Practice Modulator: Mic Pre-Amp

[W9ZSL]

OK.  Got the ground bus.  Thanks for reminding me!  Makes sense grounding at the input.  I have a .22 for the screen bypass.  I can handle that just fine.  Main question. 

I think I can put the Triad A-9J to much-better use.  It's a perfect mic input iron not line.  I have a super-quality Electro-Voice dynamic mic made for speech/vocal.  The A-9J has a 250 Ohm primary tap.  Secondary is 85K.  Response is 30-15Khz.

Given the ratio, would I be better off eliminating the 6J5 entirely and substituting a 12AX7 two-stage and feed the 6AG7 with it?  No crystal mic.  Dynamic.

[AB2EZ]

Mike

Hi!

I think the .001uF capacitor between the plate of the 6SJ7 and ground is going to roll off the high frequency response of the 6SJ5 stage... starting at around 1060Hz (3dB roll off point).

The |impedance| of that capacitor is around 150k ohms at 1060Hz... which is equal to the value of the 220k ohm 6SJ5 load resistor in parallel with the 470k ohm 6V6 grid leak resistor.

Correction:

On further thought... with the negative feedback provided by the 750k ohm resistor... the net effect is to make the input to the 6V6 look like a relatively low impedance. The audio frequency plate current of the 6SJ7 gets multiplied by the 750k ohm "transimpedance" to produce the output voltage of the 6V6, ... provided the open loop voltage gain of the 6V6 stage is significantly greater than (1/ the feedback ratio) = approximately: (f /1060Hz) x 750k ohms/150k ohms = (f/1060 Hz) x 5

The open loop voltage gain of the 6V6 is in the ballpark of 20 with a 5k ohm load impedance. In this design, the load impedance presented by the primary of the driver transformer will be around 5kohms... maybe a little more at lower audio frequencies... and a little less (and also reactive) at audio frequencies above 5kHz, with the .01uF capacitors you plan to put between each side of the secondary of the driver transformer and ground.  

So the frequency response of the 6SJ7 stage in combination with the 6V6 stage is flat to a frequency f that is somewhat greater than 1060Hz ... but not that much greater. Maybe something like a 2500Hz 3dB rolloff frequency.

Bottom line: I think you will want to decrease the capacitance of the capacitor whose value is shown as .001uF to something like 330pF.

Stu

[AB2EZ]

I decided to simulate the pre-amp using LTSpiceIV.

Its bandwidth is greater than I estimated in my post above.

Attached Slide 1 shows the schematic. I set the transconductance of the 6SJ7 to 1650 micromhos. I set the transconductance of the 6V6 to 4100 micromhos.

For the simulation, the input voltage to the 6SJ7 grid has a 1V amplitude.

Slide 2 shows the frequency response from the input of the 6SJ7 to the output of the 6V6. The voltage gain at low frequencies is 59.6dB. The 3dB rolloff frequency is approximately 5.6KHz (where the gain is 56.6dB). Changing C1 from 0.001uF to 0.0005uF extends the 3dB rolloff frequency to approximately 11.2kHz.

Slide 3 shows the plate current (in dB relative to 1 ampere) of the 6SJ7 (green curves) and the current flowing through the 750,000 transimpedance resistor.

At low frequencies, the current through the transimpedance resistor is about 1.9dB lower than the plate current of the 6SJ7...because the 6V6 stage is not acting as an ideal transimpedance amplifier due to its limited open loop voltage gain (which is 4100 micromhos x 5000 ohms = 20.5)

At 5.6kHz, the current through the 750,000 ohm transimpedance resistor has dropped by around 3dB, while the plate current of the 6SJ7 has not dropped (as expected). This reflects the decreasing feedback ratio due to the decreasing impedance of the 0.001uF capacitor.

Note that 20log(1650 micromhos x 750,000 ohms) = 61.85dB. The low frequency gain of the amplifier is 59.6dB. This 2.25dB difference corresponds, primarily, to the portion of the 6SJ7 plate current that is not passing through the 750,000 ohm resistor... as shown in Slide 3. The remainder of the difference corresponds to the amplitude of the voltage at the input to the 6V6... which is not zero, because the 6V6 has an open loop voltage gain that is not large enough to drive its input voltage to zero.

Stu

[AB2EZ]

Note:

If you remove the 5000 ohm resistive load, and replace it with a 0.005uF capacitor (-j5000 ohms at 6366Hz)... then the simulation shows that the frequency response has a 12dB peak at 5.2kHz

See attached slides 4 and 5

Therefore, the combination of: the grid inputs of the 811's and a .01uF capacitor from each side of the secondary of the driver transformer to ground may not be the best choice for the load across the secondary of the driver transformer.

Adding the 5000 ohm resistor (R2) back into the circuit... in parallel with C2... flattens the frequency response out again. See slide 7. But, there is still more phase shift at 6kHz than there is with a purely resistive load (see my previous post).

Stu

[WD5JKO]

  Good discussion, and I like the analysis Stu.

This audio driver is very similar to what I use in my Gonset G-76 AM transceiver. This RIG uses a single 6DQ5 RF tube modulated by a pair of 6DQ6's in class B triode mode. Gonset and a few other companies used sweep tubes as modulators this way. They ground the cathode, ground Grid 1, and put AC drive to the screen grids. The advantage is no bias supply needed, and no screen supply needed. These are like little 811's or littler 809's.

A single 6CM6 is the driver. The 6CM6/6AQ5/6V6 are all in the same family with almost identical characteristics. They each have a unique base, and pinout though. I played with the Gonset design, and ended up using the plate load for the 12AX7 a combination of inverse feedback from the 6CM6 plate, and one 6DQ6 plate.

More detail, the audio driver(XMIT)/Audio output tube (RCV) V13, here I beefed up the DC operating point by reducing the cathode resistor R58 from 470 ohms to 330 ohms. The cathode bypass capacitor C98) was bumped up to 100uf as well. Since V12B and V13 are in a NFB loop together, we don't want any R-C network phase shifting going on, so I bumped up the audio coupling cap C97 to 0.1uf. The combination of higher class A operating point, and the NFB results in considerably higher grid swing to the modulator tubes before distortion sets in.

I added a NFB loop from either V14 plate or V15 plate back to V12-B pin 9 (plate). No series cap was used, and this brought up V12-B plate up from 90v to around 120v. The result was around 3db NFB. This might not be much, but the modulated envelope shape was changing somewhat from 50% up to 100%+ modulation. With a triangle wave I could see the slope change even at midband (1 Khz). This small amount of NFB made a noticeable difference on how it looks, and sounds.

My point here is that a single ended audio driver like what is presented at the beginning of this thread can do a wonderful job. I also broaden the scope a bit to show how sweep tubes triode connected the Gonset way can still provide decent quality and depth of modulation.

I show two plots of the rig running at 50% modulation. One shows the distortion products (-40db = 1%), and the other IMD. The numbers creep up a bit at 80%, and then go non linear as the negative peak limiter diodes kick in.

Jim
Wd5JKO

[AB2EZ]

Jim

Hi!

It is interesting that, in this approach, the screens of the modulator tubes provide an approximately resistive load across the output of the driver transformer.

Stu

[W9ZSL]

  Wow Stu, NICE.  I'm going to have to pick my way through all the technical stuff so it eventually sinks in!  I took inventory and need only 4 parts, one of which would be the .0005 cap for the feedback.  Rather than use the Triad A-9J to provide a balanced line input, I'm going to use it for a mic pre-amp, probably a 5763 or 12AU7 outboard to drive the 6SJ7.  Then again, if the 6J5 has enough gain, I could use it there.  I'm open to suggestions for a pre-amp but the more I think about it, the more using the Triad for a dynamic mic input makes sense.   I can make the line input unbalanced, keep the 6J5 as is so a crystal mic (EV-911) can also be used.  I DEFINITELY want the highest "FI" possible.  

Was in Broadcasting since 1961 and recording since 1966.  No doubt.  I cloned an Ampex MX-10 mixer in the day and that was my first home board.  Started as a 6 in, 2 out which eventually ended up being 8 x 2.  Everything was HB except the recorders including a Fairchild Reverb clone.  Everything sang. Sweet isn't even the word!

It used an EF-86 / 6267 mic pre-amp feeding a 6267 mixer and 1/2 of a 12AU7 per channel.  Get this, the output came off the cathodes of both, went into a 25K "master" and some matching but no xfmr on the output.  Awesome sound.  The A-9J would be a perfect match to an EF-86.  It would drive the 6SJ7.  Ampex used a whopping .33 cap followed by a 47K to the mix bus.  I may have to get a heftier power tranny.  Mine is 90 ma.

Yeah, the Triad is 30-15,000 Hz. If I clone the Ampex MX-10 pre-amp, that would sing!  No doubt!  The clone is on the first photo.  A few years later, same space and 4-track.  Note a pair of modified Ampex MX-10 mixers.  This was followed with 8-Track.  Pix 3 is what I had in the mid-90s.  I'm now 16-Track and 40-Track capable.  Lord, where will it end!

[W9ZSL]

Three more pix.  Full Compass in Middleton, WI started as a studio  I built a lot of the gear, much of the board, nearly all the cabinetry including our HB 1" 8-Track Ampex recorder.  Got such a deal on AG-350 electronics.  If you look in the far left rack, notice the white unit?  It's in the previous pix of my 90's studio.  The unit with the meter and two knobs directly under the Revox is the original mono HB limiter.  It still is in service but stereo now.

I was commissioned to build a mixer for a Minneapolis Ford dealership in-house studio shown in #2.  I can HB.  Now if only I didn't have cataracts in my right eye.  Oh well.  Gotta build some masterpieces, crap or go blind.  Saw the Doc yesterday.  The crap part is working fine, thank you.  Eye, not so much!

[W9ZSL]

Feedback cap 500 pf CERAMIC I'm sure?