Coupling of EF86 mic input stage

[rotation]

Hi,

I would like to know what is a proper way for RC coupling of EF86 based mic input stage to tubes like ECC82,83 etc.
I found several examples, one of them is "Four channel input mixing preamplifier" from Mullard's book "Tube circuits for audio amplifiers":
technicalscientific.com/data/Mullard_3_PA01.pdf
The other is 47 type coupled with R332k and C22n.
I would like to know if values of resistors from them, 332k and 470k, are typical for this kind of tubes? And does is matter if i put capacitor first and then resistors?
Mullard book says 470k is reducing crosstalk between channels and preventing grids of second stage to short when pot is at a minimum in mixer, while Winston o'Boogie says that this is load resisstor for EF86 stage in 47. I didn't find other examples with explanations, except for triode to triode coupling.

Miha

 

[Winston OBoogie]

Winston o'Boogie says that this is load resisstor for EF86 stage in 47.

Miha

Not quite.  It's part of the feedback around the E88CC, the 1M6 resistor being the other part of the loop.
It does (the 330K) also play into the loading on the EF86 but it's there primarily as part of a F.B. loop to reduce the gain and distortion of the E88CC.

Mostly, you would just simply cap couple the signal from an EF86 anode to the grid of the next stage. 

Edit: the Mullard circuit you mentioned uses the resistors on the output of each stage as mixing and isolation resistors.  Think about what would happen (loading) if they weren't there...

 

[PRR]

> technicalscientific.com/data/Mullard_3_PA01.pdf

That's a show-off fancy-pants paper. The Fig 1 power-amp is direct-coupled; the mixer takes four inputs with low interaction and stunningly low gain. It's Mullard pushing tubes, sketching new ideas so the industry does not stagnate. (Little did they know what the next 10 years would bring....)

You load a tube with a resistor to drop "about half" of the supply voltage, typically 50K-220K. The next grid must be DC-grounded, but you may use a resistor up-to 1 Meg so you don't short the signal. Then to get from the plate at high DC voltage to grid at zero DC voltage, you use a capacitor. The R-C product of this C and all the R sets your lowest useful bass frequency.

For simple systems, use 0.01uFd and 1 Meg. That is flat to 17Hz at -3dB, 34Hz at -1dB, which is flat enough for simple systems. For systems which may have many-many stages, and must be fault-free to the lowest audio frequencies, you may want to go up by a factor of 10 to get -3dB at 2Hz and "dead flat" for piano, organ, bass.

However. Usually you do not couple a pentode direct to another gain stage. If pentode has gain of 150, and next stage is gain of 40, total gain is 6,000. You sometimes have 10mV signals which will be boosted to 60V, but do you need 60V? More often you have 30mV and 100mV signals which "would be boosted" to 200V or 600V, and a pentode working with 250V supply can NOT do that. So you have a Gain Control after your pentode. (Alternatively, you run NFB around both stages to trade-off gain for other performance; but this goes beyond the level of your question.)

 

[PRR]

http://pmillett.com/Books/RDH3.pdf  -- 15 MEGAbyte PDF file, but worth the wait; right/long-click and Save As... to your machine.

Page 5 in book (page 10 in PDF file): Pentode Voltage Amplifiers.

See also Chapter 11 and 12.

 

[Winston OBoogie]

http://pmillett.com/Books/RDH3.pdf  -- 15 MEGAbyte PDF file, but worth the wait; right/long-click and Save As... to your machine.

Caution! that particular PDF contains far too many words.

 

[rotation]

http://pmillett.com/Books/RDH3.pdf  -- 15 MEGAbyte PDF file, but worth the wait; right/long-click and Save As... to your machine.

Caution! that particular PDF contains far too many words.

Yeah, that's why it took me so long to say thanks.

I checked suggested material before asking and it's very good for learning basics. I couldn't figure out 332k resistor because i got confused by other thread where Winston O'B talked about it in other context..
So, i didn't add pot between the stages at the end, i guess i shouldn't do it because of ECC88 feedback and global feedback.
Another thing i done is disconnecting global feedback; i got more gain this way but it sounded very very bad. Then i added a pot in series with it and i again prefered original sound. I also played with 1M6 and this was the only place where i could get some more gain with good sound...

I would like to add DI input to this preamp, so i searched a little. Normally i should wire it to G1 of EF86 but i couldn't find any info about how much signal pentode like this one can take. Anyone?

Another thing is that i don't want to use another expensive switch and run wires all over chasis for making input attenuator. I know i could use 3p toggle switch with relays or even rotary switch with relays. But as my input transformer can take big signal i thought about attenuating at the output. I think this is much better plan if i know i will never overdrive the input.
One option for this is NYD's output attenuator with expensive switch and many resistors of strange values in series which i don't like. The other is suggestion by Winston O'B (i hope i'm correct this time), 1k pot and some resistors (don't have schematic, NYD suggested 220R in series at each line before the pot). PRR suggested only 1k pot ala API.
The last two options are very interesting for me because i can use them for many other outputs i want to control, and beside this i don't have to mess with precious signal at the input. Could someone please explain if simple 1K pot is ok or i should add resistors before it?

I found some explanations about 1K pot method, it's PRR's:

"So a 1K pot will work fine for most things. The 1K-0.9K load on the source is fine for most "600" outputs. The 0-270 ohm output impedance will be fine for most inputs. If you know the input works better (flatter, or more iron-sound) with true 600 ohm source, a 400-500 ohm build-out resistor will get the pot output impedance in the ballpark.

> if you don't mind unbalancing the output impedances

IF the output is true floating, as most transformer-out pro gear is, then unbalanced pads will NOT unbalance the midband output! That is the great thing about true-floating outputs. In fact you can ground the "hot" pot-wiper output, take a signal from the "cold" pot-end, and it will be fine.

Now I have to admit that no transformer winding is ever perfectly floating. There is stray capacitance everywhere. At high frequencies it will be un-balanced. I'm not sure it matters. Even attempting to measure the amount of unbalance will change the situation."

I didn't understand the part about "grounding the hot wiper output, take a signal from cold pot-end, and it will be fine". So i meassured it and it's thruth. Nice.


Miha

 

[PRR]

> i couldn't find any info about how much signal pentode like this one can take. Anyone?

It's right there in the EF86 datasheet. Yes, you do need a slide-rule or similar to convert gain and max output to max input.

 

[Winston OBoogie]

http://pmillett.com/Books/RDH3.pdf  -- 15 MEGAbyte PDF file, but worth the wait; right/long-click and Save As... to your machine.

Caution! that particular PDF contains far too many words.

Yeah, that's why it took me so long to say thanks.

Miha, my comment was NOT in any way aimed at you.  It was in reference to another post in another thread, sorry for the confusion. 


Along with your output pot, I think you should put a pad on the input myself, even if it's just a D.P.D.T. toggle.  If you used two toggles, you could have one toggle be the On/Off and the other select two options, say -10dB and -20dB.  The 2nd toggle would be connecting and disconnecting parallel resistors to the ones on the 1st toggle.

I don't know what type of music you're recording but, even if the gain of the 47 is set at 36dB, you could quite easily be beyond the headroom of the amp on loud sources.

I don't advise applying more feedback for lower gain on this amp.  Even at 36dB, it's a different sounding beast than it is at 40 and 46dB.  You'd also need to fiddle with your caps across the feedback/gain defining resistors.  Personally, I don't like amps that change character too much with gain, I'd rather have a noise penalty by using a pad.  Of course, some folks like the changes and use it as a means of tone/colour control. 

Like yourself, on a few amps,  I've raised the 1M6 output valve feedback resistor for a bit more gain but I don't do it any more. 

 

[Winston OBoogie]

Bumped because I added a bit more to the above.

Also:

"The other is suggestion by Winston O'B (i hope i'm correct this time), 1k pot and some resistors (don't have schematic, NYD suggested 220R in series at each line before the pot). PRR suggested only 1k pot ala API."




I don't know which suggestion of mine you're referring to (maybe the use of a pot as an input or output volume where the pot is a variable shunt to a balanced line?) but, if you're just wanting to hang a pot off the back end and run the signal out unbalanced, a 1K or even 500R audio taper would be fine.

 

[rotation]

[/quote]


Miha, my comment was NOT in any way aimed at you.  It was in reference to another post in another thread, sorry for the confusion. 

[/quote]

Hi,

i didn't understand it wrong, no problem. Well, i understood it as a kind of bump, i thought it was nice from you.. I'm quite bussy this days and don't have time for fast replys.

Thread about 1k pot is missing some drawings. As i understand it pot there is used as a variable shunt to balanced line. NYD is suggesting 220R on each line before 1k pot if we want to mantain balanced line..

I didn't think about using two dpdt switches as input attenuator. But i see i will have to put something in front of preamp. I will also think about toggle switches with 3 poles, maybe this could work for -10 and 20, will think about when i come home.

I have another question; why designers of this preamp used 7:1 output transformer? Maybe this ratio was not so rare at that time, but today it is hard to find and expensive. For this reason i'm thinking about using another output tube with lower Ri (for lower output imp.). Would this work if i have correct bias?

Miha

 

[emrr]

I'd imagine someone calculated 7:1 as optimum for the expected operating condition, but I don't know that.  In practice, I don't think you'll find a huge difference going with a more common ratio.  Some things will compensate, and having not even looked at the feedback implementation, I'll be bold enough to suggest it is and will handle any minor changes for the most part. 

We have at least one modern example.  There was an RCA BC-2B preamp kit available for awhile, which used a 15K:600 output transformer.  The website that made the preamp 'famous' claims that ratio, but the real thing is in fact more like 25K-30K:600.  Both work.  In simple terms, the lighter load with the higher ratio allows more voltage swing from the tube, but the increased ratio also means we have more step down voltage loss.  So, it mostly comes out in the wash.  And there are always other factors......

 

[Winston OBoogie]

I'd imagine someone calculated 7:1 as optimum for the expected operating condition, but I don't know that.   In practice, I don't think you'll find a huge difference going with a more common ratio. 

Yes, I agree.  The 7:1 ratio was picked to have a fairly negligible loading effect on the output valve.   
The worst case load on the secondary of the OPT when the amps were used in the desks would be 350 ohms.  Multiply that by 49 (turns squared) and you get a little over 17K reflected on the primary.  20K:600 would be fine .  If your OPT doesn't need a 600R termination then you could also try a lower ratio.   


E88CC with both sides in // and at the current used in the '47 is a fairly low impedance as is.  You could use a different valve but you'll be looking at something a lot beefier to obtain anything that is vastly lower in impedance.  ECC99 is a bit beefier and is a nice linear valve as well.  You could probably do away with some of the feedback around it as it's harmonic distortion and spread is better than an E88CC or 6922 in my experience.  It also has less Mu so you could also juggle the local feedback around it to achieve the same open global loop gain.     I've used that myself in a design but I ended up with something like 32mA through it which is a bit of a hog for a mic amp output to say the least.  Don't have the plots on hand but, in a '47 type circuit, it'd be less than 2K impedance off the top of my head.


Edit: not really a fan of adding extra stages (sometimes, even if the specs look better with one) and would rather not use a cathode follower if I can help it but, a cathode follower after the E88CC would give you more options as far as output transformer?



2nd Edit as I wrote the valve number incorrectly - it's correct now.

 

[PRR]

You can always consider a small power tube as output driver. Triode-wired 6V6 runs about 2K plate impedance.

> I ended up with something like 32mA through it which is a bit of a hog

Yeah, that's where you might run a 6V6, and it is why you want a high load impedance IF you have free choice. To put +18dBm in 600 ohms you need 14mA peak secondary current. To get low-low THD you want tube idle current much higher than signal current. At 1:1 ratio this suggests over 30mA standing current. At 20:1 you only need 1.5mA -but- to make your 8.6V peak secondary voltage you need 173V primary swing, high for typical supplies. 5:1 to 10:1 are typical compromises.

> a cathode follower after the E88CC would give you more options as far as output transformer?

If the impedance is really your problem, this is an answer.

But that 1:1 ratio transformer still must run the tube at over 14mA idle current to make +18dBm. The power relations around a cathode follower are not significantly different from a plate-loaded stage. If the cathode load is not similar to a "happy" plate load (several K or at least 3:1 ratio to 600r) then the THD won't be magically improved either. And at higher impedances above 10K, the internal L-C lumps like to resonate, and this may be less dramatic with a high-impedance source like a plate.

As WO'B knows, a "good" mike preamp needs a careful outline of Requirements, and then a careful detailed design, and then some careful compromise with costs and practical issues (16 channels of 32mA tubes will melt your lunch).

OTOH, 1 or 2 channels in a casual home studio do NOT need over-thinking. Wire up a tube so you can connect mike and tape-deck without hurting anything. If not enough gain, add another tube. 600 ohm lines, balanced lines, transformers are more about BIG studios with LONG lines than about typical modern small studios.

> far too many words.

Understood. But at least I did not point to the 4th.

 

[Winston OBoogie]

Nothing really to add, good advice from P.R.R.

Except:

To get low-low THD you want tube idle current much higher than signal current.

Thanks for pointing this out.  A lot of people don't seem to get it.  That and using a high ratio of supply to signal voltage.

Getting individual stages as linear as possible by themselves is where it's at.  This equally applies to discrete S.S.

Understood. But at least I did not point to the 4th.

  I believe Mrs Fritz Langford-Smith pointed at the 4th as she vetoed a 5th.

 

[rotation]

> far too many words.

Understood. But at least I did not point to the 4th.

I just printed big part of RDH3. That and your replys will keep me bussy for a while

Miha

 

[rotation]

E88CC with both sides in // and at the current used in the '47 is a fairly low impedance as is.  You could use a different valve but you'll be looking at something a lot beefier to obtain anything that is vastly lower in impedance.  ECC99 is a bit beefier and is a nice linear valve as well.  You could probably do away with some of the feedback around it as it's harmonic distortion and spread is better than an E88CC or 6922 in my experience.  It also has less Mu so you could also juggle the local feedback around it to achieve the same open global loop gain.     I've used that myself in a design but I ended up with something like 32mA through it which is a bit of a hog for a mic amp output to say the least.  Don't have the plots on hand but, in a '47 type circuit, it'd be less than 2K impedance off the top of my head.


Edit: not really a fan of adding extra stages (sometimes, even if the specs look better with one) and would rather not use a cathode follower if I can help it but, a cathode follower after the E88CC would give you more options as far as output transformer?

I don't want to add another stage here, another output tube would be ok.
I tried to calculate resistor for 32mA through ECC99 but i don't know how i calculate voltage drop i get after anode resistor. Well, to make it simple i took actual 150V from original 8k2/ECC88 and change resistor to 4k7 i get 32mA. From this two numbers i get 150,4V. Am i correct? I know there are also other parameters i should take in account but my understanding is not that far at the moment.
Reason for this change is that i would like to change the sound a little bit, learn something new and also get more options about the output transformer. Beside this ECC99 costs less than good ECC88.
Some of my gear have 600 ohm input and my output needs to be capable of driving it. It also needs to be balanced because in some cases (in friend's studio) i'm recording to 24ch tape recorder which is about 12 meters from preamp.

Miha

 

[rotation]

This calculation was all wrong. Last night i rechecked all with this example:
http://en.wikipedia.org/wiki/Triode
and compared my results to ECC99 datasheet:
http://lib.store.yahoo.net/lib/thetubestore/JJ-ECC99.pdf

It looks like i'm closer now, with 9k resistor i get 32mA of current and voltage on triodes is 144V. This corresponds to about -4,2V grid DC bias
With this numbers i get 2,3W for each triode which is very close to typical characteristics (like the rest).
Am i closer this time?

Miha

 

[Winston OBoogie]

I assume your H.T. is about 300V

It looks like i'm closer now, with 9k resistor i get 32mA of current and voltage on triodes is 144V.

With a 9K anode load R and 144V on the // triodes, you're about half of your stated 32mA I.

Is this a built circuit or are you plotting it?

Did you forget that there are two triodes in // ?

Toobs are pretty forgiving of things.  Play with it.   You can put the anode voltage at exactly halfway or put it slightly above or below. On each stage, I'm usually not at the halfway point, but only by a little bit.

Having two stages (input and output), you can make them be in anti-phase to each other if you want to cancel some distortion.  You could also play around with the operating points to get more/less flavour.    Salt to taste...

 

[rotation]

Hi,

my HT is actually a little lower but in caluclations i assumed it is 290V. I haven't rewire it for ECC99 because i'm still waiting for my order. Anyway, i'm learning something even by calculating.

Did you forget that there are two triodes in // ?

Well, i got very confused because of this. I still don't know what happens with voltage after resistor. In example of Red i get back to 290V if i calculate voltage with Ohm's law. But i meassured 150V or something on anodes. I know voltage in // circuits doesn't divide while current does. I don't know...

You can put the anode voltage at exactly halfway or put it slightly above or below. On each stage, I'm usually not at the halfway point, but only by a little bit.

I'm not sure what you mean here. Are you talking about // triodes or input and output stage?

I assume your H.T. is about 300V


Having two stages (input and output), you can make them be in anti-phase to each other if you want to cancel some distortion. 

How would you make them be in antiphase, do you mean making push-pull or maybe adding cathode follower where signal doesn't change phase compared to pentode where it does?

 

[Winston OBoogie]

i'm learning something even by calculating.

That's good.

But i meassured 150V or something on anodes. I know voltage in // circuits doesn't divide while current does. I don't know...


Yes, current will divide.  Hopefully equally but not always so.  In this case, equally enough I think.

You can put the anode voltage at exactly halfway or put it slightly above or below. On each stage, I'm usually not at the halfway point, but only by a little bit.

I'm not sure what you mean here. Are you talking about // triodes or input and output stage?



I'm talking about any class 'A' stage. 


How would you make them be in antiphase, do you mean making push-pull or maybe adding cathode follower where signal doesn't change phase compared to pentode where it does?

Yes the pentode V.A. inverts but so does the output stage in your amp.
Each stage in the amp is an inverting stage and each stage in the pair is in anti-phase with the other stage.  You can play with the operating points of one or both to cancel some non-linearity - or not if you prefer.


Ciao