Another Tube Mic Preamp

[RicharD]

Somebody make him STOP!

The 6BQ7A circuit is a scrub.  It was a good learning experiment but it had too many weaknesses.  1st off, it wasn't really quiet enough.  It had the potential to deliver devastating output levels.  The plain ordinary boot strapped cathode follower's performance swung wildly depending on output load.

>I think the 6BQ7 is under-appreciated. But it would not be my go-to for a high-level stage with high-ratio input iron.
OK... let's start with some "go to" tubes.  Let's keep it sexy and use octals.  I picked a 6SL7 and a 6SN7. 

With a LOT of help form ISOtone, we decided a new topology was in order.  Working from the output to input, the first obstacle to overcome is varying output loads.  After looking at a certain lady's schematic (that I probably shouldn't have a copy of), the White Cathode Follower seemed like a good place to start.  I opted for a symmetrical WCF over the text book version.  This delivers better output control.  The circuit display little or no difference between a 10k load and a 600 ohm load.

The 6SL7 delivers more gain than I need + I saw no need for a 2nd gain stage.  Due to my filament supply (12VDC), I found myself with an extra 6SL7, or 1/2 bottle per stage.  A little TubeCad research and we discovered that the Totem Pole SRPP circuit might be just the answer.  I can now lower the gain of the 6SL7 and improve the power supply noise rejection ratio.  I also get more input threshold, enough so that my input transformer saturates long before I I overdrive the input of V1.  The only expense is the extra triode I wanted to use anyway.

Cool!  So here is the new schematic with measured telemetry:


Here are the TubeCad design parameters:


Here are the analyzer results:

[OldHouseScott]

Dang, that looks nice, especially without any global NFB. The slight droop on the high end probably helps smooth things out as well, especially if you're tracking to digital. Did you do phantom, pad, and polarity reversal as well, or is it not that far along yet? What iron did you use, PS included?

[RicharD]

Thanks!  This is built on the protoboard at the moment.  For the sake of analysis, all of the bells and whistles are left off the circuit diagram.  I'm actually considering adding some NFB, but at the moment it seems right w/o any. 

Right now it has:
48V regulated phantom power.
12V regulated DC filaments and control voltage.
An opamp driver for the analog VU meters hijacked from JLM audio.
I/O transformers are CineMag.
PT & choke are Hammond.
(All the transformer specs are in the Dif-prE thread, the last MP3 schematic.)

Eventually it will have:
Time delayed output relay
Polarity phase switch
-20dB mic pad

Other thangs I'm considering:
Low cut switch
Output clamping
peak indicator

[PRR]

> without any global NFB.

There's a TON of 2-stage NFB hidden inside the WCF. Probably that first stage also.

> slight droop on the high end

May be the high-resistance 1Meg volume pot. When set anywhere between the ends the wiper looks like a high resistance. There's capacitance everywhere. We usually don't go over 250K for a hi-fi volume pot. Since V1 output is OTOO 13K, it could probably drive 250K just fine.

The 0.47u after the volume pot is physically large, adding stray capacitance. Since this version of WCF has 890K input impedance, 0.01u would work, 0.05u would be way-ample. 0.5u is money and excess high-sucking stray capacitance.

The WCF is not optimum, probably because the B+ is way more than it needs to be for this stage. The 15K should be 1.0-1.4 times the cathode resistor, say 1.5K. 4V peak in 600 ohms is respectable for one direct-coupled tube but not generous for studio practice. .

What would it take to get +20dBu, 11V peak? 600 ohm load, 6K plate resistance, roughly 110V across the tube, so 11V+110V idle voltage. Double that and add 10% for bias and slop, 270V B+. 11V peak in 600 is 18mA, the WCF won't get out of Class A, so the idle current must be >9mA. Safe? 110V*9mA is 1 Watt plate dissipation which is plenty safe for 6SN7. In fact we could go over 400V B+ without melting the plates (heater-cathode insulation is a problem) and get near 20V peak, +25dBu. 350V at 14mA idle gets +22dBu.

However the real overload may be in V1 (or the transformer), since the output system is all loss.

[DummyLoad]

PRR, i think that 1000K is a typo - SRPP design was based on a 100K load.

[DummyLoad]

output transformer is cinemag - CMOT-2H 

richard, for WCF try Rk 180Ω, Ra 270Ω, 1/2W package will work for both - should get you around 14.5mA w/ 16Vpp Pdiss should be around 2.5W.

[RicharD]

The volume pot is a 100k audio taper.  That is a typo.  The Vout Max listed is actually the max voltage at which the OPT is still undistorted across the audio band.  IIRC, I actually saw output in excess of 10Vrms but it HEAVILY saturates and distorts the OPT at frequencies less than 50Hz.  That value would be worth double checking when I get the board away from my PC and back in the lab.  I don't have a problem increasing the current to the WCF.  It's just across the edge of what TubeCad called insufficient plate current.

[RicharD]

>try Rk 180Ω, Ra 270Ω,

I come up with 29mA using those values.  This completely changes the values of Ra2 and Rk2.  I get a whopping .5V more output and an expense of almost 200 ohms of output impedance.  

I come up with

Ra=2k2, Rk=330, Ra2=18k, and Rk2=16k

This puts me at 20.5mA keeping my B+ at 350V.

[nateflanigan]

An octal mic pre sounds fun, I look forward to seeing how it turns out.

[DummyLoad]

What would it take to get +20dBu, 11V peak? 600 ohm load, 6K plate resistance, roughly 110V across the tube, so 11V+110V idle voltage. Double that and add 10% for bias and slop, 270V B+. 11V peak in 600 is 18mA, the WCF won't get out of Class A, so the idle current must be >9mA. Safe? 110V*9mA is 1 Watt plate dissipation which is plenty safe for 6SN7. In fact we could go over 400V B+ without melting the plates (heater-cathode insulation is a problem) and get near 20V peak, +25dBu. 350V at 14mA idle gets +22dB

we are building a symmetrical WCF not the classic textbook WCF - two entirely different circuits. PRR, plugging your values into tube cad with the classic WCF and i get very similar result to what you do - plugging the same values into a symmetrical WCF, current doubles, Vo is actually less. the symmetrical WCF breaks the WCF into into two current paths. we are mixed fruit here - apples and oranges.

[PRR]

What is a "symmetrical WCF"?

[RicharD]

>What is a "symmetrical WCF"?

I'm home with a killer migraine... I'll do my best to explain it.  I'm gonna hijack Broskie.

Here is the diagram for the "textbook" WCF:

Here is the diagram for the "syymetrical" or "constant current" WCF:

If I understand this correctly (and we all know I probably don't), the major difference is the 2 resistors Ra2 & Rk2 which set a constant DC bias voltage between the cathode of the upper tube and the plate of the lower tube.  What I believe this does is lock the output impedance regardless of load.  The trade off is double the current.  (considering my initial circuit)  Using TubeCad, when I swing Rl between 100 ohms and 1k, the output stays constant at -4/+4.2v.  Current is 7.8mA.  When I switch it over to the textbook circuit, current is now 3.9mA but my output varies wildly, -0/+0.41v @Rl=110 ohms to -4/+3.7v @Rl=1k  I assume this is why my circuit even works with Ra set so high.  BTW:  it is safe to assume I selected Ra = 2x Rp.  Probably not right for this circuit considering there are 2 triodes involved.  In retrospect, it seems like Ra should = Rp.  Running with that notion, if Ra=6k8, then I come up with a Rk=180 ohms.  The textbook circuit runs at 10.75mA and swings 12v.  The constant current circuit runs at 21.5mA and swings 14v.  As I said earlier, I'm nopt opposed to raising current, but I fear 21mA which puts me at 1.5W on the plate (50%) might be noisy.  My fears could be completely unfounded, but in previous experiments,  I've had 12AU7's hissing like angry snakes at higher currents.  My original circuit may not be optimal, but 8v of swing should be adequate, but not exactly generous.  OTOH a tube that's loafing tends to be quieter.  In this case PSRR is 7dB greater between the 15k scenario and the 6k8 scenario above.

[RicharD]

Headache = better, therefore time to experiment.

I decided to pick a middle ground between Vout and current.  Sticking with the idea Ra=Rp, I selected a 6k8 resistor (standard value below 7700 ohms).  With Rk=470 ohms, I get a current of 15mA and a Vout of 10v.  For now I'd like to stick with B+ = 350V.  Of course I had to change the voltage divider for the constant current & and power supply dropping resistor (PS not shown).  I'm still set up in the bedroom and not the lab so I'm using a DMM and my PC for my telemetry so consider Rl test to be my sound card's input impedance.  The limiting factor here is my sound card get unhappy above +3dBu, so to measure Vout max at 20Hz, I slapped a 500:500 Daven T pad set at -26dB between the output of the protoboard and the sound card input.  Of course I also disconnected the VU meters. 

frequency response is just about the same.  0 signal SPL improved but that might be due to different level settings since the previous experiment.  None the less, -74dB is quite acceptable for a hay-wired protoboard circuit.  THD got a bit wonky between the 2 channels.  I'm not sure why this happened.  I'm gonna blame it on hay-wiring and/or used tubes pulled from the bucket.  With the T-pad in place, I set the circuit wide open and ran the input signal up until the 20Hz sine wave started to distort, then backed off until clean.  This measured at 4.8Vrms or 15.8dBu.  At 1kHz, I measured >12Vrms undistorted which I called plenty.

I'd like to call this circuit good.  The only change I want to make is to replace the 100k pot with a 50k pot.  Unfortunately I do not have any 50k audio taper pots, only linear taper and linear taper pots suck in this application.

Here is the new schematic:

Here are the TubeCad calculations:

Here are the measured results:

[PRR]

> 12AU7's hissing like angry snakes at higher currents

Shouldn't be.

For low noise Voltage, the classic design is a hot triode. TV set inputs used a 12AT7 or 6BQ7 working at 8mA-15mA. In a TV store, all sets tuned to a weak station, you could see which ones cheaped-out with low current by the excess snow.

> I'd like to call this circuit good.

It's good.

It's extravagant: it bumps small signals up to BIG signals and has to knock them down again. The output stage power efficiency is under 5%. But you can afford the extravagance.

> Broskie

I knew his name would come up.

The two resistors make the _supply_ current invariant (and much higher). Broskie feels this is very important.

They also provide a minimum load; the WCF does get a bit twerkly un-loaded.

Except at the UN-loaded condition, I don't see how those bleeders affect output impedance.

I still feel the 15K-2200-6K8 plate resistor is large. (And I mis-remembered the rule of thumb.) 1/gM is the general size. The exact optimum has a correction for load. Alex did a long analysis:
http://www.cavalliaudio.com/cj/docs/WCFOptimization.pdf
Page 9 treats the case where the top resistor is "large".

If you are going to get it 'wrong', going high "works" but is wasteful. Going low turns it into a simple CF and cuts peak output current in half.

[RicharD]

>1/gM is the general size.

1/gm is 380 ohms.     With a 100 ohm Rk, current is 35mA (17.5 for a text book WCF).  I get more swing and my PT can afford the current.  Oh I see where this is going.  Vtube is now very close 1/2 B+.  That may come in very handy.  I could now drop B+ of V1 10ish volts and direct couple the 2 stages.  I lose the pot, 2 caps, and 2 resistors all in the main signal path in lieu of controlling gain with global NFB.  Can you smell the hair atop my head burning?


>I knew his name would come up.
TubeCad is cool.  I play with it as if it were a TI calculator and this was 1976.  Broskie has published some fun articles.

>The two resistors make the _supply_ current invariant (and much higher).
They're doubling the current.  EEEk, I already running them above wattage spec.  I may has misspoken about them affecting output impedance, they have noneza.  What I am seeing is no change in Vout as load changes, and Vout is locked at a considerable  lower voltage that the textbook WCF at 600 ohms, drastically less at 10k.  Going to 380/100 ohm design, the 2 resistors are about 10k at 175V.  That's 3 watts of hot resistor times 4 (stereo)...... sheesh.  Off with their head!  I'll hang a 10k off the output so there's never no load.

[PRR]

> direct couple the 2 stages.

No. You need the adjustable loss, you can't DC ground the bottom of the pot, and the cap to effectively AC-ground it would be impossibly large.

[Merlin]

1/gm is 380 ohms.   With a 100 ohm Rk...

Only the top-most resistor needs to be 1/gm. You can still control quiescent current with the bias resistors.

[RicharD]

>No. You need the adjustable loss......

I was thinking about getting rid of the pot, the 2 coupling caps, and the 2 grid bias resistors.  Yes gain is now through the roof, but I would attempt to control gain by adding a global feedback loop.  I guess this isn't feasible.

>Only the top-most resistor needs to be 1/gm. You can still control quiescent current with the bias resistors.

Yes, I understand this.  I was simply citing the 100 ohm Rk to establish a complete scenario.  This sets me at 17mA and gives me 18v of swing into 600 ohms.  If I were to go 470/470, that puts me right back where I started with 10v of swing into 600 ohms.  Apparently the resistor values of the voltage divider Broskie calls Ra2 and Rk2 are supposed to be = to the impedance of the tubes + their respective resistors.  Setting the WCF where I am supposed to makes these resistor values in the neighborhood of 10k which means they are gonna hafta blow off an excessive amount of heat.  Seems like a complete waste, but maybe I'm missing something.  For the moment, I am running 470/100 w/o the Ra2 & Rk2 resistors.  I have not had a chance to really run the circuit as of yet.

[PRR]

> these resistor values ... are gonna hafta blow off an excessive amount of heat.  Seems like a complete waste...

There's subtleties. The WCF is more complicated than it looks. It isn't push-pull for high load impedances. A dummy load is a crude fix. These DC loads respond to tube voltages and give a different action. Some day I might look at that. But it is indeed a LOT of heat, essentially doubling total current consumption and heat.

While the constant power supply drain might lessen the work of the PS, clearly the PS must be twice as hefty. Is that a good trade-off? In commercial work, it sucks. In gilded-lily DIY, it might lessen residual corruption through the rail.

> control gain by adding a global feedback loop

Well, OH Scott appreciates no GNFB (though the 2-gain-stage "local" NFB inside the WCF is astonishing). Many people favor that.

And as I guess you saw: it's inverting, the only place to aply NFB is at the input grid, and putting the NFB network between a weak source and the first grid is bound to lower input Z and raise resistance hiss. I've seen it done, but that was a PA quality mixer.

Also, while you have a bit too much gain now, I don't think you have so much too-much to maintain significant NFB at high gains. Which means it will run tight at low gain and loose at high gain. And while tastes differ, I'd want the tight sound on harpsichord and the loose sound on Fender Twin, which isn't how it works.

Alternatively you can make the second stage inverting gain, and take a loop from output plate to input cathode. That's how most classic designs did it. The 2nd stage must have enough power to overwhelm the 1st stage cathode, and you already have issues. With four triodes, you sure can make it work; that's two more than classic mike preamps used (but they didn't push 1:1 OTs either).

So while I feel it is extravagant to boost every/anything 50dB and then pot-away the excess over what your tape can swallow, I can't see anything "wrong" with the idea.

[RicharD]

>While the constant power supply drain might lessen the work of the PS, clearly the PS must be twice as hefty. Is that a good trade-off? In commercial work, it sucks. In gilded-lily DIY, it might lessen residual corruption through the rail.

My PT is rated for 100mA which is ample to keep the voltage divider in the circuit, but I see no benefit, only trouble sourcing 10W precision resistors and potential hot spots should I follow through into PCB land.  At this point, they're out of the circuit and I'm following your advice and knocking the plate resistor down to 470 ohms and using a 100 ohm Rk which ironically puts me right at the same current I had before with the 6k8 plate and the voltage divider in place.  Test results show no change in frequency response or THD.... actually THD isn't wonky on 1 channel anymore.  I now blame wonkyness on pulling too much current through the divider resistors.

>And as I guess you saw: it's inverting, the only place to aply NFB is at the input grid
Not at first.  It wasn't until I had clip leads in hand that I discovered global NFB wasn't going to happen.  I was hoping to get rid of all that RC business around the pot that is causing my HF roll off.  Even though my pot is in fact a 100k and again I followed your advice and changed the cap after the pot to a .047uF, I still have hi end roll off.  I have verified that this is in fact the source of the problem whereas when I probe the top of the pot, there is no roll off but when I probe the wiper of the pot, the roll off matches the roll off measured at the output (actually a little worse).  Putting a .01uF in it's place starts to affect the low end.  What's popping my bubble is the fact that the roll off is effectively the same for a .47 as a .047.  Using 1/2piRC, a .47 has me looking for 50ish ohms and a .047 has me looking for 500ish ohms to cause a roll off at 7kHz.  I did a little tail chasing swapping cap values to no avail.  I say tail chasing because after thinking about it, it doesn't make any sense.  Now considering the Zin of the WCF at 890k, this has me looking for 25pF.   Hmmmm.... stray capacitance on the protoboard perhaps?  The secondary cap is adjacent to the pot and then there's roughly 6" of #22 unshielded wire running up to the grid input of the WCF.  Reversing the cap and wire orientation caused no change.  It's kinda driving me nuts.

[RicharD]

Putting a 120pF across the pot sorta fixes the high end roll off.  Pretty much what I didn't wanna do was add more capacitance.  I say sorta because with the pot at 9:00, I now get a hi end bump above 10k.  10:00 to 3:00 looks good.  Dimed and I still experience some roll off.  I guess adding a bright switch isn't the end of the world.  I was already considering adding a low cut switch which of course adds a capacitor and resistor.  Looks like compromise time, but by using switches, I get choices.

The left or blue graph is brite in low out.  The pink right graph is low in hi out.

[PRR]

> Using 1/2piRC, a .47 has me looking for 50ish ohms and a .047

Not the marked value. These are physically large caps. There's dozens of pFd from the outside of the cap to the rest of the world.

There's more in the leads to/from the pot. And in the WCF grid though that should bootstrap to just a few pFd.

It would be good to have the pot RIGHT between V1 and V2 (this may not be the case on your breadboard).

Before you go nuts: measure from first GRID to WCF output. Get the transformer(s) out of the picture. Flatten, or at least document, the through-tubes response. Transformers have their own problems. Take that as a separate step. A good IT will mainly need the suggested (or empirical) termination.

[RicharD]

>It would be good to have the pot RIGHT between V1 and V2 (this may not be the case on your breadboard).

A picture is worth 1000 words. My layout leaves something to be desired.  Had I been paying attention, Ida laid it out SL-SN / SL-SN but no.  I typically pull the controls forward purely for safety, and push all the B+ to the back.  It'll be interesting to see how the measured results change from protoboard to prototype to real McCoy PCB (should I make it that far).  Prototype will happen.  I just bought my local Rat Shack outta Vector board.   

>Get the transformer(s) out of the picture.

I really should do that although I'm certain you nailed the problem in your first post regardless of my bad data.  Response measures flat at the top of the pot and drops at the wiper.  I'm experimenting with CineMag transformers so the "funky transformer" variable is low.

Now to completely change the subject, I've notice it is almost a universal practice to put the polarity phase switch in front of the input transformer.  Is there any compelling reason why one couldn't flip flop the output transformer instead?  Seems like there would be less thud when switching the output what with phantom power and amplification of the input.

[kagliostro]

Hi Richard

May be there are many reason to see the polarity phase switch at the input, reasons that I don't know

but I think that a polarity switch  in the output will decrease the wire length and put out the same switch from the input side of the signal, giving circuit less ability to "absorb" noise

Jensen, however, put the switch in the input path

http://www.jensen-transformers.com/as/as016.pdf

for those that can be interested, here their page with a lot of interesting stuff

http://www.jensen-transformers.com/apps_sc.html

Kagliostro

 

[RicharD]

>May be there are many reason to see the polarity phase switch at the input, reasons that I don't know but I think that a polarity switch  in the output will decrease the wire length

I'm thinking it's more to do with reversing the mic itself vs reversing the circuit, but I don't know.  This is why I asked.  As for wire length, there's already a long cable hanging off both ends.  Noise would be more about unshielded junctions.  I'm gonna do my polarity switches with relays.

I'm hoping to have the master schematic finished soon but I have my day job and a couple of amps to diagnose so R&D needs to pause.

[kagliostro]

Noise would be more about unshielded junctions.

I was thinking about to "absorb" noise inside the preamplifier

from the PS

a switch in the input will be more hard to shield than a wire

I'm gonna do my polarity switches with relays.

and if you use shielded relays you can dribble the shielding problem (nice)

Kagliostro

[RicharD]

>>put the polarity phase switch in front of the input transformer

It's all about changing the characteristics of the mic and not the amp.  In a perfect world, you could place a microphone anywhere you like and it would be in phase with the other mics in the room.  Unfortunately it doesn't work like that.  Having a phase switch increases your chances of mic placement being usable by 50%.  It's good practice for all your equipment to be in phase.

I copied and attached the Jensen input schematic from the link Kagliostro posted.  I am surprised that with the phantom power off, the 2 - 6k81 resistors are shunted to ground through a 200 ohm resistor.  It seems like this would effectively turn off the input and not allow you to use dynamic microphones or higher end microphones which have their own power supply.  Maybe I'm not understanding the circuit but it seems odd.  I was planning on using a DPDT relay between the 6k81 resistors and the input jack to completely eliminate any connection to the phantom power supply.

I understand the -20dB mic pad although it could become redundant whereas most better condenser mics have a -20dB pad built in.  I'm still weighing the benefits because for me, it adds more relays and LED's.  I also question the value of the impedance switch.  For the moment, both of these features are not in my working schematic.

 

[kagliostro]

In the Jensen schematic you posted their council is to twist the wires

how do you solve that in a PCB ?

other thing, maybe it is OT

have you seen this schematic ?

http://www.jensen-transformers.com/as/as088.pdf

what do you think about ?

Kagliostro

[RicharD]

>how do you solve that in a PCB ?

They're specifying twisted/shielded wires between the input jack and the input transformer.  Once you're physically on the PCB, it is prudent to make sure there is a ground plane on the opposite side of the board from all main signal paths.  I try to keep as many traces off the component side of the board as possible leaving it available for the ground plane.

The 2nd link you posted is the foundation for a typical grounded cathode differential amplifier.  I'm gonna keep running with totem pole & White cathode follower.  Lab results are awesome and I'm too far into this project to turn back now.

[PRR]

> a typical grounded cathode differential amplifier.

That will not work properly unless the transformer is MADE for this use.

The usual transformer has high stray capacitance at one end (near the center core). You wire that end to audio ground/common. The other, low-C end, goes to grid.

That normal transformer, to two grids, one grid gets full treble and the other end has treble loss. The sum will be droopy, and could even null-out.


> 2 - 6k81 resistors are shunted to ground through a 200 ohm resistor.  It seems like this would effectively turn off the input and not allow you to use dynamic microphones or higher end microphones which have their own power supply.

It adds a 13K shunt impedance across the mike. Since the mike is 100-300 ohms, it causes 0.2dB added loss and reduction of S/N. The Phantom value was choosen to cause "neligible load".

It adds 3.4K of DC leakage to ground. Dynamics don't care. A few radio-station ribbon mikes were build with grounded center-taps and won't even know there's resistors also. External power pack mike makers are well aware that the mike input could be tied high with 3K4 or low with 3K4 or even a dead short center-tap, and DC-block the side facing the console.

There must be smoke detectors which are both battery and hard-120V wired with common signaling. When one detector smells smoke, it puts a signal-tone on the 120V line which the other detectors sense so that they all sing together. The line may have other loads: just 2 detectors or maybe big lamps and heaters (in areas where a dedicated circuit is not required). In a fire, the 120V might be zero V. A simple capacitor can pass the signal-tone onto the power line no matter what is really on the power line. If extreme isolation were needed (Tesla-coil monitor?) you could isolate with transformer, opto-isolator. If instead of a 2KHz or 50KHz tone you needed a 20Hz-20KHz range, you could use bigger caps or transformer.

The 200 ohms is there to slow the up/down transistion. You probably do want to go to ground (via resistor) when "off" or a capacitor-input preamp could store 48V for a longer time than it takes to change microphones. While 48P should not hurt most mikes, the off switch "should" discharge the system for those few times when you NEED it dead. (Like large street-power capacitors have built-in discharge resistors or other shorting system.)

[OldHouseScott]

I would keep the phantom and phase switches and drop the pad and mic-Z switches.

The phantom circuit works as shown. The two resistors represent a compromise of circuit loading and current limiting. A dead short from pins 2 and 3 to ground will flow only 14mA total. As a load to the mic, they are only 6k8 single-ended or 13k6 differential presented in parallel to the input impedance of the mic preamp. SS preamps usually present no more than 10k input impedance, often no more than 2k to 3k, though tube preamps can be higher, so the phantom resistors will bring down the total input impedance. But, since most mics output impedances are 100 ohms or so, the input impedance is usually still at least an order of magnitude higher. This is perfectly acceptable in today's voltage transfer type impedance matching circuits.

The 200 ohm resistor and 220uF cap provide an additional filter for the phantom, and also provide for a short ramp-up of the voltage to the mic. In the off position, the 200 ohm resistor then drains the cap. A pretty neat little trick.

Edit: I see PRR beat me to the punch.

[RicharD]

Ah.... of course.  Good thing I have no shame otherwise I'd be really embarrassed right now.  So this turns a loud POP into a little pfft when switching phantom power on and off.  No need for a timed muting circuit just to switch the phantom in.  I had something designed using a pair of 555's but it was totally hokey.  I am keeping my 555 warm up timer.

[PRR]

The "impedance" switch is a frill, essential to Jensen's thinking.

A high-ratio input can be optimized for zero source impedance or some finite source impedance, but not both at once. The main difference is the top-end ring: it can be damped by a finite source or damped in the winding design.

Classic microphones have finite source impedance, 100-300 ohms. Mike transformers are often optimized for that. Deane was the wizard.

Some beefy condenser mikes have output impedances lower than 50 ohms (though intended for 200+ ohm loads).

When the normal-Z transformer faces a low-low-Z mike, the top end may ring-up. Maybe a few dB in the 50K-30K area.

Jensen's plan allows you to add 130 ohms to pad-up a <50 ohm mike to a 150-200 ohm source for transformers which "expect" such sources.

I agree that Richard may omit this for today.

There's -other- "impedance" switches to throw 1K, 300, 100 across the mike. While it has been a looong time since mikes were "matched" (not really wise with transformers), some people like some mikes heavily-loaded. Myself, I'd be thinking about an in-line switchable shunt, values suited to the specific mike, and kept with the mike. But preamp sellers like more knobs so this is often added to a preamp.

PAD: I rarely needed pads in 10th-row recording, only once with a modern transistor (high-level) mike preamp, but occasionally with older transformer-input preamps. Richard may be working closer than I typically did, and with generally louder sources. (An in-room operatic or jazz extreme-peak is well below the flavor zone of a close-mike Fender Twin or rock/alt percussion set.)

[RicharD]

http://www.sotxampco.com/MP4/MP4-KiCad-Schematic.pdf

Here is the master schematic.  The file is too large to attach.  I've also made the jump to KiCad.  Express is schweet for fast sketching but does not lend itself to designing a PCB.  All parts are annotated and the netlist has been generated.  Next comes the tedious task of associating footprints.  There's a whopping 181 parts in this circuit.  It'd be beneficial at this point to make sure everything is right.

[kagliostro]

Lab results are awesome and I'm too far into this project to turn back now.

Richard my intention wasn't to put a new idea for the pre

sorry for the Off Topic

I was waiting for something like PRR answer

That will not work properly unless the transformer is MADE for this use.

only want to know your opinion on that circuit approach

Thanks

Kagliostro

[RicharD]

http://www.sotxampco.com/MP4/MP4-Temp.pdf

In case anyone is curious, the link above show current progress on the MP4 project  (pages 7&8 are the good stuffs).  It's gonna be tight but I think I can get the whole enchalada into a 10" 2U rack chassis.  The PCB's (7) are all but done.  Just need to verify a few dimensions and do the ground plane.  For lack of anything better, the chassis will be a Par Metal 14 Series cabinet.  The 14 Series jumps from 10" to 16" deep.  At the moment, there is no case fan which I don't think is gonna fly.  It is a ventilated top.  The only place I can see a fan fitting is mounted to the top above the meter board.  Not really ideal.  If I do make the leap to a 16" deep chassis, I can lengthen the 2 audio boards and move the input transformers onto them from the input card.  I can then narrow the input card and squeeze a fan in back where it belongs.  16" seems awfully deep for  stereo mic preamp.  I've already laid the PT and audio boards sideways to keep the chassis down to a 2U.  This is anything but inexpensive so I need to make certain everything is gonna work before I send the board artwork off.  Holy guacamole those Sifam VU meters are buckish!  Comments and critiques are encouraged & greatly appreciated.

-Richard

[woolly]

Wow! I wondered what happened to ya. Quite a project, I'm glad you're doing it and not me. 

 Very impressive. It should be awesome, sure hope it is successful. Best of luck.

[kagliostro]

A question about Power Transformer / Transformers for this 6SL7 + 6SN7 project

for Phantom Power the mic consumption is in the order of 1 / 2 mA

so availability of 10 / 15 mA will be fine

6SL7   is rated 300mA at 6.3v and 150mA at 12.6v

6SN7   is rated 600mA at 6.3v (12SN7 is rated 300mA at 12.6v)

so 1A at 6.3v or 500mA at 12.6v will do the job

The question is:

What about B+ consumption ?

can someone specify which is the consumption for each tube in this circuit ?
(so I will know the dimensions of the smallest PT will do the job)

Many thanks

Kagliostro

[RicharD]

The input stage draws 1.1mA and the output stage draws 17.5mA for a total of 18.6mA per channel.  If you are going to use 12V filaments, you have to build a stereo pair whereas the 6SL7 and the 6SN7 have different filament current draws.  This circuit easily adapts to using a 12AX7 and 12AU7.  The SRPP' input stage is fine as it sits with a 12AX7.  Current draw drops a little to .95mA.  To use a 12AU7 in the White Cathode follower output stage, you must change Rk to a 150 ohm so you don't fry the plate.  It's current should be 15mA.

[kagliostro]

Hi Richard  

The input stage draws 1.1mA and the output stage draws 17.5mA for a total of 18.6mA per channel.  If you are going to use 12V filaments, you have to build a stereo pair whereas the 6SL7 and the 6SN7 have different filament current draws.  This circuit easily adapts to using a 12AX7 and 12AU7.  The SRPP' input stage is fine as it sits with a 12AX7.  Current draw drops a little to .95mA.  To use a 12AU7 in the White Cathode follower output stage, you must change Rk to a 150 ohm so you don't fry the plate.  It's current should be 15mA.

Many thanks for the info about currents

and also about the use of 12ax7 and 12au7 (I hesitated to ask the data for this modification because I had fear of giving too much disturbance)

Kagliostro

[RicharD]

Distractions are good.  I'm in a holding pattern at the moment anyway awaiting the jacks and switches to I can physically verify their dimensions.  I am hoping to send off the board art in about a week.   

[kagliostro]

Hi Richard

What about this project ?

Did you find the time to finish it ?

can we see a photo of the result ?

Kagliostro

[RicharD]

I'm kinda dead in the water.  I just can't justify throwing any additional cash at the project right now.  I have a daughter in college so money is tight, not to mention business is slow.  I started a perf board version but I screwed up and used the old I/O board layouts.  I do have most of the parts except for the forward controls and a bunch of connectors. DL gave me a rack mount PC box to put it in.  Nothing is nailed down on the perf boards.  Here's what it looks like.



Here is the final schematic, PCB layouts, and preliminary chassis layout.
http://www.sotxampco.com/MP4/MP4-2-Temp.pdf

Here are the design docs and bread board measured telemetry.
http://www.sotxampco.com/MP4/MP4-Design.pdf

The measured telemetry is outstanding considering this was done on the hay-wired bread board.  Response is flat, Distortion is <.1%, and the noise floor is -75dB.  I would expect to see some improvements when properly wired and properly enclosed.  I'd love to finish this trip but there's no wind in my sails.

[kagliostro]

Hi Richard

thanks for the info and photo

nice and cheap box that, you can cover the front with an aluminium panel and it will be perfect

My son has finished studying few years ago but does not have a permanent job till now
every now and then has a committed working, these days is on tour with Max Pezzali, not as a sound engineer, he assemble and disassemble the equipment on stage (not the maximum for an agrarian expert, graduate in Japan language  who attended a special school for sound engineer), he like to do but the work is hard and the pay is not so much so I must help him every month and this last time I've job problems too.

so I can well understand the reason you were not able to finish the preamp till now

What kind of college your daughter has chosen ?

Kagliostro

[RicharD]

My eldest is here in Austin at The University of Texas.  I don't know where my little one will go, but I think she wants to fly far from the nest if possible. 

It's kind of an expensive hobby this thermionics is.  Fortunately I can bread board around existing parts on hand to keep my brain busy.

[kagliostro]

In south of Italy they say:

"I figli so piezz e core" (slang)

"I figli sono pezzi di cuore" (italian)

children are pieces of the heart

I think it's the same all over the world

Kagliostro