Triode-connected tetrodes vs: conventional-configured in mod service - Ques:

[K1JJ]

I've been wondering about this lately....   Many of us run tetrodes in triode-connected service for power amplifier push-pull modulators.  In the past, I have connected the screens and control grids together and drove them this way for both 4-1000A's and 813's. The 813 rig especially looked perfect when driven with a MOSFET driver and was a simple arrangement.

Anyway, the question:  Looking at 4X1 specs for modulator service, they suggest  running them as tetrodes with about 500V on the screens, (regulated) use about -70v grid bias and to drive the grids only. It has been done this way for many years in BC modulators and PA systems - and, the Eimac engineers designed the tube to operate this way.  HOW can we possibly depart from this config and go from a stable 400 volts on the screen to tying the screen  to the control grid and swinging it all over the place with audio?

You wud think that the tube wud be most linear as designed for tetrode service and become non-linear when swinging the screen, way out of its design.  The screen has a tremendous effect on plate current, so I don't understand why there is not noticeable distortion departing from the original tube operating design of Eimac.


The reason I'm wondering is that the 4X1's in my modulator look pretty decent triode connected. But yesterday I ran them tetrode-connected and they looked even better. I put 400V regulated on the screens,  -60 bias via diodes in the fil center tap,  and drove the grids with my SS amp and backwards xfmr.  The modulators were MUCH easier to drive at about 11 watts! (as expected) and the response looked great. Still, the difference was not that big compared to what you would think, considering the huge difference in screen operating parameters. (From regulated to swinging around with audio)

I’ll run some more detailed tests today, but I’m leaning towards keeping them tetrode- connected and working out a smaller audio driver as a result.

Tom, K1JJ

[WA1GFZ]

Triode connecting the tube makes for a simple circuit more stable lower gain at the expense of greater drive. Easier control circuits. Tetrode has more gain and then the screen voltage issue.
I played with 807s in my CDC and found the lower I ran the screen voltage the greater the clean the plate swing until I ran out of drive. Also need to keep an eye on g1 dissipation when running low screen voltage.

[K1JJ]

Yep, those are the results of running terode vs: triode configured... no doubt.

But I wonder how can a tube stay linear when in one instance the screen is regulated stable and in the next instance everything is the same except the screen is now swung all over the place with audio?  You would think one is linear and the other would depart from it or vice versa.  

GG operation is similar to triode connected except being cathode driven, so we know it works, but it just seems like a "hambone" tactic for modulator use when Eimac designed and recommends the screen to be fixed and commercials have always done it this way for years.  I'm just trying to understand the tech reasons of how it could possibly work well... since I've used it myself for years...


I understand designing a tube's elements and placement are a difficult task when trying to make a tube truly linear. By simply tying the screen to the grid to make it one BIG control grid seems like an arbitary idea that wud fly in the face of linear design.

T

[K1JJ]

Another point -  Why is the 6146 tetrode connected such a lousy modulator?

I know Dino/WA1KNX wrote an artcile how he JSed with the % of audio fed to the screen and grid to linearize the modulator. Why is this required at all? Did Eimac make a poor tube design that required JSing like this?

I notice the 4X1 plate to plate impedance is only about 7,000 ohms in tetrode service (with 4KV) so it's not an issue of high plate impedance causing a problem vs: triodes with lower impedance.

T

[N2DTS]

I dont know much about tubes, but it seems to me that many tubes were designed for some operation, and JSing them into triodes might not give the same results.

The only way to tell is to measure the distortion products.
I dont have any sort of distortion tester, but they are out there.

Some tubes were really abused, the Gonset G76 used a pair of 6dq6's as modulators, in a zero bias setup driving the screen grids with audio. They made enough power, but who knows about the distortion...

You want clean audio, try some AB1 hifi type tubes sometime, although they dont make them for your power levels...
In the care and feeding of screen grid tubes book, there is a chart of old eimac tubes and their distortion products.
Unfortunetly, it does not list modern tubes or other brand tubes.
Some of the tubes had very good spec's.

As I said, I dont know much about tubes, but it seems most very hi fi amps used hi fi tubes in class AB1.
Or class A....

I dont think I ever heard of a class B hifi amp...

Now, for ham radio, distortion can be (or is) not quite as important as broadcast, and I think something has to be quite distorted before anyone notices in most cases.

When I switch between the 100th (class B) and the 4cx250b (AB1) mod decks, I could hear a difference, but no one at the other end could tell....

Brett

[KM1H]

If the screen is gonnected to G1 then it aint acting as a screen any longer, its just another grid that the cathode electrons have to go thru.

The 6146 is a RCA design, never used by Eimac.  Since its capable of -40db IMD3 in SSB service which it wasnt designed for, Id expect similar as an AB1 modulator.  It would be interesting to curve trace them and see if they are even remotely matched, especially if they are hamfest floor sweepings.

Carl
KM1H

[N2DTS]

A long time ago I tried 6146's as modulators, regulating the screen voltage, and they did not sound good, which makes no sense as they do the same job at RF in loads of ssb rigs, some of which did very good AM.
The ft101 and 102 used them I think, the 102 I know did as I had a few of those.
The only difference is they are not in push pull in RF service...

But... I dont think I ever heard a ham rig that used them as modulators sound really good.
Most B+W 5100's have a grunge sound to them.

Yet you can get great sound out of something like a 4-400, 807/1625, etc, as long as you run them as tetrodes.
I never heard anyone mention a broadcast rig running 4-400's or 4-1000's in triode connection....

Brett

 

[KC2IFR]

Hmmmm,
U know Tom.....I think it depends on the toob........as an example a lot of folks use the 813 triode connected with no problem. I use 5881's (6L6) triode connected as cathode followers to drive the 572B's in my Johnson 500.
Im not smart enough to give u a detailed explanation! As someone said.......if aint broke....dont ask any questions!

[KE4LRL]

Timely Topic - In order put a odd modulation transformer to use I have been contemplating push pull parallel 807's as a 180 to 200 watt modulator. (I have ran across a modulator circuit using 807 connected for high mu class B triodes.)   As noted above the only trade off I have noticed is the higher drive requirement vs a bias and screen supply.  Any comments or suggestion ?

[K1JJ]

Carl:  I think you hit it by saying a "screen" becomes just another control grid when it is tied to  G1.

I wonder when does a screen become a control grid?  It certainly is a screen when it has a fixed voltage on it, but maybe it is a control grid as soon as a signal is placed on it, regardless if it also has a fixed voltage too.

As suggested, I think the only way to fine tune the whole system is to run some critical linearity tests on the whole rig and vary the various parameters. A trapezoid on the scope may be a good starting point.

I did change the class C bias back to about the recommended -200V and it was MUCH easier to drive the final, but it now shows some slight plate color, indicating the efficiency dropped a little as expected.  The final now shows about the same faint orange as the modulators. I will try different class C biases in the linearity testing later too.

BTW, Here's a great article written by Dino/WA1KNX about fine tuning the linearity of a rig via the RF final's screen modulating parameters.

http://www.amwindow.org/tech/htm/scrnmod.htm

T

[KL7OF]

Tom...When you said "Triode connected they looked pretty good and tetrode connected they looked even better" .... What are you looking at and how are you looking at it?  As far as WHY the triode connected tetrode  can be linear ...My comment would be...Is it still (as) linear?  I guess it is linear enuf huh?  The screen just becomes part of the grid structure and swinging it with audio it behaves like the grid....

[KM1H]

Tom, as soon as you connect the screen to the control grid it is in parallel with it. The only voltage it sees is the - bias and the audio drive.

The 6146 was used in SSB service all the way back to the KWM-1 and was used in the KWM-2 and all the S Line by Collins. Soon many others followed.

Since it is used PP in a modulator is why I suggested curve tracing. There has to be a reason for reported problems, I doubt if the tube design has anything to do with it. Probably everone has copied the same circuit and simply duplicated the problem. Set up as per the RCA spec sheet and get out the instruments.

Carl
KM1H

[KE4LRL]

A couple of other  points to ponder:

When connected as a tetrode the screen becomes a drive control and (i believe) is also effected by the load impedance.
When connected as a triode ( grids in parallel as noted above) how do the two grids interact, if the two are additive perhaps the capacitance between the to grids interacts.

Jeff

[N2DTS]

I dont think a trapizoid is going to show you anything.
Nor would a scope, except for gross defects maybe.

To really find out you need a distortion analizer, or some very good audio ears...

Yes, when you run them as triodes, the control and screen grid get tied together, the control grid likely does 90% of the control, the lack of high voltage on the screen makes the tube zero bias or close to it.

I suspect using tubes that were not designed for such service might not be very clean when run that way, unlike some tubes that were designed to run that way, like the 3-500z , 3cx800, etc(zero bias triodes).

Brett


Brett

[N3DRB The Derb]

http://webace.com.au/%7Eelectron/tubes/screens.htm

some good reading.

[Gito]

HI

If a tetrode tube has it screen connected to the control grid, its not a tetrode anymore,but its a triode.
It's not the same tube anymore ,but a different  tube with different bias, different driving power,a triode needs a bigger driver power,and maybe a different load impedance.
And zero bias class B Triode  makes it easy to build  modulator .
So a 4X1 tetrode connected as a Triode is " not" a 4X1 anymore(not working as it was design)

Gito.
.

[KE4LRL]

I will take a look at the atrtice (Iwant to get a good handle on this theory). 

Most of the older articles I have run across only suggest the triode connection with beam tetrode 
tubes.  Could be that there is a structural similarity between a triode and  a beam tetrode that accommodates the configuration.

Experimenting with the ratio of audio applied to the two grids may prove interesting. 

[KC2IFR]

I will say it again......how the toob performs depends on the toob. If specs say it wont work......maybe someone should tell the toob......

[K1JJ]

Good reading, Derb.   I'm 1/2 way thru it and see the old buzzards of yester-year had a good handle on the technology of their times...

I'm starting to lean towards using tetrode config in the future for tetrodes instead of triode connected.
BTW, Carl and Brett: I'll bet those 6146 modulators in many rigs had poor screen regulation and maybe even bad grid bias reg. That will kill any tetrode set-up.  Check the Valiant and see what they did - I'd be curious.


Tetrode config on the modulator 4X1's -  update:

Well, I took out the 680 uf bypass cap on the 4x1 modulator screen grids and replaced it with a 4 uf.  The screens now have regulated 400V and the bias is about 90 diodes in the fil CT (-45V) for a stiff bias.  The screen spike problem is gone - fixed! I ran some tests with the audio sweep gen and also some voice checks. I think it's the best config so far!  It's also easy to drive with about 10 watts of audio or so.    The screens are drawing only about 15ma on peaks, while the Eimac specs say 95 ma is for full power - this tells me there's plenty of headroom left.  The grid is drawing no current at all. I'm idling them at about 300ma total and the meter barely move under voice peaks.  This tells me we are running close to class A, but really about push-pull AB1. That's a great mode for cleanliness.  This is on medium power tap, so it will increase the modulator requirements once I bump it up to full strap.

I'm real pleased with the tetrode configuration so far. I think it's gonna stay that way.

Still have some minor issues to iron out with the input grid tuning, but almost finished with the rig for now. (until I get the next modification bug up my ass... )

More later -

T

[N2DTS]

Yes, I guess if you dont strap it, you are likely in AB1, a good place to be!
Increasing the plate voltage on the modulator will give even more power output in the AB1 zone.

I did some tests with the old ppp 100th mod deck, turning the plate volts up to over 2500 volts, and the drive requirements went down and it sounds better than at 2000 volts.

I run out of bias volts at 2700 volts though....

Brett
 

[WA1GFZ]

Tom, the screen needs to be stable so the gain doesn't change as the plate voltage changes...unless you want it to. Screen grid was installed to increase gain.