Sensing Plate and Screen Currents

[R.G.]

I ran into a new way to do current sense as I slogged through the tester circuits and the bias assistant.

TI makes an isolated shunt resistor amplifier with a built-in floating power supply, the AMC3300. Feed it 5V on the low voltage side and the chip makes power supply for its floating side. The floating side senses a voltage across a resistor of up to +/-250mV, and transfers that across a 4000V isolation barrier. Super neat! But about $10.00 a pop. Ouch.

I found a second and cheaper solution. Silicon Labs makes the Si8920 which does much the same thing, sense a couple of hundred millivolts and transfer it across a 1000V barrier to a low voltage side. They're ~$3.00 or so, but don't contain a power supply for the high side. But ReCom makes 1W 5V to 5V isolated DC-DC converter that transfers 200ma (that is, total of 1W) across a 3000V barrier. They cost $1.60 each.

So it's possible to sense a -floating- current by putting a sense resistor in the plate and screen leads and using an SI8920 to ground reference it for about $5 each. Not perfect, but better.

And, in re pdf64's comments on current sensing, this kind of setup WILL sense a plate short to heater instead of ground. Likewise a screen melt, shorting to other electrodes. I'm still slogging through the circuits, but this kind of circuit opens up some much more advanced protection circuits.

[roberto]

Great! Using voltage dividers we can sense whatever current we want without going too low on the series resistance.

[luix]

Those are interesting IC.

Unlucky the AMC3301 claim an input bias current between -40uA and -20uA, this makes this IC not suitable for signal triode and pentode current measurement.

I was studying the AD8479 to make a continuous measurement of anode current but with this IC raising the shunt resistor will raise the error.

I think that the solution will not be simple...

[R.G.]

You're right, luix, the bias currents are too large for measuring plate and screen voltage for signal tubes.

However, it's a matter of objectives. For characterizing signal tubes, you would not use this device, and perhaps you would not bother to try to separate screen from plate current for signal pentodes. My objective is partly to measure plate and screen currents for power tubes to characterize them, but also for a protection scheme.

The reasoning goes like this: if you think it's worth spending some amount of money and time to make component failures not kill other parts of the amp, How do you choose which ones to protect? My thinking is that you list all the parts in the amp, and start protecting the most expensive ones first, continuing until you run out of money, time, patience and practicality.

I mentioned this in an earlier post. Excepting the outer enclosure, speakers, etc. the most expensive parts to replace are usually the PT, OT, possibly the choke, and then the power supply rectifiers and filters. So my my thinking, it's worth some effort to keep those from dying. The PT is relatively easy to protect, as fuses will stop overcurrents from killing the PT if you use enough of them and properly selected values. This same approach is possible with an OT, but trickier because of the unpredictability of the audio signal.

Protecting the OT and possibly rectifiers is less straightforward if your power tubes can draw overcurrents that don't go through the cathodes, as pdf64 mentioned. That pretty much says that to sense overcurrents in power tubes, you really need to sense the actual plate current. For power tubes, bias current is far less of a consideration as the currents are so much larger.

I reasoned that something on the order of US$5 per tube might possibly, maybe, be usable for live protection of the PT and OT in an amp. The general idea is that if you can sense currents, you can have logic make decisions about whether to let the amp continue with what it's doing or shut it down. The "shut it down" comes in the form of a power MOSFET in series in the plate and screen supply lines, or since power MOSFETs are cheap, one for each. This approach offers a way to stop collateral damage if a tube goes rogue and shorts plate to heater, or screen to whatever.

Anyway, that's the idea...