Switching bias current circuit

You don't need a switch. Speed wants simplicity. But to get simple you must express the electrical operation neatly.
Like, source or sink? To gnd, virtual gnd or lean and consistent over some common mode range?

Might be a tad slow but I'd look at "NPN and PNP matched pairs" as simplest current mirror, couple of resistors to set the levels and a 50-ohm sig gen to toggle between.

Such specialty chips I have seen in old Intersil UHF-1X (10V, 8/4Ghz) and THAT Corp (36V, 500/300Mhz fT). May be others.

dick_freebird said:

You don't need a switch. Speed wants simplicity. But to get simple you must express the electrical operation neatly.
Like, source or sink? To gnd, virtual gnd or lean and consistent over some common mode range?

Might be a tad slow but I'd look at "NPN and PNP matched pairs" as simplest current mirror, couple of resistors to set the levels and a 50-ohm sig gen to toggle between.

Such specialty chips I have seen in old Intersil UHF-1X (10V, 8/4Ghz) and THAT Corp (36V, 500/300Mhz fT). May be others.

Click to expand...

hi freebird,

thanks for your reply!

What we need is a rectangle current pulse (with 2mA DC component) that biases the chip (sourcing "high side", the chip is connected to GND on the other port).
I attached sketches of two designs that came to my mind so far. The high phase should be around 10ms, so period would be 20ms.

I haven't worked much with BJTs, and when I think of current mirror, I have this simple design with one diode connected device and another one with the two bases connected. As far as I remember, these devices are both of the same kind, so either NPN or PNP. How does the design you had in mind with one NPN and one PNP look like?
Also, I would like to ask how I can toggle the BJTs with a signal generator, I think the generator will be a voltage source?

Best,
Manuel

Confusingly, the diagram tells 20 ns while the text says 20 ms period. Presume you know that the difference matters in terms of implementation.

Other parameters that matter are bias node impedance and required current waveform accuracy.

Switching a current source between nodes of different voltage level involves settling transients. In high speed circuit design, we prefer non-saturated transistor switches for currents, typically differential pairs.

Very short pulses seem to be a common tendency in a programmable unijuction transistor. It's really an arrangement of two transistors similar to an SCR. You might be able to adapt it to your project.

Below is 105 MHz frequency, pulses so brief they seem more like spikes. You can place the load close to 0V ground but the supply wiring may need additional components, or perhaps you can contrive an inverted arrangement of the transistors.

FvM said:

Confusingly, the diagram tells 20 ns while the text says 20 ms period. Presume you know that the difference matters in terms of implementation.

Other parameters that matter are bias node impedance and required current waveform accuracy.

Switching a current source between nodes of different voltage level involves settling transients. In high speed circuit design, we prefer non-saturated transistor switches for currents, typically differential pairs.

Click to expand...

Hey Fvm,

thanks for mentioning the typho!

Its ns, just as in the original post.

BradtheRad said:

Very short pulses seem to be a common tendency in a programmable unijuction transistor. It's really an arrangement of two transistors similar to an SCR. You might be able to adapt it to your project.

Below is 105 MHz frequency, pulses so brief they seem more like spikes. You can place the load close to 0V ground but the supply wiring may need additional components, or perhaps you can contrive an inverted arrangement of the transistors.

View attachment 191946

Click to expand...

Hi Brad,
that sounds interesting, I'll look into it!

Best,
Manuel

UJT is saturated bipolar switch. Switching speed is rarely faster than us range.