Parallel Triodes for AC Coupled CF

[Mondy]

Hello!

I am putting together an AC coupled Cathode Follower based on a 6922 tube as below...

As the 6922 has parallel heater connections only, I would like to use the other triode in the bulb in parallel. Could anyone advise how I should connect the second triode? Do I just connect anode to anode, grid to grid etc? What changes do I need to make to the values of the resistors to retain correct operating points? (R15 is the resistance I found I needed to remove the bass response peak with the transformer the CF is driving)

Also, Is there any advantages/dissadvantges to paralleling the second triode in this circuit?

Cheers
Ray

 

[ruffrecords]

If you have a spare triode then you should look at alternative ways of using it other than simply paralleling them. Check out the White follower for instance.

Cheers

Ian

 

[Mondy]

Thanks for your help Ian!

The follower is driving a transformer 4:1 (load on the tx about 1K) so I didn't think I needed the very low output impedance the white follower offers.

 

[ruffrecords]

Thanks for your help Ian!

The follower is driving a transformer 4:1 (load on the tx about 1K) so I didn't think I needed the very low output impedance the white follower offers.

I have said this many times but it does not hurt to say it again. Output impedance is no indicator of drive capability. The White follower is a push pull circuit which means that for a given quiescent current, it can drive four times the power of a single ended stage, like a cathode follower, into the load. It just so happens that this also produces a rather low output impedance which is helpful for bass response. The problem with cathode followers is that they are single ended and unless their quiescent current is high, their drive capability is poor. I guess the big question is what is the maximum signal level you want to drive into your 1K load?

Cheers

Ian

 

[PRR]

> parallel heater connections only, I would like to use the other triode in the bulb in parallel.

I don't follow. The heater is *insulated*. Doesn't matter how you connect the heaters. Within wide limits you can the two sections for two purposes.

IMHO (in my Hasty opinion)- just parallel the sections, using the same resistors. The stage probably has ample drive (not specified) with one section, two sections makes it a little better with no brain pain and only slightly increased current.

> Check out the White

WCF stacks voltages. As this tube voltage drop may be 70V slammed, two of those under just 250V does not leave a lot of room. WCF output impedance is even lower than the CF, and he already wants to pad-up the Zout to damp a resonance. WCF will approach 0.000% THD but only if biased just-so; the CF will be >0.1% pure 2nd for any reasonable bias. And the upper cathode-heater insulation will be at +125V rather than 80V for CF with his values. All of which can go either way depending on designer preference and test spec, but I would not run to the WCF (even though I like it a lot).

 

[ruffrecords]

I had missed the part about a series resistor to 'remove a bass peak response'. 4K7 is a rather high value for this which suggests there is something else wrong that is causing the bass peak. You always get a bass peak caused by the resonance of the output cap and the primary inductance of the output transformer. 2u2 is not too low a value for an output cap but if you get a significant in band bass resonance then it implies your transformer does not have enough primary inductance. Which transformer are you using?

Cheers

Ian

 

[CJ]

how are you measuring the bass peak? with a scope?

try switching scope to DC input if not doing so already.

AC select puts a cap in series with the probe which can cause the bass peak,

we fell victim to this many moons ago when plotting transformer response curves, inductance may increase with a decrease in frequency but since frequency is also dropping, the reactance of the transformer primary tends to stay the same, ie: from 20 to 10 hz, inductance might double but frequency is halved which means no bass peak should be encountered.

getting spell check on reactance  wth?

 

[Mondy]

Thanks, Ian! The transformer is a Carnhill VTB 2291 with Pri and Sec wired in series (9600ohms > 600ohms). I ran a test using my sound card and ARTA frequency response software and there was a bump at the base end. I soldered in a pot for R15 and dialled till the response was as flat as I could get and the nearest standard is 4K7. Not a lot of info on the transformer from Carnhill other than the DCR of the windings, so hard to say on what the primary inductance is. Assuming 20Hz bottom end and 9600ohm primary, the shunt inductance shouldn’t be too bad and bass end is plenty. Do you have experience with this transformer?

I thought the same as you that 4K7 seemed a little high and most of the schematics I have come across on the interweb don’t have any series resistance into the transformer at all and most have a similar value for the output cap. Do you always try to correct any resonance peaks with tube designs and transformers or do you normally just ignore them?

Thanks PRR, your thoughts were pretty much the same as mine on the topology (but mine without the wisdom!). I do not want or need 0.0000x % distortion and the reason why I am playing with tubes is for this reason, I like a little 2nd harmonic even if it is just a tiny smidge as in a CF config.  But what I do need is clarity. I am driving about 30 foot of balanced cable at the worst case into about 1Kohm and If I use as a straight DI, worst case input would be around 1.0v rms.

What I meant with the 6922 heaters being in parallel is that you cannot just light one triode up and leave the other off as the heaters in the bulb are just wired in parallel (unlike a 12AX7/ECC83 where you can do series or parallel heaters). If I left one triode out of use it would eventually die so I thought it would be better to parallel the triodes. Do I just wire the triodes anode to anode, grid to grid and cathode to cathode and keep all the resistors the same value or will I have to reduce the cathode resistors R5 and R2 by half to keep the same operating points?

CJ, I have not hung a scope off the output yet, but I would have never of thought of the probe cap causing any funny business. That is very handy information to know and thanks!

 

[johnheath]

Yes - as PRR mentioned just couple the anodes, grids and cathodes… the values of the resistors and caps may need a tweak and then you have two triodes in parallel.

If you want you can take a look at the Redd 47 schematic. They used a 6922 before the output transformer… even though it is not a ac cathode follower.

Parallel or series heater is referred to if you use ac or dc on the heaters… you can use either with 6,3V on the 6922. The 12AX7 you mention needs 12VDC or 6,3VAC.


Best regards

/John

 

[emrr]

Do you have a secondary load?  That may solve the bump if you don't.  Lowest load condition you expect around 1K?  Load the secondary with 1K2 and see what it looks like, with the 4K7 removed.

My experience suggests the transformer is a proper old school design intended for matching conditions, with a lot of inductance, and the 4K7 is bringing it more into it's designated source Z range.  If that's true, another paralleled section will only make it worse.  In fact, a 10K winding only really needs 1mfd or so with a plate coupled triode to get practical bass response, a cathode follower is gravy, yet can add problems like you see.  In other words, you may need a crappier transformer to retain all other values. 

It is old and apparently forgotten knowledge that a transformer driven with a substantially lower source Z will tend to develop a smiley face EQ curve, boosts at the top and bottom of it's effective response range.  This was during a time when transformers were truly designed for matching source Z.  Since the 1970's, many transformers are designed assuming a bridging source Z, and are unsuitable for use in 'classic' tube circuits with matching values.  There are modern cases of both types. There are a handful of legacy transformers still being made that work with matching conditions, this Carnhill is possibly one, and anything in the 800 and 850 series from Hammond are other examples. 

Really, the 4K7 is undoing everything you think you want the cathode follower for. 

 

[Mondy]

The 12AX7 you mention needs 12VDC or 6,3VAC.

Damn, I never knew that!!

Thanks for the info. I see that the REDDI 47 schem ties the anodes and cathodes together but keeps the grids separate, each being fed from the input via its own 1K  grid stopper. This looks as though it makes sense and I will do the same unless anyone has any objections!

Can you recommend any tweaks to the resistor values that I have at present to better suit this parallel arrangement?

Thanks Doug, the way I tested with Arta was to run the output from the CF into the intended input on my mixer and then use the output of the mixer as the source for the software.  First I checked the mixer channel alone  in ARTA and got a lovely flat response with no bass peaks. Then I ran the same signal into the CF and then into the mixer and there I saw the bass hump from the transformer resonance. So really the transformer is running into its intended load.

Really, the 4K7 is undoing everything you think you want the cathode follower for. 

Agreed! I have been scratching my backside over this thought for quite a while!

 

[emrr]

Forgot to ask, maybe I missed it, what frequency and size of resonance? 

The transformers intended load may truly be 600 ohms, or at least something well lower than your mixer.  It should be loaded experimentally and evaluated without the 4K7 first.  Adding the 4K7 without first adjusting the secondary load is backwards procedure. 

 

[ruffrecords]

Thanks, Ian! The transformer is a Carnhill VTB 2291 with Pri and Sec wired in series (9600ohms > 600ohms). I ran a test using my sound card and ARTA frequency response software and there was a bump at the base end. I soldered in a pot for R15 and dialled till the response was as flat as I could get and the nearest standard is 4K7. Not a lot of info on the transformer from Carnhill other than the DCR of the windings, so hard to say on what the primary inductance is. Assuming 20Hz bottom end and 9600ohm primary, the shunt inductance shouldn’t be too bad and bass end is plenty. Do you have experience with this transformer?

I use that transformer a lot in my tube designs but generally wired 2K4:600 and with a 4.7uF coupling cap. Wired for 9K6 primary its primary winding inductance is over 100 Henries.. With 2.2uF that resonates at 10Hz so you might see the beginning of this above 20Hz. If you changed the cap to 4.7uf then the resonance would be around 5Hz.

Cheers

Ian

 

[Mondy]

Thanks for the info Ian, appreciated!

2K4:600 and with a 4.7uF coupling cap.

Wouldn't you get a bass bump around 15Hz with this combination (assuming around 25H primary L when connected in parallel)?

 

[PRR]

> getting spell check on reactance  wth?

Right-Click, "Add to dictionary"

It's not a word most FaceBookers would use deliberately.
______________________

> test using my sound card

What is the output capacitor in this sound card?

Since you don't want a Precision design, (as EMRR says) I suggest you leave the "4.7K" in pencil until you have the tube hot, and the actual loads you will use. Tuning the R-L-C bump can mean much larger C (100X C only shifts F by 10X) or some R. But 5K on a "10K" load seems wrong.

 

[ruffrecords]

Thanks for the info Ian, appreciated!


Wouldn't you get a bass bump around 15Hz with this combination (assuming around 25H primary L when connected in parallel)?

That is correct. In practice you barely see it. The 25H I have only measured at 100Hz. In this type of transformer, inductance tends to increase at lower frequencies so the resonance is lower than you would expect. This is a series resonant circuit so at resonance it looks like a very low impedance - pretty much the dc resistance of the transformer primary - I don't remember the exact value of this for the VTB2291 wired 2K4:600 but I think is is around 100 ohms, maybe less.  The output impedance of the tube amplifier I use to drive it is about 150 ohms. It is quite happy driving a 2K4 load but seriously struggles with 100 ohms. The result is its response is well down at resonance which largely offsets the peak. The real solution is to move the resonance lower still. A 22uF cap would take it down to 7Hz or less but 22uF film caps are big and expensive.

Cheers

ian

 

[Mondy]

It was a while ago so I thought I check the circuit again. Using ARTA Freq respose check, the bass bump measured 3 dB at approx 8Hz and it took 3K7 (not 4K7 as I posted in the OP schem) to reduce the bump to as close as flat as possible.

I do not get the chance to check Doug's suggesting of loading the secondary with additional 1K2 resistor, will have to check that next when I get the chance.

Do you think this is worth correcting?

 

[ruffrecords]

Do you think this is worth correcting?

No.

Cheers

Ian

 

[abbey road d enfer]

Parallel or series heater is referred to if you use ac or dc on the heaters… you can use either with 6,3V on the 6922. The 12AX7 you mention needs 12VDC or 6,3VAC.

???
12AX7 (as well as 12AU7 and 12AT7) heaters can be wired in parallels or in series, regardless of the nature of the voltage, AC or DC. Some people tend to favour DC for series connections, because 12Vac would potentially be noisier,though.

 

[abbey road d enfer]

. I ran a test using my sound card and ARTA frequency response software and there was a bump at the base end. I soldered in a pot for R15 and dialled till the response was as flat as I could get and the nearest standard is 4K7.

While you were at it, have you measured the low-frequency distortion? I would think it would be quite high with such a high drive impedance.