Valve mic preamp design incoherent rambling

[GussyLoveridge]

Okay. So with a mic source of 150Ω reflected through a 1:10 you end up with 150Ω x 10(10) = 15K?

Am I concerned with the resistances of the primary or secondary coils? What about the grid stopper if applied and the grid leak?

 

[Winston OBoogie]

Yes that's correct for a perfect 1:10 transformer: 150 ohm primary source is reflected to the secondary as 15K

It's useful to know and calculate the reflected primary dcr and know the secondary dcr. for sure
The primary dcr would be subject to the same 1:10 turns (100 X impedance) transformation as was your 150 ohm typical mic source. So a figure of 15 ohms primary dcr would be reflected to the secondary of the same 1:10 transformer as 1.5k.
Let's assume your secondary dcr is 2k so, what appears at the secondary of your transformer is:

(150 ohm mic source X 100) + (15 ohm pri. dcr X 100) + 2K sec. dcr =

15K + 1.5K + 2K = 18.5K

All this helps determine if and by how much noise might be affected, and how much loss due to the addition of the dcr in series with the 15K reflected impedance there is.

I haven't personally ever needed a grid stopper on the first stage of a transformer coupled mic amp, but there could indeed be a need for one if you have long cabling between the transformer and the first grid. My advise is don't have long cabling.
If you have various bits of switching (for a Hi-Z input for instance) between the transformer and tube grid, then use a grid stopper of as low as value as cures whatever problem you have. No need to put a 2K stopper there when a 220 ohm does the trick.

Edit: forgot this bit. On the grid leak question. The transformer itself, or a combination of it and any secondary loading/termination you have, is your grid leak.

 

[ruffrecords]

Good stuff Winston. Transformers also work both ways, so whatever load you place across the seconday gets reflected back to the input to define your input impedance. So a typical 150K secondary load is reflected to the primary as a 1500 ohms input impedance ( modified by the transformer dcr values as before).

Cheers

Ian

 

[Winston OBoogie]

Yes thanks Ian, I forgot to mention about it working both ways.

Whether we may want it or not, transformers are bi-directional.

 

[GussyLoveridge]

Okay - for the degenerated stage I'm getting numbers that look like this:

I know these numbers aren't exact right now but I'm just trying to get all the math correct.

F = 1/(2πRC)
F = 1/ (2*3.14*18,500*0.784^-10)
F = ~110 kHz

And for the fully bypassed stage:

F = 1/(2πRC)
F = 1/ (2*3.14*18,500*1.28^-10)
F = ~67 kHz

I think that this is what we should expect. Because there is less gain with the degenerated stage we have less Miller capacitance therefore a higher frequency pole?

For the DCR of the transformer secondary acting as my Grid Leak - Can I just measure this with my multimeter once I have purchased my choice of input transformer? I don't see this stated on the datasheets. Actually there is a line for Static Resistance of Each secondary quoted on the Lundahl datasheet: 125Ω.

So reflected back through the 1:10 this would look like a 12k5 Ω loading to the microphone, which for a 150Ω source would be an adequate loading? Ah, nope. I'm going the wrong way...shit.

I think I am mixing something up here. I'm sure of it.

 

[ruffrecords]

Okay - for the degenerated stage I'm getting ....
I think that this is what we should expect. Because there is less gain with the degenerated stage we have less Miller capacitance therefore a higher frequency pole?

Exactly

For the DCR of the transformer secondary acting as my Grid Leak - Can I just measure this with my multimeter once I have purchased my choice of input transformer? I don't see this stated on the datasheets. Actually there is a line for Static Resistance of Each secondary quoted on the Lundahl datasheet: 125Ω.

So reflected back through the 1:10 this would look like a 12k5 Ω loading to the microphone, which for a 150Ω source would be an adequate loading? Ah, nope. I'm going the wrong way...shit.

I think I am mixing something up here. I'm sure of it.

Yup. The sum of the mic R plus the transformer dcr is reflected to the primary. This is the source impedance seen by the tube. This is also the R in your Miller capacitance calculation.

Going the other way you typicaly ass a 150K from grid to ground of the first tube. This is reflected back to the primary as (150K + secondary dcr) / 100. Add the resuklt to the primary dcr and that is the load seen by the mic itself.

Cheers

ian

 

[GussyLoveridge]

Exactly

Yup. The sum of the mic R plus the transformer dcr is reflected to the primary. This is the source impedance seen by the tube. This is also the R in your Miller capacitance calculation.
mic
Going the other way you typicaly ass a 150K from grid to ground of the first tube. This is reflected back to the primary as (150K + secondary dcr) / 100. Add the resuklt to the primary dcr and that is the load seen by the mic itself.

Cheers

Ian

Please excuse me, I'm feeling a little dense.

If the 150K grid to ground is different from the DCR, the where is it coming from? I can see a suggested 150K (resistance only) suggestion for a zobel (maybe?) in the literature for the Jensen transformer. Looking at your Classic Solo I see both some resistance Rz (from your zobel) as well as a 1M to ground which I think are both loading the secondary? When I look at the NYDave One Bottle in one instance there is a 220K as part of the zobel and in another later schematic there is a 100K in the zobel plus another 1M to ground (Though this looks like it is setting the input impedance for when there is a line in, not coming through the mic input transformer. The Gyraf G9 has a bunch more stuff going on there...

Just so I'm not totally missing the boat here. I'm reading the term DCR as the DC resistance of the coils. There would be DC resistance of the coils on the primary, and assumedly a greater DC resistance of the coils on the secondary because of the higher number of turns.

Also for clarity - I'm calling the side of the transformer that attached to the microphone input the primary and the side attached to the grid of the tube the secondary. Is it the convention that we always call the side we're driving the primary? I know I should probably know these things at this point.

 

[ruffrecords]

Please excuse me, I'm feeling a little dense.

If the 150K grid to ground is different from the DCR, the where is it coming from? I can see a suggested 150K (resistance only) suggestion for a zobel (maybe?) in the literature for the Jensen transformer. Looking at your Classic Solo I see both some resistance Rz (from your zobel) as well as a 1M to ground which I think are both loading the secondary?

The recommended secondary load for the Jensen (and also the Cinemag and Sowter) 1:10 mic transformers is 150K. A Zobel is usually a resistor in series with a capacitor which is required for some types of transformer but not these three. In my classic solo the parallel 150K and 1M is 130k, a little lower than the recommended value. My boards mostly include provision for a series R and C for a Zobel network in case people decide to use other transformers that might need one.

When I look at the NYDave One Bottle in one instance there is a 220K as part of the zobel and in another later schematic there is a 100K in the zobel plus another 1M to ground (Though this looks like it is setting the input impedance for when there is a line in, not coming through the mic input transformer. The Gyraf G9 has a bunch more stuff going on there...

I cannot comment on other people's designs but the principles are the same. The load attached to the secondary is reflected to the primary divided by the turns ratio squared.

Just so I'm not totally missing the boat here. I'm reading the term DCR as the DC resistance of the coils. There would be DC resistance of the coils on the primary, and assumedly a greater DC resistance of the coils on the secondary because of the higher number of turns.

Yes, that is right. As far as the secondary is concerned its load is the secondary dc resistance in series with the Zobel network. The sum of these is reflected to the primary. That reflected impedance appears in series with the dc resistance of the primary is seen by the mic as its load.

Also for clarity - I'm calling the side of the transformer that attached to the microphone input the primary and the side attached to the grid of the tube the secondary. Is it the convention that we always call the side we're driving the primary? I know I should probably know these things at this point.

That is also right.

Cheers

Ian

 

[abbey road d enfer]

For the DCR of the transformer secondary acting as my Grid Leak - Can I just measure this with my multimeter once I have purchased my choice of input transformer? I don't see this stated on the datasheets. Actually there is a line for Static Resistance of Each secondary quoted on the Lundahl datasheet: 125Ω.

So reflected back through the 1:10 this would look like a 12k5 Ω loading to the microphone, which for a 150Ω source would be an adequate loading? Ah, nope. I'm going the wrong way...shit.

I think I am mixing something up here. I'm sure of it.

Yes, I believe you are mixed up. The DCR of the secondary is almost irrelevant to the reflected Z at the input, being in series with much higher resistances (impedances). Let's assume you load the sec with 100k. If the xfmr was perfect (no DCR), te refelected Z at the pri would be exactly 1k.
If you include all the DCR's (15r at the pri and 125r at the sec) you end up with 1001.25 for the reflected and 15r in series for a total of 1016.25
The actual DC value of the grid leak is 125r. For normal biasing, the grid leak resistance can be anything between zero and the recommanded value.
Higher values can be used for grid-leak bias, where there is no cathode resistor. The negative grid bias is the result of "stray" electrons hitting the grid. It generally involves grid resistors of several megohms.

 

[GussyLoveridge]

Okay - I'm going to take all this and work toward a more complete schematic. I can't thank you all enough for the valuable information and for your patience. I am feeling slightly more confident about being able to get something relatively in the ballpark of workable.

I will post a little more here before too long.

In the meantime - Does anyone have a 6072 symbol for Eagle? It seems odd that it was not in the library that I have received. I'm brand new to Eagle so forgive me if I'm not describing this correctly.

 

[abbey road d enfer]

In the meantime - Does anyone have a 6072 symbol for Eagle? It seems odd that it was not in the library that I have received. I'm brand new to Eagle so forgive me if I'm not describing this correctly.

You can use a standard dual triode symbol. 12AX7, ECC83...

 

[GussyLoveridge]

Roger.

 

[Winston OBoogie]

Hey Gus and others,
the issue of noise in tubes came up earlier. There's a list on the site linked below which has results from tested examples of the usual suspects and a couple others wrt noise. There are a couple of examples with 2 parallel triodes as well as comparison against a low noise J-Fet.
The tests were made with reference to a typical magnetic cartridge impedance so quite a lot lower source impedance than we have on our transformer secondaries. Still, the results are scaleable or give you a good general idea of things.

No guarantees in tube work though so, a lot of the time you just end up needing to measure yourself, or try a few of the tube you use until you find the quietest one for the front end.

https://tavishdesign.com/pages/audio-tube-noise-measurments

 

[ruffrecords]

Those noise results also illustrate just how important it is to use a transformer as the firs stage of a tube mic pre. The noise from a 150 ohm resistor in a 20KHz bandwidth at room temperature is about -131dB which is close to 0.2uV. The free 10 times gain from a 1:10 transformer raises this noise to 2uV which is much the same at the EIN of most triodes in the same bandwidth. This would give your mic pre a noise figure of 3dB which is an excellent result.

Cheers

Ian

 

[solkatten]

Many of the vintage design used pentode because of low miller C. Even though pentodes has more inherent noise, it can be combined with a high ratio input transfomer like 1:20 or higher and the end result was low noise. Parralleling tubes increases the problem with capacintance... But that´s just another way to skinn the cat

 

[Winston OBoogie]

Yep, paralleling triodes for a noise improvement increases your issues with capacitance. But if you're gonna go down that route, I suppose it's not much further to then shove a cascode on top to get rid of most of that C. Gain is then akin to a pentode. Is it worth the extra effort? I'm not sure. I just played around with something like that but it was something of a rat's nest and I had to dismantle it before it did serious injury to me or the cat.

Part of me thinks, for the most part the usual topologies are fine, noise wise, for 90% of what they get used for. And if I want super low noise I'd be better using J-Fets on the bottom of that cascode, or not using tubes in the first place.

 

[solkatten]

Yep, paralleling triodes for a noise improvement increases your issues with capacitance. But if you're gonna go down that route, I suppose it's not much further to then shove a cascode on top to get rid of most of that C. Gain is then Gm X (1+ Gm') so akin to a pentode. Is it worth the extra effort? I'm not sure. I just played around with something like that but it was something of a rat's nest and I had to dismantle it before it did serious injury to me or the cat.

Part of me thinks, for the most part the usual topologies are fine, noise wise, for most all that they get used for.
And if I want super low noise I'd be better using J-Fets on the bottom of that cascode.

Agreed...

I think the best reason to use a pentode is that u can make switable between triode and pentode. To be able to vary gain and harmonic content. Pentodes just sounds cool

I think in the real world you often don´t need super high gain or super low noise. In many occasions high headroom is desirable because of hot mic signal. Headroom is a true virtue of tubes.

 

[Winston OBoogie]

I think pentodes are cool too Playing with the screen grid, as you say, will affect the characteristics. As will where you place the loadline - through or slightly below the knee for pentode tone, or above the knee for more triode like characteristics.

Most all of the recording and broadcasting places in Europe in the 1950's - '60's seem to have used mic amps with pentodes as the first tube. I'm sure there are plenty of exceptions, but from what I've seen, seems to be mostly that way.

P.S. I made a silly mistake in my above post re. the gain of a cascode. I edited it out so I don't look like a complete idiot

But gain, with same type tubes in cascode config., is approx: Gm of lower tube X Ra.
Oops...

 

[solkatten]

I think pentodes are cool too Playing with the screen grid, as you say, will affect the characteristics. As will where you place the loadline - through or slightly below the knee for pentode tone, or above the knee for more triode like characteristics.

Most all of the recording and broadcasting places in Europe in the 1950's - '60's seem to have used mic amps with pentodes as the first tube. I'm sure there are plenty of exceptions, but from what I've seen, seems to be mostly that way.

P.S. I made a silly mistake in my above post re. the gain of a cascode. I edited it out so I don't look like a complete idiot

But gain, with same type tubes in cascode config., is approx: Gm of lower tube X Ra.
Oops...

Yep many of the brilliant post war engineers used pentodes in the front end of mic preamp, not because they sound cool, but because they peformed well

 

[abbey road d enfer]

IMost all of the recording and broadcasting places in Europe in the 1950's - '60's seem to have used mic amps with pentodes as the first tube.

I think the use of pentodes goes hand in hand with superior winding techniques. It allows the use of high ratio xfmrs, cause low Miller cap.
I would like to see a comparison of a pentode with a cascode. Noise ought to be better with the latter... I don't see anything against cascode, except maybe the need for slightly higher B+.