6EM7 Mic Preamp

[EDIT] Updated schematic, then changed pot to 250k.


I used the plate curves found in the RCA Datasheet (300k PDF)

I used resistors on the input and output to represent the transformers. Input tx is a Nytronics that I got from jrmintz, impedance ratio 120:31,600. Output tx is UTC A-44, 4k:600.

Here it is:

Increase R5 150K

No feedback-you be cool.

Start with that and adjust later.

[quote author="CJ"]Increase R5 150K

No feedback-you be cool.

Start with that and adjust later.[/quote]

R5 is just a stand-in for the 15K secondary of the input tx, not an actual resistor.

Do you think I'll need a grid a resistor?

a grid resistor would also be part of your input trans termination, so it might need to be tweaked with the transformer you use. if any.

The source impedance should be modeled as a resistor in series with the signal, not shunting it. So the order should be signal source, 15K resistor in series from source to grid, large resistor from grid to ground. The grid resistor is necessary for purposes of the simulation (to provide a DC reference for the grid) but may not be needed in real life since the xfmr secondary fixes the DC point of the grid essentially at 0V.

No feedback-you be cool.

Click to expand...

What does that mean?

Do you think I'll need a grid a resistor?

Click to expand...

Depends on the transformer; some are designed to work an unterminated grid. The 15095A is actually a line output transformer so it's hard to say without experimenting.

Oh, and a ~500K pot is a pretty high value to work into a triode grid if you want to pass full-bandwidth audio. Without digging out the data for this particular tube, and calculating the gain, I can't say for sure if it'll be a problem in this case but it is something to consider.

i would increase c2 to 0.22 or 0.47

[quote author="NewYorkDave"]The source impedance should be modeled as a resistor in series with the signal, not shunting it.[/quote]

Ahhhhhh, that is making more sense now, or I'm closer to making sense of it.

[quote author="NewYorkDave"]

No feedback-you be cool.

Click to expand...

What does that mean?[/quote]

I have no idea, but I like being called cool. So, thanks CJ!

[quote author="NewYorkDave"]Oh, and a ~500K pot is a pretty high value to work into a triode grid if you want to pass full-bandwidth audio. Without digging out the data for this particular tube, and calculating the gain, I can't say for sure if it'll be a problem in this case but it is something to consider.[/quote]

Looking at it now, my reasoning for that doesn't apply now, with the 1M hanging off the second grid. The first plate will always see the higher impedance, so the four times rule-of-thumb is taken care of. I'll change it to 100KA.

[quote author="ioaudio"]i would increase c2 to 0.22 or 0.47[/quote]

I only have 0.5 200VDC or 1uF 250VDC. I think I'll go with 1uF.

Many thanks for the responses everyone!

Hi,

My first post here and I don't want to seem impolite, but I'm looking forward to hearing your results on this. I've been thinking about building a mic pre using one or another of the dissimilar twin triodes. I know they have been used for audio amps, but don't know how the noise/microphonics will be for high gain circuits.

From tubecad it looks like you should be around 40dB of gain not counting the iron stepup and down.

Michael

Hi Michael,

Welcome, and thanks for expressing interest. The tubecad website is great. I refer to it alot, especially the Grounded Cathod Amplifier article

Edit: I totally miscalculated the gain somehow.

I gotta give credit to PRR for the inspiration and starting point:

[quote author="PRR"]let's see.... 300V supply, 10K DC resistor, 20mA, puts 100V on the plate. We can cleanly swing a low-Z load to 10mA and 30mA, 10mA peak. With a 2:1 transformer, 2K4 primary, we can swing 24V peak primary. We have 12Vpk 20mA peak at the secondary, +21dBm in 600 ohms. The incremental plate resistance here is about 1K, so damping factor is 2.4, so output impedance is 250 ohms, resonably low. It looks like about 14Vp-p on the grid, and 7Vpk in for 24Vpk out is a gain of 3.4, which agrees with Mu about 5 and this ratio of Rl/Rp. Gain to the load is 2:1 less or 1.7. The small-side of 6EM7 can easily give voltage gain of 40 (use 12AT7 tables). Total gain from first grid to 600 ohm load is 70, or 37dB. 1:10 input iron makes it 57dB. That may be too much for some uses, put a pot between stages. How far down can we turn the pot before the input tube clips first? Taking voltage-amp peak output as 20% of supply, the small-side can make almost 60V peak, we can turn-down 18dB. After that, the input stage clips before the output stage reaches +21dBm.

Working the tube richer, with the plate around 60V, buys a little lower Rp and a little more gain. But we may have trouble holding it there. 80V-100V on the plate may be more stable.

You could also work things at lower B+. Some TV sets worked as low as 110V B+. But this works much better with transformer-coupled (DC in the iron) than RC-coupled (iron just for matching).

Working 300V B+, 4K resistor, 50mA current, 2k4 load, we could pull 50V 20mA peak, 25V 40mA peak in the load, +27dBm. Gain rises to 4, Rp drops to 800 ohms. If we don't need +28dBm, the small improvement is probably not worth the added supply power. Note also how little the performance changes when plate resistor and current changes by a factor of 2: we do NOT need precision parts to get consistent results. 4/3.4=1.4dB. Your idiot parts-assistant got 7K4 instead of 4K7 resistors? Slap 'em and ship 'em... customers will never know.[/quote]

Hi Skipwave,

Thanks for the welcome. A cogent analysis by PRR. Hope the tube turns out to be quiet.

I'm hoping to build 8 channels or so of mic pre, that's how I happened onto this site, and these dissimilar dual triodes were one of the things I was considering. Along with differential cascode, and the 12AY7 one from RDH4 and ....... So many designs, so little time.

Michael

will always see the higher impedance

Click to expand...

Nope. See below:



With a 100K pot and a 1M grid resistor, the plate of the preceding stage will see 90.9K with the pot turned all the way up. So you want to arrange for the preceding stage to have an output impedance of around 10K (or less) if at all possible. The output impedance is the tube's plate resistANCE in parallel with its plate resistOR. Or use a bigger pot if the input capacitance of the next stage is low enough to allow it.

Plate resistance 33k, plate resistor 100k, gives 24.8k output impedance.

So a 400k pot will give the closest to ideal impedance to bridge the first plate into. The grid to plate capacitance of V2 is 10pF, is that in parallel with the grid to cathode&heater capacitance (which is 7pF)?

If it is just the 10pF, that gives a frequency of 40KHz. If its the 10pF in parallel with the 7pF, that gives 4pF and moves the frequency point up to 99KHz to start with. I was using the OpAmpLabs calculator. Either one seems like its outside the audible range, but the 40KHz might have audible artifacts... maybe, but that's only with the pot all the way up so no volume, as you dial the resistance down and the volume up, the frequency moves higher into the inaudible range.

So looking out from the first tube you see the plate resistance in parallel with the plate resistor. So that's 25K, lets say. Then that is in parallel with the Rload. For simplicity let's drop the 1M resistor and just use a 100K pot. So now the output impedance of the tube is looking into a 100K load. Something smaller than ideal, but at 4 times output impedance it might work.

Now let's hook the wiper of the pot to the grid of the next section and stand there looking backwards compared to our former view. From the grid 100K in parallel with 25K when the wiper is set at the top of the pot for max volume. So looking back from the grid a source impedance of 20K is seen. As the volume is turned down part of the pot goes in series with the output impedance of the first stage. So at the mid resistance setting we see (25K+50K) in parallel with 50K, or 30K. If the pot is set at 75% then 75Kll75K would be seen, 37.5K.

Unfortunately, I think the 10pF grid to plate capacitance of the second stage is multiplied by the mu, around 4.5 in this case. Add a few picofarads for strays and it's something like 50pF. In the worst case source impedance, 37.5K that should give a -3dB at around 85kHz. I think that might be higher than the -3dB point for the first tube.

Caveat, I could be totally wrong on all points.

Michael

[quote author="Michael A"]Unfortunately, I think the 10pF grid to plate capacitance of the second stage is multiplied by the mu, around 4.5 in this case. [/quote]

Thanks Michael. You're right, I totally forgetting about that.

[quote author="Michael A"] -3dB at around 85kHz[/quote]

So the capacitor in series with the resistance creates a rolloff with 85KHz as the knee?

Based on what I had read about capacitors and frequency I thought that a cap in series with a resistor created a high pass, so I thought I was going to get a rise above the frequency calculated.

> I could be totally wrong on all points.

Not checking your math, it all looks fine.

In a system like this, you can simplify to death and not be far off. The grid resistor will be much larger than the pot, ignore it. The tube side will be lower Z than the pot, assume the top of the pot is nearly grounded. The worst-case wiper impedance is when the wiper is as far as possible from both ends, or mid-rotation on a linear pot. For a 400K pot, worst-case is 100K. For a 100K pot, midpoint looks like 25K, but if you account for the tube it may be 33K.

> the 10pF grid to plate capacitance of the second stage is multiplied by the mu, around 4.5 in this case

By the grid to plate gain. In many practical triode amps, this is 2/3 to 1/2 of Mu. In this case, we are really underloading the fat tube, so it will be nearly Mu. I like "4" because I can do that on my fingers.

Since the pot wire may be long, I'd assume a bit more capacitance... being really lazy, I might round-up to 100pFd and see how much trouble I bought.

100pFd and 400K pot (100K worst-case) is 17KHz.... hmmmm. Can't run a long wire off the wiper. Even if the pot was right on the 6EM7, it's mighty droopy. 100pFd and 33K gives 50KHz. Not bad for simple tube gear, and could be almost an octave better with tight layout.

More to the point: there isn't a heap of gain here. Mu*Mu is 270, of which you can only use about half: 35*4= 140

Another issue: if the output has gain of 4, and you want large output, you can't put much pot attenuation in front of the output. The driver will overload before the output.

The 6EM7 and kin should be used as intended: a one-bottle Power Output Stage. If you need gain control, do it before the pair. If you need low noise, the gain control goes after the first stage. This means we need another bottle in front of the 6EM7, but that gives a pretty killer mike-amp.

He speaks! Thanks again for the idea and the guidance. This is my first attempt at designing anything. It's been a great experience. Kinda like translating an ancient map to buried treasure.... or that image just sprung from all the x's I made on the plate curves sheets. :green:

[quote author="PRR"]100pFd and 400K pot (100K worst-case) is 17KHz.... hmmmm. Can't run a long wire off the wiper.[/quote]

How long is long? I've got about 6 in. right now and could only possibly reduce that to 5 with the chassis I'm working in. Is there a fairly accurate way to calculate wire capacitance, or is it only worthwhile to measure the actual (cuz I can't)?

[quote author="PRR"]a one-bottle Power Output Stage. This means we need another bottle in front of the 6EM7, but that gives a pretty killer mike-amp.[/quote]

I definitely could see this evolving into something like that. I have plenty of space for another bottle in my box. OTOH, I calculate a gain of 56db from source to 600ohm load. That ain't bad.

We'll know soon enough how it performs... almost finished.

It's alive! I just fired up this pre and did a brief test. It works. Tomorrow I will find out how well.

My unregulated B+ supply works perfectly, 3 RC poles and metering 299.5V going to the plate resistors. The plates are idling exactly where they should be. This tube gives off some heat, but with a 1A heater I expected that.

Will look forward to seeing what the usable gain and frequency response turns out to be. What pot value did you go with? Or for that matter, do you have an as built schematic?

Cg-p is a shunt capacitance, not a series one.

Michael

I realized immediately upon testing this thing that I didn't have enough negative grid bias for V1. The preamp started clipping when I spoke into a dynamic. I raised the value of the cathode R to 2.2k, and now its happy. Plate sits at 200V, 2/3rds of B+ and as far up as I want to push it, grid is biased to about -2.5V. Plus I got some more gain in the transaction, mu went up to 68, rp fell to 30k. Total gain from source to 600ohm load is 59db. Wow. And its dead quiet, even testing with the lid off the box.

I think I've struck a nice balance here, with the switchable 20db pad it should be fairly flexible. Now I understand how there are endless trade-offs in the design process, and this is a simple design!

Clown singing through a 57 sounded pretty nice. I'm going to take it over to the rehearsal space and test it out on a few things tomorrow.

Did you go with the 400k pot? Have you tested the frequency response? Sounds like a good outcome.

Thanks,
Michael