Quickie as cathode follower help needed

[cpaul]

A while back Dr. Toobz had posted a thread about wiring Quickie as a cathode follower (I can't find the thread right now).  He was using a choke in the cathode (150H, 3700 ohm!).  While I love my PJCCS version of the quickie, its gain is too high for the amps I typically use it with and it seems to trip one of my amp's protection circuits every once in a while, perhaps from the noises it puts out (need to check for DC offset, but I don't think that's it).  Also the microphonics are often a distraction, so I decided to breadboard a cathode follower version to see what it's like.

I don't have a suitable choke right now so used a straight resistance as load.  I wired the B+ batteries directly to the plate (wired as triode) and wired the output to the cathode pin.  "Below" that pin I wired in a 4.7k resistor to ground (no bypass cap).  I thought I should keep the circuit as close to the 5k resistance of the standard Quickie, and chose 4.7k because it was the closest thing I had on hand.  I included the 470k resistor at the output.  Filament batteries were wired per Quickie.

The sound is VERY promising and there are no microphonics or odd noises (at one point there were some pops at low volume but I think that was a wiring issue).  I can't get the sound level even between the channels but that may just be the junk box pot I'm using.  The main issue is that the sound level is WAY lower than the Quickie.  I think of Q as having relatively low gain, so it should be close to the cathode follower version (assuming the CF is pretty close to unity gain).  What I'm getting, however, is well below Q's output, and I think well below unity gain (well below what it would be if I wired a CD directly to the amp with no attenuation).  I will try to measure output at some point just to see.

Any ideas?  Should I be using a different resistance in the tail of the cathode? Any  reason why the 3S4 would end up well below unity gain as a cathode follower?

Thanks for any input!

 

[cpaul]

Hmmm.  I think I may have wired in some 220k resistors as grid stoppers on pin 3 (from the input pot).  I will lower them to 1k or below if that's the case.  I don't think there's any signal current through these, though, so I wouldn't expect any real impact on sound output level.

 

[Paul Birkeland]

While what you have wired up will pass sound, it's far from an optimized cathode follower.  I would imagine that if you measured the voltage across each 4.7K resistor, you see less than 5V?

The ~4K cathode load is OK, and connecting the B+ to the Quickie plates is good too.  What you need to do for proper operation is to elevate the voltage of the grid on each channel to let the tubes draw proper current. 

The target would be 2mA of current.  2mA of current through a 4.7K resistor is 9.4V at each filament.  With your 36V B+, this leaves 26V of plate to cathode voltage and the triode curves for the tube show us that we want about 2V of bias here for 2mA of current.  What would probably get you close enough to be in the ballpark would be to bias up the grid to +9V available from the first 9V battery in the stack (the battery with its black wire connected to ground).  Connect a wire to the + output of that battery, then use a 249K resistor from +9V to each grid.  You'll then want a small cap from the volume pot to each grid (0.1uF is fine). 

 

[Paul Joppa]

(I see PB has posted while I was writing. His suggestion is basically the same as mine, except that if you move the ground as I suggest, you don't need the 249K resistor or the input capacitor. I have not edited this post, other than to add this parenthetical note.)

Your intuition is correct, the operating point is far from optimal - that cathode resistor severely limits the quiescent current. Ideally, you would want a choke with 1000 ohms DC resistance (i.e. the same as the current cathode resistor). Since you want 100 henries or more, such a choke would be quite large - I am not aware of a suitable choke in current production.

What you want is a negative voltage power supply for the load, and a positive one for the plate. It should be possible to place the ground at the + end of the lowest 9-v battery, with the cathode load resistance negative end going to the negative end of that battery. Then you have +27v at the plate, and -9v at the negative end of the cathode load. In that case, you could use the 4000 ohm resistance, the same 150H 4700 ohm choke, or a modified CCS as a cathode load. This is just speculation right now, but I'll look into it a little more carefully.

As for gain, the quickie in stock form has a gain around 2.2 into a very high impedance load. Into 50K, it's probably closer to 2.0 or about +6dB. (This is because the load resistor of 4K is about the same as the plate resistance). In cathode follower form, putting the 4K resistor in the cathode, that gain is further reduced by negative feedback - in this case, to about 2/3 or -3.5dB. So you should expect a gain reduction of 9.5dB.

With a choke or CCS, the stock gain is closer to 4 (+12dB) and the CF gain would be 4/5, about -2dB - giving a difference of 14dB.

Hope that helps!

 

[cpaul]

Ah, Paul (PB?) and Paul (J), you have helped a lot.  I have never done much studying on cathode followers.  I had assumed I'd want the same total resistance through the circuit, but if I'm at a lower current, I'm not getting much swing.  Would have made sense to run through the curves, but I've never understood how things change with a CF (and plate curves don't seem to be too relevant?).  Anyway, I'll measure the 4.7k resistor for good, um, measure. 

So I get the idea of elevating the grid relative to the cathode for a ~9v bias (and 2mA current) and I hear you both to say that will increase the sound output due to a greater swing.  I can imagine an improvement to the sound under those conditions, and that's saying something given what I've already heard. 

I'll try Paul J's idea first as I seem to grok that better, though I see they aren't much different.  I know you (Paul J) have outlined how to move the PJCCS to the cathode but it'll take me some time to grok that (was it npn instead of pnp or the other way around - I know it's out there so that's a rhetorical question!).  That's a whole new project for the future...but probably well worth it.

Thanks so much for the help!!!
Carl

 

[Paul Joppa]

The circuit is as described above for resistor or choke load. However, the PJCCS is a little more complicated. I tried to write up a simple change, but it's not that simple. Post when (and if) anyone is ready and I'll get it sorted out.

 

[cpaul]

Is it simpler if there were a true negative supply to work with?

 

[Paul Joppa]

By moving the ground to an intermediate point in the battery stack, you do have a true negative supply of 9 volts, along with the remaining 27-volt positive supply.

 

[cpaul]

Yes, I get that.  I was thinking about the stock Quickie with a grounded cathode and figured it would have to be more negative than 9v, but not for any real reason.

 

[cpaul]

Finally had time to wire up the recommended changes today due to the holiday.  Kept the cathode tied to the bottom of the 9v stack but moved the ground with input and output tied to the top of the bottom battery for the 9v bias.  3.9k resistor "loads" on the cathodes.  2.2uF cap on the output with 470k to ground (the elevated ground).

It works great!  Thanks so much for your input PJ and PB!  I have to say, this gives the Quickie with ccs a run for the money, at least when using a higher powered solid state amp (I haven't tried anything else yet).  The key thing for me is that with a higher powered amp, I get very little in the way of microphonics or other noises with the cathode follower version.  I get a tiny bit of hum but that may be because my RCA's are junk bin parts with no isolation from the top plate.  So there may be ground currents flowing in the plate?  I'm REALLY glad I tried this and I highly recommend it.  Truth be told, I don't really notice any reduction in gain.  I still have plenty to work with. My only question is will my useful battery life be at all reduced since my plate sits at only 26-27v instead of 36v.  A small drop in battery voltage would drop the absolute voltage at the plate to pretty low levels.  I know the tube still has 36v going through it due to the -9v cathode, so maybe it's only relative voltage that matters.  So should it be the same battery life?

Time to consider an official cathode follower version of Quickie?  My answer would be yes... 8)

This build started as a breadboard but I find the clip lead approach to be less than ideal.  I liked what I heard enough in the breadboard stage to go ahead and start a semi-finished build. I used a piece of galvanized tin I had around, junk box parts, and the bottom third of the wooden box I built for the original Quickie.  The original box was too tall so I cut it down and kept the bottom piece.  Perfect for this project.  Left switch is the filament batteries and right is the inputs.  I kept the gist of the Quickie layout for ease of the build. One photo shows my green Quickie in the background for comparison.  Anyway, here are some photos to whet your whistle:

 

[Paul Joppa]

In stock form, the cathode is at 2v and the plate is at 27v, net 25v plate to cathode. In CF form, the cathode is about 11v and the plate is 36v, net 25v plate to cathode. Current in both cases is 2mA. The same plate to cathode voltage and the same current means the same battery life.

 

[cpaul]

That's what I suspected.  Thx Paul.

 

[Bonzo]

So tempting!
I have a nice T-amp which has a high input sensitivity...
Thank you for having shared the project!

 

[cpaul]

I certainly recommend it!  If you have trouble understanding what I did, I hand-drew a schematic which I'm happy to share.  BUT...I don't want to do that unless the Bottlehead folks think it's OK since it's mostly their work.

 

[cpaul]

Today I listened again to the CF Quickie and at some point started hearing distortion and the sound may have gotten louder.  I checked batteries and the B+ stack was at over 8v per battery, not really a problem.  The filament batteries, though, were at .9v or even lower.  I'm guessing that was the problem.  Which brings up 2 questions:

1. For parallel filament heating, the Sylvania tube I'm using states 1.4v DC as the target.  At 10% variance from that, the low end for the 3S4 would be 1.26v, yet I was running them well below that and the amp sounds good.  I think I read PJ or PB in a thread somewhere mentioning 1v is basically the end of life with the 3S4.  But that's well below the usual minimum (10% below as the lowest you should use).  The issue I've read about with very low heater/filament voltages is cathode poisoning, though according to some direct heating cathodes are less susceptible.  So is there a potential problem operating the 3S4 down to 1v?

2. What would be the cause of distortion in this CF version under very low filament voltages?  This CF version of Quickie seems close to A2 operation with the bias below 1v, maybe as low as 0.5v, and it also seems like it hinges on both sets of batteries.  If the filament batteries drop too low and the 9v is too high, is it possible it would go into A2 and draw grid current?  Could that be what I was hearing?  And if so, is there any danger in running the amp to this point?

Thanks!
Carl

 

[Paul Joppa]

The 3S4 and related tubes were designed for carbon-zinc batteries - alkalines have a much greater lifetime but are otherwise similar. So they were intended to work well with a much wider range of filament voltages, and to not be damaged by low filament voltage.

In Quickie, the plate voltage is low, 36v compared to rated 90v. Cathode poisoning happens when the plate voltage drives positive ions into the filament, so the low voltage means less energetic ions - and less damage. In addition, Quickie operates at about 2mA plate current, compared to rated maximum 9mA. That means the space charge remains adequate to protect from poisoning down to a lower filament voltage than normal - I estimate another 10% beyond the usual range for these tubes.

Grid current is not usually a problem since the alloy used for the grid in these battery tubes is chosen to have a contact potential that delays grid current until the grid is more positive than the usual AC-heated tube - perhaps +0.5 to +1.0 volts IIRC.

 

[cpaul]

Wow, thanks Paul.  That's a very helpful answer, and things I definitely did not know.  A group of audio engineering friends were discussing the possibility of A2 operation because of the rough numbers I'd thrown out when talking about the amp.  And when I try to measure voltages to figure out exactly where the amp is operating, I'm getting odd results.  I'll assume there's no harm if it did happen, except perhaps to sound quality...

As for the filament, I have a pile of alkaline batteries from another device that have 1.2-1.3v left and I was hoping to use them up.  The first pair I used lasted something like 15-20 hours (maybe more) which I was happy with.  But knowing that meant dropping to 1v or possibly below, I started wondering about potential harm to the filaments.  I'll assume there isn't a problem.  Thankfully we're not talking about replacing a pair of Western Electric 300Bs in any case.

I do need to try this CF version with a full set of new/charged filament and B+ batteries to see how it sounds at much better voltages.

 

[Paul Birkeland]

Very low battery voltage indeed isn't great for the tube, but it sounds terrible so it's not like you're going to spend hours and hours operating the preamp in that region.

The datasheet for the 3S4 makes it look pretty resistant to drawing grid current.

 

[cpaul]

Thanks PB.  If you're willing and able, I'd love to know how the data sheet indicates resistance to drawing grid current.  It's not something I've ever worried about before so I don't know how to assess that.  Either way, thanks for the input.

 

[Paul Birkeland]

The presence of any grid current information, especially on triode curves for a pentode, is pretty rare outside of powergrid transmitting tubes. 

Still, you might try clip leading in one extra 9V battery to see what happens.  An extra 9V on the plate certainly won't hurt anything.