[ARC5] Grid Block Keying an ARC-5

[Kenneth G. Gordon]

On 13 Jan 2014 at 19:48, Tom Bridgers wrote:

> Hi Ken,
> Would you be willing to elaborate on "now I know why"?  Which
> I take to mean something else was required to eliminate the chirp.

Hi, Tom:

What I meant was that it took something like 50 years for me to learn 
many of the tricks to elminate chirp with the ARC-5 transmitters, and 
MOPAs in general, many of which Dave Stinson so graciously provided 
with his series of e-mails and discussions on the proper ways to use 
an ARC-5 transmitter.

There are two collections of his wisdom on this subject on this page:

http://www.w7ekb.com/glowbugs/Military/arc5pages.htm

See 1) and 2) at the link above.

Since then I also learned many other things about building cleanly-
keying simple MOPA transmitters from other sources.

> If you do that something else (to eliminate the chirp), then what
> benefits does one get by using grid block keying an ARC-5?

Well, IMHO, Blocked-Grid or grid-blocked (not grid block) keying of 
any transmitter is that keying method which results in the least 
"change" from the unkeyed to the keyed condition of the circuits in 
question (in other words, it seems to me to be the "easiest" on the 
keyed circuit), and secondly, the current through the key contacts is 
always very low, which reduces at-the-key sparking and locally-heard 
key clicks, and 3) it makes it very easy to "differential" key the 
rig.

However, proper implemenation of GB keying requires a negative DC 
bias supply of sometimes as much as several hundred volts in the case 
of large power tubes like the 304TL.

Even so, unless other methods to eliminate or reduce chirp are 
followed in addition to blocked grid keying, the BG keying itself is 
not going to completely elminate chirps either.

One big problem with chrips, of course, is that, in general, the 
methods used to reduce chirps can increase clicks, and vice versa.

And, of course, much of the problem with chirps in the ARC-5 
transmitters can be traced to power supply issues of one kind or 
another, usually, but not always, involving unstable voltage sources. 
And a BIG reason for chirp in any transmitter is defective capacitors 
in various circuits. I.e. those with significant ESR.

One thing most people, even technically competent hams, don't know is
that for a screen-gridded tube, the drift (and chirp is simply very 
fast drift) caused by dropping voltage (like a voltage under a sudden 
change of load) is in opposite directions for 
the effects of plate and screen voltages.

I.e., a reduction in screen voltage causes a frequency shift in one 
direction, and a reduction in plate voltage causes a frequency shift 
in the opposite direction.

I first became aware of this phenomenon by reading about it in an old
RSGB Handbook, which I still have.

With most screen-gridded tubes, there is a "sweet spot" of the 
combination of screen and plate voltage relationships where the one 
essentially cancels out the other and frequency shift is minimized.

Although this effect is most noticeable for ECOs using tubes like the
6AG7, or 802, or 837 (as used in many Navy transmitters) it is also 
true to a lesser extent for amplifiers. We don't notice it with amps 
though, but for VFOs we sure do.

In the case of the ARC-5, and although I cannot prove this, I have an
idea that David's insistence on 200 VDC for the oscillator, coupled 
with his equal insistence on no more that 550 VDC for the 1625 plate 
(coupled with 270 VDC for their screens) achieves somewhat the same 
effect. After all, we really shouldn't look so much at each 
individual circuit as at the entire transmitter as a whole.

Being aware of this effect, it is somewhat less important to build 
"granite-stiff" or regulated power supplies, although of course, the 
stiffer the better in order to eliminate as many variables in the 
equation as possible.

If I had known of David's methods 50 years ago, I would have used 
those.

Using GB keying with the ARC-5 is not difficult. It requires a 
reasonably well-filtered negative bias supply, one resistor, and 
sometimes a bypass capacitor or two, if they aren't already there.

When I did this with my T-19 ARC-5 transmitter, that transmitter had 
already been hacked pretty much to death. It had been given to me. I 
had to "restore" much of it before I could get it to work at all. C-
58, that triple 0.05 mfd can on the back wall, had been removed, and 
most of the wiring was gone. It had been modified for cathode keying 
of the final amp stage.

So here is what I did:

I built a simple bias supply using a small filament transformer, 
turned around "backwards" so that its secondary was connected to the 
normal 12V filament voltage source, a rectifier, and a simple CRC pi-
filter using something like 20 mfd caps and a 1k to 3K resistor on 
its 115 VAC "output".

Doing this will provide you with a good -120 to -150 VDC bias supply.

Then I lifted the ground end of T-53A, the oscillator grid coil, from 
ground and ran that to the hot side of the key-jack.

I lifted the ground-end of R-74, the 15K 1625 grid resistor from 
ground and connected that to the same hot side of the key jack as the 
oscillator coil connection.

Then, again at the hot side of the key jack, I installed one end of a 
100K 1/2 watt resistor, connecting the free end to a small terminal 
strip, from which I ran a wire to the negative bias supply.

I then grounded the final cathodes permanently.

Now, when the key was up, the -120VDC cut all the tubes off 
completely. When the key was down, that shorted the 100K resistor to 
ground, and returned both stages to their original state.

It worked perfectly, but as I said, it did not totally eliminate the 
chrip either, although it did seem to help.

But remember, this was over 50 years ago now.

Ken W7EKB