# Bias tee with very small low cut-off frequency

I need to measure the noise spectrum of a diode under certain bias. The noise frequency I am interested is in the range of several Hz to tens or hundreds of kHz. I am thinking of using a bias tee with fairly small low cut-off frequency. But the low cut-off frequencies of bias tees I am looking at are still a little high (for example, in http://www.picosecond.com/product/category.asp?pd_id=2 the minimum low cut-off frequency is 3.5kHz). I am also thinking of DIY such a bias tee. For example, the post I need to design a Bias-T circuit for a diode laser - help choosing component values discussed how to design a bias tee working around 10MHz with 25kHz bandwidth. However, in my situation, I at least need it to go from hundred Hz to tens of kHz.

My questions is : 1) is there a commercial product available somewhere else that fits my needs? 2) If there is no such commercial product, how to DIY?

Thanks.

A single capacitor is all you need. The term for this is a "DC blocking capacitor". You commonly see them in audio circuits but they are used elsewhere.

The value of the capacitor is dependent on the input impedance of whatever you are feeding after the blocking cap. The lower the input impedance, the higher the value of cap that is required. Also, the lower the frequency response you need, the higher the value of cap.

A cap of only 10 uF is good if the input impedance is fairly high (100's of K ohms). But if the impedance is lower then you might need 68 uF or even 100 uF.

Unfortunately there is often some amount of trial and error needed when selecting the proper capacitor value-- especially if you do not accurately know the input impedance.

• Sorry I probably didn't say it clearly in the problem description. I also need to apply a DC bias to the device so a capacitor is not enough. Maybe a bias tee is still needed. Also what about the bandwidth of a capacitor? Is it wide enough (from several Hz to hundreds of kHz)?
– shva
Aug 12, 2012 at 17:57
• @shva, if you are testing the diode forward-biased, you can bias it with a high-impedance current source. If you are testing the diode reverse-biased, you can apply the bias voltage through a large-ish resistor (1k to 10k maybe). In either case, the bias circuit then has high enough impedance that you can connect to the diode with a dc blocking capacitor like David suggests without significantly disturbing the circuit. Aug 13, 2012 at 16:10
• @ThePhoton, are real-world current sources really high-impedance at MHz frequencies? Jun 20, 2018 at 22:12
• @ignacio, you could certainly design one that way, for example by including a series inductor at the output. Jun 21, 2018 at 0:55