Amps
[Top] [All Lists]

[Amps] more on the HFS1000G RFPP amplifier

To: <amps@contesting.com>
Subject: [Amps] more on the HFS1000G RFPP amplifier
From: jtml@lanl.gov (John T. M. Lyles)
Date: Mon, 18 Nov 2002 10:51:05 -0700
I have the RFPP HF1000G amplifier purring along here, in daily 
operation now. It was a very good find and I am quite impressed with 
the ruggedness of the componentry and general spaciousness of the RF 
compartment.

As Phil Clements noted, the choke input original HV filter limits it 
to about 2650 VDC in the stock machine. So the max RFout I can reach 
before serious saturation is about 1100 W. I run it up to 1200 with 
no ill effects, but it is NOT  considered linear anymore. I ran power 
sweeps with the 8753E network analyzer Thursday, and the gain does 
change over the range of power.

The choke input supply bounces as we sweep the rf drive with our 
network analyzer, and this can be seen at the beginning of the sweep 
as a little hickup. Going to Cap filtering only would improve this, 
although step start would be required as well as an appropiate 5-20 
ohm series resistance to the tube to limit energy in case of a flash 
over somewhere.

Our 1000G is now at 4.5 - 5 MHz, into a tighly shielded load. Changed 
some components in the input and output networks.  It then drives a 
1:4 stepup transmission line transformer on #43 toroid, then a small 
pi network using a couple of vac variables and a miniductor coil, 
then a tap into a highly loaded (ferrite cores) 1/4 lambda cavity. I 
get 2 kV of RF at the top of the cavity across the resonating caps, 
through all this stuff. The ultimate plan was to tie the anode of the 
8877 directly to the load cavity, through only a blocker cap, and let 
the 2500 Ohm cavity resonance be across the tube, perfect for the 1 
kW operating point. However, we got distracted (engineers disagreed) 
and wanted to have some good taps to read power, so a Werlatone 2kW 
wideband directional coupler feeds a calibrated hp 437A powermeter 
for forward power and reflected power.

I can leave the 8877 conducting all morning, at 210 mA ZSAC, using 
the original cathode resistor bias arrangement. Outlet stack 
temperature is about 50 deg C then, but with the monster Mclean 
blower it doesn't mind it at all. We just keep the room heat turned 
down, and let the 1000G become a 650 watt room heater! When the RF is 
on, the heat is less (except when I closed the loop, see below).

Because I didn't have the schematic when I cut into it, I was only 
able to reverse engineer that there was a big resistor (several 
actually) after the RF bypass Cap at the cathode. I didn't see a 
zener in the original RFPP design. Besides, 210 mA for 2650 VDC seems 
to fly higher than the 8.2 V zener would produce. I think mine has 
more like 4 V bias due to the resistor alone!

Finally, I added an ALC circuit to it last week, as I needed it to 
hold fixed forward power while sweeping the frequency +/- 100 Khz or 
more around center freq. The original HFS1000G had some sort of 
leveling loop which controlled the gain of the solid stage driver 
stage. Since I didn't plan to use that (due to having its own tuning 
response which might further limit the overall bandwidth) I drive it 
with an ENI 240L 50 watt solid state broadband amplifier. The hp 
8753E network analyzer has an AM input aux connector on the rear 
apron. I built a small feedback loop with proportional and integral 
terms, driven by a JFD detector diode off the forward coupler. This 
has a couple of OP27 op amps, and it drives this AM input to close 
the loop around the entire chain. A reference voltage from a ten turn 
pot feeds the error amp. Before closing the loop, the response was 
sometimes -6dB on the edges. When I close the loop, the ripple over 
200 kHz span is a fraction of dB. Of course the HFS1000G power supply 
really groans over the sweep, and Ip gets up to 800 mA for an instant 
as the output is locked at 1 kW regardless of reflected power or 
bandwidth of the PA. I wouldn't recommend this for amateur use into 
an antenna! But it does point out to the general heftiness of the 
RFPP amplifier. I could never do this with the Johnson Thunderbolt 
that I was using before this.

For CW in class C, one might choose 12 - 15 VDC cathode bias. Years 
ago we used a 63 volt active bias circuit for the 1500 Watt FM (class 
C) transmitters that BE sold. It was an MJ2500 darlington transistor, 
with 1N4758 zener. Emitter connected to cathode side of tube, and 
collector to a 5 Ohm R to ground.

So, choose your bias with zener if you like, or a transistor and 
zener for a bit more protection. And use some R in parallel if that 
device should open. Go for it. You'll like the amplifier you get I 
think.

73
John
K5PRO

>  > I am about to activate an RFPP-1000 amp which uses
>>  1 - 8877 @ 2750v. (exactly what '5PRO recently described.)
>
>I have converted several of these.
>If you remove the choke, add 30 mfd or so filter capacitor,
>and step-start, you will be able to set the taps on the HV
>transformer to 4 kv under load. You will then need a 12 volt
>zener for proper bias. I use the 50 watt units for simplicity.
>The 8877 seems to shift into "overdrive" @ 3800-3900 volts
>on the anode. Less drive required, and c. 2200 watts output,
>but you will need the 12 volt zener to keep ZSAC below 180 ma.
>Be sure to set the heater voltage @ the tube pins to 5 volts or a little
>less....some of these units came tapped @ 104 volts on the filament
>primary when the unit was used on a 208 volt service.
>I hope this helps.
>(((73)))
>Phil, K5PC
>
>
>>
>>  My last remaining conversion change is to chose the Cathode
>>  bias voltage to use. Here's what has been published:
>>
>>  K8RA-8877:   10Volt/1watt zener &  2N3055
>>                            " for 8.2v cathode voltage ".
>>                           Text indicates " bias for 18ma idle current."
>>                           ( That can't be for the tube idle plate
>>  current, can it ??)
>>
>>  W8ZR-8877:  12 volt/50watt 1N2812
>>                           150ma.  idle plate current
>>
>>  Ameritron-AL-1500:     7.5 Zener
>>
>>  RA's-8877 in QST:   8.2 volt 50watt Zener for SSB
>>                                      22  volt 50watt Zener for CW
>>  (Ref: Radio Handbook W6SAI / 1996 ARRL Handbook.)
>>
>>  So my questions are:
>>  ================
>>  -What bias voltage is needed for SSB ? for CW ?
>>
>>  -Advantages / Diadvanages of using a 50w Zener vs.
>>    a 2N3055 & a 5 or 10watt Zener ?(Other than $$$).
>>
>>  Charlie - N1RR
>  >

-- 

<Prev in Thread] Current Thread [Next in Thread>