wonder if the better answer is to build a DBM in
glass, or the differential pair?

1. You've never outlined the necessity of the double-balance in a
mixer. The non-balanced type has worked fine in the original
WW2 "handie-talkie" and on into the BC-1000 VHF manpack
transceiver and lots of battery-operated consumer radios.
Unbalanced mixers were used in the Korean War era PRC-8
series using subminiature battery tubes. For both the Tx and
Rx sections. Also the PRC-6 handy-talky, also VHF.

2. A balanced mixer of any kind is not necessarily a relief from
spurious responses. The choice of frequencies to mix will do
that...for any mixer type. Note: The intermodulation products
are a different situation and depend on the characteristics of
the mixer.

3. I'm not convinced that "battery tubes" wil "withstand an EMP."
It's become an urban myth that "all solid-state electronics is
destroyed by EMP but tubes/valves miraculously survive."
Not absolutely true...but I can't quibble with urban myths so
I've just met the MIL STDs with attention to detail on the
probable EMP effects which then passed the EMP simulator.

4. Designing a circuit using battery powered, directly-heated
filaments as a differential pair is going to be difficult...unless you
have a separate "A" battery supply for that differential pair.
Since the cathodes ARE the filaments, not separate as in
indirectly-heated tubes, those cathode-filaments are going to
be elevated or, if run near common, will require a "B-" supply
for the long-tailed pair's large "cathode" resistor.

5. Battery packs are almost in the unobtanium category except
for the single, lower voltage variety. You could use DC-DC
converters but those are now all solid-state and that doesn't
meet the "EMP requirement." Electro-mechanical vibrators
could generate the higher B+ (or B-) but those are terribly
inefficient, short-lived, and get bulky with transformers that
must be at low AC frequencies. Primary batteries such as
the carbon-zinc variety don't last long, maybe several years
if kept very cold to slow down the internal chemistry...all those
being made 30 to 40 years ago are now NG.

6. You CAN use techniques for suppressing ESD (electrostatic
discharge) to protect from EMP effects, then go ahead and
work with solid-state devices with some assurance of
surviveability. But, you MUST know the EMP characteristics
and do a thorough design task analysis on every part. Anyone
using battery-filament tubes should do the same thing although
I haven't any idea if anyone has done that.