Measuring the input impedance of the RTL2832U direct sampling input pins
The direct sampling mod allows you to listen to the HF frequencies between 0 – 14 MHz on an RTL-SDR by simply connecting an antenna directly to the ADC pins on the RTL2832U chip. Until recently the impedance of these pins was unknown, but most people assumed that it was about 300 Ohms.
Now Martin (G8JNJ) has recently updated his webpage with some interesting results that he and another experimenter (Robert Symanek) obtained regarding the input imedpance. Robert found that using a Minicircuits T16-6T-KK81 transformer with 50 Ohms on the primary and 800 Ohms on the secondary gave a 16dB to 19dB higher signal than a simpler 200 Ohm transformer.
This inspired Martin to do a proper measurement of the input impedance. Martin’s measurements found that the differential input impedance of the RTL2832U is approximately 3,330 Ohms when the input is enabled, and this would require a 66:1 transformer. However, Martin writes that a wideband transformer like this probably does not exist, but that the T16-6T-KK81 with terminating resistors added is probably a good choice.
The reason for adding the terminating resitor is to ensure that the transformer secondary ‘sees’ the correct load impedance so that the input match to 50 Ohm coax is good.
It will work without a terminating reistance, but as you increase the impedance transformation ratio the transfromer windings will have a tendency to be self-resonant on some frequencies, which will change with different antennas and lengths of coax. The end result is that you may find the sensitivity is better on some frequencies than others. With a 4:1 transformer and 220 Ohm load the frequency response is very flat without any distinct ‘lumps and bumps’ in the amplitude / frequency response curve.
Although the use of a higher ratio transformer will increase the overall signal level ‘seen’ by the dongle, this is often not the single most important factor on the LF and HF bands. Overloading (due to the 8 bit sampling and associated 40-50dB max dynamic range) is one issue, and with most wire antennas the receive performance is usually defined by the signal to noise ratio rather than absolute signal strength. As the natural noise level is fairly high (and even higher in urban locations) it is often the placement of the antenna that becomes the critical factor.
I have used a 10m tall broadband vertical antenna with my dongle on the HF bands and I often find that I can get better results with a 10 or 20dB attenuator placed between the dongle and the antenna, as this reduces overloading due to strong broadcast stations, especially at night when propagation conditions tend to be enhanced.
if the Z ratio is 66:1 (3330/50) then the ratio of a transformer should have sqrt(66):1 ratio which is about 8:1.
Can you find a 8:1 that has good numbers as the T16-6T-KK81 ?????????????????????????????? Than use it . It’s frequency .03 to 75 MHZ and it has the lowest insertion and input loss of every that I looked at .
I think, though, that the ratios usually discussed are the Z ratios. So, e.g., a “9:1” transformer has a Z ratio of 9:1 (not 81:1) and a turns ratio of 3:1.
Here’s a 1:60 transformer normally used for end fed half wave antennas: https://pa3hho.wordpress.com/end-fed-antennes/multiany-band-end-fed-english/
I’m using one and it works well. It’s not really wide band, but it would be fun to try it with direct sampling.
What about cascading two or more transformers (secondary of one to primary of the next)? Wouldn’t two 4:1 transformers make a 16:1 transformer, two 9:1 make 81:1, and 3 4:1 make 64:1?
The reason the T16-6T-KK81 works so well is the insertion and input loss are low . If you cascade two or more transformers the insertion and input loss will go up . The most impotent is getting the impedance to match , you can’t put let say 50 ohms to 300 ohms , you get impedance miss match .
The question is whether the gain from the better impedance matching exceeds the additional insertion loss. I imagine that the answer is “yes”, or the transformer would not be of much use for impedance matching in any situation.
I found that the resistor losses you about 4db .