EMRFD Message Archive 7392
Message Date From Subject 7392 2012-04-02 07:00:45 lb3hc RF voltage detector design? I am working on some ideas for a LDMOS linear amp design and I would like to design a voltage detector for shutting down the input drive and bias if voltages in some of these points get over a certain level:
-Over capacitors in output filter
-Other places deemed necessary
I am going to read the inputs with a microprosessor via an AD converter or alternatively having a set of comparators with reference levels sensing voltage levels. If one of the comparators triggers, the shutdown signal is issued. (Logical OR).
I was thinking of using a diode detector and a rectifier as well as a low pass filter.
However do you have better ideas? Ideally the circuits should not influence the circuit that is probed. It should also be differential so I can sense voltages over a coil in a filter circuit for example. If the detectpr was high impedance with AC coupling this could possibly be achieved without the measuring circuit affecting the circuit or component that is probed too much.
7393 2012-04-02 19:47:36 Gary, WB9JPS Re: RF voltage detector design? For some ideas, download the schematics for the Elecraft KPA500. They use a directional coupler, then a diode and capacitor to sense the input power. That's followed by an opamp buffer, and off to the ADC. Works like a champ. There are other sense points in there, too. This amp is absolutely bulletproof; I've made every operational mistake (!) with no harm at all.
7394 2012-04-03 06:29:41 Tim Re: RF voltage detector design? I think you're in luck - because in each of the cases for transistor protecti 7395 2012-04-03 19:38:47 William Carver Re: RF voltage detector design? I'm working on a small KW-level tuner with a series C-shunt L-series C
T-network. Like a lot of my projects, I started it with a "goodie", a
beautiful little ceramic vacuum capacitor. Fine, except its rated a 3KV
working, 5KV test. It was tested to 6.5 KV, but if sparks jump across
the plates it would be toast.
After modeling I realized a ground-based voltage measurement would not
do: I "HAD" to think of how to measure the voltage across the
Unlike a power amplifier, minimum capacitance is not needed. If you look
at the conditions under which you want very low capacitance, you will
see that even QRP power levels can generate kilvolts RMS across the
variable. So a pF or two of additional capacitance would not compromise
the tuner function.
So I came up with the idea....and just an idea, I have not tried it
yet.... to use an AD8310 log-detector for each capacitor. They have
differential inputs, so I'm thinking two capacitive dividers, one from
each side of the tuning capacitor, will give me an output proportional
to the log of the RF voltage across it.
I'm thinking of two 1 pF 10KV ceramic capacitors, one to each side of
the tuning capacitor, 1000 pF feedthrough capacitors on the wall of the
small "voltage detector" box, supplemented with a second adjustable
divider to the chip. No DC path needed for the chip.
One of the second-stage dividers would be adjusted to produce minimum
output when the tuning capacitor was short circuited. With the short
removed it then would act like a logarithmic RF voltmeter, one for each
cap. I'm thinking I'd have a yellow LED light at 3KV, a red light at
4.9KV, and unkey the transmitter at 5.0KV.
Rough thinking at this point. If anyone knows "hey, that won't work",
I'd be interested.
On Tue, 2012-04-03 at 10:59 +0000, Tim wrote:
> I think you're in luck - because in each of the cases for transistor
7396 2012-04-03 21:28:18 Russell Shaw Re: RF voltage detector design? 7397 2012-04-04 07:20:43 William Carver Re: RF voltage detector design? On Wed, 2012-04-04 at 14:14 +1000, Russell Shaw wrote:
>Perhaps my explanation wasn't clear. The AD8310 has differential RF
> > So I came up with the idea....and just an idea, I have not tried it
> > yet.... to use an AD8310 log-detector for each capacitor. They have
> > differential inputs, so I'm thinking two capacitive dividers, one
> from each side of the tuning capacitor, will give me an output
> proportional to the log of the RF voltage across it.
> You're getting the difference between two voltage magnitudes, but if
> there's a phase shift between them (ie 90deg), the result will be
> wrong (need vector subtraction). Any waveform distortion adds to the
> error too.
> Make a 100:1 capacitive voltage divider using a 10kV 1pF and an
> ordinary 100pF ceramic. A twisted-wire gimmic capacitor could be used
> too. Connect the divider directly across the capacitor to be measured,
> rectify the voltage across the 100pF cap, and figure out a way to
> measure that DC whilst it is superimposed on a maybe somewhat high AC
inputs and the two capacitive dividers feed the two inputs on a single
AD8310: it IS working with those as vector quantities. The output is a
dc proportional to the log of the RF voltage across the capacitor as
long as the two dividers are equal.
But your idea is very good Russell, and a helluva lot less expensive.
Since the idea is only to protect the capacitor, is the wide dynamic
range of the AD8310 really needed?
If two 1 pF capacitors are tied to each end of the variable, then to
each end of a single 500 pF capacitor, 5 KV across the variable would
develop 5 volts across the 500 pF, a bit easier for an inexpensive
shottkey diode detector to handle.
For operation down to 160m that 500 pF could be shunted with a 5K
resistor for diode dc current flow and negligible effect on the
capacitive divider, and 100K resistors from each end of the diode
detector to the monitoring/display circuitry.
7398 2012-04-04 17:09:20 Russell Shaw Re: RF voltage detector design? 7399 2012-04-04 17:27:36 Russell Shaw Re: RF voltage detector design? 7400 2012-04-04 17:44:48 Russell Shaw Re: RF voltage detector design? 7401 2012-04-04 19:39:15 Alex P Re: RF voltage detector design? How are you handling the temperature drift?
7402 2012-04-04 20:33:15 Russell Shaw Re: RF voltage detector design? 7403 2012-04-04 23:03:56 William Carver Re: RF voltage detector design? > If the waveform being measured is not very sinusoidal, reversing theThe capacitor waveform is quite sinusoidal since its heavily lowpass
> diodes for a negative measurement might be of interest.
> Beware of capacitance between the whole circuit and ground. It could
> affect the circuit under measurement if the capacitances are small.
> If the loading isn't too great, the rectifier could drive the LED of
> an optocoupler and you could measure current at the opto output
filtered in the transmitter (even my 20W exciter has 7 pole lowpass
filters). If the detector load was tens of uA the diode would probably
be quite linear over a volt or so.
With volts of RF I don't think threshold temperature change in the
shottkey rectifier would create much error. If it did, a second diode
into a differential amplifier could be used.
I need to do some SPICE modelling using real diode models. I can add in
stray capacitance to ground and resistors to ground-reference
Thanks for the input/ideas.
Bill - W7AAZ
7404 2012-04-06 12:12:32 Tim Re: RF voltage detector design? Sounds like a differential HV scope probe. Such things do exist, e.g. Probemaster 4241A, and it costs a kilobuck+.
However... if peak voltage across the capacitor is all that matters... I could imagine a simple diode peak voltage detector (probably with a voltage divider preceding it) ACROSS the capacitor, with it's output optocoupled to a shutdown circuit. 15kV and 20kV optoisolators are stock devices at Digikey and Mouser, about $5-$7 a pop.
I used to work with 7-story Van De Graff's, and it was very common even back then to have a "high voltage platform" and everything from there was optocoupled to the operators. The optocoupling in some cases was a telescope or insulated periscope to let you look on the meter face up on the platform!
I myself in my shack have tried series-C antenna matching networks but usually come back to L networks where the shaft of the variable capacitor is at ground. This invariably makes them low-pass networks. Which is a shame because for my 160M antenna I think I want a high-pass network to keep the local BCB blowtorch out. But... at night when the local BCB station goes to low power and when I actually am on 160M,... the low-pass matching network works OK.
7405 2012-04-07 16:36:52 Bill Carver Re: RF voltage detector design? Lowpass woud be preferable to the highpass topology from a harmonic rejection standpoint, athough into a reactive load that topology does NOT guarantee harmonic rejection. But the component values for the C-L-C lowpass network are much much larger than my 650 pF capacitors.
It would simplify the voltmeter by having one side at ground, but it's not that big a deal to make a differential detector. I'm still in Colorado so haven't done a SPICE simulation but I'm pretty sure the diode detector will work. A slight disadvantage is somewhat slower speed than the log detectors, but I'm not sure it makes much difference: if the overvoltage trips the envelope of the keyed waveform can't be interrupted instantly anyway. Although I guess I could go into the "grid overcurrent trip" circuit and that WOULD at least take the amplifier offline. Talk about key clicks!
Optoisolators would be a great idea and I've done linear isolators, but calibration and drift make them more trouble than they're worth and the detectors are asked to produce more current and higher dc voltage than I'm interested in.