Hi Susan,
used a signal generator and scope to check the phase of all the windings. Have now got a working Zeus. Thanks very much for your help.
I'm just starting to do a few measurements. With 32V rails, a 50ohm source, 32 ohm load what voltage gain would you expect the Zeus75 to have? I've wired the input transformer in series and the output has the 4:1 configuration, i.e. four output windings in parallel.
Thanks,
James
used a signal generator and scope to check the phase of all the windings. Have now got a working Zeus. Thanks very much for your help.
I'm just starting to do a few measurements. With 32V rails, a 50ohm source, 32 ohm load what voltage gain would you expect the Zeus75 to have? I've wired the input transformer in series and the output has the 4:1 configuration, i.e. four output windings in parallel.
Thanks,
James
JFET Heat Sink
It looks like the low temp. solder will work just fine for bonding these little packages to a more substantial heat sink. I used Tix solder (an indium alloy) and Tix flux. It's sold here by jewelery making supply houses, gunsmith supplies, and some other hobby suppliers.
I cut some squares of sheet copper about 3mm thick to the dimensions of a T0-247 package. I brushed a bit of flux in the center of the piece and clamped across the piece in a vice. The surface was facing up at this point. I put a small u-shaped piece of the Tix, a little less than 1cm total on the copper. I carefully tinned the back of the JFET with the Tix. I only had to touch the iron to the piece, as the solder melts at about 135 degrees C. I set the FET on top of the solder U, and applied a pencil torch to the back of the copper. As soon as the FET started to move, I pushed it down on the copper and removed the torch. It hardened before my finger got too hot - within seconds. The bond looks good and tight.
Sheldon
It looks like the low temp. solder will work just fine for bonding these little packages to a more substantial heat sink. I used Tix solder (an indium alloy) and Tix flux. It's sold here by jewelery making supply houses, gunsmith supplies, and some other hobby suppliers.
I cut some squares of sheet copper about 3mm thick to the dimensions of a T0-247 package. I brushed a bit of flux in the center of the piece and clamped across the piece in a vice. The surface was facing up at this point. I put a small u-shaped piece of the Tix, a little less than 1cm total on the copper. I carefully tinned the back of the JFET with the Tix. I only had to touch the iron to the piece, as the solder melts at about 135 degrees C. I set the FET on top of the solder U, and applied a pencil torch to the back of the copper. As soon as the FET started to move, I pushed it down on the copper and removed the torch. It hardened before my finger got too hot - within seconds. The bond looks good and tight.
Sheldon
Hi Sheldon,
See your making progress with the Lovoltech fets - how will you fix the jfet with copper pad attached to a heatsink - using bolts ? - is there enough room for bolt holes in the copper?
No deformation of copper took place in the heating - jfet is flat to copper surface? I assume you tested fets after & they are ok.
I am waiting for my fets & will be keeping an eye on your progress - please post here your steps so we can all benefit from your experiences.
Regards
John
See your making progress with the Lovoltech fets - how will you fix the jfet with copper pad attached to a heatsink - using bolts ? - is there enough room for bolt holes in the copper?
No deformation of copper took place in the heating - jfet is flat to copper surface? I assume you tested fets after & they are ok.
I am waiting for my fets & will be keeping an eye on your progress - please post here your steps so we can all benefit from your experiences.
Regards
John
jkeny said:
The copper is thick, so no way it will bend. I ground the surface flat before I soldered too. The copper is about as thick as a regular T0-247 package, and similar dimensions, so I can use a standard mica insulator. I'll drill a hole in the copper so that a standard insulating washer fits and mount the assembly just like one would mount a T0-247 power transistor. So far, I just tested resistances. No change from before soldering.,
Sheldon
Hi Susan, thanks a lot for sharing your excellent and easy-to-build design! One question, if I were to build an amplifier system including the PP line drive and the power stage, fed by a CD player, where is the best place to put in a volume pot in your design? Your opinion?
Best Regards
Mark
Best Regards
Mark
Hi James,
Simplest setup:
HP8903B at 1KHz 50 ohm source impedance
Input transformer 1+1:10+10 Sowter
Output Transformer 2:1 and 4:1 configurations
Load 8 ohms.
Set level of HP = 1.00 Vac
Across input transformer primaries - series = 0.955 Vac
Across input transformer secondaries = 9.47 Vac
Output transformer 2:1 = 4.21 Vac
Output transformer 4:1 = 2.29 Vac
Notes:
======
50 ohm output impedance results in small signal drop, however this source impedance is necessary to match the Rterm value used across the input transformer's secondaries. Mismatch here results in either restricted bandwidth or peaking just before rolloff.
Input transformer gain is 10x across both secondaries. Using headphone type driver with primaries in parallel the gain here is doubled (Rterm may need to be changed).
Output transformer "gain" is 2:1 or 4:1 but actual output is reduced somewhat by the load and the follower "less than unity" (which is one of the key stabilization mechanisms of the amplifier when driving the reactive speaker load).
I use an 8 ohm load, for 32 ohms the reduction will be less.
Hope this helps.
Best wishes,
Susan.
nemestra said:
Simplest setup:
HP8903B at 1KHz 50 ohm source impedance
Input transformer 1+1:10+10 Sowter
Output Transformer 2:1 and 4:1 configurations
Load 8 ohms.
Set level of HP = 1.00 Vac
Across input transformer primaries - series = 0.955 Vac
Across input transformer secondaries = 9.47 Vac
Output transformer 2:1 = 4.21 Vac
Output transformer 4:1 = 2.29 Vac
Notes:
======
50 ohm output impedance results in small signal drop, however this source impedance is necessary to match the Rterm value used across the input transformer's secondaries. Mismatch here results in either restricted bandwidth or peaking just before rolloff.
Input transformer gain is 10x across both secondaries. Using headphone type driver with primaries in parallel the gain here is doubled (Rterm may need to be changed).
Output transformer "gain" is 2:1 or 4:1 but actual output is reduced somewhat by the load and the follower "less than unity" (which is one of the key stabilization mechanisms of the amplifier when driving the reactive speaker load).
I use an 8 ohm load, for 32 ohms the reduction will be less.
Hope this helps.
Best wishes,
Susan.
Hi Mark,
Thanks for your interest.
For the PP line driver are you are thinking of using the Sowter 9063, or some other transformer? The Sowters are expensive but they are a huge improvement over the previous transformers which in themselves were good.
The VMOS PP was envisioned for people who already have a favored preamp, probably tube, perhaps an Aikido or some such.
If the CD player has 24 bit DACs can you use digital attenuation (not recommended for lower resolution parts)?
Otherwise in the PP circuit one can use a balanced attenuator per channel (switched resistors to maintain the accuracy of the two phases).
Something like the DACT switches (which they refer to as balanced stereo)...
http://www.dact.com/html/attenuators.html
... probably the 500K version to give a 60K input impedance.
Not sure how the matching or the high impedance is going to affect the performance as I have not tried this particular configuration.
One could try with a conventional ganged pot to check on the basic operation and ensure bandwidth is not affected (I only have 20K volume pots to hand but can try some fixed resistor divider combinations).
Putting a pot in series with the input would also work as a volume control, but the rising impedance will affect the bandwidth (however this may still be enough particularly as you are using CD as the source).
Other option is SE and switched attenuator, I hope to have a chance to try out the VMOS parts in this configuration next weekend.
Best wishes,
Susan.
Zhan said:
Thanks for your interest.
For the PP line driver are you are thinking of using the Sowter 9063, or some other transformer? The Sowters are expensive but they are a huge improvement over the previous transformers which in themselves were good.
The VMOS PP was envisioned for people who already have a favored preamp, probably tube, perhaps an Aikido or some such.
If the CD player has 24 bit DACs can you use digital attenuation (not recommended for lower resolution parts)?
Otherwise in the PP circuit one can use a balanced attenuator per channel (switched resistors to maintain the accuracy of the two phases).
Something like the DACT switches (which they refer to as balanced stereo)...
http://www.dact.com/html/attenuators.html
... probably the 500K version to give a 60K input impedance.
Not sure how the matching or the high impedance is going to affect the performance as I have not tried this particular configuration.
One could try with a conventional ganged pot to check on the basic operation and ensure bandwidth is not affected (I only have 20K volume pots to hand but can try some fixed resistor divider combinations).
Putting a pot in series with the input would also work as a volume control, but the rising impedance will affect the bandwidth (however this may still be enough particularly as you are using CD as the source).
Other option is SE and switched attenuator, I hope to have a chance to try out the VMOS parts in this configuration next weekend.
Best wishes,
Susan.
Hi Alfonso,
Okay, looking at Fig 53 one should add a 25 ohm resistor in series with each output as per Fig 2 to decouple the load capacitance from the feedback.
For the Sowter 1+1:10+10 transformers most standard op-amps don't have the drive capability even though they might nominally be shown driving 600 ohm circuits in their app notes.
I use OP275s which work although the bass is a bit light (which wasn't a problem as I was driving the low bass separately).
But the OPA552 has high drive capability so should work okay.
From the TI/BB datasheet I note that the OPA552 is designed for use in configurations with gains of 5 or greater - from which I assume they are a little "twitchy".
I have just got my hands on some of Nat Semi's super dupa new LM4562NA opamps which I will try out this weekend.
Otherwise the TI headphone driver works well particularly if using toroid transformers for experimenting.
I must admit to still be a little uncertain as to your exact implementation, is it possible for you to show us a schematic?
Thanks.
Best wishes,
Susan.
P.S. I have ordered a pair of AD815s to try out.
apelizzo said:
Okay, looking at Fig 53 one should add a 25 ohm resistor in series with each output as per Fig 2 to decouple the load capacitance from the feedback.
For the Sowter 1+1:10+10 transformers most standard op-amps don't have the drive capability even though they might nominally be shown driving 600 ohm circuits in their app notes.
I use OP275s which work although the bass is a bit light (which wasn't a problem as I was driving the low bass separately).
But the OPA552 has high drive capability so should work okay.
From the TI/BB datasheet I note that the OPA552 is designed for use in configurations with gains of 5 or greater - from which I assume they are a little "twitchy".
I have just got my hands on some of Nat Semi's super dupa new LM4562NA opamps which I will try out this weekend.
Otherwise the TI headphone driver works well particularly if using toroid transformers for experimenting.
I must admit to still be a little uncertain as to your exact implementation, is it possible for you to show us a schematic?
Thanks.
Best wishes,
Susan.
P.S. I have ordered a pair of AD815s to try out.
Hi Susan,
thanks again for your help
You are right, they are "twitchy" so I will put them away and I'll try not to make Zeus unnecessarily complicated.
I'm reviewing my approach to the driver section. One of the guidelines in building Zeus is what you mentioned at the beginning "the wire with gain" and therefore it has to be as simple as possible and also I prefer not to use "special" parts. I'll follow your design of two stages amp using 10K - 10K input transformer - pp mosfet - output/input transformer - pp mosfet - output transformer. I will use a 38VCC for this amp and so I assume that in order to drive it properly the pre-amp has to provide ~30V to the power stage. I assume to have 1Vpp signal at the input from a mixer output, the input transformer has gain 2 and the second transformer should have 1:15 or 1:20 ratio.
I would appreciate if you can correct me with this approach.
Best regards,
Alfonso
thanks again for your help
You are right, they are "twitchy" so I will put them away and I'll try not to make Zeus unnecessarily complicated.
I'm reviewing my approach to the driver section. One of the guidelines in building Zeus is what you mentioned at the beginning "the wire with gain" and therefore it has to be as simple as possible and also I prefer not to use "special" parts. I'll follow your design of two stages amp using 10K - 10K input transformer - pp mosfet - output/input transformer - pp mosfet - output transformer. I will use a 38VCC for this amp and so I assume that in order to drive it properly the pre-amp has to provide ~30V to the power stage. I assume to have 1Vpp signal at the input from a mixer output, the input transformer has gain 2 and the second transformer should have 1:15 or 1:20 ratio.
I would appreciate if you can correct me with this approach.
Best regards,
Alfonso
Hi Susan (et al.),
I'm working on a ribbon microphone concept right now, and I'm curious about how something like the Zeus line driver might work as a head/preamp connected directly to a low-impedance ribbon element (perhaps omitting the input transformer?).
Would this topology be well suited for such low-power duty? What might be its advantages/disadvantages over more standard head amps and preamps?
I'm working on a ribbon microphone concept right now, and I'm curious about how something like the Zeus line driver might work as a head/preamp connected directly to a low-impedance ribbon element (perhaps omitting the input transformer?).
Would this topology be well suited for such low-power duty? What might be its advantages/disadvantages over more standard head amps and preamps?
Zeus VMOS Line Driver
Dear All,
Sorry for the delay in replying, I have been busy working on the following having a couple of days specifically set aside to work on audio.
http://www.audiophonics.com/audiophonics-zeus-line-driver-vmos-pp-1.html
Got it up and running with a new pre-amp at ten o'clock this evening.
Will reply to outstanding questions tomorrow.
Best wishes,
Susan.
Dear All,
Sorry for the delay in replying, I have been busy working on the following having a couple of days specifically set aside to work on audio.
http://www.audiophonics.com/audiophonics-zeus-line-driver-vmos-pp-1.html
Got it up and running with a new pre-amp at ten o'clock this evening.
Will reply to outstanding questions tomorrow.
Best wishes,
Susan.
Hi Sheldon,
Yes, it weighs 28 lbs (12.7 kgs).
Yes, the little toroid is a common mode choke.
It is my intention that the 24V supply will come from the system controller (which I haven't built yet) but for the time being I am using a 24V 3A switcher (similar type to those used for laptop PCs).
Whilst I am not overly keen on using switched mode power supplies for pure audio setups so much of the other stuff (TV, DVD, VCR etc.) have them that one more isn't going to make much difference.
Also I hope to have my demo system finished soon and using switchers means a lot less weight to carry.
Of course the Line Amp will happily run off a pair of deep cycle 12V lead acid batteries.
Best wishes,
Susan.
Sheldon said:You'll get your exercise lifting that one.
Yes, it weighs 28 lbs (12.7 kgs).
Yes, the little toroid is a common mode choke.
It is my intention that the 24V supply will come from the system controller (which I haven't built yet) but for the time being I am using a 24V 3A switcher (similar type to those used for laptop PCs).
Whilst I am not overly keen on using switched mode power supplies for pure audio setups so much of the other stuff (TV, DVD, VCR etc.) have them that one more isn't going to make much difference.
Also I hope to have my demo system finished soon and using switchers means a lot less weight to carry.
Of course the Line Amp will happily run off a pair of deep cycle 12V lead acid batteries.
Best wishes,
Susan.
Hi Alfonso,
Your approach is fine.
Don't forget that with the followers the actual gain is a little less per stage than the unloaded turns ratio for the transformers.
I assume you are thinking of an integrated two stage Zeus with a middle PP step up transformer rather than two separate stages?
Whilst this is simpler (on paper) the trick is getting both wide bandwidth and high step up ratios in the single transformer as one is dealing with relatively large inter-winding capacitances.
This requires a complex multi-interleaved transformer with both halves identically matched.
At the moment I am working with 3 primary and 4 secondary sections per side on a 1-1/4" x 78 dual chamber bobbin.
Using the VMOS fets I am only getting about 63kHz which is a little on the low side for a modern system. I will be retesting with standard NMOS types which have a lower impedance and this may be enough to get up over 100 kHz.
For first stage power (assuming a standard 50/60 c/s bridge rectifier supply) one can derive the half voltage for the first stage using a mains transformer with secondaries in series and adding a single diode to the center tap.
=================
With reference to my previous post I would note that with the separate Line Driver and Power Amplifier configuration this issue is sidestepped.
On the driver side the transformer is optimized for high current wide bandwidth and is quad filar wound with a gaped EI core. (Unless one also wants to have the option of low impedance headphone driving one can use a single chamber bobbin and easily wind this by hand.)
The power stage input step-up is optimized for bandwidth and being driven from a low impedance source but is of much smaller size (and small enough to be able to use Mumetal laminations which halves measured midband distortion levels). Because the transformer's bobbin is small and relatively few turns are used the inter winding capacitance is low so bandwidth is over 200 kHz (note: this is a square law effect because its derived from surface area).
Also with the separate line driver one has a nominal 4 times gain, rather than just the 2x from the integrated version if using the same input transformer.
Perhaps I am being over pedantic about the bandwidth however I am trying to get the best possible signal out of vinyl and SACD as well as standard sources.
Hope this helps.
Best wishes,
Susan.
apelizzo said:
Your approach is fine.
Don't forget that with the followers the actual gain is a little less per stage than the unloaded turns ratio for the transformers.
I assume you are thinking of an integrated two stage Zeus with a middle PP step up transformer rather than two separate stages?
Whilst this is simpler (on paper) the trick is getting both wide bandwidth and high step up ratios in the single transformer as one is dealing with relatively large inter-winding capacitances.
This requires a complex multi-interleaved transformer with both halves identically matched.
At the moment I am working with 3 primary and 4 secondary sections per side on a 1-1/4" x 78 dual chamber bobbin.
Using the VMOS fets I am only getting about 63kHz which is a little on the low side for a modern system. I will be retesting with standard NMOS types which have a lower impedance and this may be enough to get up over 100 kHz.
For first stage power (assuming a standard 50/60 c/s bridge rectifier supply) one can derive the half voltage for the first stage using a mains transformer with secondaries in series and adding a single diode to the center tap.
=================
With reference to my previous post I would note that with the separate Line Driver and Power Amplifier configuration this issue is sidestepped.
On the driver side the transformer is optimized for high current wide bandwidth and is quad filar wound with a gaped EI core. (Unless one also wants to have the option of low impedance headphone driving one can use a single chamber bobbin and easily wind this by hand.)
The power stage input step-up is optimized for bandwidth and being driven from a low impedance source but is of much smaller size (and small enough to be able to use Mumetal laminations which halves measured midband distortion levels). Because the transformer's bobbin is small and relatively few turns are used the inter winding capacitance is low so bandwidth is over 200 kHz (note: this is a square law effect because its derived from surface area).
Also with the separate line driver one has a nominal 4 times gain, rather than just the 2x from the integrated version if using the same input transformer.
Perhaps I am being over pedantic about the bandwidth however I am trying to get the best possible signal out of vinyl and SACD as well as standard sources.
Hope this helps.
Best wishes,
Susan.
Re: 35W Output Transformer
Hi Alfonso,
Hum, not quite so easy to answer.
I chose the size to be able to run full range with a nominal power of 35 Watts into 8 ohms.
It is based on a 200 VA 50 Hz rated mains transformer (my first prototypes were actually from PSU transformer kits with the mains primaries stripped off) and used a factor of four (as a minimum) to ensure it would operate satisfactorily down to 20Hz.
At the moment I get about 12 Watts in 4:1 step down and about 40 Watts in 2:1 step down. But the distortion is factored by mosfet bias and simply clipping into the power rail.
http://www.susan-parker.co.uk/zeus-original-irfp150-pics-fft-1khz.htm
I haven't seen this transformer fully saturate and turn the signal into a square wave (although I have seen this in a toroid). EIs tend to saturate gracefully and the actual point of saturation is really the point at that the frequency and current becomes sufficient to introduce such distortion that is considered unacceptable.
I will do some tests with a 2:1 transformer as soon as I am able.
Note that with the 0.8 mm 20 AWG magnet wire the maximum bias will be about 500 mA as compared to the 750 mA I use with the 1.0 mm 18 AWG size.
Best wishes,
Susan.
Hi Alfonso,
apelizzo said:
Hum, not quite so easy to answer.
I chose the size to be able to run full range with a nominal power of 35 Watts into 8 ohms.
It is based on a 200 VA 50 Hz rated mains transformer (my first prototypes were actually from PSU transformer kits with the mains primaries stripped off) and used a factor of four (as a minimum) to ensure it would operate satisfactorily down to 20Hz.
At the moment I get about 12 Watts in 4:1 step down and about 40 Watts in 2:1 step down. But the distortion is factored by mosfet bias and simply clipping into the power rail.
http://www.susan-parker.co.uk/zeus-original-irfp150-pics-fft-1khz.htm
I haven't seen this transformer fully saturate and turn the signal into a square wave (although I have seen this in a toroid). EIs tend to saturate gracefully and the actual point of saturation is really the point at that the frequency and current becomes sufficient to introduce such distortion that is considered unacceptable.
I will do some tests with a 2:1 transformer as soon as I am able.
Note that with the 0.8 mm 20 AWG magnet wire the maximum bias will be about 500 mA as compared to the 750 mA I use with the 1.0 mm 18 AWG size.
Best wishes,
Susan.
Hi Bill,
I am working (amongst everything else) on a RIAA preamp with the intention to incorporate Zeus follower principles.
I don't see any reason why your application can't benefit.
A big advantage in using the Zeus configuration is the very low noise level as all amplification is done with wire.
However the downside is the necessity of the transformers which will add bulk.
What is the basic sensitivity i.e. what is typical mV or uV output for average sound pressure levels?
Best wishes,
Susan.
Bill F. said:
I am working (amongst everything else) on a RIAA preamp with the intention to incorporate Zeus follower principles.
I don't see any reason why your application can't benefit.
A big advantage in using the Zeus configuration is the very low noise level as all amplification is done with wire.
However the downside is the necessity of the transformers which will add bulk.
What is the basic sensitivity i.e. what is typical mV or uV output for average sound pressure levels?
Best wishes,
Susan.
Thanks for the reply, Susan.
One of the advantages of ribbon microphones is that they usually need only a step-up transformer in the signal path before the preamp, as opposed to the rather complex head amps built into most high-end condenser microphones. (Here's a Neumann U67 for example.)
However, even after the transformer, ribbon mikes generally are putting out something on the order of -54dBv referenced to 1v/pa, so they need a preamp with a good deal of gain. Obviously then, a low-noise preamp is very important.
The Zeus follower appears to harmonize with the transformer-for-gain paradigm of the ribbon microphone, and if a preamp is going to be required, I'd like to put it ahead of the cabling. That's what got me wondering about how it might work if fed directly from the ribbon element. Bulk is not a factor in my application.
I'm shooting for the highest possible sensitivity in my design, and I'm anticipating something in the neighborhood of -55dBv re. 1v/pa straight off a ~1-ohm ribbon element before any transformer gain.
(Edit: To answer your question more directly, that's about 2mV in a 94dB soundfield)
Thanks for any guidance you can give.
Bill
One of the advantages of ribbon microphones is that they usually need only a step-up transformer in the signal path before the preamp, as opposed to the rather complex head amps built into most high-end condenser microphones. (Here's a Neumann U67 for example.)
However, even after the transformer, ribbon mikes generally are putting out something on the order of -54dBv referenced to 1v/pa, so they need a preamp with a good deal of gain. Obviously then, a low-noise preamp is very important.
The Zeus follower appears to harmonize with the transformer-for-gain paradigm of the ribbon microphone, and if a preamp is going to be required, I'd like to put it ahead of the cabling. That's what got me wondering about how it might work if fed directly from the ribbon element. Bulk is not a factor in my application.
I'm shooting for the highest possible sensitivity in my design, and I'm anticipating something in the neighborhood of -55dBv re. 1v/pa straight off a ~1-ohm ribbon element before any transformer gain.
(Edit: To answer your question more directly, that's about 2mV in a 94dB soundfield)
Thanks for any guidance you can give.
Bill
Hi Bill,
Hum, yes I see the complexity.
Similar to a moving coil cartridge.
Okay, that's not too dissimilar to the phono cartridges I am designing for, although to start with I am working with a custom moving magnet one (made specifically for me).
If I understand correctly I would see no reason not to treat the ribbon as the first transformer element and feed the gates directly (I assume you are thinking of PP rather than SE). After all it is in itself a transformer mechanical acoustic to electrical energy.
However I also see no reason not to exploit the very low impedance of the ribbon element and the very high impedance of the mosfets and feed a suitable step up transformer where the Zeus configuration allows for optimization of the transformer characteristics.
Sowter have a 1:40 part (although without a center tap which would have to be faked with a pair of resistors) - # 8074 1:40 0.25 ohm Ribbon microphone Transformer.
I suspect that the opportunity to get in some substantial "noise free" gain right at the beginning is the best way to go.
Where you get the bonus is in then using the zeus stage to give another round of ultra low noise amplification so the signal should be sufficiently large to be processed normally.
What bandwidth are you looking for?
The Sowter # 8074 is quoted as having a bandwidth of 10 Hz to 50 kHz, but in my experience with the VMOS fets this might go higher.
Would this be enough?
Best wishes,
Susan.
Bill F. said:
Hum, yes I see the complexity.
Similar to a moving coil cartridge.
Okay, that's not too dissimilar to the phono cartridges I am designing for, although to start with I am working with a custom moving magnet one (made specifically for me).
If I understand correctly I would see no reason not to treat the ribbon as the first transformer element and feed the gates directly (I assume you are thinking of PP rather than SE). After all it is in itself a transformer mechanical acoustic to electrical energy.
However I also see no reason not to exploit the very low impedance of the ribbon element and the very high impedance of the mosfets and feed a suitable step up transformer where the Zeus configuration allows for optimization of the transformer characteristics.
Sowter have a 1:40 part (although without a center tap which would have to be faked with a pair of resistors) - # 8074 1:40 0.25 ohm Ribbon microphone Transformer.
I suspect that the opportunity to get in some substantial "noise free" gain right at the beginning is the best way to go.
Where you get the bonus is in then using the zeus stage to give another round of ultra low noise amplification so the signal should be sufficiently large to be processed normally.
What bandwidth are you looking for?
The Sowter # 8074 is quoted as having a bandwidth of 10 Hz to 50 kHz, but in my experience with the VMOS fets this might go higher.
Would this be enough?
Best wishes,
Susan.