Thanks for posting that.

I guess you used the other amp as a rail splitting reference because the LF353 doesn't have the -ve supply in its common mode range? The common mode range down to near the -ve supply is one of the reasons I picked the CA3130 for mine, that and the offset null simplified bias power setting.

The decision to use a bias resistor stemmed from my general dissatisfaction with RFCs (mainly due to frequency response and resonance issues) but immediately the elegance and circuit simplification that this allowed became apparent. Since the intended application was solely the measurement of low RF levels, the AC coupled input was not considered to be a limitation. The bolometer was set up using standard 5% carbon film resistors in the divider network though using a pot to set the bias level through a precision 100 ohm resistor may decrease the error by a few percent...an improvement I have not yet bothered with. The maximum power that can be measured is limited to the DC bias power though in practice, error increases around 75% of the bias power.

Bulb selection is critical since the filament resistance has to be twice the input impedance. A No. 387 lamp (28V 40mA midget flanged base bulb) was found to exhibit a 100 ohm resistance around 300mV and since the goal was to measure power levels under 0dBm, it seemed like a good candidate. Any bulb with a cold resistance lower than and a hot resistance higher than the target value would work. The schematic below should answer most questions about the actual circuit.

http://dxer.silverchris.ca/gallery2/d/1086-2/117-1716_IMG.JPG

Thanks!

It occurred to me this morning (in the shower, I seem to do my best thinking there...) that I should try using small incandescent lamps as bolometer sensors - rather like hot-wire Barretter detectors, but the response time is not so critical. Arranging for the detector impedance to be held at 50 ohms put me off a little, but I could see it working without too much effort. How do you set it up, tune the bias power for best return loss across a range of input powers, or just Ohms law on the bias supply? It does seem it would constrain the filament impedance at the bias power level, which means bulb selection is important depending on the power ranges you are interested in.

Sending some fraction of the RF through an ohmic resistor which delivers the bias power and is decoupled solves the "RFC problem" nicely, thanks for that! I was trying to make my initial unit as easy to calibrate as possible (i.e. using DC), so I used the separate resistor. If you don't mind AC-only use and some extra calibration steps that scheme allows more versatility in sensor choice. I guess now I have a bolometer that can be precisely calibrated I can use it to make a transfer power standard and precision attenuator and use them to calibrate the FSD of another bolometer... I can also finally calibrate my AD8307 power meter with some confidence!

What is a "No. 387" lamp and where do I get one? I have some grain-of-wheat lamps I bought in bulk a while back, I'll measure their properties and see how suitable they are.

I am working on a simple MCU-based digital display with zeroing for the bolometer to turn it into a direct-reading unit for the bench. The basic code should be adaptable to any such bolometer, just input the bias "heater impedance" coefficient.

Regards,

Alan

Great work on the bolometer...just the solution for measuring optical power I was looking for.

For RF applications, I've been using a No. 387 incandescent lamp as my bolometer element. An op-amp feedback loop is utilized to always maintain the filament at a constant resistance (100 ohms) through a 100 ohm DC input resistor. RF is effectively fed in parallel with the bulb and resistor, thereby presenting a constant 50 ohm load. Power is measured by calculating the voltage delta in the feedback loop with a digital meter, just like in your design. The dynamic range is about -30 to 0dBm though below -20dB, the error exceeds a few percent.