[R-390] killer and mica caps

Wed Nov 2 12:22:20 EDT 2016

A while ago I proposed the Supertex LND150 depletion-mode high-voltage MOSFET to protect the filters.

It occurred to me that under some circumstances, RF signal current into the filter might be enough to cause pinch-off, so I checked.

I'm happy to say it's fine.

Details:

I don't actually have an LND150 installed.  Neither did I connect a current probe - I didn't feel like removing the IF deck.

Instead, I measured the impedance and voltage.

I used an HP 400 wideband AC voltmeter, with a calibrated 10x attenuator probe to minimize capacitance.  I put the probe on V501's plate.

I measured impedance with the radio off, using a generator and series resistor and adjusting the resistor so the voltage at the filter input was half the generator output.

My 8kHz filter input impedance is 20-30K in a broad frequency band centered on 455kHz, but tuning carefully, I detected one sharp notch on one skirt.  At the very bottom, the impedance fell to 5K.  I assume the other filters are similar.

The R-390A IF signal chain is designed so that loud signals make the final stage, V504, overload first, so that the fall in cathode current will make the CARRIER meter read upscale to give the operator a straightforward overload indication in MGC mode.  I decided that if the LND150 didn't interfere with this, it was okay.

Now I turned on the radio and measured V501's RF plate voltage with signal applied at the antenna jack.

With the radio in MGC mode with RF GAIN at maximum, I applied just enough signal to overload V504.

V501 had about 0.5V of RF.  At 5K impedance, this is 0.1mA rms, 3dB above one tenth of the LND150's guaranteed minimum Idss.

Therefore, worst-case, the LND150 has 17dB of headroom before tripping.

As RF GAIN is reduced, more input is needed to overload the deck.  I repeated the experiment at several settings.

It turns out that worst case is RF GAIN about 7.  At that point, it takes 1.0V (6dB more than 0.5V) at V501 to overload V504.

So the LND150 has 11dB of guaranteed worst-case headroom.

This was at the point of overload.  When I turned my generator to maximum, filter input rose to 10V.  Therefore, it may be possible for a signal to make the LND150 limit, but only well after V504 is already limiting.

Finally, I switched to AGC mode just to see what the filter input is under normal operation.

It was hard to measure due to LF noise swamping the meter, but I believe it's 10mV or less.

Regards,

Dave Wise

________________________________
From: David Wise
Sent: Monday, October 17, 2016 12:38 PM
To: R390 (E-mail)
Subject: Re: [R-390] killer and mica caps

I've used the LND150 N-channel depletion-mode MOSFET as a current limiter before.  It's darn handy - and almost the only one of its kind.  It can block 500V, Idss is a couple mA, and it's very fast.

The impedance seen by the filter and its trim cap is the parallel combination of pentode V501's plate resistance and choke L505's impedance.  Both are high, so the additional kilohm or so of Rds(on) is insignificant.

The filter input coil is low-resistance, so tacking a diode across will not save it.

The R648/URR works around the problem by elevating the filter input coil to B+.  I don't like it.

Regards,

Dave Wise

________________________________
From: Larry H <dinlarh at att.net>
Sent: Thursday, October 13, 2016 5:47 PM
To: David Wise
Subject: Re: [R-390] killer and mica caps

Hi Dave,  This certainly has possibilities. I had no idea that a high voltage N-channel depletion mode (normally-on) transistor DMOS FET existed with this speed.  I'm not sure what the 1k ohm resistance would do to the signal path.

I was just thinking that since the filter inputs are grounded, a diode clamp might work.

Regards, Larry

________________________________
From: David Wise <David_Wise at Phoenix.com>
To: Larry H <dinlarh at att.net>
Sent: Thursday, October 13, 2016 2:08 PM
Subject: Re: [R-390] killer and mica caps

How about an LND150 current limiter?  Acts instantly.

Dave Wise
________________________________________

From: R-390 <r-390-bounces at mailman.qth.net<mailto:r-390-bounces at mailman.qth.net>> on behalf of Larry H <dinlarh at att.net<mailto:dinlarh at att.net>>
Sent: Monday, October 10, 2016 8:53 PM
To: k6fsb at arrl.net<mailto:k6fsb at arrl.net>; R-390 Forum
Subject: Re: [R-390] killer and mica caps

Sorry Renee, There are no fuses that I know of that would open before the filter would be damaged.

Chris, no need to hide, this is a good learning tool.

I also wanted to define what I meant by test the cap before installing.  Apply the rated voltage (not higher than) and look for leakage with a sensitive amp meter (around 50 micro amps full scale) in series with a 10k ohm resistor.  If there is any current flowing at all, do not use it for C553.

As for measuring the distortion, in this situation, you would need to be able to measure the IMD of the IF unit.  This takes special equipment, but in any case, this is a sensitive cap for minimizing IMD.

Clearing 'self healing' event possibilities can not be done today, as was pointed out, they normally occur because of contamination (and other issues) in the manufacturing process, and it takes a varying amount of time for insulation penetration/breakdown to occur.  However, you could test them as the manufacture does, with a recording scope monitoring for current spikes while measuring for leakage for an extended time.  If you apply twice the rated voltage and you see no leakage, that does not mean that an event will not occur in a few months or a couple years.

I am all for finding a longer life cap for C553.  Does anyone know of any tests done by independent testers?  I'm not too keen on believing this kind of data provided by a manufacturer.

Regards, Larry
______________________________________________________________
R-390 mailing list
Home: http://mailman.qth.net/mailman/listinfo/r-390
Help: http://mailman.qth.net/mmfaq.htm
Post: mailto:R-390 at mailman.qth.net<mailto:R-390 at mailman.qth.net>

This list hosted by: http://www.qsl.net<http://www.qsl.net/>
Please help support this email list: http://www.qsl.net/donate.html