813 anode choke

[G3PIJpeter]

I am about to start work on a partially-stripped Labgear LG300. The VFO and wideband multiplier were removed years ago and so I intend to retain the 813 PA but drive it with a GW4GTE Multirock on all bands 80-10 including WARC and 5 MHz.
The (valve) buffer-driver is straightforward enough to design but the PA choke is another matter. Most of the advice from a web search seems to be "... Wind 4 inches of 26g onto a 3/4-inch former and then 'suck it and see' ..."
However, this approach can involve fuses blowing and fire breaking out. So . . . does anyone out there have a sure-fire tried-and-tested design for a suitable 813 anode choke (around 90 uH) that does not have series resonances (and attendant pyrotechnics) near these bands?
- Peter (G3PIJ)

[Herald1360]

Do you have any way to "suck it and see" at low power? Only time I tried this sort of thing was with a HP Vector Impedance Meter......

[G3PIJpeter]

Yes - I can run the PA at reduced volts via a variac on the EHT supply but then the choke will still have to be removed, modified and refitted and retested - possibly many times. I just wondered if anyone had been down this path before and could save me some time (and possible frustration). - Peter

[Pamphonica]

Are we talking about the small anode anti-parasitic choke or the main anode choke?
The small choke is about 100 ohms 2W with 6 wound turns over it. The main choke is normally the multi-stage wave-wound type like this one on Ebay.
http://www.ebay.co.uk/itm/Trio-Kenwo...item3a7f70059d
I don't think it's realistic to hand-wind the main choke.
I hope this helps.
Jeremy

[G3PIJpeter]

Originally Posted by Pamphonica

Are we talking about the small anode anti-parasitic choke or the main anode choke?
Jeremy

The main anode choke as RFC1
- Peter

[Pamphonica]

OK - that's the multi-stage chappie then. A friend built a 2x813 amp and bought an anode choke on Ebay from, I think, an FT101 for about £10-£12. It worked fine.
regards
Jeremy

[G3PIJpeter]

Most of these chokes will not be OK on WARC frequencies, especially 24.9 MHz. My wish is to build the choke - the information I lack is turns/diameter information.
Peter G3RZP has already suggested elsewhere the following (which may be of use to others on this forum): "... To get a 'feel' if things will work, a good test is to put a resistor equal to the anode load impedance from anode to earth - with the valve in place, and obviously, no volts on it! Then feed in at the antenna socket from a signal generator and using a valve voltmeter or 'scope or spectrum analyser with FET probe, measure than anode volts on each band when tuned for maximum volts. ..."
I think I shall wind the usual 4-inch solenoid on a 3/4-inch former and take it from there using Peter's approach, which has the great advantage of not running the risk of pyrotechnics (as the usual 'hot' method does).
Many thanks for all suggestions - Peter

[Radio Wrangler]

The technique was to build the choke as a number of dissimilar sections with a view to not having the whole choke resonant at any frequency in any of the bands.

When things really go wrong, resonance can create multiplied voltages at placs along the sections and produce Tesla coil results for a while.

The microwave gang discovered the benefits of winding conical coils for use over wide bands without gaps, which is really carrying the idea of a string of progressively scaled segments to the limit. A sweeper and an oscilloscope would be a great help, or do you know anyone with a network analyser?

David

[G3PIJpeter]

Originally Posted by Radio Wrangler

The technique was to build the choke as a number of dissimilar sections ... discovered the benefits of winding conical coils for use over wide bands without gaps ... do you know anyone with a network analyser? David

I have come across web references that say separate sections show up just as many resonance problems (for an 80-10 plus WARC PA) as a straight solenoid. The 1960s RSGB handbook multi-section choke was for pre-WARC transmitters. One of the main problems is the distributed capacitance, which one author gets around by using fewer turns and a ferrite core (!) at high power.There is so much conflicting information out there! But I haven't seen any designs for HF 'Xmas tree' chokes as you suggest. I do know someone who has offered the use of a network analyser but the 'RZP method will be my chosen approach to start with before I get desperate. This subject may be a bit 'off-piste' but I am surprised there are no sure-fire home-made designs out there - or a pirated copy from one of the commercial rigs (some of which suggest reduced power on 24.9 MHz!).
Many thanks for interesting points to follow up - Peter G3PIJ

[G6Tanuki]

When I designed a high-power grounded-grid amp to cover 2 to 40MHz continuously, I resorted to using a two-stage choke wound on a ceramic former, with a vacuum-sealed relay to switch in extra inductance below 22MHz.

[G3PIJpeter]

Brilliant!! I have a small vacuum relay which will be just the job - exactly as http://www.rfparts.com/vc2t-26-5.html around 1.5 inches tall. With the relay in close proximity to RF fields, do you feed the energising coil through RFCs (pi-wound 1 mH types)? - Peter

[G6Tanuki]

With my vacuum-relay choke-changer I didn't use any RFCs in the relay coil feed: one terminal of the coil was earthed directly, and the other was fed through a 1000pF feedthrough from a small diode-matrix whose input was the appropriate lines from a Ledex (remember them?) which drove the tank-coil tap-changer switch, so that when you changed to a frequency in the lower-frequency band the relay was energised and switched to open so removing the short on the "low frequency" choke winding.

I did it this way for 2 reasons: firstly the amp spent most of its time on the HF end of its range so the relay remained un-energised, and secondly when on the "LF" range there was then only the capacitance of the short piece of braid from the junction of the 2 chokes to the relay-terminal to worry about.

I had the "HF" choke mounted horizontally between two ceramic standoffs, with the "LF" choke vertical on a ceramic feed-through and the vacuum-relay on a PTFE slab mounted on a further 2 ceramic standoffs. 3.5Kv on the anode makes you nervous about insulation.

There was also a rather complicated interlock which removed the HT from the amplifier before allowing the Ledex to start turning, and only reapplied HT once it had stopped; I dread to think of the arcs that would have happened if it tried tap-changing with full drive.

[G3PIJpeter]

Originally Posted by G6Tanuki

... I didn't use any RFCs in the relay coil feed: one terminal of the coil was earthed directly, and the other was fed through a 1000pF feedthrough ... when you changed to a frequency in the lower-frequency band the relay was energised and switched to open so removing the short on the "low frequency" choke winding.

Absolutely clear - thank you. I have a monster two-gang ceramic rotary switch lined up for the nine pi-network taps (80-10 + WARC and 5 MHz) - the rear gang can switch the relay as your Ledex arrangement. I might just get away with the existing LG300 choke for the LF bands, with an extra smaller solenoid for the HF bands. As stated before, I shall first check for resonances with the 'RZP cold method before winding up the EHT on a variac. - Peter

[G3PIJpeter]

Final word - http://www.w8ji.com/rf_plate_choke.htm is about as good as it gets. Many thanks for interesting discussion. - Peter

[amornummi]

Some pix here of my 80m only 813 grounded grid amp.(with normal g1 and g2 voltages)
Important to make sure the grid +screen bypassed to ground properly up to 100Mc/s.
Use shortest possible copper strip anode -blocking cap - tuning cap .Make so that path from tuning cap to bottom end of valve is minimum.
Pix below have resonance about 120 Mc/s,above possible parasitic capabilities of 813. Observed by sweeper into return loss bridge with small wire loop into test port.

The only safe way to have an all band rfc is to short out half of it on the hf bands. The fatal resonances are too variable due stray capacity. In my experience pie sections always burn up, much better is a single layer , spaced with monofilament fishing line.This amp , being qrp is not so spaced.

Input pi network last

[G3PIJpeter]

Originally Posted by amornummi

Some pix here of my 80m only 813 grounded grid amp.... Pix below have resonance about 120 Mc/s,above possible parasitic capabilities of 813. .... The only safe way to have an all band rfc is to short out half of it on the hf bands.

As suggested by G3RZP earlier, I have applied a sig gen to the aerial socket, resonated the Pi network to the corresponding frequency and monitored the voltage on the anode of the 813 (with a 5k resistor from anode to deck). The voltage stayed at a consistent level throughout all bands 80-10+ WARC, suggesting that there were no series resonances within these bands.

The attached pic shows the LG300 RFC wound in segments - it is possible that the designers managed to avoid the usual resonance at 25 Mc/s without intending to. If I do hit problems later on, then I shall use a vacuum relay to switch a shunt into the existing RFC, or to insert a smaller winding. At the moment, I am constructing a MultiRock II to drive the whole thing (see S9plus.com) - via EF91 / 5763 / 5763.

- Peter

[G3PIJpeter]

I have now re-modified my part-cannibalized LG300 to cover Top Band and 60 metres only. The full saga of the DDS - EF91-5763 driver is in a back number of 'Signal'. The circuit in the 1955 ARRL handbook for a 2-30 MHz 813 PA transmitter gives the anode choke as a National R-175A which has an inductance of 175 microhenries.

Does anyone know the inductance of the anode choke (E.5032 from memory) fitted in the standard LG300?

I am thinking that its impedance at 1.8 MHz will be insufficient but would be glad to know if it is, which will save me having to wind a new choke. In any case, I shall have to construct a new pi network inductor, with extra 'doorknob' capacitors across the tune and load variables.

- Peter G3PIJ

[G4XWDJim]

Peter you have a PM

[m0cemdave]

Originally Posted by G3PIJpeter

Does anyone know the inductance of the anode choke (E.5032 from memory) fitted in the standard LG300?

I have just measured one on my bridge, which indicates 323 uH, 6.55 ohms.

[G3PIJpeter]

I knew someone would have the figure to hand. Many thanks Dave - I think that your measurement confirms that the existing choke should be OK for Top Band. All I have to do now is fabricate a rather larger pi-tank coil and switch in some extra fixed capacitors across the existing variables. I shall post some photos when completed.
73 - Peter