Topic: Are a PGA a good choice as the first gain adjusting element?

[Einar]

What happens when the signal exceeds the dynamic range of the Programmable Gain Amplifier?
The gain is controlled by the feedback loop in an opamp while the full input voltage is presented to it.
I would think that invites front end overload.

Would it not be better to switch in an attenuator before any nonlinear elements as an amplifier?

And would a high voltage blocking capacitor preceded by a bleed off resistor not be a good idea on the input of preamp 14? If a static voltage builds up in the antenna it would already "sit resting on" one of the clamping diodes. And in a thunderstorm a static buildup is to be expected in a simple rod antenna.

I may be outside my actual profession here, but the elements are in some ways similar. And by flapping my gums I hope to learn more.

[Cutty Sark Sailor]

Yep, if gain set too high, signal can saturate and flatten.  I think they designed it that way for several reasons.  They don't want signals from around the world. Bad signals get discarded. Anything over about 1.2-1.5 VP out of amp will flatten at controller also.

E field detector is a probe for E fields. You don't want to bleed them off or distort them.  It operates on capacitance to air.  Dielectric between probe and ground. Nothing else should be connected. Only ground is the ground back at the controller.  Signal out of the so called 'pre-amp' is actually reduced to about 1/16 before leaving for the amplifier. Preamp is first stage of the very tight 50Khz filtering circuitry formed by it, the coax, and components in the amp.

[JonathanW]

What happens when the signal exceeds the dynamic range of the Programmable Gain Amplifier?
The gain is controlled by the feedback loop in an opamp while the full input voltage is presented to it.
I would think that invites front end overload.

Would it not be better to switch in an attenuator before any nonlinear elements as an amplifier?

And would a high voltage blocking capacitor preceded by a bleed off resistor not be a good idea on the input of preamp 14? If a static voltage builds up in the antenna it would already "sit resting on" one of the clamping diodes. And in a thunderstorm a static buildup is to be expected in a simple rod antenna.

I may be outside my actual profession here, but the elements are in some ways similar. And by flapping my gums I hope to learn more.

With respect to a switchable attenuator, not necessarily.  Since the signals in question are subsequently amplified further and all stages basically have the same power/ground rails (and components), the front end is not where clipping/compression is most likely to occur.  Further, in my experience even with signals from lightning virtually overhead, drastic reduction in overall system gain generally keeps signals in the BO system below clipping levels.  With that in mind, adding in a switchable attenuator would probably add system noise unnecessarily.  In general, it's usually a good idea to use the least amount of circuitry prior to the first input-stage amp as possible.

With a direct strike, of course, all bets are off

As far as a bleed-off resistor is concerned: I doubt that the input signals in question are generally very large, and I think static DC offset would need to be very near one rail or the other for distortion to occur.  It's probably not a terrible idea from an engineering perspective, but probably solving a problem that hasn't really reared its head in any significant way.

On the other hand, the resistor combined with the capacitance of the system (which actually varies from system to system, thanks to variance in the capacitance of the E-field probe) might introduce additional poles/zeroes; you'd need to be careful not to cut off low frequency signals of interest, or cause oscillation.

[Cutty Sark Sailor]

And in many respects, the system is not looking at "waveforms"... other than it needs to more or less match the waveforms of other 'locating' station signals.  it's looking at "Time of Arrival", somewhere on the leading stroke discharge pulse. Hopefully that will be 'ground wave' component, not 'skywave' reflection. So clipping, etc isn't really an issue. Signals are further filtered for freqs contained, and that data may be distorted by 'square waving'.... and will become a bit more important as "TOGA" (Time of Group Arrival) relating to specific frequency bands and pulses from the entire sferic sequence is more closely analyzed.... bad signals will be tossed. The server programming is quite capable of that, and more is being added. It's basically of no consequence to the network, as a whole.
One other thing that hasn't been implemented, is an "Automatic" mode, where the server controls the local system related to what it sees. That's a goal since day one, and in fact can be switched in with limited success by the local operator. It's simply not fully implemented on the server. Probably be a while.

[JonathanW]

See, here's the problem.  After spending time on Facebook, you're used to being able to "like" comments.  You can't do that here.

With that in mind, consider Mike's comment, above, "liked".

[Einar]

I see that my posting implied E-field measurements. While actually the second half of my post was and the first was the one that was relevant to the topic.

Going OT on the first post of a thread! That is probably a personal record. 
Anyway bear with me, I'm trying to understand how this works.

One of the reasons I brought this up is that I see several stations not registering strikes closer than appx. 50Km. I would think they were the best sources of timing information since the signal have been subject to less of the uncertain effects of the atmosphere. 

I did believe it was common that the server control the gain thus making it sort of an AGC. That would to me seem ideal, as the server "knows" what stations will be close to a thunderstorm, and the local disturbances of a station. And to do that, it might need to turn it all the way down to zero gain through the whole amplifier chain. In a thunderstorm the receiver is after all located inside the tank circuit of the transmitter! I would think attenuation rather than gain would be in order. I thought the manual GC was just to make it possible for the user to optimize the antenna and placement.

That is why I thought the PGA as first stage might be a problem, since even if gain is turned all the way down to zero, it might be driven to clipping. And what usually happens then is it takes some time to come out of it. That recovery time will destroy the timing information and thus render the station unable to supply any meaningful information.

Anyway it all may be a moot point if the grid of stations is small enough. Then data overload on the servers might be more of a problem than signal overload in some stations. Signal processing on each station may be how they do or will deal with that.

Did any

[JonathanW]

My first question would be, do we actually know that signals from strikes close in to the stations drive the first stage of either E- or H-field amps into clipping?

I had a storm go overhead last night,  and by reducing my gain to 5x1 on both amps,  kept most received signals well below clipping level.   What's more,  my station routinely picks up strikes less than 10 km distant with this procedure.

This is also a station that, with a gain of 32x2 on both E- and H-field, routinely picks up signals from up to 5000 km (the record for the station is over 5500 km).  So it's not a question of significant attenuation due to antenna siting, at least for H-field--once the E-field is mounted outside, I imagine the gain on that will be reduced (or the probe reduced in size).