Impedance matching questions

** Not true.

Doing that would mean throwing away half the power of a transmitter.

... Phil

** Care to estimate how much capacitance even 1km of coax or twisted pair mic cable has ??

You can render the line resistive across the audio band and beyond just by adding one resistor to the receiving end that matches the cable impedance.

Ham radio is for hams.

.... Phil

Only a few application require "impedance matching" of some sort, e.g. R.F. The undue weight given to this concept is quite possibly due to the historic use in telephone lines. Pretty much all significant electrical theory came from Bell Labs, and this is what ended up in all the books and university courses. Telephone lines need amplification in both directions at once. This is achieved by splitting up a two wire subscriber line to four wire each way signals by way of a special network. If the impedances are not matched, the forward gain path and the reverse path will loop and result in oscillation.

Kevin Aylward

- SuperSpice

Any impedance (R+jX) mismatch will cause a wave to be reflected.

With lossless lines (e.g. open wire lines on HF), this is not much of an issue, since the reflected wave is re-reflected at source anyway, so sooner or later all the power is finally transferred to the load. An Antenna Tuning Unit (ATU) can be used at the transmitter end to the load (e.g. antenna) make the ATU+feeder+antenna into a single resonant circuit.

With lossy transfer lines, such as coaxial cable at VHF/UHF, such arrangements will cause a significant true power loss, when the forth and back moving reflected waves are partially absorbed in the cable losses at each pass.

On average, the signal levels for device to device signal transfer in audio, the power levels are quite sufficient for high SNR without optimal power transfer, thus, if the source has a medium output impedance and the load has a high impedance, the signal transfer is OK.

The only point in which the impedance match _might_ have some significance is the match between a microphone/pickup and first amplifier stage. For high impedance microphones, such as piezo or electrostatic microphones, you should have a quite high amplifier impedance. For low impedance dynamic microphones, a quite low amplifier input impedance should be used.

While in RF, the use of power match is mainly an issue of the first amplifier thermal noise, in audio circuits, the impedance selection (source/load) may be detected by the methods, in which the interference can penetrate your audio circuits. For instance a high impedance circuit is suspectable to capacitively coupled interference. A low impedance circuit has issues with magnetic interference, whereas all circuits are suspectable to ground potential issues (balansed/unbalansed).

One should remember that if a load impedance differs significantly from the load resistance that the device was initially designed for, there might be some frequency response issues.

That is not necessary true, unless you have a separate matched secondary coil or at least take the signal output from a low tap in the inductor.

In order to understand why the very bad ferrite antenna works at all, one must understand the issue of band noise and why a very bad antenna is still quite useful.

With that assumption, you will repeat the same mistakes as the first Trans-Atlantic telegraph operators did in 1858. They treated the cable as an RC network instead of a transmission line.

In telephony loading (Pupin) coils were used to boost the higher (3 kHz) audio frequencies at a cost of high frequency drop above telephony frequencies.

For some decades, the wireless access systems (WiFi, GSM etc) have been far more significant to the society compared to traditional audio applications.

Maximum power transfer does not result in highest S/N.

Power transfer is irrelevant. Audio systems are sensitive to voltage, not power (well excepting the amp driving the speaker, but max power from a 1m ohm output resistance is not the goal).

This is still the opposite of "matching" as in making source and load impedances equal, i.e. to match them.

In all cases for mics, the load impedance should be at least, say, 5 times the source impedance to avoid signal loss, i.e. unmatched.

A typical 150 ohm dynamic mic wants a load of around 1k to 2K.

In RF, and in LF, best noise performance is pretty much invariably, when the source and load are not matched. Matching in RF, given that a source impedance actually exists, is used to get maximum power (when required), or the correct frequency response.

Yes.

Kevin Aylward

- SuperSpice

When the antenna pre-amp impedance is not matched to the antenna (+feedline) impedance, some of the received power is reflected back and reradiated back to the space, thus reducing the actual power entering the amplifier. The thermal power generated by the amplifier is constant (for a 3 dB 300 K amplifier at -174 dBm/Hz), the less signal power is entering the amplifier, the lower the SNR.

he antenna source impedance always exists, but it may be complex (with a significant X component).

Maybe still more important is the match of a transmitter to its feedline. A bad match makes the final amplifier feel pretty bad.

--

Tauno Voipio (OH2UG)

Yes, but that doesn't mean the transmitter output impedance equals the feedline impedance. It merely means the transmitter has be *designed* to drive that impedance. "Matching" can mean different things, depending on context.

Jeroen Belleman

For audio frequencies? Forget the whole thing.

Yes.

The equivalent input noise (ein) of an amplifier is dependant on the source impedance.

There is input current noise, and input voltage noise. As a rough approximation the source impedance wants to be vn/in for lowest noise. For example, using a transformer to achieve this from a fixed source impedance might be done. This will result in mismatch.

There is nothing new with this idea. It is well known that mismatching impedances will usually result in better S/N, despite power transfer being lower.

Yes.

No. See above.

Yes. However, maximising power transfer does not, usually, maximise S/N.

Kevin Aylward

- SuperSpice

You don't have to take the signal from a tap, Using the entire loop driving a high impedance jfet input gets you the most voltage but maybe not the best S/N ratio. And using an unloaded ferrite antenna gives you the highest Q factor which may be 200 or more. But that produces a narrow bandwidth of maybe 5KHz at 1 Mhz (F=1Mhz, f1-f2 = 5K, Q=200). This makes the radio hard to tune and requires very small adjustments to the tuning cap. So, it seems the optimum Q factor might be 100 so the bandwidth is wider and easier to tune with wider tollerance on the tuning capacitor setting. But the question is, what does lowering the Q do to the S/N ratio? There must be some optimum Q factor for best S/N ratio and bandwidth?

--- news://freenews.netfront.net/ - complaints: snipped-for-privacy@netfront.net ---

Can you explain how Q would affect the SNR? Are you talking about noise from the environment or noise in the front end of the radio? Considering noise in the radio getting the best SNR would require the highest voltage at the input I would think. So highest Q gives highest SNR.

For environmental noise I would think the highest possible Q would exclude more noise as out of band so maximizing Q would maximize SNR. Are there other factors?

--

Rick

** A dead short still gives the lowest noise.

** Having a source impedance close to that ratio will give you the best "noise figure" - usually the number is non critical and sources have their own noise ( more than thermal) anyhow.

However, the effective input impedance of the same stage should be many times higher than its optimum noise impedance - allowing almost the full signal voltage to be present at the input.

**How about an example:

200ohm mic feeding a 1:5 step up transformer making the source impedance 5kohms.

A NE5534 op-amp, connected non inverting, is close to optimum at this impedance - but the *actual* input impedance is set by a load resistor on the secondary of the transformer - likely of about 50kohms.

So, the mic sees an impedance of 2000 ohms and 91% of its unloaded output signal appears at the input.

... Phil

Ahmmm.....

Yes. typically, it is automatic to achieve this when choosing the optimum bias current for a bipolar input amplifier.

Equalising voltage and current noise gets you:

re = RS/sqrt(beta)

i.e. Set Ic=1/40.re

rin = beta.re

So, if beta is 100, rin=10 X RS

Kevin Aylward

- SuperSpice

The selection of loaded and unloaded Q is mainly critical for unpowered crystal receivers, inn which the _loaded_ Q should be sufficient for selecting the wanted station but sufficiently attenuate nearby stations. OTOH, the loaded should be as low as possible, to deliver some actual rectified audio power into the headphones.

On the superhet, the main concern is keeping the image frequencies as well as harmonics of other input frequencies as well oscillator harmonics. If the RF amplifier and mixer are not strong enough, strong input signals can be mixed to the wanted RF frequency or directly into the IF frequency. For a typical AM BC receiver, a front end loaded Q should be quite sufficient.

Any attenuation between the antenna and first amplifier will directly degrade the SNR of the whole system by the same amount of dB. For this reason, on VHF and above, it is essential to keep the losses for any input filters and switching diodes to a minimum to get the best system SNR.

The filter losses depends on the unloaded (Qu) to loaded Q (Ql) ratio

loss[dB] = 20 log ( 1 / (1-Ql/Qu))

So with Qu = 200 and Ql=40, there would be a loss of 2 dB and hence a SNR reduction by 2 dB, which would be acceptable for AM BC applications.

For VHF/UHF applications a 20 times unloaded Q would give about 0.5 dB losses, which would acceptable in most cases.

For many antenna sites with a lot of transmitters all around, you might need loaded Q values well above 100, so the unloaded Q would need to be in the thousands. Thus you may to use 1/4 or 1/2 wavelength linear resonators, which are coated by silver to minimize losses.

The other option would be to use small helical resonators with not so good unloaded Q, but the SNR and hence system sensitivity will degrade with several decibels, especially, if multiple filter units needs to be cascaded , in order to protect the first antenna preamplifier.

On a sunny day (Sat, 11 Oct 2014 22:02:41 +0200) it happened jeroen Belleman wrote in :

Actually it does.